Introduction Due to the nuance of the relationship
Due to the nuance of the relationship between race, victim sex, victim-offender
relationship, and violence, Lauritsen and White (2001) suggested that “it is important to
recognize that the relationship between an individual’s racial and ethnic characteristics
and their risk for violent victimization is not simple. Rather it depends on the sex of the
victim and whether the violent incident involves a stranger or nonstranger offender”
(p.44). However, this dynamic has often been omitted from violence literature. Rather
many studies have controlled for the percent of the population that is Black instead of
attending to the racial dynamics that underlie society and violence in the United States.
When race-disaggregated trends of violence against women are investigated,
major gaps in the crime decline, race, and gender literature are exposed. For example,
Lauritsen (2012) illustrated that although the rape and sexual assault trends for Latinas,
Black women, and white women have been similar for the last 40 years, the gaps between
them have varied occasionally. The variation in the gap could indicate different processes
at work. These types of findings are important in that they denote the importance of
attending to gender, race, and ethnicity but are often overlooked in later studies.
Ultimately, the crime decline literature has at times denoted the varying magnitude of the
decline by race and gender subgroups, but has not produced many intersectional
approaches to explaining the disparate decline (Heimer & Lauritsen, 2008; Lauritsen,
2012; Lauritsen & Heimer, 2010; Levitt, 2004; Parker, 2008). As a consequence, the
disparate declines and trends by race, ethnicity, and gender are not well understood. The
race and inequality literature tends to lack a gender-lens, a framework denoting the role
that gender plays in how things are enacted and perceived, while at the same time the
gender and inequality literature tends to lack a race-lens, a framework denoting the role
of race (McCall, 2005). This study argues that it is important to consider violence against
women from a gender x race-lens or an intersectional-lens.
HYPOTHESES AND EMPIRICAL SUPPORT
Before discussing the ways to apply an intersectionality perspective to the gender
inequality hypothesis, it is pertinent to review the various hypotheses that exist within the
literature. As noted in the introduction, the nature of the relationship between gender
inequality and violence against women has not been clearly defined within the empirical
literature (Whaley et al., 2013). In fact, evidence exists for multiple hypotheses. The
original hypothesis and the one many organizations work under suggests that as gender
inequality decreases and the genders become more equal violence against women will
decrease (Whaley et al., 2013). This is referred to as the amelioration hypothesis.
Peterson and Bailey (1992) found some support for the amelioration hypothesis. They
investigated the relationship between various forms of inequality (general, racial, and
gender) and rape rates for United States metropolitan areas in 1980. The rape rates were
obtained from the UCR in 1980. General inequality was measured using the GINI index
which represents income inequality in the total population. Racial inequality was
measured by the difference in median income for Black and white families. Gender
inequality was measured in terms of education using the ratio of men to women who
completed 4 years of high school, who completed 4 years of college, and who completed
5 or more years of college. It was also measured in terms of income using a ratio of men
to women’s median income for all people, people who completed 4 years of high school,
and people who completed 4 years or more of college. Gender inequality was also
measured by the percent of people employed in managerial and professional positions
who were women and by a ratio of men to women who live in poverty. Multiple models
were estimated with most containing one form of gender inequality and one containing a
representative of each type. The GINI index was found to be positively and significantly
related to rape rapes in each model. This finding suggests that places with greater general
income inequality exhibited higher rape rates. Racial inequality in income was significant
and positive in the models that included gender inequality by ratio of high school
graduates only and gender inequality in percentage of women in professional occupations
only. When all forms of gender inequality were included (income, high school graduation,
and profession occupations), gender inequality in median income was significantly and
positively related to rape rates. This finding suggests that where gender inequality in
income is greater, rape rates will also be higher and this is consistent with the
amelioration hypothesis.
Another hypothesis, the backlash hypothesis, suggests an opposite relationship.
Specifically, the backlash hypothesis suggests that as gender inequality decreases and the
genders become more equal, violence against women will increase (Whaley et al., 2013).
It is important to note that the term “backlash” is sometimes used to imply a specific
mechanism causing the increase of violence. For instance, sometimes backlash is defined
as a process by which men (or other dominant group) react violently to challenges to their
dominant status. In this study, as with many past studies, the mechanism of the
relationship between gender inequality and violence against women is not being
evaluated. Rather this study examines the relationship between gender inequality and
violence against women which therefore only allows for some speculation as to the
mechanism of the relationship. Backlash, herein, is used to represent a negative
relationship between changes in gender inequality and changes in violence against
women.
Within the same article described above, Peterson and Bailey (1992) also found
support for a backlash hypothesis when further investigating the relationship between
gender inequality in occupations and rape rates. Specifically, they separated professional
occupations into 1) executives, administrators, and managers 2) public officials and
administrators 3) engineers and natural scientists 4) health diagnostic occupations 5)
elementary and secondary teachers. They found that where the percent of executives,
administrators, and managers that are women were higher rape rates were also higher
suggesting backlash. However, they also found that where the percent of health diagnostic
occupations that are women were higher rape rates were lower. This was true of
elementary and secondary teachers as well. Since being an elementary or secondary
teacher is a women-typed occupation, the finding of a negative relationship could be
interpreted as backlash as well. The authors suggested that “the more that women remain
in traditional ‘female’ occupations, the greater their ‘reward’ – freedom from rape” (p.
172). However, the significant finding for health diagnostic occupations appears to
support amelioration, since health diagnostic occupations are traditionally men-typed
occupations.
The Peterson and Bailey (1992) study is illustrative for three primary reasons.
First, the analysis assesses the relationship between gender inequality and rape in
multiple ways. They use several forms of inequality including: education, income, and
occupation. In addition, they estimate models with each form individually as well as a
model with all of the forms. Second, this study is illustrative because it finds support for
both the ameliorative and backlash hypotheses. Due to these mixed findings, the authors
conclude that their “analysis by no means settles the issue of rape and the structural
position of women vis-à-vis men” (p. 174). Lastly, this study thoroughly investigated
different elements of inequality. For example, as discussed above, they investigated more
specific occupation categories to gain an understanding of the relationship between
gender parity in occupations and rape rates. Ultimately, together these three elements
exemplify how complicated the relationship between gender inequality and rape may be.
Specifically, it is possible that common indicators of gender disparity do not capture the
most important aspects of gender inequality. This could be one explanation for why some
measures of gender inequality were significantly related to rape and others were not. For
instance, inequality in educational attainment was not related to rape rates in any of the
Peterson and Bailey (1992) models. Through this study, it was also demonstrated that
some of the measures of gender inequality may be confounding other relationships. In
particular, Peterson and Bailey (1992) found that although in the main model there was
no significant relationship between gender inequality in occupation type and rape rates,
when looking at specific types of occupations the findings were mixed. Indeed, they
found support for backlash and amelioration when looking at specific types of
occupations. By thoroughly investigating the relationship between gender inequality and
rape, Peterson and Bailey (1992) illustrated how complicated the relationship may be and
demonstrated important ways to attempt to uncover more information.
Some scholars have noted that much of the research on gender inequality and
violence against women, including Peterson and Bailey (1992), have acted under the
axiom that amelioration and backlash are contradictory hypotheses (Bailey, 1999;
Whaley, 2001; Whaley et al., 2013). However, these hypotheses may be corollaries of one
another contingent either on time or the level of gender inequality. For instance, Bailey
(1999) and Whaley (2001) suggested and found some evidence that backlash occurs for a
period of time leading to amelioration. According to Bailey (1999), prior to his analysis
no studies had investigated the role of changes in gender inequality and changes in the
rape rate. Indeed, I was unable to locate any studies prior to Bailey (1999) that
investigated the relationship between changes in gender inequality and changes in any
form of violence against women. Bailey (1999) investigated the relationship between
gender inequality (using male and female differences in educational attainment,
occupation type, and income as indicators) and rape rates for central cities. He found
support for the backlash hypothesis between occupation type and rape rates for 1980 and
1990 (cross-sectional models), but found an ameliorative relationship between changes in
income inequality and changes in rape rates from 1980 to 1990 (change model). In the
end, Bailey suggested that although his analysis was an “improvement over previous
investigations this analysis does not fully resolve important feminist arguments regarding
the socioeconomic status of women and rape” (1999, p. 59).
Whaley (2001) presented itself as refining the temporal theoretical relationship
between change in gender inequality and violence against women. She hypothesized that
initially during times of higher gender inequality, women experience lower levels of rape
as a “reward, “but that in the long term, gender inequality prompts cultural changes that
build support for rape, prompting an increase. The reverse was hypothesized as well. To
use the hypotheses already introduced, Whaley (2001) hypothesized that backlash occurs
first and that after a sustained period of change amelioration can occur. Using 5-, 10-, and
20-year lagged indicators, Whaley (2001) was also unable to provide strong support for
her hypothesis; however, most of the significant relationships were in the proposed
direction. For example, cross-sectional analyses suggested that higher levels of inequality
in holding executive occupations were associated with lower levels of rape supporting the
backlash hypothesis. The model using a 5-year lag from 1970 to 1975 suggested backlash
also. Yet in accord with the hypothesis, the model using a 10-year lag between 1980 and
1990 found support for the ameliorative hypotheses. Specifically, decreases in gender
inequality in holding executive occupations were associated with decreases in rape.
Therefore, findings related to holding executive occupations appeared to support backlash
cross-sectionally and amelioration over a 10-year period. However, gender inequality in
executive occupations was not significantly related to rape when looking at change from
1980 to 1985, 1970 to 1980, and 1970 to 1990. Therefore, within this analysis, the results
reflect a complicated relationship.
Additionally, Whaley et al. (2013) suggested and found some evidence that the
nature of the relationship between gender inequality and violence against women may be
nonlinear and related to the level of gender inequality. Specifically, they investigated the
relationship between gender inequality (in terms of educational attainment, median
income, and employment status) and various rates of homicides for 1990 and 2000 from
the SHR (male-female, male-male, female-male, and female-female). Using negative
binomial regressions, Whaley et al. (2013) found that gender equality (represented by a
factor score of the three variables) was positively related to male offender – female victim
homicide rates and that the squared form of gender equality was negatively related in
1990. These significant findings suggest that the relationship between gender equality and
male-female homicide is nonlinear. Gender equality and its squared term were not
significantly related to any other homicide rates except female-offender and
femalevictim. For the female-female homicide rate, gender equality was significantly and
positively related in the model that did not include the robbery rate, which was included
as a “robust control for other factors related to criminal violence” (Whaley et al., 2013, p.
743). In 2000, gender equality and its squared form were significantly related to
malefemale homicide when the robbery rate was not included but only the squared form
was significantly related (negatively) when the robbery rate was included. The same was
true for male-male homicide rates. Just gender equality is significantly and positively
related to the female-female rate when the robbery rate was not included. In all other
instances, gender equality and its squared form were not significantly related to homicide
rates. Whaley et al. (2013) interpreted the effect of adding the robbery rate as suggesting
that unmeasured factors associated with violent crime may be mediating the relationship
or that there may be an issue with multicollinearity. As a whole, they found considerable
support for their hypothesis that backlash processes are likely to function at “low to
intermediate levels” of gender equality and that amelioration occurs when equality is high
for male-female homicide rates (p. 747).
Xie, Heimer, and Lauritsen (2012a) used longitudinal methods to assess whether
there was a relationship between non-fatal intimate partner violence and absolute/relative
indicators of women’s status. They created five five-year pooled periods using MSAlevel
data (which includes the 40 largest MSAs) from the NCVS for 1980 to 2004. To
thoroughly assess their hypotheses, they also disaggregated the non-fatal violence against
women data by victim-offender relationship using the following listed categories:
stranger, intimate partner, and other known. The four hypotheses included amelioration
(which was posited to function more for intimate partner violence than non-intimate
violence), backlash (which was also posited to function more for intimate partner
violence than non-intimate violence), routine activities (which suggests that increases in
absolute status also increases exposure and could lead to an increase in stranger and
known victimization), and protective (which suggests that an increase in absolute status
may increase the accessibility of protective mechanisms and decrease victimization).
Absolute women’s status was operationalized by the percent of women who were
employed, percent of voting age women who voted in the November presidential and
congressional elections, and an index of income-educational attainment, which included
median income and percent who completed four years of college or more. Relative status
was measured using the rate of women’s labor force participation minus that of men, the
level of income-educational attainment for women minus that of men, and the rate of
voter turnout for women minus that of men. Their results found support for each of the
hypotheses except for amelioration. They found a negative relationship between women’s
absolute economic and political status and intimate partner violence which supports the
protective hypothesis. They also found the negative relationship between inequality and
intimate partner victimization which supports the backlash hypothesis. The routine
activities hypothesis was supported through a positive relationship between absolute
indicators of labor force participation and victimization by strangers and non-intimate
persons known to the victim. This longitudinal analysis does not suggest that backlash,
amelioration, protective, nor routine activities hypotheses are corollary by explaining the
relationship between inequality, absolute status, and violence against women at different
periods. However, they found evidence that the nature of the relationship is contingent on
the victim-offender relationship.
The hypotheses and findings suggesting that backlash and amelioration are
corollary are particularly important considering that much of the empirical quantitative
literature on gender inequality has been cross-sectional and assumed a linear relationship.
By design much of the extant literature is unable to address these possibilities (Whaley et
al., 2013). The evidence supporting corollary processes indicate that future analyses
should continue to consider change over time and the structural form of the relationship
as the proposed dissertation will do. In addition, Xie et al. (2012a) presented evidence
that the victim-offender relationship is important to also consider in future analyses.
In addition, while not an inequality hypothesis, several scholars have investigated the role
of absolute status on trends of violence against women. Changes in absolute status are
increases or decreases in the status of a group with no consideration or comparison to
other groups. It is possible that there is a relationship between absolute status and
violence against women and that any findings suggesting a relationship
When race-disaggregated trends of violence against women are investigated,
major gaps in the crime decline, race, and gender literature are exposed. For example,
Lauritsen (2012) illustrated that although the rape and sexual assault trends for Latinas,
Black women, and white women have been similar for the last 40 years, the gaps between
them have varied occasionally. The variation in the gap could indicate different processes
at work. These types of findings are important in that they denote the importance of
attending to gender, race, and ethnicity but are often overlooked in later studies.
Ultimately, the crime decline literature has at times denoted the varying magnitude of the
decline by race and gender subgroups, but has not produced many intersectional
approaches to explaining the disparate decline (Heimer & Lauritsen, 2008; Lauritsen,
2012; Lauritsen & Heimer, 2010; Levitt, 2004; Parker, 2008). As a consequence, the
disparate declines and trends by race, ethnicity, and gender are not well understood. The
race and inequality literature tends to lack a gender-lens, a framework denoting the role
that gender plays in how things are enacted and perceived, while at the same time the
gender and inequality literature tends to lack a race-lens, a framework denoting the role
of race (McCall, 2005). This study argues that it is important to consider violence against
women from a gender x race-lens or an intersectional-lens.
HYPOTHESES AND EMPIRICAL SUPPORT
Before discussing the ways to apply an intersectionality perspective to the gender
inequality hypothesis, it is pertinent to review the various hypotheses that exist within the
literature. As noted in the introduction, the nature of the relationship between gender
inequality and violence against women has not been clearly defined within the empirical
literature (Whaley et al., 2013). In fact, evidence exists for multiple hypotheses. The
original hypothesis and the one many organizations work under suggests that as gender
inequality decreases and the genders become more equal violence against women will
decrease (Whaley et al., 2013). This is referred to as the amelioration hypothesis.
Peterson and Bailey (1992) found some support for the amelioration hypothesis. They
investigated the relationship between various forms of inequality (general, racial, and
gender) and rape rates for United States metropolitan areas in 1980. The rape rates were
obtained from the UCR in 1980. General inequality was measured using the GINI index
which represents income inequality in the total population. Racial inequality was
measured by the difference in median income for Black and white families. Gender
inequality was measured in terms of education using the ratio of men to women who
completed 4 years of high school, who completed 4 years of college, and who completed
5 or more years of college. It was also measured in terms of income using a ratio of men
to women’s median income for all people, people who completed 4 years of high school,
and people who completed 4 years or more of college. Gender inequality was also
measured by the percent of people employed in managerial and professional positions
who were women and by a ratio of men to women who live in poverty. Multiple models
were estimated with most containing one form of gender inequality and one containing a
representative of each type. The GINI index was found to be positively and significantly
related to rape rapes in each model. This finding suggests that places with greater general
income inequality exhibited higher rape rates. Racial inequality in income was significant
and positive in the models that included gender inequality by ratio of high school
graduates only and gender inequality in percentage of women in professional occupations
only. When all forms of gender inequality were included (income, high school graduation,
and profession occupations), gender inequality in median income was significantly and
positively related to rape rates. This finding suggests that where gender inequality in
income is greater, rape rates will also be higher and this is consistent with the
amelioration hypothesis.
Another hypothesis, the backlash hypothesis, suggests an opposite relationship.
Specifically, the backlash hypothesis suggests that as gender inequality decreases and the
genders become more equal, violence against women will increase (Whaley et al., 2013).
It is important to note that the term “backlash” is sometimes used to imply a specific
mechanism causing the increase of violence. For instance, sometimes backlash is defined
as a process by which men (or other dominant group) react violently to challenges to their
dominant status. In this study, as with many past studies, the mechanism of the
relationship between gender inequality and violence against women is not being
evaluated. Rather this study examines the relationship between gender inequality and
violence against women which therefore only allows for some speculation as to the
mechanism of the relationship. Backlash, herein, is used to represent a negative
relationship between changes in gender inequality and changes in violence against
women.
Within the same article described above, Peterson and Bailey (1992) also found
support for a backlash hypothesis when further investigating the relationship between
gender inequality in occupations and rape rates. Specifically, they separated professional
occupations into 1) executives, administrators, and managers 2) public officials and
administrators 3) engineers and natural scientists 4) health diagnostic occupations 5)
elementary and secondary teachers. They found that where the percent of executives,
administrators, and managers that are women were higher rape rates were also higher
suggesting backlash. However, they also found that where the percent of health diagnostic
occupations that are women were higher rape rates were lower. This was true of
elementary and secondary teachers as well. Since being an elementary or secondary
teacher is a women-typed occupation, the finding of a negative relationship could be
interpreted as backlash as well. The authors suggested that “the more that women remain
in traditional ‘female’ occupations, the greater their ‘reward’ – freedom from rape” (p.
172). However, the significant finding for health diagnostic occupations appears to
support amelioration, since health diagnostic occupations are traditionally men-typed
occupations.
The Peterson and Bailey (1992) study is illustrative for three primary reasons.
First, the analysis assesses the relationship between gender inequality and rape in
multiple ways. They use several forms of inequality including: education, income, and
occupation. In addition, they estimate models with each form individually as well as a
model with all of the forms. Second, this study is illustrative because it finds support for
both the ameliorative and backlash hypotheses. Due to these mixed findings, the authors
conclude that their “analysis by no means settles the issue of rape and the structural
position of women vis-à-vis men” (p. 174). Lastly, this study thoroughly investigated
different elements of inequality. For example, as discussed above, they investigated more
specific occupation categories to gain an understanding of the relationship between
gender parity in occupations and rape rates. Ultimately, together these three elements
exemplify how complicated the relationship between gender inequality and rape may be.
Specifically, it is possible that common indicators of gender disparity do not capture the
most important aspects of gender inequality. This could be one explanation for why some
measures of gender inequality were significantly related to rape and others were not. For
instance, inequality in educational attainment was not related to rape rates in any of the
Peterson and Bailey (1992) models. Through this study, it was also demonstrated that
some of the measures of gender inequality may be confounding other relationships. In
particular, Peterson and Bailey (1992) found that although in the main model there was
no significant relationship between gender inequality in occupation type and rape rates,
when looking at specific types of occupations the findings were mixed. Indeed, they
found support for backlash and amelioration when looking at specific types of
occupations. By thoroughly investigating the relationship between gender inequality and
rape, Peterson and Bailey (1992) illustrated how complicated the relationship may be and
demonstrated important ways to attempt to uncover more information.
Some scholars have noted that much of the research on gender inequality and
violence against women, including Peterson and Bailey (1992), have acted under the
axiom that amelioration and backlash are contradictory hypotheses (Bailey, 1999;
Whaley, 2001; Whaley et al., 2013). However, these hypotheses may be corollaries of one
another contingent either on time or the level of gender inequality. For instance, Bailey
(1999) and Whaley (2001) suggested and found some evidence that backlash occurs for a
period of time leading to amelioration. According to Bailey (1999), prior to his analysis
no studies had investigated the role of changes in gender inequality and changes in the
rape rate. Indeed, I was unable to locate any studies prior to Bailey (1999) that
investigated the relationship between changes in gender inequality and changes in any
form of violence against women. Bailey (1999) investigated the relationship between
gender inequality (using male and female differences in educational attainment,
occupation type, and income as indicators) and rape rates for central cities. He found
support for the backlash hypothesis between occupation type and rape rates for 1980 and
1990 (cross-sectional models), but found an ameliorative relationship between changes in
income inequality and changes in rape rates from 1980 to 1990 (change model). In the
end, Bailey suggested that although his analysis was an “improvement over previous
investigations this analysis does not fully resolve important feminist arguments regarding
the socioeconomic status of women and rape” (1999, p. 59).
Whaley (2001) presented itself as refining the temporal theoretical relationship
between change in gender inequality and violence against women. She hypothesized that
initially during times of higher gender inequality, women experience lower levels of rape
as a “reward, “but that in the long term, gender inequality prompts cultural changes that
build support for rape, prompting an increase. The reverse was hypothesized as well. To
use the hypotheses already introduced, Whaley (2001) hypothesized that backlash occurs
first and that after a sustained period of change amelioration can occur. Using 5-, 10-, and
20-year lagged indicators, Whaley (2001) was also unable to provide strong support for
her hypothesis; however, most of the significant relationships were in the proposed
direction. For example, cross-sectional analyses suggested that higher levels of inequality
in holding executive occupations were associated with lower levels of rape supporting the
backlash hypothesis. The model using a 5-year lag from 1970 to 1975 suggested backlash
also. Yet in accord with the hypothesis, the model using a 10-year lag between 1980 and
1990 found support for the ameliorative hypotheses. Specifically, decreases in gender
inequality in holding executive occupations were associated with decreases in rape.
Therefore, findings related to holding executive occupations appeared to support backlash
cross-sectionally and amelioration over a 10-year period. However, gender inequality in
executive occupations was not significantly related to rape when looking at change from
1980 to 1985, 1970 to 1980, and 1970 to 1990. Therefore, within this analysis, the results
reflect a complicated relationship.
Additionally, Whaley et al. (2013) suggested and found some evidence that the
nature of the relationship between gender inequality and violence against women may be
nonlinear and related to the level of gender inequality. Specifically, they investigated the
relationship between gender inequality (in terms of educational attainment, median
income, and employment status) and various rates of homicides for 1990 and 2000 from
the SHR (male-female, male-male, female-male, and female-female). Using negative
binomial regressions, Whaley et al. (2013) found that gender equality (represented by a
factor score of the three variables) was positively related to male offender – female victim
homicide rates and that the squared form of gender equality was negatively related in
1990. These significant findings suggest that the relationship between gender equality and
male-female homicide is nonlinear. Gender equality and its squared term were not
significantly related to any other homicide rates except female-offender and
femalevictim. For the female-female homicide rate, gender equality was significantly and
positively related in the model that did not include the robbery rate, which was included
as a “robust control for other factors related to criminal violence” (Whaley et al., 2013, p.
743). In 2000, gender equality and its squared form were significantly related to
malefemale homicide when the robbery rate was not included but only the squared form
was significantly related (negatively) when the robbery rate was included. The same was
true for male-male homicide rates. Just gender equality is significantly and positively
related to the female-female rate when the robbery rate was not included. In all other
instances, gender equality and its squared form were not significantly related to homicide
rates. Whaley et al. (2013) interpreted the effect of adding the robbery rate as suggesting
that unmeasured factors associated with violent crime may be mediating the relationship
or that there may be an issue with multicollinearity. As a whole, they found considerable
support for their hypothesis that backlash processes are likely to function at “low to
intermediate levels” of gender equality and that amelioration occurs when equality is high
for male-female homicide rates (p. 747).
Xie, Heimer, and Lauritsen (2012a) used longitudinal methods to assess whether
there was a relationship between non-fatal intimate partner violence and absolute/relative
indicators of women’s status. They created five five-year pooled periods using MSAlevel
data (which includes the 40 largest MSAs) from the NCVS for 1980 to 2004. To
thoroughly assess their hypotheses, they also disaggregated the non-fatal violence against
women data by victim-offender relationship using the following listed categories:
stranger, intimate partner, and other known. The four hypotheses included amelioration
(which was posited to function more for intimate partner violence than non-intimate
violence), backlash (which was also posited to function more for intimate partner
violence than non-intimate violence), routine activities (which suggests that increases in
absolute status also increases exposure and could lead to an increase in stranger and
known victimization), and protective (which suggests that an increase in absolute status
may increase the accessibility of protective mechanisms and decrease victimization).
Absolute women’s status was operationalized by the percent of women who were
employed, percent of voting age women who voted in the November presidential and
congressional elections, and an index of income-educational attainment, which included
median income and percent who completed four years of college or more. Relative status
was measured using the rate of women’s labor force participation minus that of men, the
level of income-educational attainment for women minus that of men, and the rate of
voter turnout for women minus that of men. Their results found support for each of the
hypotheses except for amelioration. They found a negative relationship between women’s
absolute economic and political status and intimate partner violence which supports the
protective hypothesis. They also found the negative relationship between inequality and
intimate partner victimization which supports the backlash hypothesis. The routine
activities hypothesis was supported through a positive relationship between absolute
indicators of labor force participation and victimization by strangers and non-intimate
persons known to the victim. This longitudinal analysis does not suggest that backlash,
amelioration, protective, nor routine activities hypotheses are corollary by explaining the
relationship between inequality, absolute status, and violence against women at different
periods. However, they found evidence that the nature of the relationship is contingent on
the victim-offender relationship.
The hypotheses and findings suggesting that backlash and amelioration are
corollary are particularly important considering that much of the empirical quantitative
literature on gender inequality has been cross-sectional and assumed a linear relationship.
By design much of the extant literature is unable to address these possibilities (Whaley et
al., 2013). The evidence supporting corollary processes indicate that future analyses
should continue to consider change over time and the structural form of the relationship
as the proposed dissertation will do. In addition, Xie et al. (2012a) presented evidence
that the victim-offender relationship is important to also consider in future analyses.
In addition, while not an inequality hypothesis, several scholars have investigated the role
of absolute status on trends of violence against women. Changes in absolute status are
increases or decreases in the status of a group with no consideration or comparison to
other groups. It is possible that there is a relationship between absolute status and
violence against women and that any findings suggesting a relationship
When race-disaggregated trends of violence against women are investigated,
major gaps in the crime decline, race, and gender literature are exposed. For example,
Lauritsen (2012) illustrated that although the rape and sexual assault trends for Latinas,
Black women, and white women have been similar for the last 40 years, the gaps between
them have varied occasionally. The variation in the gap could indicate different processes
at work. These types of findings are important in that they denote the importance of
attending to gender, race, and ethnicity but are often overlooked in later studies.
Ultimately, the crime decline literature has at times denoted the varying magnitude of the
decline by race and gender subgroups, but has not produced many intersectional
approaches to explaining the disparate decline (Heimer & Lauritsen, 2008; Lauritsen,
2012; Lauritsen & Heimer, 2010; Levitt, 2004; Parker, 2008). As a consequence, the
disparate declines and trends by race, ethnicity, and gender are not well understood. The
race and inequality literature tends to lack a gender-lens, a framework denoting the role
that gender plays in how things are enacted and perceived, while at the same time the
gender and inequality literature tends to lack a race-lens, a framework denoting the role
of race (McCall, 2005). This study argues that it is important to consider violence against
women from a gender x race-lens or an intersectional-lens.
HYPOTHESES AND EMPIRICAL SUPPORT
Before discussing the ways to apply an intersectionality perspective to the gender
inequality hypothesis, it is pertinent to review the various hypotheses that exist within the
literature. As noted in the introduction, the nature of the relationship between gender
inequality and violence against women has not been clearly defined within the empirical
literature (Whaley et al., 2013). In fact, evidence exists for multiple hypotheses. The
original hypothesis and the one many organizations work under suggests that as gender
inequality decreases and the genders become more equal violence against women will
decrease (Whaley et al., 2013). This is referred to as the amelioration hypothesis.
Peterson and Bailey (1992) found some support for the amelioration hypothesis. They
investigated the relationship between various forms of inequality (general, racial, and
gender) and rape rates for United States metropolitan areas in 1980. The rape rates were
obtained from the UCR in 1980. General inequality was measured using the GINI index
which represents income inequality in the total population. Racial inequality was
measured by the difference in median income for Black and white families. Gender
inequality was measured in terms of education using the ratio of men to women who
completed 4 years of high school, who completed 4 years of college, and who completed
5 or more years of college. It was also measured in terms of income using a ratio of men
to women’s median income for all people, people who completed 4 years of high school,
and people who completed 4 years or more of college. Gender inequality was also
measured by the percent of people employed in managerial and professional positions
who were women and by a ratio of men to women who live in poverty. Multiple models
were estimated with most containing one form of gender inequality and one containing a
representative of each type. The GINI index was found to be positively and significantly
related to rape rapes in each model. This finding suggests that places with greater general
income inequality exhibited higher rape rates. Racial inequality in income was significant
and positive in the models that included gender inequality by ratio of high school
graduates only and gender inequality in percentage of women in professional occupations
only. When all forms of gender inequality were included (income, high school graduation,
and profession occupations), gender inequality in median income was significantly and
positively related to rape rates. This finding suggests that where gender inequality in
income is greater, rape rates will also be higher and this is consistent with the
amelioration hypothesis.
Another hypothesis, the backlash hypothesis, suggests an opposite relationship.
Specifically, the backlash hypothesis suggests that as gender inequality decreases and the
genders become more equal, violence against women will increase (Whaley et al., 2013).
It is important to note that the term “backlash” is sometimes used to imply a specific
mechanism causing the increase of violence. For instance, sometimes backlash is defined
as a process by which men (or other dominant group) react violently to challenges to their
dominant status. In this study, as with many past studies, the mechanism of the
relationship between gender inequality and violence against women is not being
evaluated. Rather this study examines the relationship between gender inequality and
violence against women which therefore only allows for some speculation as to the
mechanism of the relationship. Backlash, herein, is used to represent a negative
relationship between changes in gender inequality and changes in violence against
women.
Within the same article described above, Peterson and Bailey (1992) also found
support for a backlash hypothesis when further investigating the relationship between
gender inequality in occupations and rape rates. Specifically, they separated professional
occupations into 1) executives, administrators, and managers 2) public officials and
administrators 3) engineers and natural scientists 4) health diagnostic occupations 5)
elementary and secondary teachers. They found that where the percent of executives,
administrators, and managers that are women were higher rape rates were also higher
suggesting backlash. However, they also found that where the percent of health diagnostic
occupations that are women were higher rape rates were lower. This was true of
elementary and secondary teachers as well. Since being an elementary or secondary
teacher is a women-typed occupation, the finding of a negative relationship could be
interpreted as backlash as well. The authors suggested that “the more that women remain
in traditional ‘female’ occupations, the greater their ‘reward’ – freedom from rape” (p.
172). However, the significant finding for health diagnostic occupations appears to
support amelioration, since health diagnostic occupations are traditionally men-typed
occupations.
The Peterson and Bailey (1992) study is illustrative for three primary reasons.
First, the analysis assesses the relationship between gender inequality and rape in
multiple ways. They use several forms of inequality including: education, income, and
occupation. In addition, they estimate models with each form individually as well as a
model with all of the forms. Second, this study is illustrative because it finds support for
both the ameliorative and backlash hypotheses. Due to these mixed findings, the authors
conclude that their “analysis by no means settles the issue of rape and the structural
position of women vis-à-vis men” (p. 174). Lastly, this study thoroughly investigated
different elements of inequality. For example, as discussed above, they investigated more
specific occupation categories to gain an understanding of the relationship between
gender parity in occupations and rape rates. Ultimately, together these three elements
exemplify how complicated the relationship between gender inequality and rape may be.
Specifically, it is possible that common indicators of gender disparity do not capture the
most important aspects of gender inequality. This could be one explanation for why some
measures of gender inequality were significantly related to rape and others were not. For
instance, inequality in educational attainment was not related to rape rates in any of the
Peterson and Bailey (1992) models. Through this study, it was also demonstrated that
some of the measures of gender inequality may be confounding other relationships. In
particular, Peterson and Bailey (1992) found that although in the main model there was
no significant relationship between gender inequality in occupation type and rape rates,
when looking at specific types of occupations the findings were mixed. Indeed, they
found support for backlash and amelioration when looking at specific types of
occupations. By thoroughly investigating the relationship between gender inequality and
rape, Peterson and Bailey (1992) illustrated how complicated the relationship may be and
demonstrated important ways to attempt to uncover more information.
Some scholars have noted that much of the research on gender inequality and
violence against women, including Peterson and Bailey (1992), have acted under the
axiom that amelioration and backlash are contradictory hypotheses (Bailey, 1999;
Whaley, 2001; Whaley et al., 2013). However, these hypotheses may be corollaries of one
another contingent either on time or the level of gender inequality. For instance, Bailey
(1999) and Whaley (2001) suggested and found some evidence that backlash occurs for a
period of time leading to amelioration. According to Bailey (1999), prior to his analysis
no studies had investigated the role of changes in gender inequality and changes in the
rape rate. Indeed, I was unable to locate any studies prior to Bailey (1999) that
investigated the relationship between changes in gender inequality and changes in any
form of violence against women. Bailey (1999) investigated the relationship between
gender inequality (using male and female differences in educational attainment,
occupation type, and income as indicators) and rape rates for central cities. He found
support for the backlash hypothesis between occupation type and rape rates for 1980 and
1990 (cross-sectional models), but found an ameliorative relationship between changes in
income inequality and changes in rape rates from 1980 to 1990 (change model). In the
end, Bailey suggested that although his analysis was an “improvement over previous
investigations this analysis does not fully resolve important feminist arguments regarding
the socioeconomic status of women and rape” (1999, p. 59).
Whaley (2001) presented itself as refining the temporal theoretical relationship
between change in gender inequality and violence against women. She hypothesized that
initially during times of higher gender inequality, women experience lower levels of rape
as a “reward, “but that in the long term, gender inequality prompts cultural changes that
build support for rape, prompting an increase. The reverse was hypothesized as well. To
use the hypotheses already introduced, Whaley (2001) hypothesized that backlash occurs
first and that after a sustained period of change amelioration can occur. Using 5-, 10-, and
20-year lagged indicators, Whaley (2001) was also unable to provide strong support for
her hypothesis; however, most of the significant relationships were in the proposed
direction. For example, cross-sectional analyses suggested that higher levels of inequality
in holding executive occupations were associated with lower levels of rape supporting the
backlash hypothesis. The model using a 5-year lag from 1970 to 1975 suggested backlash
also. Yet in accord with the hypothesis, the model using a 10-year lag between 1980 and
1990 found support for the ameliorative hypotheses. Specifically, decreases in gender
inequality in holding executive occupations were associated with decreases in rape.
Therefore, findings related to holding executive occupations appeared to support backlash
cross-sectionally and amelioration over a 10-year period. However, gender inequality in
executive occupations was not significantly related to rape when looking at change from
1980 to 1985, 1970 to 1980, and 1970 to 1990. Therefore, within this analysis, the results
reflect a complicated relationship.
Additionally, Whaley et al. (2013) suggested and found some evidence that the
nature of the relationship between gender inequality and violence against women may be
nonlinear and related to the level of gender inequality. Specifically, they investigated the
relationship between gender inequality (in terms of educational attainment, median
income, and employment status) and various rates of homicides for 1990 and 2000 from
the SHR (male-female, male-male, female-male, and female-female). Using negative
binomial regressions, Whaley et al. (2013) found that gender equality (represented by a
factor score of the three variables) was positively related to male offender – female victim
homicide rates and that the squared form of gender equality was negatively related in
1990. These significant findings suggest that the relationship between gender equality and
male-female homicide is nonlinear. Gender equality and its squared term were not
significantly related to any other homicide rates except female-offender and
femalevictim. For the female-female homicide rate, gender equality was significantly and
positively related in the model that did not include the robbery rate, which was included
as a “robust control for other factors related to criminal violence” (Whaley et al., 2013, p.
743). In 2000, gender equality and its squared form were significantly related to
malefemale homicide when the robbery rate was not included but only the squared form
was significantly related (negatively) when the robbery rate was included. The same was
true for male-male homicide rates. Just gender equality is significantly and positively
related to the female-female rate when the robbery rate was not included. In all other
instances, gender equality and its squared form were not significantly related to homicide
rates. Whaley et al. (2013) interpreted the effect of adding the robbery rate as suggesting
that unmeasured factors associated with violent crime may be mediating the relationship
or that there may be an issue with multicollinearity. As a whole, they found considerable
support for their hypothesis that backlash processes are likely to function at “low to
intermediate levels” of gender equality and that amelioration occurs when equality is high
for male-female homicide rates (p. 747).
Xie, Heimer, and Lauritsen (2012a) used longitudinal methods to assess whether
there was a relationship between non-fatal intimate partner violence and absolute/relative
indicators of women’s status. They created five five-year pooled periods using MSAlevel
data (which includes the 40 largest MSAs) from the NCVS for 1980 to 2004. To
thoroughly assess their hypotheses, they also disaggregated the non-fatal violence against
women data by victim-offender relationship using the following listed categories:
stranger, intimate partner, and other known. The four hypotheses included amelioration
(which was posited to function more for intimate partner violence than non-intimate
violence), backlash (which was also posited to function more for intimate partner
violence than non-intimate violence), routine activities (which suggests that increases in
absolute status also increases exposure and could lead to an increase in stranger and
known victimization), and protective (which suggests that an increase in absolute status
may increase the accessibility of protective mechanisms and decrease victimization).
Absolute women’s status was operationalized by the percent of women who were
employed, percent of voting age women who voted in the November presidential and
congressional elections, and an index of income-educational attainment, which included
median income and percent who completed four years of college or more. Relative status
was measured using the rate of women’s labor force participation minus that of men, the
level of income-educational attainment for women minus that of men, and the rate of
voter turnout for women minus that of men. Their results found support for each of the
hypotheses except for amelioration. They found a negative relationship between women’s
absolute economic and political status and intimate partner violence which supports the
protective hypothesis. They also found the negative relationship between inequality and
intimate partner victimization which supports the backlash hypothesis. The routine
activities hypothesis was supported through a positive relationship between absolute
indicators of labor force participation and victimization by strangers and non-intimate
persons known to the victim. This longitudinal analysis does not suggest that backlash,
amelioration, protective, nor routine activities hypotheses are corollary by explaining the
relationship between inequality, absolute status, and violence against women at different
periods. However, they found evidence that the nature of the relationship is contingent on
the victim-offender relationship.
The hypotheses and findings suggesting that backlash and amelioration are
corollary are particularly important considering that much of the empirical quantitative
literature on gender inequality has been cross-sectional and assumed a linear relationship.
By design much of the extant literature is unable to address these possibilities (Whaley et
al., 2013). The evidence supporting corollary processes indicate that future analyses
should continue to consider change over time and the structural form of the relationship
as the proposed dissertation will do. In addition, Xie et al. (2012a) presented evidence
that the victim-offender relationship is important to also consider in future analyses.
In addition, while not an inequality hypothesis, several scholars have investigated the role
of absolute status on trends of violence against women. Changes in absolute status are
increases or decreases in the status of a group with no consideration or comparison to
other groups. It is possible that there is a relationship between absolute status and
violence against women and that any findings suggesting a relationship
When race-disaggregated trends of violence against women are investigated,
major gaps in the crime decline, race, and gender literature are exposed. For example,
Lauritsen (2012) illustrated that although the rape and sexual assault trends for Latinas,
Black women, and white women have been similar for the last 40 years, the gaps between
them have varied occasionally. The variation in the gap could indicate different processes
at work. These types of findings are important in that they denote the importance of
attending to gender, race, and ethnicity but are often overlooked in later studies.
Ultimately, the crime decline literature has at times denoted the varying magnitude of the
decline by race and gender subgroups, but has not produced many intersectional
approaches to explaining the disparate decline (Heimer & Lauritsen, 2008; Lauritsen,
2012; Lauritsen & Heimer, 2010; Levitt, 2004; Parker, 2008). As a consequence, the
disparate declines and trends by race, ethnicity, and gender are not well understood. The
race and inequality literature tends to lack a gender-lens, a framework denoting the role
that gender plays in how things are enacted and perceived, while at the same time the
gender and inequality literature tends to lack a race-lens, a framework denoting the role
of race (McCall, 2005). This study argues that it is important to consider violence against
women from a gender x race-lens or an intersectional-lens.
HYPOTHESES AND EMPIRICAL SUPPORT
Before discussing the ways to apply an intersectionality perspective to the gender
inequality hypothesis, it is pertinent to review the various hypotheses that exist within the
literature. As noted in the introduction, the nature of the relationship between gender
inequality and violence against women has not been clearly defined within the empirical
literature (Whaley et al., 2013). In fact, evidence exists for multiple hypotheses. The
original hypothesis and the one many organizations work under suggests that as gender
inequality decreases and the genders become more equal violence against women will
decrease (Whaley et al., 2013). This is referred to as the amelioration hypothesis.
Peterson and Bailey (1992) found some support for the amelioration hypothesis. They
investigated the relationship between various forms of inequality (general, racial, and
gender) and rape rates for United States metropolitan areas in 1980. The rape rates were
obtained from the UCR in 1980. General inequality was measured using the GINI index
which represents income inequality in the total population. Racial inequality was
measured by the difference in median income for Black and white families. Gender
inequality was measured in terms of education using the ratio of men to women who
completed 4 years of high school, who completed 4 years of college, and who completed
5 or more years of college. It was also measured in terms of income using a ratio of men
to women’s median income for all people, people who completed 4 years of high school,
and people who completed 4 years or more of college. Gender inequality was also
measured by the percent of people employed in managerial and professional positions
who were women and by a ratio of men to women who live in poverty. Multiple models
were estimated with most containing one form of gender inequality and one containing a
representative of each type. The GINI index was found to be positively and significantly
related to rape rapes in each model. This finding suggests that places with greater general
income inequality exhibited higher rape rates. Racial inequality in income was significant
and positive in the models that included gender inequality by ratio of high school
graduates only and gender inequality in percentage of women in professional occupations
only. When all forms of gender inequality were included (income, high school graduation,
and profession occupations), gender inequality in median income was significantly and
positively related to rape rates. This finding suggests that where gender inequality in
income is greater, rape rates will also be higher and this is consistent with the
amelioration hypothesis.
Another hypothesis, the backlash hypothesis, suggests an opposite relationship.
Specifically, the backlash hypothesis suggests that as gender inequality decreases and the
genders become more equal, violence against women will increase (Whaley et al., 2013).
It is important to note that the term “backlash” is sometimes used to imply a specific
mechanism causing the increase of violence. For instance, sometimes backlash is defined
as a process by which men (or other dominant group) react violently to challenges to their
dominant status. In this study, as with many past studies, the mechanism of the
relationship between gender inequality and violence against women is not being
evaluated. Rather this study examines the relationship between gender inequality and
violence against women which therefore only allows for some speculation as to the
mechanism of the relationship. Backlash, herein, is used to represent a negative
relationship between changes in gender inequality and changes in violence against
women.
Within the same article described above, Peterson and Bailey (1992) also found
support for a backlash hypothesis when further investigating the relationship between
gender inequality in occupations and rape rates. Specifically, they separated professional
occupations into 1) executives, administrators, and managers 2) public officials and
administrators 3) engineers and natural scientists 4) health diagnostic occupations 5)
elementary and secondary teachers. They found that where the percent of executives,
administrators, and managers that are women were higher rape rates were also higher
suggesting backlash. However, they also found that where the percent of health diagnostic
occupations that are women were higher rape rates were lower. This was true of
elementary and secondary teachers as well. Since being an elementary or secondary
teacher is a women-typed occupation, the finding of a negative relationship could be
interpreted as backlash as well. The authors suggested that “the more that women remain
in traditional ‘female’ occupations, the greater their ‘reward’ – freedom from rape” (p.
172). However, the significant finding for health diagnostic occupations appears to
support amelioration, since health diagnostic occupations are traditionally men-typed
occupations.
The Peterson and Bailey (1992) study is illustrative for three primary reasons.
First, the analysis assesses the relationship between gender inequality and rape in
multiple ways. They use several forms of inequality including: education, income, and
occupation. In addition, they estimate models with each form individually as well as a
model with all of the forms. Second, this study is illustrative because it finds support for
both the ameliorative and backlash hypotheses. Due to these mixed findings, the authors
conclude that their “analysis by no means settles the issue of rape and the structural
position of women vis-à-vis men” (p. 174). Lastly, this study thoroughly investigated
different elements of inequality. For example, as discussed above, they investigated more
specific occupation categories to gain an understanding of the relationship between
gender parity in occupations and rape rates. Ultimately, together these three elements
exemplify how complicated the relationship between gender inequality and rape may be.
Specifically, it is possible that common indicators of gender disparity do not capture the
most important aspects of gender inequality. This could be one explanation for why some
measures of gender inequality were significantly related to rape and others were not. For
instance, inequality in educational attainment was not related to rape rates in any of the
Peterson and Bailey (1992) models. Through this study, it was also demonstrated that
some of the measures of gender inequality may be confounding other relationships. In
particular, Peterson and Bailey (1992) found that although in the main model there was
no significant relationship between gender inequality in occupation type and rape rates,
when looking at specific types of occupations the findings were mixed. Indeed, they
found support for backlash and amelioration when looking at specific types of
occupations. By thoroughly investigating the relationship between gender inequality and
rape, Peterson and Bailey (1992) illustrated how complicated the relationship may be and
demonstrated important ways to attempt to uncover more information.
Some scholars have noted that much of the research on gender inequality and
violence against women, including Peterson and Bailey (1992), have acted under the
axiom that amelioration and backlash are contradictory hypotheses (Bailey, 1999;
Whaley, 2001; Whaley et al., 2013). However, these hypotheses may be corollaries of one
another contingent either on time or the level of gender inequality. For instance, Bailey
(1999) and Whaley (2001) suggested and found some evidence that backlash occurs for a
period of time leading to amelioration. According to Bailey (1999), prior to his analysis
no studies had investigated the role of changes in gender inequality and changes in the
rape rate. Indeed, I was unable to locate any studies prior to Bailey (1999) that
investigated the relationship between changes in gender inequality and changes in any
form of violence against women. Bailey (1999) investigated the relationship between
gender inequality (using male and female differences in educational attainment,
occupation type, and income as indicators) and rape rates for central cities. He found
support for the backlash hypothesis between occupation type and rape rates for 1980 and
1990 (cross-sectional models), but found an ameliorative relationship between changes in
income inequality and changes in rape rates from 1980 to 1990 (change model). In the
end, Bailey suggested that although his analysis was an “improvement over previous
investigations this analysis does not fully resolve important feminist arguments regarding
the socioeconomic status of women and rape” (1999, p. 59).
Whaley (2001) presented itself as refining the temporal theoretical relationship
between change in gender inequality and violence against women. She hypothesized that
initially during times of higher gender inequality, women experience lower levels of rape
as a “reward, “but that in the long term, gender inequality prompts cultural changes that
build support for rape, prompting an increase. The reverse was hypothesized as well. To
use the hypotheses already introduced, Whaley (2001) hypothesized that backlash occurs
first and that after a sustained period of change amelioration can occur. Using 5-, 10-, and
20-year lagged indicators, Whaley (2001) was also unable to provide strong support for
her hypothesis; however, most of the significant relationships were in the proposed
direction. For example, cross-sectional analyses suggested that higher levels of inequality
in holding executive occupations were associated with lower levels of rape supporting the
backlash hypothesis. The model using a 5-year lag from 1970 to 1975 suggested backlash
also. Yet in accord with the hypothesis, the model using a 10-year lag between 1980 and
1990 found support for the ameliorative hypotheses. Specifically, decreases in gender
inequality in holding executive occupations were associated with decreases in rape.
Therefore, findings related to holding executive occupations appeared to support backlash
cross-sectionally and amelioration over a 10-year period. However, gender inequality in
executive occupations was not significantly related to rape when looking at change from
1980 to 1985, 1970 to 1980, and 1970 to 1990. Therefore, within this analysis, the results
reflect a complicated relationship.
Additionally, Whaley et al. (2013) suggested and found some evidence that the
nature of the relationship between gender inequality and violence against women may be
nonlinear and related to the level of gender inequality. Specifically, they investigated the
relationship between gender inequality (in terms of educational attainment, median
income, and employment status) and various rates of homicides for 1990 and 2000 from
the SHR (male-female, male-male, female-male, and female-female). Using negative
binomial regressions, Whaley et al. (2013) found that gender equality (represented by a
factor score of the three variables) was positively related to male offender – female victim
homicide rates and that the squared form of gender equality was negatively related in
1990. These significant findings suggest that the relationship between gender equality and
male-female homicide is nonlinear. Gender equality and its squared term were not
significantly related to any other homicide rates except female-offender and
femalevictim. For the female-female homicide rate, gender equality was significantly and
positively related in the model that did not include the robbery rate, which was included
as a “robust control for other factors related to criminal violence” (Whaley et al., 2013, p.
743). In 2000, gender equality and its squared form were significantly related to
malefemale homicide when the robbery rate was not included but only the squared form
was significantly related (negatively) when the robbery rate was included. The same was
true for male-male homicide rates. Just gender equality is significantly and positively
related to the female-female rate when the robbery rate was not included. In all other
instances, gender equality and its squared form were not significantly related to homicide
rates. Whaley et al. (2013) interpreted the effect of adding the robbery rate as suggesting
that unmeasured factors associated with violent crime may be mediating the relationship
or that there may be an issue with multicollinearity. As a whole, they found considerable
support for their hypothesis that backlash processes are likely to function at “low to
intermediate levels” of gender equality and that amelioration occurs when equality is high
for male-female homicide rates (p. 747).
Xie, Heimer, and Lauritsen (2012a) used longitudinal methods to assess whether
there was a relationship between non-fatal intimate partner violence and absolute/relative
indicators of women’s status. They created five five-year pooled periods using MSAlevel
data (which includes the 40 largest MSAs) from the NCVS for 1980 to 2004. To
thoroughly assess their hypotheses, they also disaggregated the non-fatal violence against
women data by victim-offender relationship using the following listed categories:
stranger, intimate partner, and other known. The four hypotheses included amelioration
(which was posited to function more for intimate partner violence than non-intimate
violence), backlash (which was also posited to function more for intimate partner
violence than non-intimate violence), routine activities (which suggests that increases in
absolute status also increases exposure and could lead to an increase in stranger and
known victimization), and protective (which suggests that an increase in absolute status
may increase the accessibility of protective mechanisms and decrease victimization).
Absolute women’s status was operationalized by the percent of women who were
employed, percent of voting age women who voted in the November presidential and
congressional elections, and an index of income-educational attainment, which included
median income and percent who completed four years of college or more. Relative status
was measured using the rate of women’s labor force participation minus that of men, the
level of income-educational attainment for women minus that of men, and the rate of
voter turnout for women minus that of men. Their results found support for each of the
hypotheses except for amelioration. They found a negative relationship between women’s
absolute economic and political status and intimate partner violence which supports the
protective hypothesis. They also found the negative relationship between inequality and
intimate partner victimization which supports the backlash hypothesis. The routine
activities hypothesis was supported through a positive relationship between absolute
indicators of labor force participation and victimization by strangers and non-intimate
persons known to the victim. This longitudinal analysis does not suggest that backlash,
amelioration, protective, nor routine activities hypotheses are corollary by explaining the
relationship between inequality, absolute status, and violence against women at different
periods. However, they found evidence that the nature of the relationship is contingent on
the victim-offender relationship.
The hypotheses and findings suggesting that backlash and amelioration are
corollary are particularly important considering that much of the empirical quantitative
literature on gender inequality has been cross-sectional and assumed a linear relationship.
By design much of the extant literature is unable to address these possibilities (Whaley et
al., 2013). The evidence supporting corollary processes indicate that future analyses
should continue to consider change over time and the structural form of the relationship
as the proposed dissertation will do. In addition, Xie et al. (2012a) presented evidence
that the victim-offender relationship is important to also consider in future analyses.
In addition, while not an inequality hypothesis, several scholars have investigated the role
of absolute status on trends of violence against women. Changes in absolute status are
increases or decreases in the status of a group with no consideration or comparison to
other groups. It is possible that there is a relationship between absolute status and
violence against women and that any findings suggesting a relationship
When race-disaggregated trends of violence against women are investigated,
major gaps in the crime decline, race, and gender literature are exposed. For example,
Lauritsen (2012) illustrated that although the rape and sexual assault trends for Latinas,
Black women, and white women have been similar for the last 40 years, the gaps between
them have varied occasionally. The variation in the gap could indicate different processes
at work. These types of findings are important in that they denote the importance of
attending to gender, race, and ethnicity but are often overlooked in later studies.
Ultimately, the crime decline literature has at times denoted the varying magnitude of the
decline by race and gender subgroups, but has not produced many intersectional
approaches to explaining the disparate decline (Heimer & Lauritsen, 2008; Lauritsen,
2012; Lauritsen & Heimer, 2010; Levitt, 2004; Parker, 2008). As a consequence, the
disparate declines and trends by race, ethnicity, and gender are not well understood. The
race and inequality literature tends to lack a gender-lens, a framework denoting the role
that gender plays in how things are enacted and perceived, while at the same time the
gender and inequality literature tends to lack a race-lens, a framework denoting the role
of race (McCall, 2005). This study argues that it is important to consider violence against
women from a gender x race-lens or an intersectional-lens.
HYPOTHESES AND EMPIRICAL SUPPORT
Before discussing the ways to apply an intersectionality perspective to the gender
inequality hypothesis, it is pertinent to review the various hypotheses that exist within the
literature. As noted in the introduction, the nature of the relationship between gender
inequality and violence against women has not been clearly defined within the empirical
literature (Whaley et al., 2013). In fact, evidence exists for multiple hypotheses. The
original hypothesis and the one many organizations work under suggests that as gender
inequality decreases and the genders become more equal violence against women will
decrease (Whaley et al., 2013). This is referred to as the amelioration hypothesis.
Peterson and Bailey (1992) found some support for the amelioration hypothesis. They
investigated the relationship between various forms of inequality (general, racial, and
gender) and rape rates for United States metropolitan areas in 1980. The rape rates were
obtained from the UCR in 1980. General inequality was measured using the GINI index
which represents income inequality in the total population. Racial inequality was
measured by the difference in median income for Black and white families. Gender
inequality was measured in terms of education using the ratio of men to women who
completed 4 years of high school, who completed 4 years of college, and who completed
5 or more years of college. It was also measured in terms of income using a ratio of men
to women’s median income for all people, people who completed 4 years of high school,
and people who completed 4 years or more of college. Gender inequality was also
measured by the percent of people employed in managerial and professional positions
who were women and by a ratio of men to women who live in poverty. Multiple models
were estimated with most containing one form of gender inequality and one containing a
representative of each type. The GINI index was found to be positively and significantly
related to rape rapes in each model. This finding suggests that places with greater general
income inequality exhibited higher rape rates. Racial inequality in income was significant
and positive in the models that included gender inequality by ratio of high school
graduates only and gender inequality in percentage of women in professional occupations
only. When all forms of gender inequality were included (income, high school graduation,
and profession occupations), gender inequality in median income was significantly and
positively related to rape rates. This finding suggests that where gender inequality in
income is greater, rape rates will also be higher and this is consistent with the
amelioration hypothesis.
Another hypothesis, the backlash hypothesis, suggests an opposite relationship.
Specifically, the backlash hypothesis suggests that as gender inequality decreases and the
genders become more equal, violence against women will increase (Whaley et al., 2013).
It is important to note that the term “backlash” is sometimes used to imply a specific
mechanism causing the increase of violence. For instance, sometimes backlash is defined
as a process by which men (or other dominant group) react violently to challenges to their
dominant status. In this study, as with many past studies, the mechanism of the
relationship between gender inequality and violence against women is not being
evaluated. Rather this study examines the relationship between gender inequality and
violence against women which therefore only allows for some speculation as to the
mechanism of the relationship. Backlash, herein, is used to represent a negative
relationship between changes in gender inequality and changes in violence against
women.
Within the same article described above, Peterson and Bailey (1992) also found
support for a backlash hypothesis when further investigating the relationship between
gender inequality in occupations and rape rates. Specifically, they separated professional
occupations into 1) executives, administrators, and managers 2) public officials and
administrators 3) engineers and natural scientists 4) health diagnostic occupations 5)
elementary and secondary teachers. They found that where the percent of executives,
administrators, and managers that are women were higher rape rates were also higher
suggesting backlash. However, they also found that where the percent of health diagnostic
occupations that are women were higher rape rates were lower. This was true of
elementary and secondary teachers as well. Since being an elementary or secondary
teacher is a women-typed occupation, the finding of a negative relationship could be
interpreted as backlash as well. The authors suggested that “the more that women remain
in traditional ‘female’ occupations, the greater their ‘reward’ – freedom from rape” (p.
172). However, the significant finding for health diagnostic occupations appears to
support amelioration, since health diagnostic occupations are traditionally men-typed
occupations.
The Peterson and Bailey (1992) study is illustrative for three primary reasons.
First, the analysis assesses the relationship between gender inequality and rape in
multiple ways. They use several forms of inequality including: education, income, and
occupation. In addition, they estimate models with each form individually as well as a
model with all of the forms. Second, this study is illustrative because it finds support for
both the ameliorative and backlash hypotheses. Due to these mixed findings, the authors
conclude that their “analysis by no means settles the issue of rape and the structural
position of women vis-à-vis men” (p. 174). Lastly, this study thoroughly investigated
different elements of inequality. For example, as discussed above, they investigated more
specific occupation categories to gain an understanding of the relationship between
gender parity in occupations and rape rates. Ultimately, together these three elements
exemplify how complicated the relationship between gender inequality and rape may be.
Specifically, it is possible that common indicators of gender disparity do not capture the
most important aspects of gender inequality. This could be one explanation for why some
measures of gender inequality were significantly related to rape and others were not. For
instance, inequality in educational attainment was not related to rape rates in any of the
Peterson and Bailey (1992) models. Through this study, it was also demonstrated that
some of the measures of gender inequality may be confounding other relationships. In
particular, Peterson and Bailey (1992) found that although in the main model there was
no significant relationship between gender inequality in occupation type and rape rates,
when looking at specific types of occupations the findings were mixed. Indeed, they
found support for backlash and amelioration when looking at specific types of
occupations. By thoroughly investigating the relationship between gender inequality and
rape, Peterson and Bailey (1992) illustrated how complicated the relationship may be and
demonstrated important ways to attempt to uncover more information.
Some scholars have noted that much of the research on gender inequality and
violence against women, including Peterson and Bailey (1992), have acted under the
axiom that amelioration and backlash are contradictory hypotheses (Bailey, 1999;
Whaley, 2001; Whaley et al., 2013). However, these hypotheses may be corollaries of one
another contingent either on time or the level of gender inequality. For instance, Bailey
(1999) and Whaley (2001) suggested and found some evidence that backlash occurs for a
period of time leading to amelioration. According to Bailey (1999), prior to his analysis
no studies had investigated the role of changes in gender inequality and changes in the
rape rate. Indeed, I was unable to locate any studies prior to Bailey (1999) that
investigated the relationship between changes in gender inequality and changes in any
form of violence against women. Bailey (1999) investigated the relationship between
gender inequality (using male and female differences in educational attainment,
occupation type, and income as indicators) and rape rates for central cities. He found
support for the backlash hypothesis between occupation type and rape rates for 1980 and
1990 (cross-sectional models), but found an ameliorative relationship between changes in
income inequality and changes in rape rates from 1980 to 1990 (change model). In the
end, Bailey suggested that although his analysis was an “improvement over previous
investigations this analysis does not fully resolve important feminist arguments regarding
the socioeconomic status of women and rape” (1999, p. 59).
Whaley (2001) presented itself as refining the temporal theoretical relationship
between change in gender inequality and violence against women. She hypothesized that
initially during times of higher gender inequality, women experience lower levels of rape
as a “reward, “but that in the long term, gender inequality prompts cultural changes that
build support for rape, prompting an increase. The reverse was hypothesized as well. To
use the hypotheses already introduced, Whaley (2001) hypothesized that backlash occurs
first and that after a sustained period of change amelioration can occur. Using 5-, 10-, and
20-year lagged indicators, Whaley (2001) was also unable to provide strong support for
her hypothesis; however, most of the significant relationships were in the proposed
direction. For example, cross-sectional analyses suggested that higher levels of inequality
in holding executive occupations were associated with lower levels of rape supporting the
backlash hypothesis. The model using a 5-year lag from 1970 to 1975 suggested backlash
also. Yet in accord with the hypothesis, the model using a 10-year lag between 1980 and
1990 found support for the ameliorative hypotheses. Specifically, decreases in gender
inequality in holding executive occupations were associated with decreases in rape.
Therefore, findings related to holding executive occupations appeared to support backlash
cross-sectionally and amelioration over a 10-year period. However, gender inequality in
executive occupations was not significantly related to rape when looking at change from
1980 to 1985, 1970 to 1980, and 1970 to 1990. Therefore, within this analysis, the results
reflect a complicated relationship.
Additionally, Whaley et al. (2013) suggested and found some evidence that the
nature of the relationship between gender inequality and violence against women may be
nonlinear and related to the level of gender inequality. Specifically, they investigated the
relationship between gender inequality (in terms of educational attainment, median
income, and employment status) and various rates of homicides for 1990 and 2000 from
the SHR (male-female, male-male, female-male, and female-female). Using negative
binomial regressions, Whaley et al. (2013) found that gender equality (represented by a
factor score of the three variables) was positively related to male offender – female victim
homicide rates and that the squared form of gender equality was negatively related in
1990. These significant findings suggest that the relationship between gender equality and
male-female homicide is nonlinear. Gender equality and its squared term were not
significantly related to any other homicide rates except female-offender and
femalevictim. For the female-female homicide rate, gender equality was significantly and
positively related in the model that did not include the robbery rate, which was included
as a “robust control for other factors related to criminal violence” (Whaley et al., 2013, p.
743). In 2000, gender equality and its squared form were significantly related to
malefemale homicide when the robbery rate was not included but only the squared form
was significantly related (negatively) when the robbery rate was included. The same was
true for male-male homicide rates. Just gender equality is significantly and positively
related to the female-female rate when the robbery rate was not included. In all other
instances, gender equality and its squared form were not significantly related to homicide
rates. Whaley et al. (2013) interpreted the effect of adding the robbery rate as suggesting
that unmeasured factors associated with violent crime may be mediating the relationship
or that there may be an issue with multicollinearity. As a whole, they found considerable
support for their hypothesis that backlash processes are likely to function at “low to
intermediate levels” of gender equality and that amelioration occurs when equality is high
for male-female homicide rates (p. 747).
Xie, Heimer, and Lauritsen (2012a) used longitudinal methods to assess whether
there was a relationship between non-fatal intimate partner violence and absolute/relative
indicators of women’s status. They created five five-year pooled periods using MSAlevel
data (which includes the 40 largest MSAs) from the NCVS for 1980 to 2004. To
thoroughly assess their hypotheses, they also disaggregated the non-fatal violence against
women data by victim-offender relationship using the following listed categories:
stranger, intimate partner, and other known. The four hypotheses included amelioration
(which was posited to function more for intimate partner violence than non-intimate
violence), backlash (which was also posited to function more for intimate partner
violence than non-intimate violence), routine activities (which suggests that increases in
absolute status also increases exposure and could lead to an increase in stranger and
known victimization), and protective (which suggests that an increase in absolute status
may increase the accessibility of protective mechanisms and decrease victimization).
Absolute women’s status was operationalized by the percent of women who were
employed, percent of voting age women who voted in the November presidential and
congressional elections, and an index of income-educational attainment, which included
median income and percent who completed four years of college or more. Relative status
was measured using the rate of women’s labor force participation minus that of men, the
level of income-educational attainment for women minus that of men, and the rate of
voter turnout for women minus that of men. Their results found support for each of the
hypotheses except for amelioration. They found a negative relationship between women’s
absolute economic and political status and intimate partner violence which supports the
protective hypothesis. They also found the negative relationship between inequality and
intimate partner victimization which supports the backlash hypothesis. The routine
activities hypothesis was supported through a positive relationship between absolute
indicators of labor force participation and victimization by strangers and non-intimate
persons known to the victim. This longitudinal analysis does not suggest that backlash,
amelioration, protective, nor routine activities hypotheses are corollary by explaining the
relationship between inequality, absolute status, and violence against women at different
periods. However, they found evidence that the nature of the relationship is contingent on
the victim-offender relationship.
The hypotheses and findings suggesting that backlash and amelioration are
corollary are particularly important considering that much of the empirical quantitative
literature on gender inequality has been cross-sectional and assumed a linear relationship.
By design much of the extant literature is unable to address these possibilities (Whaley et
al., 2013). The evidence supporting corollary processes indicate that future analyses
should continue to consider change over time and the structural form of the relationship
as the proposed dissertation will do. In addition, Xie et al. (2012a) presented evidence
that the victim-offender relationship is important to also consider in future analyses.
In addition, while not an inequality hypothesis, several scholars have investigated the role
of absolute status on trends of violence against women. Changes in absolute status are
increases or decreases in the status of a group with no consideration or comparison to
other groups. It is possible that there is a relationship between absolute status and
violence against women and that any findings suggesting a relationship
When race-disaggregated trends of violence against women are investigated,
major gaps in the crime decline, race, and gender literature are exposed. For example,
Lauritsen (2012) illustrated that although the rape and sexual assault trends for Latinas,
Black women, and white women have been similar for the last 40 years, the gaps between
them have varied occasionally. The variation in the gap could indicate different processes
at work. These types of findings are important in that they denote the importance of
attending to gender, race, and ethnicity but are often overlooked in later studies.
Ultimately, the crime decline literature has at times denoted the varying magnitude of the
decline by race and gender subgroups, but has not produced many intersectional
approaches to explaining the disparate decline (Heimer & Lauritsen, 2008; Lauritsen,
2012; Lauritsen & Heimer, 2010; Levitt, 2004; Parker, 2008). As a consequence, the
disparate declines and trends by race, ethnicity, and gender are not well understood. The
race and inequality literature tends to lack a gender-lens, a framework denoting the role
that gender plays in how things are enacted and perceived, while at the same time the
gender and inequality literature tends to lack a race-lens, a framework denoting the role
of race (McCall, 2005). This study argues that it is important to consider violence against
women from a gender x race-lens or an intersectional-lens.
HYPOTHESES AND EMPIRICAL SUPPORT
Before discussing the ways to apply an intersectionality perspective to the gender
inequality hypothesis, it is pertinent to review the various hypotheses that exist within the
literature. As noted in the introduction, the nature of the relationship between gender
inequality and violence against women has not been clearly defined within the empirical
literature (Whaley et al., 2013). In fact, evidence exists for multiple hypotheses. The
original hypothesis and the one many organizations work under suggests that as gender
inequality decreases and the genders become more equal violence against women will
decrease (Whaley et al., 2013). This is referred to as the amelioration hypothesis.
Peterson and Bailey (1992) found some support for the amelioration hypothesis. They
investigated the relationship between various forms of inequality (general, racial, and
gender) and rape rates for United States metropolitan areas in 1980. The rape rates were
obtained from the UCR in 1980. General inequality was measured using the GINI index
which represents income inequality in the total population. Racial inequality was
measured by the difference in median income for Black and white families. Gender
inequality was measured in terms of education using the ratio of men to women who
completed 4 years of high school, who completed 4 years of college, and who completed
5 or more years of college. It was also measured in terms of income using a ratio of men
to women’s median income for all people, people who completed 4 years of high school,
and people who completed 4 years or more of college. Gender inequality was also
measured by the percent of people employed in managerial and professional positions
who were women and by a ratio of men to women who live in poverty. Multiple models
were estimated with most containing one form of gender inequality and one containing a
representative of each type. The GINI index was found to be positively and significantly
related to rape rapes in each model. This finding suggests that places with greater general
income inequality exhibited higher rape rates. Racial inequality in income was significant
and positive in the models that included gender inequality by ratio of high school
graduates only and gender inequality in percentage of women in professional occupations
only. When all forms of gender inequality were included (income, high school graduation,
and profession occupations), gender inequality in median income was significantly and
positively related to rape rates. This finding suggests that where gender inequality in
income is greater, rape rates will also be higher and this is consistent with the
amelioration hypothesis.
Another hypothesis, the backlash hypothesis, suggests an opposite relationship.
Specifically, the backlash hypothesis suggests that as gender inequality decreases and the
genders become more equal, violence against women will increase (Whaley et al., 2013).
It is important to note that the term “backlash” is sometimes used to imply a specific
mechanism causing the increase of violence. For instance, sometimes backlash is defined
as a process by which men (or other dominant group) react violently to challenges to their
dominant status. In this study, as with many past studies, the mechanism of the
relationship between gender inequality and violence against women is not being
evaluated. Rather this study examines the relationship between gender inequality and
violence against women which therefore only allows for some speculation as to the
mechanism of the relationship. Backlash, herein, is used to represent a negative
relationship between changes in gender inequality and changes in violence against
women.
Within the same article described above, Peterson and Bailey (1992) also found
support for a backlash hypothesis when further investigating the relationship between
gender inequality in occupations and rape rates. Specifically, they separated professional
occupations into 1) executives, administrators, and managers 2) public officials and
administrators 3) engineers and natural scientists 4) health diagnostic occupations 5)
elementary and secondary teachers. They found that where the percent of executives,
administrators, and managers that are women were higher rape rates were also higher
suggesting backlash. However, they also found that where the percent of health diagnostic
occupations that are women were higher rape rates were lower. This was true of
elementary and secondary teachers as well. Since being an elementary or secondary
teacher is a women-typed occupation, the finding of a negative relationship could be
interpreted as backlash as well. The authors suggested that “the more that women remain
in traditional ‘female’ occupations, the greater their ‘reward’ – freedom from rape” (p.
172). However, the significant finding for health diagnostic occupations appears to
support amelioration, since health diagnostic occupations are traditionally men-typed
occupations.
The Peterson and Bailey (1992) study is illustrative for three primary reasons.
First, the analysis assesses the relationship between gender inequality and rape in
multiple ways. They use several forms of inequality including: education, income, and
occupation. In addition, they estimate models with each form individually as well as a
model with all of the forms. Second, this study is illustrative because it finds support for
both the ameliorative and backlash hypotheses. Due to these mixed findings, the authors
conclude that their “analysis by no means settles the issue of rape and the structural
position of women vis-à-vis men” (p. 174). Lastly, this study thoroughly investigated
different elements of inequality. For example, as discussed above, they investigated more
specific occupation categories to gain an understanding of the relationship between
gender parity in occupations and rape rates. Ultimately, together these three elements
exemplify how complicated the relationship between gender inequality and rape may be.
Specifically, it is possible that common indicators of gender disparity do not capture the
most important aspects of gender inequality. This could be one explanation for why some
measures of gender inequality were significantly related to rape and others were not. For
instance, inequality in educational attainment was not related to rape rates in any of the
Peterson and Bailey (1992) models. Through this study, it was also demonstrated that
some of the measures of gender inequality may be confounding other relationships. In
particular, Peterson and Bailey (1992) found that although in the main model there was
no significant relationship between gender inequality in occupation type and rape rates,
when looking at specific types of occupations the findings were mixed. Indeed, they
found support for backlash and amelioration when looking at specific types of
occupations. By thoroughly investigating the relationship between gender inequality and
rape, Peterson and Bailey (1992) illustrated how complicated the relationship may be and
demonstrated important ways to attempt to uncover more information.
Some scholars have noted that much of the research on gender inequality and
violence against women, including Peterson and Bailey (1992), have acted under the
axiom that amelioration and backlash are contradictory hypotheses (Bailey, 1999;
Whaley, 2001; Whaley et al., 2013). However, these hypotheses may be corollaries of one
another contingent either on time or the level of gender inequality. For instance, Bailey
(1999) and Whaley (2001) suggested and found some evidence that backlash occurs for a
period of time leading to amelioration. According to Bailey (1999), prior to his analysis
no studies had investigated the role of changes in gender inequality and changes in the
rape rate. Indeed, I was unable to locate any studies prior to Bailey (1999) that
investigated the relationship between changes in gender inequality and changes in any
form of violence against women. Bailey (1999) investigated the relationship between
gender inequality (using male and female differences in educational attainment,
occupation type, and income as indicators) and rape rates for central cities. He found
support for the backlash hypothesis between occupation type and rape rates for 1980 and
1990 (cross-sectional models), but found an ameliorative relationship between changes in
income inequality and changes in rape rates from 1980 to 1990 (change model). In the
end, Bailey suggested that although his analysis was an “improvement over previous
investigations this analysis does not fully resolve important feminist arguments regarding
the socioeconomic status of women and rape” (1999, p. 59).
Whaley (2001) presented itself as refining the temporal theoretical relationship
between change in gender inequality and violence against women. She hypothesized that
initially during times of higher gender inequality, women experience lower levels of rape
as a “reward, “but that in the long term, gender inequality prompts cultural changes that
build support for rape, prompting an increase. The reverse was hypothesized as well. To
use the hypotheses already introduced, Whaley (2001) hypothesized that backlash occurs
first and that after a sustained period of change amelioration can occur. Using 5-, 10-, and
20-year lagged indicators, Whaley (2001) was also unable to provide strong support for
her hypothesis; however, most of the significant relationships were in the proposed
direction. For example, cross-sectional analyses suggested that higher levels of inequality
in holding executive occupations were associated with lower levels of rape supporting the
backlash hypothesis. The model using a 5-year lag from 1970 to 1975 suggested backlash
also. Yet in accord with the hypothesis, the model using a 10-year lag between 1980 and
1990 found support for the ameliorative hypotheses. Specifically, decreases in gender
inequality in holding executive occupations were associated with decreases in rape.
Therefore, findings related to holding executive occupations appeared to support backlash
cross-sectionally and amelioration over a 10-year period. However, gender inequality in
executive occupations was not significantly related to rape when looking at change from
1980 to 1985, 1970 to 1980, and 1970 to 1990. Therefore, within this analysis, the results
reflect a complicated relationship.
Additionally, Whaley et al. (2013) suggested and found some evidence that the
nature of the relationship between gender inequality and violence against women may be
nonlinear and related to the level of gender inequality. Specifically, they investigated the
relationship between gender inequality (in terms of educational attainment, median
income, and employment status) and various rates of homicides for 1990 and 2000 from
the SHR (male-female, male-male, female-male, and female-female). Using negative
binomial regressions, Whaley et al. (2013) found that gender equality (represented by a
factor score of the three variables) was positively related to male offender – female victim
homicide rates and that the squared form of gender equality was negatively related in
1990. These significant findings suggest that the relationship between gender equality and
male-female homicide is nonlinear. Gender equality and its squared term were not
significantly related to any other homicide rates except female-offender and
femalevictim. For the female-female homicide rate, gender equality was significantly and
positively related in the model that did not include the robbery rate, which was included
as a “robust control for other factors related to criminal violence” (Whaley et al., 2013, p.
743). In 2000, gender equality and its squared form were significantly related to
malefemale homicide when the robbery rate was not included but only the squared form
was significantly related (negatively) when the robbery rate was included. The same was
true for male-male homicide rates. Just gender equality is significantly and positively
related to the female-female rate when the robbery rate was not included. In all other
instances, gender equality and its squared form were not significantly related to homicide
rates. Whaley et al. (2013) interpreted the effect of adding the robbery rate as suggesting
that unmeasured factors associated with violent crime may be mediating the relationship
or that there may be an issue with multicollinearity. As a whole, they found considerable
support for their hypothesis that backlash processes are likely to function at “low to
intermediate levels” of gender equality and that amelioration occurs when equality is high
for male-female homicide rates (p. 747).
Xie, Heimer, and Lauritsen (2012a) used longitudinal methods to assess whether
there was a relationship between non-fatal intimate partner violence and absolute/relative
indicators of women’s status. They created five five-year pooled periods using MSAlevel
data (which includes the 40 largest MSAs) from the NCVS for 1980 to 2004. To
thoroughly assess their hypotheses, they also disaggregated the non-fatal violence against
women data by victim-offender relationship using the following listed categories:
stranger, intimate partner, and other known. The four hypotheses included amelioration
(which was posited to function more for intimate partner violence than non-intimate
violence), backlash (which was also posited to function more for intimate partner
violence than non-intimate violence), routine activities (which suggests that increases in
absolute status also increases exposure and could lead to an increase in stranger and
known victimization), and protective (which suggests that an increase in absolute status
may increase the accessibility of protective mechanisms and decrease victimization).
Absolute women’s status was operationalized by the percent of women who were
employed, percent of voting age women who voted in the November presidential and
congressional elections, and an index of income-educational attainment, which included
median income and percent who completed four years of college or more. Relative status
was measured using the rate of women’s labor force participation minus that of men, the
level of income-educational attainment for women minus that of men, and the rate of
voter turnout for women minus that of men. Their results found support for each of the
hypotheses except for amelioration. They found a negative relationship between women’s
absolute economic and political status and intimate partner violence which supports the
protective hypothesis. They also found the negative relationship between inequality and
intimate partner victimization which supports the backlash hypothesis. The routine
activities hypothesis was supported through a positive relationship between absolute
indicators of labor force participation and victimization by strangers and non-intimate
persons known to the victim. This longitudinal analysis does not suggest that backlash,
amelioration, protective, nor routine activities hypotheses are corollary by explaining the
relationship between inequality, absolute status, and violence against women at different
periods. However, they found evidence that the nature of the relationship is contingent on
the victim-offender relationship.
The hypotheses and findings suggesting that backlash and amelioration are
corollary are particularly important considering that much of the empirical quantitative
literature on gender inequality has been cross-sectional and assumed a linear relationship.
By design much of the extant literature is unable to address these possibilities (Whaley et
al., 2013). The evidence supporting corollary processes indicate that future analyses
should continue to consider change over time and the structural form of the relationship
as the proposed dissertation will do. In addition, Xie et al. (2012a) presented evidence
that the victim-offender relationship is important to also consider in future analyses.
In addition, while not an inequality hypothesis, several scholars have investigated the role
of absolute status on trends of violence against women. Changes in absolute status are
increases or decreases in the status of a group with no consideration or comparison to
other groups. It is possible that there is a relationship between absolute status and
violence against women and that any findings suggesting a relationship
When race-disaggregated trends of violence against women are investigated,
major gaps in the crime decline, race, and gender literature are exposed. For example,
Lauritsen (2012) illustrated that although the rape and sexual assault trends for Latinas,
Black women, and white women have been similar for the last 40 years, the gaps between
them have varied occasionally. The variation in the gap could indicate different processes
at work. These types of findings are important in that they denote the importance of
attending to gender, race, and ethnicity but are often overlooked in later studies.
Ultimately, the crime decline literature has at times denoted the varying magnitude of the
decline by race and gender subgroups, but has not produced many intersectional
approaches to explaining the disparate decline (Heimer & Lauritsen, 2008; Lauritsen,
2012; Lauritsen & Heimer, 2010; Levitt, 2004; Parker, 2008). As a consequence, the
disparate declines and trends by race, ethnicity, and gender are not well understood. The
race and inequality literature tends to lack a gender-lens, a framework denoting the role
that gender plays in how things are enacted and perceived, while at the same time the
gender and inequality literature tends to lack a race-lens, a framework denoting the role
of race (McCall, 2005). This study argues that it is important to consider violence against
women from a gender x race-lens or an intersectional-lens.
HYPOTHESES AND EMPIRICAL SUPPORT
Before discussing the ways to apply an intersectionality perspective to the gender
inequality hypothesis, it is pertinent to review the various hypotheses that exist within the
literature. As noted in the introduction, the nature of the relationship between gender
inequality and violence against women has not been clearly defined within the empirical
literature (Whaley et al., 2013). In fact, evidence exists for multiple hypotheses. The
original hypothesis and the one many organizations work under suggests that as gender
inequality decreases and the genders become more equal violence against women will
decrease (Whaley et al., 2013). This is referred to as the amelioration hypothesis.
Peterson and Bailey (1992) found some support for the amelioration hypothesis. They
investigated the relationship between various forms of inequality (general, racial, and
gender) and rape rates for United States metropolitan areas in 1980. The rape rates were
obtained from the UCR in 1980. General inequality was measured using the GINI index
which represents income inequality in the total population. Racial inequality was
measured by the difference in median income for Black and white families. Gender
inequality was measured in terms of education using the ratio of men to women who
completed 4 years of high school, who completed 4 years of college, and who completed
5 or more years of college. It was also measured in terms of income using a ratio of men
to women’s median income for all people, people who completed 4 years of high school,
and people who completed 4 years or more of college. Gender inequality was also
measured by the percent of people employed in managerial and professional positions
who were women and by a ratio of men to women who live in poverty. Multiple models
were estimated with most containing one form of gender inequality and one containing a
representative of each type. The GINI index was found to be positively and significantly
related to rape rapes in each model. This finding suggests that places with greater general
income inequality exhibited higher rape rates. Racial inequality in income was significant
and positive in the models that included gender inequality by ratio of high school
graduates only and gender inequality in percentage of women in professional occupations
only. When all forms of gender inequality were included (income, high school graduation,
and profession occupations), gender inequality in median income was significantly and
positively related to rape rates. This finding suggests that where gender inequality in
income is greater, rape rates will also be higher and this is consistent with the
amelioration hypothesis.
Another hypothesis, the backlash hypothesis, suggests an opposite relationship.
Specifically, the backlash hypothesis suggests that as gender inequality decreases and the
genders become more equal, violence against women will increase (Whaley et al., 2013).
It is important to note that the term “backlash” is sometimes used to imply a specific
mechanism causing the increase of violence. For instance, sometimes backlash is defined
as a process by which men (or other dominant group) react violently to challenges to their
dominant status. In this study, as with many past studies, the mechanism of the
relationship between gender inequality and violence against women is not being
evaluated. Rather this study examines the relationship between gender inequality and
violence against women which therefore only allows for some speculation as to the
mechanism of the relationship. Backlash, herein, is used to represent a negative
relationship between changes in gender inequality and changes in violence against
women.
Within the same article described above, Peterson and Bailey (1992) also found
support for a backlash hypothesis when further investigating the relationship between
gender inequality in occupations and rape rates. Specifically, they separated professional
occupations into 1) executives, administrators, and managers 2) public officials and
administrators 3) engineers and natural scientists 4) health diagnostic occupations 5)
elementary and secondary teachers. They found that where the percent of executives,
administrators, and managers that are women were higher rape rates were also higher
suggesting backlash. However, they also found that where the percent of health diagnostic
occupations that are women were higher rape rates were lower. This was true of
elementary and secondary teachers as well. Since being an elementary or secondary
teacher is a women-typed occupation, the finding of a negative relationship could be
interpreted as backlash as well. The authors suggested that “the more that women remain
in traditional ‘female’ occupations, the greater their ‘reward’ – freedom from rape” (p.
172). However, the significant finding for health diagnostic occupations appears to
support amelioration, since health diagnostic occupations are traditionally men-typed
occupations.
The Peterson and Bailey (1992) study is illustrative for three primary reasons.
First, the analysis assesses the relationship between gender inequality and rape in
multiple ways. They use several forms of inequality including: education, income, and
occupation. In addition, they estimate models with each form individually as well as a
model with all of the forms. Second, this study is illustrative because it finds support for
both the ameliorative and backlash hypotheses. Due to these mixed findings, the authors
conclude that their “analysis by no means settles the issue of rape and the structural
position of women vis-à-vis men” (p. 174). Lastly, this study thoroughly investigated
different elements of inequality. For example, as discussed above, they investigated more
specific occupation categories to gain an understanding of the relationship between
gender parity in occupations and rape rates. Ultimately, together these three elements
exemplify how complicated the relationship between gender inequality and rape may be.
Specifically, it is possible that common indicators of gender disparity do not capture the
most important aspects of gender inequality. This could be one explanation for why some
measures of gender inequality were significantly related to rape and others were not. For
instance, inequality in educational attainment was not related to rape rates in any of the
Peterson and Bailey (1992) models. Through this study, it was also demonstrated that
some of the measures of gender inequality may be confounding other relationships. In
particular, Peterson and Bailey (1992) found that although in the main model there was
no significant relationship between gender inequality in occupation type and rape rates,
when looking at specific types of occupations the findings were mixed. Indeed, they
found support for backlash and amelioration when looking at specific types of
occupations. By thoroughly investigating the relationship between gender inequality and
rape, Peterson and Bailey (1992) illustrated how complicated the relationship may be and
demonstrated important ways to attempt to uncover more information.
Some scholars have noted that much of the research on gender inequality and
violence against women, including Peterson and Bailey (1992), have acted under the
axiom that amelioration and backlash are contradictory hypotheses (Bailey, 1999;
Whaley, 2001; Whaley et al., 2013). However, these hypotheses may be corollaries of one
another contingent either on time or the level of gender inequality. For instance, Bailey
(1999) and Whaley (2001) suggested and found some evidence that backlash occurs for a
period of time leading to amelioration. According to Bailey (1999), prior to his analysis
no studies had investigated the role of changes in gender inequality and changes in the
rape rate. Indeed, I was unable to locate any studies prior to Bailey (1999) that
investigated the relationship between changes in gender inequality and changes in any
form of violence against women. Bailey (1999) investigated the relationship between
gender inequality (using male and female differences in educational attainment,
occupation type, and income as indicators) and rape rates for central cities. He found
support for the backlash hypothesis between occupation type and rape rates for 1980 and
1990 (cross-sectional models), but found an ameliorative relationship between changes in
income inequality and changes in rape rates from 1980 to 1990 (change model). In the
end, Bailey suggested that although his analysis was an “improvement over previous
investigations this analysis does not fully resolve important feminist arguments regarding
the socioeconomic status of women and rape” (1999, p. 59).
Whaley (2001) presented itself as refining the temporal theoretical relationship
between change in gender inequality and violence against women. She hypothesized that
initially during times of higher gender inequality, women experience lower levels of rape
as a “reward, “but that in the long term, gender inequality prompts cultural changes that
build support for rape, prompting an increase. The reverse was hypothesized as well. To
use the hypotheses already introduced, Whaley (2001) hypothesized that backlash occurs
first and that after a sustained period of change amelioration can occur. Using 5-, 10-, and
20-year lagged indicators, Whaley (2001) was also unable to provide strong support for
her hypothesis; however, most of the significant relationships were in the proposed
direction. For example, cross-sectional analyses suggested that higher levels of inequality
in holding executive occupations were associated with lower levels of rape supporting the
backlash hypothesis. The model using a 5-year lag from 1970 to 1975 suggested backlash
also. Yet in accord with the hypothesis, the model using a 10-year lag between 1980 and
1990 found support for the ameliorative hypotheses. Specifically, decreases in gender
inequality in holding executive occupations were associated with decreases in rape.
Therefore, findings related to holding executive occupations appeared to support backlash
cross-sectionally and amelioration over a 10-year period. However, gender inequality in
executive occupations was not significantly related to rape when looking at change from
1980 to 1985, 1970 to 1980, and 1970 to 1990. Therefore, within this analysis, the results
reflect a complicated relationship.
Additionally, Whaley et al. (2013) suggested and found some evidence that the
nature of the relationship between gender inequality and violence against women may be
nonlinear and related to the level of gender inequality. Specifically, they investigated the
relationship between gender inequality (in terms of educational attainment, median
income, and employment status) and various rates of homicides for 1990 and 2000 from
the SHR (male-female, male-male, female-male, and female-female). Using negative
binomial regressions, Whaley et al. (2013) found that gender equality (represented by a
factor score of the three variables) was positively related to male offender – female victim
homicide rates and that the squared form of gender equality was negatively related in
1990. These significant findings suggest that the relationship between gender equality and
male-female homicide is nonlinear. Gender equality and its squared term were not
significantly related to any other homicide rates except female-offender and
femalevictim. For the female-female homicide rate, gender equality was significantly and
positively related in the model that did not include the robbery rate, which was included
as a “robust control for other factors related to criminal violence” (Whaley et al., 2013, p.
743). In 2000, gender equality and its squared form were significantly related to
malefemale homicide when the robbery rate was not included but only the squared form
was significantly related (negatively) when the robbery rate was included. The same was
true for male-male homicide rates. Just gender equality is significantly and positively
related to the female-female rate when the robbery rate was not included. In all other
instances, gender equality and its squared form were not significantly related to homicide
rates. Whaley et al. (2013) interpreted the effect of adding the robbery rate as suggesting
that unmeasured factors associated with violent crime may be mediating the relationship
or that there may be an issue with multicollinearity. As a whole, they found considerable
support for their hypothesis that backlash processes are likely to function at “low to
intermediate levels” of gender equality and that amelioration occurs when equality is high
for male-female homicide rates (p. 747).
Xie, Heimer, and Lauritsen (2012a) used longitudinal methods to assess whether
there was a relationship between non-fatal intimate partner violence and absolute/relative
indicators of women’s status. They created five five-year pooled periods using MSAlevel
data (which includes the 40 largest MSAs) from the NCVS for 1980 to 2004. To
thoroughly assess their hypotheses, they also disaggregated the non-fatal violence against
women data by victim-offender relationship using the following listed categories:
stranger, intimate partner, and other known. The four hypotheses included amelioration
(which was posited to function more for intimate partner violence than non-intimate
violence), backlash (which was also posited to function more for intimate partner
violence than non-intimate violence), routine activities (which suggests that increases in
absolute status also increases exposure and could lead to an increase in stranger and
known victimization), and protective (which suggests that an increase in absolute status
may increase the accessibility of protective mechanisms and decrease victimization).
Absolute women’s status was operationalized by the percent of women who were
employed, percent of voting age women who voted in the November presidential and
congressional elections, and an index of income-educational attainment, which included
median income and percent who completed four years of college or more. Relative status
was measured using the rate of women’s labor force participation minus that of men, the
level of income-educational attainment for women minus that of men, and the rate of
voter turnout for women minus that of men. Their results found support for each of the
hypotheses except for amelioration. They found a negative relationship between women’s
absolute economic and political status and intimate partner violence which supports the
protective hypothesis. They also found the negative relationship between inequality and
intimate partner victimization which supports the backlash hypothesis. The routine
activities hypothesis was supported through a positive relationship between absolute
indicators of labor force participation and victimization by strangers and non-intimate
persons known to the victim. This longitudinal analysis does not suggest that backlash,
amelioration, protective, nor routine activities hypotheses are corollary by explaining the
relationship between inequality, absolute status, and violence against women at different
periods. However, they found evidence that the nature of the relationship is contingent on
the victim-offender relationship.
The hypotheses and findings suggesting that backlash and amelioration are
corollary are particularly important considering that much of the empirical quantitative
literature on gender inequality has been cross-sectional and assumed a linear relationship.
By design much of the extant literature is unable to address these possibilities (Whaley et
al., 2013). The evidence supporting corollary processes indicate that future analyses
should continue to consider change over time and the structural form of the relationship
as the proposed dissertation will do. In addition, Xie et al. (2012a) presented evidence
that the victim-offender relationship is important to also consider in future analyses.
In addition, while not an inequality hypothesis, several scholars have investigated the role
of absolute status on trends of violence against women. Changes in absolute status are
increases or decreases in the status of a group with no consideration or comparison to
other groups. It is possible that there is a relationship between absolute status and
violence against women and that any findings suggesting a relationship
When race-disaggregated trends of violence against women are investigated,
major gaps in the crime decline, race, and gender literature are exposed. For example,
Lauritsen (2012) illustrated that although the rape and sexual assault trends for Latinas,
Black women, and white women have been similar for the last 40 years, the gaps between
them have varied occasionally. The variation in the gap could indicate different processes
at work. These types of findings are important in that they denote the importance of
attending to gender, race, and ethnicity but are often overlooked in later studies.
Ultimately, the crime decline literature has at times denoted the varying magnitude of the
decline by race and gender subgroups, but has not produced many intersectional
approaches to explaining the disparate decline (Heimer & Lauritsen, 2008; Lauritsen,
2012; Lauritsen & Heimer, 2010; Levitt, 2004; Parker, 2008). As a consequence, the
disparate declines and trends by race, ethnicity, and gender are not well understood. The
race and inequality literature tends to lack a gender-lens, a framework denoting the role
that gender plays in how things are enacted and perceived, while at the same time the
gender and inequality literature tends to lack a race-lens, a framework denoting the role
of race (McCall, 2005). This study argues that it is important to consider violence against
women from a gender x race-lens or an intersectional-lens.
HYPOTHESES AND EMPIRICAL SUPPORT
Before discussing the ways to apply an intersectionality perspective to the gender
inequality hypothesis, it is pertinent to review the various hypotheses that exist within the
literature. As noted in the introduction, the nature of the relationship between gender
inequality and violence against women has not been clearly defined within the empirical
literature (Whaley et al., 2013). In fact, evidence exists for multiple hypotheses. The
original hypothesis and the one many organizations work under suggests that as gender
inequality decreases and the genders become more equal violence against women will
decrease (Whaley et al., 2013). This is referred to as the amelioration hypothesis.
Peterson and Bailey (1992) found some support for the amelioration hypothesis. They
investigated the relationship between various forms of inequality (general, racial, and
gender) and rape rates for United States metropolitan areas in 1980. The rape rates were
obtained from the UCR in 1980. General inequality was measured using the GINI index
which represents income inequality in the total population. Racial inequality was
measured by the difference in median income for Black and white families. Gender
inequality was measured in terms of education using the ratio of men to women who
completed 4 years of high school, who completed 4 years of college, and who completed
5 or more years of college. It was also measured in terms of income using a ratio of men
to women’s median income for all people, people who completed 4 years of high school,
and people who completed 4 years or more of college. Gender inequality was also
measured by the percent of people employed in managerial and professional positions
who were women and by a ratio of men to women who live in poverty. Multiple models
were estimated with most containing one form of gender inequality and one containing a
representative of each type. The GINI index was found to be positively and significantly
related to rape rapes in each model. This finding suggests that places with greater general
income inequality exhibited higher rape rates. Racial inequality in income was significant
and positive in the models that included gender inequality by ratio of high school
graduates only and gender inequality in percentage of women in professional occupations
only. When all forms of gender inequality were included (income, high school graduation,
and profession occupations), gender inequality in median income was significantly and
positively related to rape rates. This finding suggests that where gender inequality in
income is greater, rape rates will also be higher and this is consistent with the
amelioration hypothesis.
Another hypothesis, the backlash hypothesis, suggests an opposite relationship.
Specifically, the backlash hypothesis suggests that as gender inequality decreases and the
genders become more equal, violence against women will increase (Whaley et al., 2013).
It is important to note that the term “backlash” is sometimes used to imply a specific
mechanism causing the increase of violence. For instance, sometimes backlash is defined
as a process by which men (or other dominant group) react violently to challenges to their
dominant status. In this study, as with many past studies, the mechanism of the
relationship between gender inequality and violence against women is not being
evaluated. Rather this study examines the relationship between gender inequality and
violence against women which therefore only allows for some speculation as to the
mechanism of the relationship. Backlash, herein, is used to represent a negative
relationship between changes in gender inequality and changes in violence against
women.
Within the same article described above, Peterson and Bailey (1992) also found
support for a backlash hypothesis when further investigating the relationship between
gender inequality in occupations and rape rates. Specifically, they separated professional
occupations into 1) executives, administrators, and managers 2) public officials and
administrators 3) engineers and natural scientists 4) health diagnostic occupations 5)
elementary and secondary teachers. They found that where the percent of executives,
administrators, and managers that are women were higher rape rates were also higher
suggesting backlash. However, they also found that where the percent of health diagnostic
occupations that are women were higher rape rates were lower. This was true of
elementary and secondary teachers as well. Since being an elementary or secondary
teacher is a women-typed occupation, the finding of a negative relationship could be
interpreted as backlash as well. The authors suggested that “the more that women remain
in traditional ‘female’ occupations, the greater their ‘reward’ – freedom from rape” (p.
172). However, the significant finding for health diagnostic occupations appears to
support amelioration, since health diagnostic occupations are traditionally men-typed
occupations.
The Peterson and Bailey (1992) study is illustrative for three primary reasons.
First, the analysis assesses the relationship between gender inequality and rape in
multiple ways. They use several forms of inequality including: education, income, and
occupation. In addition, they estimate models with each form individually as well as a
model with all of the forms. Second, this study is illustrative because it finds support for
both the ameliorative and backlash hypotheses. Due to these mixed findings, the authors
conclude that their “analysis by no means settles the issue of rape and the structural
position of women vis-à-vis men” (p. 174). Lastly, this study thoroughly investigated
different elements of inequality. For example, as discussed above, they investigated more
specific occupation categories to gain an understanding of the relationship between
gender parity in occupations and rape rates. Ultimately, together these three elements
exemplify how complicated the relationship between gender inequality and rape may be.
Specifically, it is possible that common indicators of gender disparity do not capture the
most important aspects of gender inequality. This could be one explanation for why some
measures of gender inequality were significantly related to rape and others were not. For
instance, inequality in educational attainment was not related to rape rates in any of the
Peterson and Bailey (1992) models. Through this study, it was also demonstrated that
some of the measures of gender inequality may be confounding other relationships. In
particular, Peterson and Bailey (1992) found that although in the main model there was
no significant relationship between gender inequality in occupation type and rape rates,
when looking at specific types of occupations the findings were mixed. Indeed, they
found support for backlash and amelioration when looking at specific types of
occupations. By thoroughly investigating the relationship between gender inequality and
rape, Peterson and Bailey (1992) illustrated how complicated the relationship may be and
demonstrated important ways to attempt to uncover more information.
Some scholars have noted that much of the research on gender inequality and
violence against women, including Peterson and Bailey (1992), have acted under the
axiom that amelioration and backlash are contradictory hypotheses (Bailey, 1999;
Whaley, 2001; Whaley et al., 2013). However, these hypotheses may be corollaries of one
another contingent either on time or the level of gender inequality. For instance, Bailey
(1999) and Whaley (2001) suggested and found some evidence that backlash occurs for a
period of time leading to amelioration. According to Bailey (1999), prior to his analysis
no studies had investigated the role of changes in gender inequality and changes in the
rape rate. Indeed, I was unable to locate any studies prior to Bailey (1999) that
investigated the relationship between changes in gender inequality and changes in any
form of violence against women. Bailey (1999) investigated the relationship between
gender inequality (using male and female differences in educational attainment,
occupation type, and income as indicators) and rape rates for central cities. He found
support for the backlash hypothesis between occupation type and rape rates for 1980 and
1990 (cross-sectional models), but found an ameliorative relationship between changes in
income inequality and changes in rape rates from 1980 to 1990 (change model). In the
end, Bailey suggested that although his analysis was an “improvement over previous
investigations this analysis does not fully resolve important feminist arguments regarding
the socioeconomic status of women and rape” (1999, p. 59).
Whaley (2001) presented itself as refining the temporal theoretical relationship
between change in gender inequality and violence against women. She hypothesized that
initially during times of higher gender inequality, women experience lower levels of rape
as a “reward, “but that in the long term, gender inequality prompts cultural changes that
build support for rape, prompting an increase. The reverse was hypothesized as well. To
use the hypotheses already introduced, Whaley (2001) hypothesized that backlash occurs
first and that after a sustained period of change amelioration can occur. Using 5-, 10-, and
20-year lagged indicators, Whaley (2001) was also unable to provide strong support for
her hypothesis; however, most of the significant relationships were in the proposed
direction. For example, cross-sectional analyses suggested that higher levels of inequality
in holding executive occupations were associated with lower levels of rape supporting the
backlash hypothesis. The model using a 5-year lag from 1970 to 1975 suggested backlash
also. Yet in accord with the hypothesis, the model using a 10-year lag between 1980 and
1990 found support for the ameliorative hypotheses. Specifically, decreases in gender
inequality in holding executive occupations were associated with decreases in rape.
Therefore, findings related to holding executive occupations appeared to support backlash
cross-sectionally and amelioration over a 10-year period. However, gender inequality in
executive occupations was not significantly related to rape when looking at change from
1980 to 1985, 1970 to 1980, and 1970 to 1990. Therefore, within this analysis, the results
reflect a complicated relationship.
Additionally, Whaley et al. (2013) suggested and found some evidence that the
nature of the relationship between gender inequality and violence against women may be
nonlinear and related to the level of gender inequality. Specifically, they investigated the
relationship between gender inequality (in terms of educational attainment, median
income, and employment status) and various rates of homicides for 1990 and 2000 from
the SHR (male-female, male-male, female-male, and female-female). Using negative
binomial regressions, Whaley et al. (2013) found that gender equality (represented by a
factor score of the three variables) was positively related to male offender – female victim
homicide rates and that the squared form of gender equality was negatively related in
1990. These significant findings suggest that the relationship between gender equality and
male-female homicide is nonlinear. Gender equality and its squared term were not
significantly related to any other homicide rates except female-offender and
femalevictim. For the female-female homicide rate, gender equality was significantly and
positively related in the model that did not include the robbery rate, which was included
as a “robust control for other factors related to criminal violence” (Whaley et al., 2013, p.
743). In 2000, gender equality and its squared form were significantly related to
malefemale homicide when the robbery rate was not included but only the squared form
was significantly related (negatively) when the robbery rate was included. The same was
true for male-male homicide rates. Just gender equality is significantly and positively
related to the female-female rate when the robbery rate was not included. In all other
instances, gender equality and its squared form were not significantly related to homicide
rates. Whaley et al. (2013) interpreted the effect of adding the robbery rate as suggesting
that unmeasured factors associated with violent crime may be mediating the relationship
or that there may be an issue with multicollinearity. As a whole, they found considerable
support for their hypothesis that backlash processes are likely to function at “low to
intermediate levels” of gender equality and that amelioration occurs when equality is high
for male-female homicide rates (p. 747).
Xie, Heimer, and Lauritsen (2012a) used longitudinal methods to assess whether
there was a relationship between non-fatal intimate partner violence and absolute/relative
indicators of women’s status. They created five five-year pooled periods using MSAlevel
data (which includes the 40 largest MSAs) from the NCVS for 1980 to 2004. To
thoroughly assess their hypotheses, they also disaggregated the non-fatal violence against
women data by victim-offender relationship using the following listed categories:
stranger, intimate partner, and other known. The four hypotheses included amelioration
(which was posited to function more for intimate partner violence than non-intimate
violence), backlash (which was also posited to function more for intimate partner
violence than non-intimate violence), routine activities (which suggests that increases in
absolute status also increases exposure and could lead to an increase in stranger and
known victimization), and protective (which suggests that an increase in absolute status
may increase the accessibility of protective mechanisms and decrease victimization).
Absolute women’s status was operationalized by the percent of women who were
employed, percent of voting age women who voted in the November presidential and
congressional elections, and an index of income-educational attainment, which included
median income and percent who completed four years of college or more. Relative status
was measured using the rate of women’s labor force participation minus that of men, the
level of income-educational attainment for women minus that of men, and the rate of
voter turnout for women minus that of men. Their results found support for each of the
hypotheses except for amelioration. They found a negative relationship between women’s
absolute economic and political status and intimate partner violence which supports the
protective hypothesis. They also found the negative relationship between inequality and
intimate partner victimization which supports the backlash hypothesis. The routine
activities hypothesis was supported through a positive relationship between absolute
indicators of labor force participation and victimization by strangers and non-intimate
persons known to the victim. This longitudinal analysis does not suggest that backlash,
amelioration, protective, nor routine activities hypotheses are corollary by explaining the
relationship between inequality, absolute status, and violence against women at different
periods. However, they found evidence that the nature of the relationship is contingent on
the victim-offender relationship.
The hypotheses and findings suggesting that backlash and amelioration are
corollary are particularly important considering that much of the empirical quantitative
literature on gender inequality has been cross-sectional and assumed a linear relationship.
By design much of the extant literature is unable to address these possibilities (Whaley et
al., 2013). The evidence supporting corollary processes indicate that future analyses
should continue to consider change over time and the structural form of the relationship
as the proposed dissertation will do. In addition, Xie et al. (2012a) presented evidence
that the victim-offender relationship is important to also consider in future analyses.
In addition, while not an inequality hypothesis, several scholars have investigated the role
of absolute status on trends of violence against women. Changes in absolute status are
increases or decreases in the status of a group with no consideration or comparison to
other groups. It is possible that there is a relationship between absolute status and
violence against women and that any findings suggesting a relationship
When race-disaggregated trends of violence against women are investigated,
major gaps in the crime decline, race, and gender literature are exposed. For example,
Lauritsen (2012) illustrated that although the rape and sexual assault trends for Latinas,
Black women, and white women have been similar for the last 40 years, the gaps between
them have varied occasionally. The variation in the gap could indicate different processes
at work. These types of findings are important in that they denote the importance of
attending to gender, race, and ethnicity but are often overlooked in later studies.
Ultimately, the crime decline literature has at times denoted the varying magnitude of the
decline by race and gender subgroups, but has not produced many intersectional
approaches to explaining the disparate decline (Heimer & Lauritsen, 2008; Lauritsen,
2012; Lauritsen & Heimer, 2010; Levitt, 2004; Parker, 2008). As a consequence, the
disparate declines and trends by race, ethnicity, and gender are not well understood. The
race and inequality literature tends to lack a gender-lens, a framework denoting the role
that gender plays in how things are enacted and perceived, while at the same time the
gender and inequality literature tends to lack a race-lens, a framework denoting the role
of race (McCall, 2005). This study argues that it is important to consider violence against
women from a gender x race-lens or an intersectional-lens.
HYPOTHESES AND EMPIRICAL SUPPORT
Before discussing the ways to apply an intersectionality perspective to the gender
inequality hypothesis, it is pertinent to review the various hypotheses that exist within the
literature. As noted in the introduction, the nature of the relationship between gender
inequality and violence against women has not been clearly defined within the empirical
literature (Whaley et al., 2013). In fact, evidence exists for multiple hypotheses. The
original hypothesis and the one many organizations work under suggests that as gender
inequality decreases and the genders become more equal violence against women will
decrease (Whaley et al., 2013). This is referred to as the amelioration hypothesis.
Peterson and Bailey (1992) found some support for the amelioration hypothesis. They
investigated the relationship between various forms of inequality (general, racial, and
gender) and rape rates for United States metropolitan areas in 1980. The rape rates were
obtained from the UCR in 1980. General inequality was measured using the GINI index
which represents income inequality in the total population. Racial inequality was
measured by the difference in median income for Black and white families. Gender
inequality was measured in terms of education using the ratio of men to women who
completed 4 years of high school, who completed 4 years of college, and who completed
5 or more years of college. It was also measured in terms of income using a ratio of men
to women’s median income for all people, people who completed 4 years of high school,
and people who completed 4 years or more of college. Gender inequality was also
measured by the percent of people employed in managerial and professional positions
who were women and by a ratio of men to women who live in poverty. Multiple models
were estimated with most containing one form of gender inequality and one containing a
representative of each type. The GINI index was found to be positively and significantly
related to rape rapes in each model. This finding suggests that places with greater general
income inequality exhibited higher rape rates. Racial inequality in income was significant
and positive in the models that included gender inequality by ratio of high school
graduates only and gender inequality in percentage of women in professional occupations
only. When all forms of gender inequality were included (income, high school graduation,
and profession occupations), gender inequality in median income was significantly and
positively related to rape rates. This finding suggests that where gender inequality in
income is greater, rape rates will also be higher and this is consistent with the
amelioration hypothesis.
Another hypothesis, the backlash hypothesis, suggests an opposite relationship.
Specifically, the backlash hypothesis suggests that as gender inequality decreases and the
genders become more equal, violence against women will increase (Whaley et al., 2013).
It is important to note that the term “backlash” is sometimes used to imply a specific
mechanism causing the increase of violence. For instance, sometimes backlash is defined
as a process by which men (or other dominant group) react violently to challenges to their
dominant status. In this study, as with many past studies, the mechanism of the
relationship between gender inequality and violence against women is not being
evaluated. Rather this study examines the relationship between gender inequality and
violence against women which therefore only allows for some speculation as to the
mechanism of the relationship. Backlash, herein, is used to represent a negative
relationship between changes in gender inequality and changes in violence against
women.
Within the same article described above, Peterson and Bailey (1992) also found
support for a backlash hypothesis when further investigating the relationship between
gender inequality in occupations and rape rates. Specifically, they separated professional
occupations into 1) executives, administrators, and managers 2) public officials and
administrators 3) engineers and natural scientists 4) health diagnostic occupations 5)
elementary and secondary teachers. They found that where the percent of executives,
administrators, and managers that are women were higher rape rates were also higher
suggesting backlash. However, they also found that where the percent of health diagnostic
occupations that are women were higher rape rates were lower. This was true of
elementary and secondary teachers as well. Since being an elementary or secondary
teacher is a women-typed occupation, the finding of a negative relationship could be
interpreted as backlash as well. The authors suggested that “the more that women remain
in traditional ‘female’ occupations, the greater their ‘reward’ – freedom from rape” (p.
172). However, the significant finding for health diagnostic occupations appears to
support amelioration, since health diagnostic occupations are traditionally men-typed
occupations.
The Peterson and Bailey (1992) study is illustrative for three primary reasons.
First, the analysis assesses the relationship between gender inequality and rape in
multiple ways. They use several forms of inequality including: education, income, and
occupation. In addition, they estimate models with each form individually as well as a
model with all of the forms. Second, this study is illustrative because it finds support for
both the ameliorative and backlash hypotheses. Due to these mixed findings, the authors
conclude that their “analysis by no means settles the issue of rape and the structural
position of women vis-à-vis men” (p. 174). Lastly, this study thoroughly investigated
different elements of inequality. For example, as discussed above, they investigated more
specific occupation categories to gain an understanding of the relationship between
gender parity in occupations and rape rates. Ultimately, together these three elements
exemplify how complicated the relationship between gender inequality and rape may be.
Specifically, it is possible that common indicators of gender disparity do not capture the
most important aspects of gender inequality. This could be one explanation for why some
measures of gender inequality were significantly related to rape and others were not. For
instance, inequality in educational attainment was not related to rape rates in any of the
Peterson and Bailey (1992) models. Through this study, it was also demonstrated that
some of the measures of gender inequality may be confounding other relationships. In
particular, Peterson and Bailey (1992) found that although in the main model there was
no significant relationship between gender inequality in occupation type and rape rates,
when looking at specific types of occupations the findings were mixed. Indeed, they
found support for backlash and amelioration when looking at specific types of
occupations. By thoroughly investigating the relationship between gender inequality and
rape, Peterson and Bailey (1992) illustrated how complicated the relationship may be and
demonstrated important ways to attempt to uncover more information.
Some scholars have noted that much of the research on gender inequality and
violence against women, including Peterson and Bailey (1992), have acted under the
axiom that amelioration and backlash are contradictory hypotheses (Bailey, 1999;
Whaley, 2001; Whaley et al., 2013). However, these hypotheses may be corollaries of one
another contingent either on time or the level of gender inequality. For instance, Bailey
(1999) and Whaley (2001) suggested and found some evidence that backlash occurs for a
period of time leading to amelioration. According to Bailey (1999), prior to his analysis
no studies had investigated the role of changes in gender inequality and changes in the
rape rate. Indeed, I was unable to locate any studies prior to Bailey (1999) that
investigated the relationship between changes in gender inequality and changes in any
form of violence against women. Bailey (1999) investigated the relationship between
gender inequality (using male and female differences in educational attainment,
occupation type, and income as indicators) and rape rates for central cities. He found
support for the backlash hypothesis between occupation type and rape rates for 1980 and
1990 (cross-sectional models), but found an ameliorative relationship between changes in
income inequality and changes in rape rates from 1980 to 1990 (change model). In the
end, Bailey suggested that although his analysis was an “improvement over previous
investigations this analysis does not fully resolve important feminist arguments regarding
the socioeconomic status of women and rape” (1999, p. 59).
Whaley (2001) presented itself as refining the temporal theoretical relationship
between change in gender inequality and violence against women. She hypothesized that
initially during times of higher gender inequality, women experience lower levels of rape
as a “reward, “but that in the long term, gender inequality prompts cultural changes that
build support for rape, prompting an increase. The reverse was hypothesized as well. To
use the hypotheses already introduced, Whaley (2001) hypothesized that backlash occurs
first and that after a sustained period of change amelioration can occur. Using 5-, 10-, and
20-year lagged indicators, Whaley (2001) was also unable to provide strong support for
her hypothesis; however, most of the significant relationships were in the proposed
direction. For example, cross-sectional analyses suggested that higher levels of inequality
in holding executive occupations were associated with lower levels of rape supporting the
backlash hypothesis. The model using a 5-year lag from 1970 to 1975 suggested backlash
also. Yet in accord with the hypothesis, the model using a 10-year lag between 1980 and
1990 found support for the ameliorative hypotheses. Specifically, decreases in gender
inequality in holding executive occupations were associated with decreases in rape.
Therefore, findings related to holding executive occupations appeared to support backlash
cross-sectionally and amelioration over a 10-year period. However, gender inequality in
executive occupations was not significantly related to rape when looking at change from
1980 to 1985, 1970 to 1980, and 1970 to 1990. Therefore, within this analysis, the results
reflect a complicated relationship.
Additionally, Whaley et al. (2013) suggested and found some evidence that the
nature of the relationship between gender inequality and violence against women may be
nonlinear and related to the level of gender inequality. Specifically, they investigated the
relationship between gender inequality (in terms of educational attainment, median
income, and employment status) and various rates of homicides for 1990 and 2000 from
the SHR (male-female, male-male, female-male, and female-female). Using negative
binomial regressions, Whaley et al. (2013) found that gender equality (represented by a
factor score of the three variables) was positively related to male offender – female victim
homicide rates and that the squared form of gender equality was negatively related in
1990. These significant findings suggest that the relationship between gender equality and
male-female homicide is nonlinear. Gender equality and its squared term were not
significantly related to any other homicide rates except female-offender and
femalevictim. For the female-female homicide rate, gender equality was significantly and
positively related in the model that did not include the robbery rate, which was included
as a “robust control for other factors related to criminal violence” (Whaley et al., 2013, p.
743). In 2000, gender equality and its squared form were significantly related to
malefemale homicide when the robbery rate was not included but only the squared form
was significantly related (negatively) when the robbery rate was included. The same was
true for male-male homicide rates. Just gender equality is significantly and positively
related to the female-female rate when the robbery rate was not included. In all other
instances, gender equality and its squared form were not significantly related to homicide
rates. Whaley et al. (2013) interpreted the effect of adding the robbery rate as suggesting
that unmeasured factors associated with violent crime may be mediating the relationship
or that there may be an issue with multicollinearity. As a whole, they found considerable
support for their hypothesis that backlash processes are likely to function at “low to
intermediate levels” of gender equality and that amelioration occurs when equality is high
for male-female homicide rates (p. 747).
Xie, Heimer, and Lauritsen (2012a) used longitudinal methods to assess whether
there was a relationship between non-fatal intimate partner violence and absolute/relative
indicators of women’s status. They created five five-year pooled periods using MSAlevel
data (which includes the 40 largest MSAs) from the NCVS for 1980 to 2004. To
thoroughly assess their hypotheses, they also disaggregated the non-fatal violence against
women data by victim-offender relationship using the following listed categories:
stranger, intimate partner, and other known. The four hypotheses included amelioration
(which was posited to function more for intimate partner violence than non-intimate
violence), backlash (which was also posited to function more for intimate partner
violence than non-intimate violence), routine activities (which suggests that increases in
absolute status also increases exposure and could lead to an increase in stranger and
known victimization), and protective (which suggests that an increase in absolute status
may increase the accessibility of protective mechanisms and decrease victimization).
Absolute women’s status was operationalized by the percent of women who were
employed, percent of voting age women who voted in the November presidential and
congressional elections, and an index of income-educational attainment, which included
median income and percent who completed four years of college or more. Relative status
was measured using the rate of women’s labor force participation minus that of men, the
level of income-educational attainment for women minus that of men, and the rate of
voter turnout for women minus that of men. Their results found support for each of the
hypotheses except for amelioration. They found a negative relationship between women’s
absolute economic and political status and intimate partner violence which supports the
protective hypothesis. They also found the negative relationship between inequality and
intimate partner victimization which supports the backlash hypothesis. The routine
activities hypothesis was supported through a positive relationship between absolute
indicators of labor force participation and victimization by strangers and non-intimate
persons known to the victim. This longitudinal analysis does not suggest that backlash,
amelioration, protective, nor routine activities hypotheses are corollary by explaining the
relationship between inequality, absolute status, and violence against women at different
periods. However, they found evidence that the nature of the relationship is contingent on
the victim-offender relationship.
The hypotheses and findings suggesting that backlash and amelioration are
corollary are particularly important considering that much of the empirical quantitative
literature on gender inequality has been cross-sectional and assumed a linear relationship.
By design much of the extant literature is unable to address these possibilities (Whaley et
al., 2013). The evidence supporting corollary processes indicate that future analyses
should continue to consider change over time and the structural form of the relationship
as the proposed dissertation will do. In addition, Xie et al. (2012a) presented evidence
that the victim-offender relationship is important to also consider in future analyses.
In addition, while not an inequality hypothesis, several scholars have investigated the role
of absolute status on trends of violence against women. Changes in absolute status are
increases or decreases in the status of a group with no consideration or comparison to
other groups. It is possible that there is a relationship between absolute status and
violence against women and that any findings suggesting a relationship
When race-disaggregated trends of violence against women are investigated,
major gaps in the crime decline, race, and gender literature are exposed. For example,
Lauritsen (2012) illustrated that although the rape and sexual assault trends for Latinas,
Black women, and white women have been similar for the last 40 years, the gaps between
them have varied occasionally. The variation in the gap could indicate different processes
at work. These types of findings are important in that they denote the importance of
attending to gender, race, and ethnicity but are often overlooked in later studies.
Ultimately, the crime decline literature has at times denoted the varying magnitude of the
decline by race and gender subgroups, but has not produced many intersectional
approaches to explaining the disparate decline (Heimer & Lauritsen, 2008; Lauritsen,
2012; Lauritsen & Heimer, 2010; Levitt, 2004; Parker, 2008). As a consequence, the
disparate declines and trends by race, ethnicity, and gender are not well understood. The
race and inequality literature tends to lack a gender-lens, a framework denoting the role
that gender plays in how things are enacted and perceived, while at the same time the
gender and inequality literature tends to lack a race-lens, a framework denoting the role
of race (McCall, 2005). This study argues that it is important to consider violence against
women from a gender x race-lens or an intersectional-lens.
HYPOTHESES AND EMPIRICAL SUPPORT
Before discussing the ways to apply an intersectionality perspective to the gender
inequality hypothesis, it is pertinent to review the various hypotheses that exist within the
literature. As noted in the introduction, the nature of the relationship between gender
inequality and violence against women has not been clearly defined within the empirical
literature (Whaley et al., 2013). In fact, evidence exists for multiple hypotheses. The
original hypothesis and the one many organizations work under suggests that as gender
inequality decreases and the genders become more equal violence against women will
decrease (Whaley et al., 2013). This is referred to as the amelioration hypothesis.
Peterson and Bailey (1992) found some support for the amelioration hypothesis. They
investigated the relationship between various forms of inequality (general, racial, and
gender) and rape rates for United States metropolitan areas in 1980. The rape rates were
obtained from the UCR in 1980. General inequality was measured using the GINI index
which represents income inequality in the total population. Racial inequality was
measured by the difference in median income for Black and white families. Gender
inequality was measured in terms of education using the ratio of men to women who
completed 4 years of high school, who completed 4 years of college, and who completed
5 or more years of college. It was also measured in terms of income using a ratio of men
to women’s median income for all people, people who completed 4 years of high school,
and people who completed 4 years or more of college. Gender inequality was also
measured by the percent of people employed in managerial and professional positions
who were women and by a ratio of men to women who live in poverty. Multiple models
were estimated with most containing one form of gender inequality and one containing a
representative of each type. The GINI index was found to be positively and significantly
related to rape rapes in each model. This finding suggests that places with greater general
income inequality exhibited higher rape rates. Racial inequality in income was significant
and positive in the models that included gender inequality by ratio of high school
graduates only and gender inequality in percentage of women in professional occupations
only. When all forms of gender inequality were included (income, high school graduation,
and profession occupations), gender inequality in median income was significantly and
positively related to rape rates. This finding suggests that where gender inequality in
income is greater, rape rates will also be higher and this is consistent with the
amelioration hypothesis.
Another hypothesis, the backlash hypothesis, suggests an opposite relationship.
Specifically, the backlash hypothesis suggests that as gender inequality decreases and the
genders become more equal, violence against women will increase (Whaley et al., 2013).
It is important to note that the term “backlash” is sometimes used to imply a specific
mechanism causing the increase of violence. For instance, sometimes backlash is defined
as a process by which men (or other dominant group) react violently to challenges to their
dominant status. In this study, as with many past studies, the mechanism of the
relationship between gender inequality and violence against women is not being
evaluated. Rather this study examines the relationship between gender inequality and
violence against women which therefore only allows for some speculation as to the
mechanism of the relationship. Backlash, herein, is used to represent a negative
relationship between changes in gender inequality and changes in violence against
women.
Within the same article described above, Peterson and Bailey (1992) also found
support for a backlash hypothesis when further investigating the relationship between
gender inequality in occupations and rape rates. Specifically, they separated professional
occupations into 1) executives, administrators, and managers 2) public officials and
administrators 3) engineers and natural scientists 4) health diagnostic occupations 5)
elementary and secondary teachers. They found that where the percent of executives,
administrators, and managers that are women were higher rape rates were also higher
suggesting backlash. However, they also found that where the percent of health diagnostic
occupations that are women were higher rape rates were lower. This was true of
elementary and secondary teachers as well. Since being an elementary or secondary
teacher is a women-typed occupation, the finding of a negative relationship could be
interpreted as backlash as well. The authors suggested that “the more that women remain
in traditional ‘female’ occupations, the greater their ‘reward’ – freedom from rape” (p.
172). However, the significant finding for health diagnostic occupations appears to
support amelioration, since health diagnostic occupations are traditionally men-typed
occupations.
The Peterson and Bailey (1992) study is illustrative for three primary reasons.
First, the analysis assesses the relationship between gender inequality and rape in
multiple ways. They use several forms of inequality including: education, income, and
occupation. In addition, they estimate models with each form individually as well as a
model with all of the forms. Second, this study is illustrative because it finds support for
both the ameliorative and backlash hypotheses. Due to these mixed findings, the authors
conclude that their “analysis by no means settles the issue of rape and the structural
position of women vis-à-vis men” (p. 174). Lastly, this study thoroughly investigated
different elements of inequality. For example, as discussed above, they investigated more
specific occupation categories to gain an understanding of the relationship between
gender parity in occupations and rape rates. Ultimately, together these three elements
exemplify how complicated the relationship between gender inequality and rape may be.
Specifically, it is possible that common indicators of gender disparity do not capture the
most important aspects of gender inequality. This could be one explanation for why some
measures of gender inequality were significantly related to rape and others were not. For
instance, inequality in educational attainment was not related to rape rates in any of the
Peterson and Bailey (1992) models. Through this study, it was also demonstrated that
some of the measures of gender inequality may be confounding other relationships. In
particular, Peterson and Bailey (1992) found that although in the main model there was
no significant relationship between gender inequality in occupation type and rape rates,
when looking at specific types of occupations the findings were mixed. Indeed, they
found support for backlash and amelioration when looking at specific types of
occupations. By thoroughly investigating the relationship between gender inequality and
rape, Peterson and Bailey (1992) illustrated how complicated the relationship may be and
demonstrated important ways to attempt to uncover more information.
Some scholars have noted that much of the research on gender inequality and
violence against women, including Peterson and Bailey (1992), have acted under the
axiom that amelioration and backlash are contradictory hypotheses (Bailey, 1999;
Whaley, 2001; Whaley et al., 2013). However, these hypotheses may be corollaries of one
another contingent either on time or the level of gender inequality. For instance, Bailey
(1999) and Whaley (2001) suggested and found some evidence that backlash occurs for a
period of time leading to amelioration. According to Bailey (1999), prior to his analysis
no studies had investigated the role of changes in gender inequality and changes in the
rape rate. Indeed, I was unable to locate any studies prior to Bailey (1999) that
investigated the relationship between changes in gender inequality and changes in any
form of violence against women. Bailey (1999) investigated the relationship between
gender inequality (using male and female differences in educational attainment,
occupation type, and income as indicators) and rape rates for central cities. He found
support for the backlash hypothesis between occupation type and rape rates for 1980 and
1990 (cross-sectional models), but found an ameliorative relationship between changes in
income inequality and changes in rape rates from 1980 to 1990 (change model). In the
end, Bailey suggested that although his analysis was an “improvement over previous
investigations this analysis does not fully resolve important feminist arguments regarding
the socioeconomic status of women and rape” (1999, p. 59).
Whaley (2001) presented itself as refining the temporal theoretical relationship
between change in gender inequality and violence against women. She hypothesized that
initially during times of higher gender inequality, women experience lower levels of rape
as a “reward, “but that in the long term, gender inequality prompts cultural changes that
build support for rape, prompting an increase. The reverse was hypothesized as well. To
use the hypotheses already introduced, Whaley (2001) hypothesized that backlash occurs
first and that after a sustained period of change amelioration can occur. Using 5-, 10-, and
20-year lagged indicators, Whaley (2001) was also unable to provide strong support for
her hypothesis; however, most of the significant relationships were in the proposed
direction. For example, cross-sectional analyses suggested that higher levels of inequality
in holding executive occupations were associated with lower levels of rape supporting the
backlash hypothesis. The model using a 5-year lag from 1970 to 1975 suggested backlash
also. Yet in accord with the hypothesis, the model using a 10-year lag between 1980 and
1990 found support for the ameliorative hypotheses. Specifically, decreases in gender
inequality in holding executive occupations were associated with decreases in rape.
Therefore, findings related to holding executive occupations appeared to support backlash
cross-sectionally and amelioration over a 10-year period. However, gender inequality in
executive occupations was not significantly related to rape when looking at change from
1980 to 1985, 1970 to 1980, and 1970 to 1990. Therefore, within this analysis, the results
reflect a complicated relationship.
Additionally, Whaley et al. (2013) suggested and found some evidence that the
nature of the relationship between gender inequality and violence against women may be
nonlinear and related to the level of gender inequality. Specifically, they investigated the
relationship between gender inequality (in terms of educational attainment, median
income, and employment status) and various rates of homicides for 1990 and 2000 from
the SHR (male-female, male-male, female-male, and female-female). Using negative
binomial regressions, Whaley et al. (2013) found that gender equality (represented by a
factor score of the three variables) was positively related to male offender – female victim
homicide rates and that the squared form of gender equality was negatively related in
1990. These significant findings suggest that the relationship between gender equality and
male-female homicide is nonlinear. Gender equality and its squared term were not
significantly related to any other homicide rates except female-offender and
femalevictim. For the female-female homicide rate, gender equality was significantly and
positively related in the model that did not include the robbery rate, which was included
as a “robust control for other factors related to criminal violence” (Whaley et al., 2013, p.
743). In 2000, gender equality and its squared form were significantly related to
malefemale homicide when the robbery rate was not included but only the squared form
was significantly related (negatively) when the robbery rate was included. The same was
true for male-male homicide rates. Just gender equality is significantly and positively
related to the female-female rate when the robbery rate was not included. In all other
instances, gender equality and its squared form were not significantly related to homicide
rates. Whaley et al. (2013) interpreted the effect of adding the robbery rate as suggesting
that unmeasured factors associated with violent crime may be mediating the relationship
or that there may be an issue with multicollinearity. As a whole, they found considerable
support for their hypothesis that backlash processes are likely to function at “low to
intermediate levels” of gender equality and that amelioration occurs when equality is high
for male-female homicide rates (p. 747).
Xie, Heimer, and Lauritsen (2012a) used longitudinal methods to assess whether
there was a relationship between non-fatal intimate partner violence and absolute/relative
indicators of women’s status. They created five five-year pooled periods using MSAlevel
data (which includes the 40 largest MSAs) from the NCVS for 1980 to 2004. To
thoroughly assess their hypotheses, they also disaggregated the non-fatal violence against
women data by victim-offender relationship using the following listed categories:
stranger, intimate partner, and other known. The four hypotheses included amelioration
(which was posited to function more for intimate partner violence than non-intimate
violence), backlash (which was also posited to function more for intimate partner
violence than non-intimate violence), routine activities (which suggests that increases in
absolute status also increases exposure and could lead to an increase in stranger and
known victimization), and protective (which suggests that an increase in absolute status
may increase the accessibility of protective mechanisms and decrease victimization).
Absolute women’s status was operationalized by the percent of women who were
employed, percent of voting age women who voted in the November presidential and
congressional elections, and an index of income-educational attainment, which included
median income and percent who completed four years of college or more. Relative status
was measured using the rate of women’s labor force participation minus that of men, the
level of income-educational attainment for women minus that of men, and the rate of
voter turnout for women minus that of men. Their results found support for each of the
hypotheses except for amelioration. They found a negative relationship between women’s
absolute economic and political status and intimate partner violence which supports the
protective hypothesis. They also found the negative relationship between inequality and
intimate partner victimization which supports the backlash hypothesis. The routine
activities hypothesis was supported through a positive relationship between absolute
indicators of labor force participation and victimization by strangers and non-intimate
persons known to the victim. This longitudinal analysis does not suggest that backlash,
amelioration, protective, nor routine activities hypotheses are corollary by explaining the
relationship between inequality, absolute status, and violence against women at different
periods. However, they found evidence that the nature of the relationship is contingent on
the victim-offender relationship.
The hypotheses and findings suggesting that backlash and amelioration are
corollary are particularly important considering that much of the empirical quantitative
literature on gender inequality has been cross-sectional and assumed a linear relationship.
By design much of the extant literature is unable to address these possibilities (Whaley et
al., 2013). The evidence supporting corollary processes indicate that future analyses
should continue to consider change over time and the structural form of the relationship
as the proposed dissertation will do. In addition, Xie et al. (2012a) presented evidence
that the victim-offender relationship is important to also consider in future analyses.
In addition, while not an inequality hypothesis, several scholars have investigated the role
of absolute status on trends of violence against women. Changes in absolute status are
increases or decreases in the status of a group with no consideration or comparison to
other groups. It is possible that there is a relationship between absolute status and
violence against women and that any findings suggesting a relationship
When race-disaggregated trends of violence against women are investigated,
major gaps in the crime decline, race, and gender literature are exposed. For example,
Lauritsen (2012) illustrated that although the rape and sexual assault trends for Latinas,
Black women, and white women have been similar for the last 40 years, the gaps between
them have varied occasionally. The variation in the gap could indicate different processes
at work. These types of findings are important in that they denote the importance of
attending to gender, race, and ethnicity but are often overlooked in later studies.
Ultimately, the crime decline literature has at times denoted the varying magnitude of the
decline by race and gender subgroups, but has not produced many intersectional
approaches to explaining the disparate decline (Heimer & Lauritsen, 2008; Lauritsen,
2012; Lauritsen & Heimer, 2010; Levitt, 2004; Parker, 2008). As a consequence, the
disparate declines and trends by race, ethnicity, and gender are not well understood. The
race and inequality literature tends to lack a gender-lens, a framework denoting the role
that gender plays in how things are enacted and perceived, while at the same time the
gender and inequality literature tends to lack a race-lens, a framework denoting the role
of race (McCall, 2005). This study argues that it is important to consider violence against
women from a gender x race-lens or an intersectional-lens.
HYPOTHESES AND EMPIRICAL SUPPORT
Before discussing the ways to apply an intersectionality perspective to the gender
inequality hypothesis, it is pertinent to review the various hypotheses that exist within the
literature. As noted in the introduction, the nature of the relationship between gender
inequality and violence against women has not been clearly defined within the empirical
literature (Whaley et al., 2013). In fact, evidence exists for multiple hypotheses. The
original hypothesis and the one many organizations work under suggests that as gender
inequality decreases and the genders become more equal violence against women will
decrease (Whaley et al., 2013). This is referred to as the amelioration hypothesis.
Peterson and Bailey (1992) found some support for the amelioration hypothesis. They
investigated the relationship between various forms of inequality (general, racial, and
gender) and rape rates for United States metropolitan areas in 1980. The rape rates were
obtained from the UCR in 1980. General inequality was measured using the GINI index
which represents income inequality in the total population. Racial inequality was
measured by the difference in median income for Black and white families. Gender
inequality was measured in terms of education using the ratio of men to women who
completed 4 years of high school, who completed 4 years of college, and who completed
5 or more years of college. It was also measured in terms of income using a ratio of men
to women’s median income for all people, people who completed 4 years of high school,
and people who completed 4 years or more of college. Gender inequality was also
measured by the percent of people employed in managerial and professional positions
who were women and by a ratio of men to women who live in poverty. Multiple models
were estimated with most containing one form of gender inequality and one containing a
representative of each type. The GINI index was found to be positively and significantly
related to rape rapes in each model. This finding suggests that places with greater general
income inequality exhibited higher rape rates. Racial inequality in income was significant
and positive in the models that included gender inequality by ratio of high school
graduates only and gender inequality in percentage of women in professional occupations
only. When all forms of gender inequality were included (income, high school graduation,
and profession occupations), gender inequality in median income was significantly and
positively related to rape rates. This finding suggests that where gender inequality in
income is greater, rape rates will also be higher and this is consistent with the
amelioration hypothesis.
Another hypothesis, the backlash hypothesis, suggests an opposite relationship.
Specifically, the backlash hypothesis suggests that as gender inequality decreases and the
genders become more equal, violence against women will increase (Whaley et al., 2013).
It is important to note that the term “backlash” is sometimes used to imply a specific
mechanism causing the increase of violence. For instance, sometimes backlash is defined
as a process by which men (or other dominant group) react violently to challenges to their
dominant status. In this study, as with many past studies, the mechanism of the
relationship between gender inequality and violence against women is not being
evaluated. Rather this study examines the relationship between gender inequality and
violence against women which therefore only allows for some speculation as to the
mechanism of the relationship. Backlash, herein, is used to represent a negative
relationship between changes in gender inequality and changes in violence against
women.
Within the same article described above, Peterson and Bailey (1992) also found
support for a backlash hypothesis when further investigating the relationship between
gender inequality in occupations and rape rates. Specifically, they separated professional
occupations into 1) executives, administrators, and managers 2) public officials and
administrators 3) engineers and natural scientists 4) health diagnostic occupations 5)
elementary and secondary teachers. They found that where the percent of executives,
administrators, and managers that are women were higher rape rates were also higher
suggesting backlash. However, they also found that where the percent of health diagnostic
occupations that are women were higher rape rates were lower. This was true of
elementary and secondary teachers as well. Since being an elementary or secondary
teacher is a women-typed occupation, the finding of a negative relationship could be
interpreted as backlash as well. The authors suggested that “the more that women remain
in traditional ‘female’ occupations, the greater their ‘reward’ – freedom from rape” (p.
172). However, the significant finding for health diagnostic occupations appears to
support amelioration, since health diagnostic occupations are traditionally men-typed
occupations.
The Peterson and Bailey (1992) study is illustrative for three primary reasons.
First, the analysis assesses the relationship between gender inequality and rape in
multiple ways. They use several forms of inequality including: education, income, and
occupation. In addition, they estimate models with each form individually as well as a
model with all of the forms. Second, this study is illustrative because it finds support for
both the ameliorative and backlash hypotheses. Due to these mixed findings, the authors
conclude that their “analysis by no means settles the issue of rape and the structural
position of women vis-à-vis men” (p. 174). Lastly, this study thoroughly investigated
different elements of inequality. For example, as discussed above, they investigated more
specific occupation categories to gain an understanding of the relationship between
gender parity in occupations and rape rates. Ultimately, together these three elements
exemplify how complicated the relationship between gender inequality and rape may be.
Specifically, it is possible that common indicators of gender disparity do not capture the
most important aspects of gender inequality. This could be one explanation for why some
measures of gender inequality were significantly related to rape and others were not. For
instance, inequality in educational attainment was not related to rape rates in any of the
Peterson and Bailey (1992) models. Through this study, it was also demonstrated that
some of the measures of gender inequality may be confounding other relationships. In
particular, Peterson and Bailey (1992) found that although in the main model there was
no significant relationship between gender inequality in occupation type and rape rates,
when looking at specific types of occupations the findings were mixed. Indeed, they
found support for backlash and amelioration when looking at specific types of
occupations. By thoroughly investigating the relationship between gender inequality and
rape, Peterson and Bailey (1992) illustrated how complicated the relationship may be and
demonstrated important ways to attempt to uncover more information.
Some scholars have noted that much of the research on gender inequality and
violence against women, including Peterson and Bailey (1992), have acted under the
axiom that amelioration and backlash are contradictory hypotheses (Bailey, 1999;
Whaley, 2001; Whaley et al., 2013). However, these hypotheses may be corollaries of one
another contingent either on time or the level of gender inequality. For instance, Bailey
(1999) and Whaley (2001) suggested and found some evidence that backlash occurs for a
period of time leading to amelioration. According to Bailey (1999), prior to his analysis
no studies had investigated the role of changes in gender inequality and changes in the
rape rate. Indeed, I was unable to locate any studies prior to Bailey (1999) that
investigated the relationship between changes in gender inequality and changes in any
form of violence against women. Bailey (1999) investigated the relationship between
gender inequality (using male and female differences in educational attainment,
occupation type, and income as indicators) and rape rates for central cities. He found
support for the backlash hypothesis between occupation type and rape rates for 1980 and
1990 (cross-sectional models), but found an ameliorative relationship between changes in
income inequality and changes in rape rates from 1980 to 1990 (change model). In the
end, Bailey suggested that although his analysis was an “improvement over previous
investigations this analysis does not fully resolve important feminist arguments regarding
the socioeconomic status of women and rape” (1999, p. 59).
Whaley (2001) presented itself as refining the temporal theoretical relationship
between change in gender inequality and violence against women. She hypothesized that
initially during times of higher gender inequality, women experience lower levels of rape
as a “reward, “but that in the long term, gender inequality prompts cultural changes that
build support for rape, prompting an increase. The reverse was hypothesized as well. To
use the hypotheses already introduced, Whaley (2001) hypothesized that backlash occurs
first and that after a sustained period of change amelioration can occur. Using 5-, 10-, and
20-year lagged indicators, Whaley (2001) was also unable to provide strong support for
her hypothesis; however, most of the significant relationships were in the proposed
direction. For example, cross-sectional analyses suggested that higher levels of inequality
in holding executive occupations were associated with lower levels of rape supporting the
backlash hypothesis. The model using a 5-year lag from 1970 to 1975 suggested backlash
also. Yet in accord with the hypothesis, the model using a 10-year lag between 1980 and
1990 found support for the ameliorative hypotheses. Specifically, decreases in gender
inequality in holding executive occupations were associated with decreases in rape.
Therefore, findings related to holding executive occupations appeared to support backlash
cross-sectionally and amelioration over a 10-year period. However, gender inequality in
executive occupations was not significantly related to rape when looking at change from
1980 to 1985, 1970 to 1980, and 1970 to 1990. Therefore, within this analysis, the results
reflect a complicated relationship.
Additionally, Whaley et al. (2013) suggested and found some evidence that the
nature of the relationship between gender inequality and violence against women may be
nonlinear and related to the level of gender inequality. Specifically, they investigated the
relationship between gender inequality (in terms of educational attainment, median
income, and employment status) and various rates of homicides for 1990 and 2000 from
the SHR (male-female, male-male, female-male, and female-female). Using negative
binomial regressions, Whaley et al. (2013) found that gender equality (represented by a
factor score of the three variables) was positively related to male offender – female victim
homicide rates and that the squared form of gender equality was negatively related in
1990. These significant findings suggest that the relationship between gender equality and
male-female homicide is nonlinear. Gender equality and its squared term were not
significantly related to any other homicide rates except female-offender and
femalevictim. For the female-female homicide rate, gender equality was significantly and
positively related in the model that did not include the robbery rate, which was included
as a “robust control for other factors related to criminal violence” (Whaley et al., 2013, p.
743). In 2000, gender equality and its squared form were significantly related to
malefemale homicide when the robbery rate was not included but only the squared form
was significantly related (negatively) when the robbery rate was included. The same was
true for male-male homicide rates. Just gender equality is significantly and positively
related to the female-female rate when the robbery rate was not included. In all other
instances, gender equality and its squared form were not significantly related to homicide
rates. Whaley et al. (2013) interpreted the effect of adding the robbery rate as suggesting
that unmeasured factors associated with violent crime may be mediating the relationship
or that there may be an issue with multicollinearity. As a whole, they found considerable
support for their hypothesis that backlash processes are likely to function at “low to
intermediate levels” of gender equality and that amelioration occurs when equality is high
for male-female homicide rates (p. 747).
Xie, Heimer, and Lauritsen (2012a) used longitudinal methods to assess whether
there was a relationship between non-fatal intimate partner violence and absolute/relative
indicators of women’s status. They created five five-year pooled periods using MSAlevel
data (which includes the 40 largest MSAs) from the NCVS for 1980 to 2004. To
thoroughly assess their hypotheses, they also disaggregated the non-fatal violence against
women data by victim-offender relationship using the following listed categories:
stranger, intimate partner, and other known. The four hypotheses included amelioration
(which was posited to function more for intimate partner violence than non-intimate
violence), backlash (which was also posited to function more for intimate partner
violence than non-intimate violence), routine activities (which suggests that increases in
absolute status also increases exposure and could lead to an increase in stranger and
known victimization), and protective (which suggests that an increase in absolute status
may increase the accessibility of protective mechanisms and decrease victimization).
Absolute women’s status was operationalized by the percent of women who were
employed, percent of voting age women who voted in the November presidential and
congressional elections, and an index of income-educational attainment, which included
median income and percent who completed four years of college or more. Relative status
was measured using the rate of women’s labor force participation minus that of men, the
level of income-educational attainment for women minus that of men, and the rate of
voter turnout for women minus that of men. Their results found support for each of the
hypotheses except for amelioration. They found a negative relationship between women’s
absolute economic and political status and intimate partner violence which supports the
protective hypothesis. They also found the negative relationship between inequality and
intimate partner victimization which supports the backlash hypothesis. The routine
activities hypothesis was supported through a positive relationship between absolute
indicators of labor force participation and victimization by strangers and non-intimate
persons known to the victim. This longitudinal analysis does not suggest that backlash,
amelioration, protective, nor routine activities hypotheses are corollary by explaining the
relationship between inequality, absolute status, and violence against women at different
periods. However, they found evidence that the nature of the relationship is contingent on
the victim-offender relationship.
The hypotheses and findings suggesting that backlash and amelioration are
corollary are particularly important considering that much of the empirical quantitative
literature on gender inequality has been cross-sectional and assumed a linear relationship.
By design much of the extant literature is unable to address these possibilities (Whaley et
al., 2013). The evidence supporting corollary processes indicate that future analyses
should continue to consider change over time and the structural form of the relationship
as the proposed dissertation will do. In addition, Xie et al. (2012a) presented evidence
that the victim-offender relationship is important to also consider in future analyses.
In addition, while not an inequality hypothesis, several scholars have investigated the role
of absolute status on trends of violence against women. Changes in absolute status are
increases or decreases in the status of a group with no consideration or comparison to
other groups. It is possible that there is a relationship between absolute status and
violence against women and that any findings suggesting a relationship
When race-disaggregated trends of violence against women are investigated,
major gaps in the crime decline, race, and gender literature are exposed. For example,
Lauritsen (2012) illustrated that although the rape and sexual assault trends for Latinas,
Black women, and white women have been similar for the last 40 years, the gaps between
them have varied occasionally. The variation in the gap could indicate different processes
at work. These types of findings are important in that they denote the importance of
attending to gender, race, and ethnicity but are often overlooked in later studies.
Ultimately, the crime decline literature has at times denoted the varying magnitude of the
decline by race and gender subgroups, but has not produced many intersectional
approaches to explaining the disparate decline (Heimer & Lauritsen, 2008; Lauritsen,
2012; Lauritsen & Heimer, 2010; Levitt, 2004; Parker, 2008). As a consequence, the
disparate declines and trends by race, ethnicity, and gender are not well understood. The
race and inequality literature tends to lack a gender-lens, a framework denoting the role
that gender plays in how things are enacted and perceived, while at the same time the
gender and inequality literature tends to lack a race-lens, a framework denoting the role
of race (McCall, 2005). This study argues that it is important to consider violence against
women from a gender x race-lens or an intersectional-lens.
HYPOTHESES AND EMPIRICAL SUPPORT
Before discussing the ways to apply an intersectionality perspective to the gender
inequality hypothesis, it is pertinent to review the various hypotheses that exist within the
literature. As noted in the introduction, the nature of the relationship between gender
inequality and violence against women has not been clearly defined within the empirical
literature (Whaley et al., 2013). In fact, evidence exists for multiple hypotheses. The
original hypothesis and the one many organizations work under suggests that as gender
inequality decreases and the genders become more equal violence against women will
decrease (Whaley et al., 2013). This is referred to as the amelioration hypothesis.
Peterson and Bailey (1992) found some support for the amelioration hypothesis. They
investigated the relationship between various forms of inequality (general, racial, and
gender) and rape rates for United States metropolitan areas in 1980. The rape rates were
obtained from the UCR in 1980. General inequality was measured using the GINI index
which represents income inequality in the total population. Racial inequality was
measured by the difference in median income for Black and white families. Gender
inequality was measured in terms of education using the ratio of men to women who
completed 4 years of high school, who completed 4 years of college, and who completed
5 or more years of college. It was also measured in terms of income using a ratio of men
to women’s median income for all people, people who completed 4 years of high school,
and people who completed 4 years or more of college. Gender inequality was also
measured by the percent of people employed in managerial and professional positions
who were women and by a ratio of men to women who live in poverty. Multiple models
were estimated with most containing one form of gender inequality and one containing a
representative of each type. The GINI index was found to be positively and significantly
related to rape rapes in each model. This finding suggests that places with greater general
income inequality exhibited higher rape rates. Racial inequality in income was significant
and positive in the models that included gender inequality by ratio of high school
graduates only and gender inequality in percentage of women in professional occupations
only. When all forms of gender inequality were included (income, high school graduation,
and profession occupations), gender inequality in median income was significantly and
positively related to rape rates. This finding suggests that where gender inequality in
income is greater, rape rates will also be higher and this is consistent with the
amelioration hypothesis.
Another hypothesis, the backlash hypothesis, suggests an opposite relationship.
Specifically, the backlash hypothesis suggests that as gender inequality decreases and the
genders become more equal, violence against women will increase (Whaley et al., 2013).
It is important to note that the term “backlash” is sometimes used to imply a specific
mechanism causing the increase of violence. For instance, sometimes backlash is defined
as a process by which men (or other dominant group) react violently to challenges to their
dominant status. In this study, as with many past studies, the mechanism of the
relationship between gender inequality and violence against women is not being
evaluated. Rather this study examines the relationship between gender inequality and
violence against women which therefore only allows for some speculation as to the
mechanism of the relationship. Backlash, herein, is used to represent a negative
relationship between changes in gender inequality and changes in violence against
women.
Within the same article described above, Peterson and Bailey (1992) also found
support for a backlash hypothesis when further investigating the relationship between
gender inequality in occupations and rape rates. Specifically, they separated professional
occupations into 1) executives, administrators, and managers 2) public officials and
administrators 3) engineers and natural scientists 4) health diagnostic occupations 5)
elementary and secondary teachers. They found that where the percent of executives,
administrators, and managers that are women were higher rape rates were also higher
suggesting backlash. However, they also found that where the percent of health diagnostic
occupations that are women were higher rape rates were lower. This was true of
elementary and secondary teachers as well. Since being an elementary or secondary
teacher is a women-typed occupation, the finding of a negative relationship could be
interpreted as backlash as well. The authors suggested that “the more that women remain
in traditional ‘female’ occupations, the greater their ‘reward’ – freedom from rape” (p.
172). However, the significant finding for health diagnostic occupations appears to
support amelioration, since health diagnostic occupations are traditionally men-typed
occupations.
The Peterson and Bailey (1992) study is illustrative for three primary reasons.
First, the analysis assesses the relationship between gender inequality and rape in
multiple ways. They use several forms of inequality including: education, income, and
occupation. In addition, they estimate models with each form individually as well as a
model with all of the forms. Second, this study is illustrative because it finds support for
both the ameliorative and backlash hypotheses. Due to these mixed findings, the authors
conclude that their “analysis by no means settles the issue of rape and the structural
position of women vis-à-vis men” (p. 174). Lastly, this study thoroughly investigated
different elements of inequality. For example, as discussed above, they investigated more
specific occupation categories to gain an understanding of the relationship between
gender parity in occupations and rape rates. Ultimately, together these three elements
exemplify how complicated the relationship between gender inequality and rape may be.
Specifically, it is possible that common indicators of gender disparity do not capture the
most important aspects of gender inequality. This could be one explanation for why some
measures of gender inequality were significantly related to rape and others were not. For
instance, inequality in educational attainment was not related to rape rates in any of the
Peterson and Bailey (1992) models. Through this study, it was also demonstrated that
some of the measures of gender inequality may be confounding other relationships. In
particular, Peterson and Bailey (1992) found that although in the main model there was
no significant relationship between gender inequality in occupation type and rape rates,
when looking at specific types of occupations the findings were mixed. Indeed, they
found support for backlash and amelioration when looking at specific types of
occupations. By thoroughly investigating the relationship between gender inequality and
rape, Peterson and Bailey (1992) illustrated how complicated the relationship may be and
demonstrated important ways to attempt to uncover more information.
Some scholars have noted that much of the research on gender inequality and
violence against women, including Peterson and Bailey (1992), have acted under the
axiom that amelioration and backlash are contradictory hypotheses (Bailey, 1999;
Whaley, 2001; Whaley et al., 2013). However, these hypotheses may be corollaries of one
another contingent either on time or the level of gender inequality. For instance, Bailey
(1999) and Whaley (2001) suggested and found some evidence that backlash occurs for a
period of time leading to amelioration. According to Bailey (1999), prior to his analysis
no studies had investigated the role of changes in gender inequality and changes in the
rape rate. Indeed, I was unable to locate any studies prior to Bailey (1999) that
investigated the relationship between changes in gender inequality and changes in any
form of violence against women. Bailey (1999) investigated the relationship between
gender inequality (using male and female differences in educational attainment,
occupation type, and income as indicators) and rape rates for central cities. He found
support for the backlash hypothesis between occupation type and rape rates for 1980 and
1990 (cross-sectional models), but found an ameliorative relationship between changes in
income inequality and changes in rape rates from 1980 to 1990 (change model). In the
end, Bailey suggested that although his analysis was an “improvement over previous
investigations this analysis does not fully resolve important feminist arguments regarding
the socioeconomic status of women and rape” (1999, p. 59).
Whaley (2001) presented itself as refining the temporal theoretical relationship
between change in gender inequality and violence against women. She hypothesized that
initially during times of higher gender inequality, women experience lower levels of rape
as a “reward, “but that in the long term, gender inequality prompts cultural changes that
build support for rape, prompting an increase. The reverse was hypothesized as well. To
use the hypotheses already introduced, Whaley (2001) hypothesized that backlash occurs
first and that after a sustained period of change amelioration can occur. Using 5-, 10-, and
20-year lagged indicators, Whaley (2001) was also unable to provide strong support for
her hypothesis; however, most of the significant relationships were in the proposed
direction. For example, cross-sectional analyses suggested that higher levels of inequality
in holding executive occupations were associated with lower levels of rape supporting the
backlash hypothesis. The model using a 5-year lag from 1970 to 1975 suggested backlash
also. Yet in accord with the hypothesis, the model using a 10-year lag between 1980 and
1990 found support for the ameliorative hypotheses. Specifically, decreases in gender
inequality in holding executive occupations were associated with decreases in rape.
Therefore, findings related to holding executive occupations appeared to support backlash
cross-sectionally and amelioration over a 10-year period. However, gender inequality in
executive occupations was not significantly related to rape when looking at change from
1980 to 1985, 1970 to 1980, and 1970 to 1990. Therefore, within this analysis, the results
reflect a complicated relationship.
Additionally, Whaley et al. (2013) suggested and found some evidence that the
nature of the relationship between gender inequality and violence against women may be
nonlinear and related to the level of gender inequality. Specifically, they investigated the
relationship between gender inequality (in terms of educational attainment, median
income, and employment status) and various rates of homicides for 1990 and 2000 from
the SHR (male-female, male-male, female-male, and female-female). Using negative
binomial regressions, Whaley et al. (2013) found that gender equality (represented by a
factor score of the three variables) was positively related to male offender – female victim
homicide rates and that the squared form of gender equality was negatively related in
1990. These significant findings suggest that the relationship between gender equality and
male-female homicide is nonlinear. Gender equality and its squared term were not
significantly related to any other homicide rates except female-offender and
femalevictim. For the female-female homicide rate, gender equality was significantly and
positively related in the model that did not include the robbery rate, which was included
as a “robust control for other factors related to criminal violence” (Whaley et al., 2013, p.
743). In 2000, gender equality and its squared form were significantly related to
malefemale homicide when the robbery rate was not included but only the squared form
was significantly related (negatively) when the robbery rate was included. The same was
true for male-male homicide rates. Just gender equality is significantly and positively
related to the female-female rate when the robbery rate was not included. In all other
instances, gender equality and its squared form were not significantly related to homicide
rates. Whaley et al. (2013) interpreted the effect of adding the robbery rate as suggesting
that unmeasured factors associated with violent crime may be mediating the relationship
or that there may be an issue with multicollinearity. As a whole, they found considerable
support for their hypothesis that backlash processes are likely to function at “low to
intermediate levels” of gender equality and that amelioration occurs when equality is high
for male-female homicide rates (p. 747).
Xie, Heimer, and Lauritsen (2012a) used longitudinal methods to assess whether
there was a relationship between non-fatal intimate partner violence and absolute/relative
indicators of women’s status. They created five five-year pooled periods using MSAlevel
data (which includes the 40 largest MSAs) from the NCVS for 1980 to 2004. To
thoroughly assess their hypotheses, they also disaggregated the non-fatal violence against
women data by victim-offender relationship using the following listed categories:
stranger, intimate partner, and other known. The four hypotheses included amelioration
(which was posited to function more for intimate partner violence than non-intimate
violence), backlash (which was also posited to function more for intimate partner
violence than non-intimate violence), routine activities (which suggests that increases in
absolute status also increases exposure and could lead to an increase in stranger and
known victimization), and protective (which suggests that an increase in absolute status
may increase the accessibility of protective mechanisms and decrease victimization).
Absolute women’s status was operationalized by the percent of women who were
employed, percent of voting age women who voted in the November presidential and
congressional elections, and an index of income-educational attainment, which included
median income and percent who completed four years of college or more. Relative status
was measured using the rate of women’s labor force participation minus that of men, the
level of income-educational attainment for women minus that of men, and the rate of
voter turnout for women minus that of men. Their results found support for each of the
hypotheses except for amelioration. They found a negative relationship between women’s
absolute economic and political status and intimate partner violence which supports the
protective hypothesis. They also found the negative relationship between inequality and
intimate partner victimization which supports the backlash hypothesis. The routine
activities hypothesis was supported through a positive relationship between absolute
indicators of labor force participation and victimization by strangers and non-intimate
persons known to the victim. This longitudinal analysis does not suggest that backlash,
amelioration, protective, nor routine activities hypotheses are corollary by explaining the
relationship between inequality, absolute status, and violence against women at different
periods. However, they found evidence that the nature of the relationship is contingent on
the victim-offender relationship.
The hypotheses and findings suggesting that backlash and amelioration are
corollary are particularly important considering that much of the empirical quantitative
literature on gender inequality has been cross-sectional and assumed a linear relationship.
By design much of the extant literature is unable to address these possibilities (Whaley et
al., 2013). The evidence supporting corollary processes indicate that future analyses
should continue to consider change over time and the structural form of the relationship
as the proposed dissertation will do. In addition, Xie et al. (2012a) presented evidence
that the victim-offender relationship is important to also consider in future analyses.
In addition, while not an inequality hypothesis, several scholars have investigated the role
of absolute status on trends of violence against women. Changes in absolute status are
increases or decreases in the status of a group with no consideration or comparison to
other groups. It is possible that there is a relationship between absolute status and
violence against women and that any findings suggesting a relationship
When race-disaggregated trends of violence against women are investigated,
major gaps in the crime decline, race, and gender literature are exposed. For example,
Lauritsen (2012) illustrated that although the rape and sexual assault trends for Latinas,
Black women, and white women have been similar for the last 40 years, the gaps between
them have varied occasionally. The variation in the gap could indicate different processes
at work. These types of findings are important in that they denote the importance of
attending to gender, race, and ethnicity but are often overlooked in later studies.
Ultimately, the crime decline literature has at times denoted the varying magnitude of the
decline by race and gender subgroups, but has not produced many intersectional
approaches to explaining the disparate decline (Heimer & Lauritsen, 2008; Lauritsen,
2012; Lauritsen & Heimer, 2010; Levitt, 2004; Parker, 2008). As a consequence, the
disparate declines and trends by race, ethnicity, and gender are not well understood. The
race and inequality literature tends to lack a gender-lens, a framework denoting the role
that gender plays in how things are enacted and perceived, while at the same time the
gender and inequality literature tends to lack a race-lens, a framework denoting the role
of race (McCall, 2005). This study argues that it is important to consider violence against
women from a gender x race-lens or an intersectional-lens.
HYPOTHESES AND EMPIRICAL SUPPORT
Before discussing the ways to apply an intersectionality perspective to the gender
inequality hypothesis, it is pertinent to review the various hypotheses that exist within the
literature. As noted in the introduction, the nature of the relationship between gender
inequality and violence against women has not been clearly defined within the empirical
literature (Whaley et al., 2013). In fact, evidence exists for multiple hypotheses. The
original hypothesis and the one many organizations work under suggests that as gender
inequality decreases and the genders become more equal violence against women will
decrease (Whaley et al., 2013). This is referred to as the amelioration hypothesis.
Peterson and Bailey (1992) found some support for the amelioration hypothesis. They
investigated the relationship between various forms of inequality (general, racial, and
gender) and rape rates for United States metropolitan areas in 1980. The rape rates were
obtained from the UCR in 1980. General inequality was measured using the GINI index
which represents income inequality in the total population. Racial inequality was
measured by the difference in median income for Black and white families. Gender
inequality was measured in terms of education using the ratio of men to women who
completed 4 years of high school, who completed 4 years of college, and who completed
5 or more years of college. It was also measured in terms of income using a ratio of men
to women’s median income for all people, people who completed 4 years of high school,
and people who completed 4 years or more of college. Gender inequality was also
measured by the percent of people employed in managerial and professional positions
who were women and by a ratio of men to women who live in poverty. Multiple models
were estimated with most containing one form of gender inequality and one containing a
representative of each type. The GINI index was found to be positively and significantly
related to rape rapes in each model. This finding suggests that places with greater general
income inequality exhibited higher rape rates. Racial inequality in income was significant
and positive in the models that included gender inequality by ratio of high school
graduates only and gender inequality in percentage of women in professional occupations
only. When all forms of gender inequality were included (income, high school graduation,
and profession occupations), gender inequality in median income was significantly and
positively related to rape rates. This finding suggests that where gender inequality in
income is greater, rape rates will also be higher and this is consistent with the
amelioration hypothesis.
Another hypothesis, the backlash hypothesis, suggests an opposite relationship.
Specifically, the backlash hypothesis suggests that as gender inequality decreases and the
genders become more equal, violence against women will increase (Whaley et al., 2013).
It is important to note that the term “backlash” is sometimes used to imply a specific
mechanism causing the increase of violence. For instance, sometimes backlash is defined
as a process by which men (or other dominant group) react violently to challenges to their
dominant status. In this study, as with many past studies, the mechanism of the
relationship between gender inequality and violence against women is not being
evaluated. Rather this study examines the relationship between gender inequality and
violence against women which therefore only allows for some speculation as to the
mechanism of the relationship. Backlash, herein, is used to represent a negative
relationship between changes in gender inequality and changes in violence against
women.
Within the same article described above, Peterson and Bailey (1992) also found
support for a backlash hypothesis when further investigating the relationship between
gender inequality in occupations and rape rates. Specifically, they separated professional
occupations into 1) executives, administrators, and managers 2) public officials and
administrators 3) engineers and natural scientists 4) health diagnostic occupations 5)
elementary and secondary teachers. They found that where the percent of executives,
administrators, and managers that are women were higher rape rates were also higher
suggesting backlash. However, they also found that where the percent of health diagnostic
occupations that are women were higher rape rates were lower. This was true of
elementary and secondary teachers as well. Since being an elementary or secondary
teacher is a women-typed occupation, the finding of a negative relationship could be
interpreted as backlash as well. The authors suggested that “the more that women remain
in traditional ‘female’ occupations, the greater their ‘reward’ – freedom from rape” (p.
172). However, the significant finding for health diagnostic occupations appears to
support amelioration, since health diagnostic occupations are traditionally men-typed
occupations.
The Peterson and Bailey (1992) study is illustrative for three primary reasons.
First, the analysis assesses the relationship between gender inequality and rape in
multiple ways. They use several forms of inequality including: education, income, and
occupation. In addition, they estimate models with each form individually as well as a
model with all of the forms. Second, this study is illustrative because it finds support for
both the ameliorative and backlash hypotheses. Due to these mixed findings, the authors
conclude that their “analysis by no means settles the issue of rape and the structural
position of women vis-à-vis men” (p. 174). Lastly, this study thoroughly investigated
different elements of inequality. For example, as discussed above, they investigated more
specific occupation categories to gain an understanding of the relationship between
gender parity in occupations and rape rates. Ultimately, together these three elements
exemplify how complicated the relationship between gender inequality and rape may be.
Specifically, it is possible that common indicators of gender disparity do not capture the
most important aspects of gender inequality. This could be one explanation for why some
measures of gender inequality were significantly related to rape and others were not. For
instance, inequality in educational attainment was not related to rape rates in any of the
Peterson and Bailey (1992) models. Through this study, it was also demonstrated that
some of the measures of gender inequality may be confounding other relationships. In
particular, Peterson and Bailey (1992) found that although in the main model there was
no significant relationship between gender inequality in occupation type and rape rates,
when looking at specific types of occupations the findings were mixed. Indeed, they
found support for backlash and amelioration when looking at specific types of
occupations. By thoroughly investigating the relationship between gender inequality and
rape, Peterson and Bailey (1992) illustrated how complicated the relationship may be and
demonstrated important ways to attempt to uncover more information.
Some scholars have noted that much of the research on gender inequality and
violence against women, including Peterson and Bailey (1992), have acted under the
axiom that amelioration and backlash are contradictory hypotheses (Bailey, 1999;
Whaley, 2001; Whaley et al., 2013). However, these hypotheses may be corollaries of one
another contingent either on time or the level of gender inequality. For instance, Bailey
(1999) and Whaley (2001) suggested and found some evidence that backlash occurs for a
period of time leading to amelioration. According to Bailey (1999), prior to his analysis
no studies had investigated the role of changes in gender inequality and changes in the
rape rate. Indeed, I was unable to locate any studies prior to Bailey (1999) that
investigated the relationship between changes in gender inequality and changes in any
form of violence against women. Bailey (1999) investigated the relationship between
gender inequality (using male and female differences in educational attainment,
occupation type, and income as indicators) and rape rates for central cities. He found
support for the backlash hypothesis between occupation type and rape rates for 1980 and
1990 (cross-sectional models), but found an ameliorative relationship between changes in
income inequality and changes in rape rates from 1980 to 1990 (change model). In the
end, Bailey suggested that although his analysis was an “improvement over previous
investigations this analysis does not fully resolve important feminist arguments regarding
the socioeconomic status of women and rape” (1999, p. 59).
Whaley (2001) presented itself as refining the temporal theoretical relationship
between change in gender inequality and violence against women. She hypothesized that
initially during times of higher gender inequality, women experience lower levels of rape
as a “reward, “but that in the long term, gender inequality prompts cultural changes that
build support for rape, prompting an increase. The reverse was hypothesized as well. To
use the hypotheses already introduced, Whaley (2001) hypothesized that backlash occurs
first and that after a sustained period of change amelioration can occur. Using 5-, 10-, and
20-year lagged indicators, Whaley (2001) was also unable to provide strong support for
her hypothesis; however, most of the significant relationships were in the proposed
direction. For example, cross-sectional analyses suggested that higher levels of inequality
in holding executive occupations were associated with lower levels of rape supporting the
backlash hypothesis. The model using a 5-year lag from 1970 to 1975 suggested backlash
also. Yet in accord with the hypothesis, the model using a 10-year lag between 1980 and
1990 found support for the ameliorative hypotheses. Specifically, decreases in gender
inequality in holding executive occupations were associated with decreases in rape.
Therefore, findings related to holding executive occupations appeared to support backlash
cross-sectionally and amelioration over a 10-year period. However, gender inequality in
executive occupations was not significantly related to rape when looking at change from
1980 to 1985, 1970 to 1980, and 1970 to 1990. Therefore, within this analysis, the results
reflect a complicated relationship.
Additionally, Whaley et al. (2013) suggested and found some evidence that the
nature of the relationship between gender inequality and violence against women may be
nonlinear and related to the level of gender inequality. Specifically, they investigated the
relationship between gender inequality (in terms of educational attainment, median
income, and employment status) and various rates of homicides for 1990 and 2000 from
the SHR (male-female, male-male, female-male, and female-female). Using negative
binomial regressions, Whaley et al. (2013) found that gender equality (represented by a
factor score of the three variables) was positively related to male offender – female victim
homicide rates and that the squared form of gender equality was negatively related in
1990. These significant findings suggest that the relationship between gender equality and
male-female homicide is nonlinear. Gender equality and its squared term were not
significantly related to any other homicide rates except female-offender and
femalevictim. For the female-female homicide rate, gender equality was significantly and
positively related in the model that did not include the robbery rate, which was included
as a “robust control for other factors related to criminal violence” (Whaley et al., 2013, p.
743). In 2000, gender equality and its squared form were significantly related to
malefemale homicide when the robbery rate was not included but only the squared form
was significantly related (negatively) when the robbery rate was included. The same was
true for male-male homicide rates. Just gender equality is significantly and positively
related to the female-female rate when the robbery rate was not included. In all other
instances, gender equality and its squared form were not significantly related to homicide
rates. Whaley et al. (2013) interpreted the effect of adding the robbery rate as suggesting
that unmeasured factors associated with violent crime may be mediating the relationship
or that there may be an issue with multicollinearity. As a whole, they found considerable
support for their hypothesis that backlash processes are likely to function at “low to
intermediate levels” of gender equality and that amelioration occurs when equality is high
for male-female homicide rates (p. 747).
Xie, Heimer, and Lauritsen (2012a) used longitudinal methods to assess whether
there was a relationship between non-fatal intimate partner violence and absolute/relative
indicators of women’s status. They created five five-year pooled periods using MSAlevel
data (which includes the 40 largest MSAs) from the NCVS for 1980 to 2004. To
thoroughly assess their hypotheses, they also disaggregated the non-fatal violence against
women data by victim-offender relationship using the following listed categories:
stranger, intimate partner, and other known. The four hypotheses included amelioration
(which was posited to function more for intimate partner violence than non-intimate
violence), backlash (which was also posited to function more for intimate partner
violence than non-intimate violence), routine activities (which suggests that increases in
absolute status also increases exposure and could lead to an increase in stranger and
known victimization), and protective (which suggests that an increase in absolute status
may increase the accessibility of protective mechanisms and decrease victimization).
Absolute women’s status was operationalized by the percent of women who were
employed, percent of voting age women who voted in the November presidential and
congressional elections, and an index of income-educational attainment, which included
median income and percent who completed four years of college or more. Relative status
was measured using the rate of women’s labor force participation minus that of men, the
level of income-educational attainment for women minus that of men, and the rate of
voter turnout for women minus that of men. Their results found support for each of the
hypotheses except for amelioration. They found a negative relationship between women’s
absolute economic and political status and intimate partner violence which supports the
protective hypothesis. They also found the negative relationship between inequality and
intimate partner victimization which supports the backlash hypothesis. The routine
activities hypothesis was supported through a positive relationship between absolute
indicators of labor force participation and victimization by strangers and non-intimate
persons known to the victim. This longitudinal analysis does not suggest that backlash,
amelioration, protective, nor routine activities hypotheses are corollary by explaining the
relationship between inequality, absolute status, and violence against women at different
periods. However, they found evidence that the nature of the relationship is contingent on
the victim-offender relationship.
The hypotheses and findings suggesting that backlash and amelioration are
corollary are particularly important considering that much of the empirical quantitative
literature on gender inequality has been cross-sectional and assumed a linear relationship.
By design much of the extant literature is unable to address these possibilities (Whaley et
al., 2013). The evidence supporting corollary processes indicate that future analyses
should continue to consider change over time and the structural form of the relationship
as the proposed dissertation will do. In addition, Xie et al. (2012a) presented evidence
that the victim-offender relationship is important to also consider in future analyses.
In addition, while not an inequality hypothesis, several scholars have investigated the role
of absolute status on trends of violence against women. Changes in absolute status are
increases or decreases in the status of a group with no consideration or comparison to
other groups. It is possible that there is a relationship between absolute status and
violence against women and that any findings suggesting a relationship
When race-disaggregated trends of violence against women are investigated,
major gaps in the crime decline, race, and gender literature are exposed. For example,
Lauritsen (2012) illustrated that although the rape and sexual assault trends for Latinas,
Black women, and white women have been similar for the last 40 years, the gaps between
them have varied occasionally. The variation in the gap could indicate different processes
at work. These types of findings are important in that they denote the importance of
attending to gender, race, and ethnicity but are often overlooked in later studies.
Ultimately, the crime decline literature has at times denoted the varying magnitude of the
decline by race and gender subgroups, but has not produced many intersectional
approaches to explaining the disparate decline (Heimer & Lauritsen, 2008; Lauritsen,
2012; Lauritsen & Heimer, 2010; Levitt, 2004; Parker, 2008). As a consequence, the
disparate declines and trends by race, ethnicity, and gender are not well understood. The
race and inequality literature tends to lack a gender-lens, a framework denoting the role
that gender plays in how things are enacted and perceived, while at the same time the
gender and inequality literature tends to lack a race-lens, a framework denoting the role
of race (McCall, 2005). This study argues that it is important to consider violence against
women from a gender x race-lens or an intersectional-lens.
HYPOTHESES AND EMPIRICAL SUPPORT
Before discussing the ways to apply an intersectionality perspective to the gender
inequality hypothesis, it is pertinent to review the various hypotheses that exist within the
literature. As noted in the introduction, the nature of the relationship between gender
inequality and violence against women has not been clearly defined within the empirical
literature (Whaley et al., 2013). In fact, evidence exists for multiple hypotheses. The
original hypothesis and the one many organizations work under suggests that as gender
inequality decreases and the genders become more equal violence against women will
decrease (Whaley et al., 2013). This is referred to as the amelioration hypothesis.
Peterson and Bailey (1992) found some support for the amelioration hypothesis. They
investigated the relationship between various forms of inequality (general, racial, and
gender) and rape rates for United States metropolitan areas in 1980. The rape rates were
obtained from the UCR in 1980. General inequality was measured using the GINI index
which represents income inequality in the total population. Racial inequality was
measured by the difference in median income for Black and white families. Gender
inequality was measured in terms of education using the ratio of men to women who
completed 4 years of high school, who completed 4 years of college, and who completed
5 or more years of college. It was also measured in terms of income using a ratio of men
to women’s median income for all people, people who completed 4 years of high school,
and people who completed 4 years or more of college. Gender inequality was also
measured by the percent of people employed in managerial and professional positions
who were women and by a ratio of men to women who live in poverty. Multiple models
were estimated with most containing one form of gender inequality and one containing a
representative of each type. The GINI index was found to be positively and significantly
related to rape rapes in each model. This finding suggests that places with greater general
income inequality exhibited higher rape rates. Racial inequality in income was significant
and positive in the models that included gender inequality by ratio of high school
graduates only and gender inequality in percentage of women in professional occupations
only. When all forms of gender inequality were included (income, high school graduation,
and profession occupations), gender inequality in median income was significantly and
positively related to rape rates. This finding suggests that where gender inequality in
income is greater, rape rates will also be higher and this is consistent with the
amelioration hypothesis.
Another hypothesis, the backlash hypothesis, suggests an opposite relationship.
Specifically, the backlash hypothesis suggests that as gender inequality decreases and the
genders become more equal, violence against women will increase (Whaley et al., 2013).
It is important to note that the term “backlash” is sometimes used to imply a specific
mechanism causing the increase of violence. For instance, sometimes backlash is defined
as a process by which men (or other dominant group) react violently to challenges to their
dominant status. In this study, as with many past studies, the mechanism of the
relationship between gender inequality and violence against women is not being
evaluated. Rather this study examines the relationship between gender inequality and
violence against women which therefore only allows for some speculation as to the
mechanism of the relationship. Backlash, herein, is used to represent a negative
relationship between changes in gender inequality and changes in violence against
women.
Within the same article described above, Peterson and Bailey (1992) also found
support for a backlash hypothesis when further investigating the relationship between
gender inequality in occupations and rape rates. Specifically, they separated professional
occupations into 1) executives, administrators, and managers 2) public officials and
administrators 3) engineers and natural scientists 4) health diagnostic occupations 5)
elementary and secondary teachers. They found that where the percent of executives,
administrators, and managers that are women were higher rape rates were also higher
suggesting backlash. However, they also found that where the percent of health diagnostic
occupations that are women were higher rape rates were lower. This was true of
elementary and secondary teachers as well. Since being an elementary or secondary
teacher is a women-typed occupation, the finding of a negative relationship could be
interpreted as backlash as well. The authors suggested that “the more that women remain
in traditional ‘female’ occupations, the greater their ‘reward’ – freedom from rape” (p.
172). However, the significant finding for health diagnostic occupations appears to
support amelioration, since health diagnostic occupations are traditionally men-typed
occupations.
The Peterson and Bailey (1992) study is illustrative for three primary reasons.
First, the analysis assesses the relationship between gender inequality and rape in
multiple ways. They use several forms of inequality including: education, income, and
occupation. In addition, they estimate models with each form individually as well as a
model with all of the forms. Second, this study is illustrative because it finds support for
both the ameliorative and backlash hypotheses. Due to these mixed findings, the authors
conclude that their “analysis by no means settles the issue of rape and the structural
position of women vis-à-vis men” (p. 174). Lastly, this study thoroughly investigated
different elements of inequality. For example, as discussed above, they investigated more
specific occupation categories to gain an understanding of the relationship between
gender parity in occupations and rape rates. Ultimately, together these three elements
exemplify how complicated the relationship between gender inequality and rape may be.
Specifically, it is possible that common indicators of gender disparity do not capture the
most important aspects of gender inequality. This could be one explanation for why some
measures of gender inequality were significantly related to rape and others were not. For
instance, inequality in educational attainment was not related to rape rates in any of the
Peterson and Bailey (1992) models. Through this study, it was also demonstrated that
some of the measures of gender inequality may be confounding other relationships. In
particular, Peterson and Bailey (1992) found that although in the main model there was
no significant relationship between gender inequality in occupation type and rape rates,
when looking at specific types of occupations the findings were mixed. Indeed, they
found support for backlash and amelioration when looking at specific types of
occupations. By thoroughly investigating the relationship between gender inequality and
rape, Peterson and Bailey (1992) illustrated how complicated the relationship may be and
demonstrated important ways to attempt to uncover more information.
Some scholars have noted that much of the research on gender inequality and
violence against women, including Peterson and Bailey (1992), have acted under the
axiom that amelioration and backlash are contradictory hypotheses (Bailey, 1999;
Whaley, 2001; Whaley et al., 2013). However, these hypotheses may be corollaries of one
another contingent either on time or the level of gender inequality. For instance, Bailey
(1999) and Whaley (2001) suggested and found some evidence that backlash occurs for a
period of time leading to amelioration. According to Bailey (1999), prior to his analysis
no studies had investigated the role of changes in gender inequality and changes in the
rape rate. Indeed, I was unable to locate any studies prior to Bailey (1999) that
investigated the relationship between changes in gender inequality and changes in any
form of violence against women. Bailey (1999) investigated the relationship between
gender inequality (using male and female differences in educational attainment,
occupation type, and income as indicators) and rape rates for central cities. He found
support for the backlash hypothesis between occupation type and rape rates for 1980 and
1990 (cross-sectional models), but found an ameliorative relationship between changes in
income inequality and changes in rape rates from 1980 to 1990 (change model). In the
end, Bailey suggested that although his analysis was an “improvement over previous
investigations this analysis does not fully resolve important feminist arguments regarding
the socioeconomic status of women and rape” (1999, p. 59).
Whaley (2001) presented itself as refining the temporal theoretical relationship
between change in gender inequality and violence against women. She hypothesized that
initially during times of higher gender inequality, women experience lower levels of rape
as a “reward, “but that in the long term, gender inequality prompts cultural changes that
build support for rape, prompting an increase. The reverse was hypothesized as well. To
use the hypotheses already introduced, Whaley (2001) hypothesized that backlash occurs
first and that after a sustained period of change amelioration can occur. Using 5-, 10-, and
20-year lagged indicators, Whaley (2001) was also unable to provide strong support for
her hypothesis; however, most of the significant relationships were in the proposed
direction. For example, cross-sectional analyses suggested that higher levels of inequality
in holding executive occupations were associated with lower levels of rape supporting the
backlash hypothesis. The model using a 5-year lag from 1970 to 1975 suggested backlash
also. Yet in accord with the hypothesis, the model using a 10-year lag between 1980 and
1990 found support for the ameliorative hypotheses. Specifically, decreases in gender
inequality in holding executive occupations were associated with decreases in rape.
Therefore, findings related to holding executive occupations appeared to support backlash
cross-sectionally and amelioration over a 10-year period. However, gender inequality in
executive occupations was not significantly related to rape when looking at change from
1980 to 1985, 1970 to 1980, and 1970 to 1990. Therefore, within this analysis, the results
reflect a complicated relationship.
Additionally, Whaley et al. (2013) suggested and found some evidence that the
nature of the relationship between gender inequality and violence against women may be
nonlinear and related to the level of gender inequality. Specifically, they investigated the
relationship between gender inequality (in terms of educational attainment, median
income, and employment status) and various rates of homicides for 1990 and 2000 from
the SHR (male-female, male-male, female-male, and female-female). Using negative
binomial regressions, Whaley et al. (2013) found that gender equality (represented by a
factor score of the three variables) was positively related to male offender – female victim
homicide rates and that the squared form of gender equality was negatively related in
1990. These significant findings suggest that the relationship between gender equality and
male-female homicide is nonlinear. Gender equality and its squared term were not
significantly related to any other homicide rates except female-offender and
femalevictim. For the female-female homicide rate, gender equality was significantly and
positively related in the model that did not include the robbery rate, which was included
as a “robust control for other factors related to criminal violence” (Whaley et al., 2013, p.
743). In 2000, gender equality and its squared form were significantly related to
malefemale homicide when the robbery rate was not included but only the squared form
was significantly related (negatively) when the robbery rate was included. The same was
true for male-male homicide rates. Just gender equality is significantly and positively
related to the female-female rate when the robbery rate was not included. In all other
instances, gender equality and its squared form were not significantly related to homicide
rates. Whaley et al. (2013) interpreted the effect of adding the robbery rate as suggesting
that unmeasured factors associated with violent crime may be mediating the relationship
or that there may be an issue with multicollinearity. As a whole, they found considerable
support for their hypothesis that backlash processes are likely to function at “low to
intermediate levels” of gender equality and that amelioration occurs when equality is high
for male-female homicide rates (p. 747).
Xie, Heimer, and Lauritsen (2012a) used longitudinal methods to assess whether
there was a relationship between non-fatal intimate partner violence and absolute/relative
indicators of women’s status. They created five five-year pooled periods using MSAlevel
data (which includes the 40 largest MSAs) from the NCVS for 1980 to 2004. To
thoroughly assess their hypotheses, they also disaggregated the non-fatal violence against
women data by victim-offender relationship using the following listed categories:
stranger, intimate partner, and other known. The four hypotheses included amelioration
(which was posited to function more for intimate partner violence than non-intimate
violence), backlash (which was also posited to function more for intimate partner
violence than non-intimate violence), routine activities (which suggests that increases in
absolute status also increases exposure and could lead to an increase in stranger and
known victimization), and protective (which suggests that an increase in absolute status
may increase the accessibility of protective mechanisms and decrease victimization).
Absolute women’s status was operationalized by the percent of women who were
employed, percent of voting age women who voted in the November presidential and
congressional elections, and an index of income-educational attainment, which included
median income and percent who completed four years of college or more. Relative status
was measured using the rate of women’s labor force participation minus that of men, the
level of income-educational attainment for women minus that of men, and the rate of
voter turnout for women minus that of men. Their results found support for each of the
hypotheses except for amelioration. They found a negative relationship between women’s
absolute economic and political status and intimate partner violence which supports the
protective hypothesis. They also found the negative relationship between inequality and
intimate partner victimization which supports the backlash hypothesis. The routine
activities hypothesis was supported through a positive relationship between absolute
indicators of labor force participation and victimization by strangers and non-intimate
persons known to the victim. This longitudinal analysis does not suggest that backlash,
amelioration, protective, nor routine activities hypotheses are corollary by explaining the
relationship between inequality, absolute status, and violence against women at different
periods. However, they found evidence that the nature of the relationship is contingent on
the victim-offender relationship.
The hypotheses and findings suggesting that backlash and amelioration are
corollary are particularly important considering that much of the empirical quantitative
literature on gender inequality has been cross-sectional and assumed a linear relationship.
By design much of the extant literature is unable to address these possibilities (Whaley et
al., 2013). The evidence supporting corollary processes indicate that future analyses
should continue to consider change over time and the structural form of the relationship
as the proposed dissertation will do. In addition, Xie et al. (2012a) presented evidence
that the victim-offender relationship is important to also consider in future analyses.
In addition, while not an inequality hypothesis, several scholars have investigated the role
of absolute status on trends of violence against women. Changes in absolute status are
increases or decreases in the status of a group with no consideration or comparison to
other groups. It is possible that there is a relationship between absolute status and
violence against women and that any findings suggesting a relationship
When race-disaggregated trends of violence against women are investigated,
major gaps in the crime decline, race, and gender literature are exposed. For example,
Lauritsen (2012) illustrated that although the rape and sexual assault trends for Latinas,
Black women, and white women have been similar for the last 40 years, the gaps between
them have varied occasionally. The variation in the gap could indicate different processes
at work. These types of findings are important in that they denote the importance of
attending to gender, race, and ethnicity but are often overlooked in later studies.
Ultimately, the crime decline literature has at times denoted the varying magnitude of the
decline by race and gender subgroups, but has not produced many intersectional
approaches to explaining the disparate decline (Heimer & Lauritsen, 2008; Lauritsen,
2012; Lauritsen & Heimer, 2010; Levitt, 2004; Parker, 2008). As a consequence, the
disparate declines and trends by race, ethnicity, and gender are not well understood. The
race and inequality literature tends to lack a gender-lens, a framework denoting the role
that gender plays in how things are enacted and perceived, while at the same time the
gender and inequality literature tends to lack a race-lens, a framework denoting the role
of race (McCall, 2005). This study argues that it is important to consider violence against
women from a gender x race-lens or an intersectional-lens.
HYPOTHESES AND EMPIRICAL SUPPORT
Before discussing the ways to apply an intersectionality perspective to the gender
inequality hypothesis, it is pertinent to review the various hypotheses that exist within the
literature. As noted in the introduction, the nature of the relationship between gender
inequality and violence against women has not been clearly defined within the empirical
literature (Whaley et al., 2013). In fact, evidence exists for multiple hypotheses. The
original hypothesis and the one many organizations work under suggests that as gender
inequality decreases and the genders become more equal violence against women will
decrease (Whaley et al., 2013). This is referred to as the amelioration hypothesis.
Peterson and Bailey (1992) found some support for the amelioration hypothesis. They
investigated the relationship between various forms of inequality (general, racial, and
gender) and rape rates for United States metropolitan areas in 1980. The rape rates were
obtained from the UCR in 1980. General inequality was measured using the GINI index
which represents income inequality in the total population. Racial inequality was
measured by the difference in median income for Black and white families. Gender
inequality was measured in terms of education using the ratio of men to women who
completed 4 years of high school, who completed 4 years of college, and who completed
5 or more years of college. It was also measured in terms of income using a ratio of men
to women’s median income for all people, people who completed 4 years of high school,
and people who completed 4 years or more of college. Gender inequality was also
measured by the percent of people employed in managerial and professional positions
who were women and by a ratio of men to women who live in poverty. Multiple models
were estimated with most containing one form of gender inequality and one containing a
representative of each type. The GINI index was found to be positively and significantly
related to rape rapes in each model. This finding suggests that places with greater general
income inequality exhibited higher rape rates. Racial inequality in income was significant
and positive in the models that included gender inequality by ratio of high school
graduates only and gender inequality in percentage of women in professional occupations
only. When all forms of gender inequality were included (income, high school graduation,
and profession occupations), gender inequality in median income was significantly and
positively related to rape rates. This finding suggests that where gender inequality in
income is greater, rape rates will also be higher and this is consistent with the
amelioration hypothesis.
Another hypothesis, the backlash hypothesis, suggests an opposite relationship.
Specifically, the backlash hypothesis suggests that as gender inequality decreases and the
genders become more equal, violence against women will increase (Whaley et al., 2013).
It is important to note that the term “backlash” is sometimes used to imply a specific
mechanism causing the increase of violence. For instance, sometimes backlash is defined
as a process by which men (or other dominant group) react violently to challenges to their
dominant status. In this study, as with many past studies, the mechanism of the
relationship between gender inequality and violence against women is not being
evaluated. Rather this study examines the relationship between gender inequality and
violence against women which therefore only allows for some speculation as to the
mechanism of the relationship. Backlash, herein, is used to represent a negative
relationship between changes in gender inequality and changes in violence against
women.
Within the same article described above, Peterson and Bailey (1992) also found
support for a backlash hypothesis when further investigating the relationship between
gender inequality in occupations and rape rates. Specifically, they separated professional
occupations into 1) executives, administrators, and managers 2) public officials and
administrators 3) engineers and natural scientists 4) health diagnostic occupations 5)
elementary and secondary teachers. They found that where the percent of executives,
administrators, and managers that are women were higher rape rates were also higher
suggesting backlash. However, they also found that where the percent of health diagnostic
occupations that are women were higher rape rates were lower. This was true of
elementary and secondary teachers as well. Since being an elementary or secondary
teacher is a women-typed occupation, the finding of a negative relationship could be
interpreted as backlash as well. The authors suggested that “the more that women remain
in traditional ‘female’ occupations, the greater their ‘reward’ – freedom from rape” (p.
172). However, the significant finding for health diagnostic occupations appears to
support amelioration, since health diagnostic occupations are traditionally men-typed
occupations.
The Peterson and Bailey (1992) study is illustrative for three primary reasons.
First, the analysis assesses the relationship between gender inequality and rape in
multiple ways. They use several forms of inequality including: education, income, and
occupation. In addition, they estimate models with each form individually as well as a
model with all of the forms. Second, this study is illustrative because it finds support for
both the ameliorative and backlash hypotheses. Due to these mixed findings, the authors
conclude that their “analysis by no means settles the issue of rape and the structural
position of women vis-à-vis men” (p. 174). Lastly, this study thoroughly investigated
different elements of inequality. For example, as discussed above, they investigated more
specific occupation categories to gain an understanding of the relationship between
gender parity in occupations and rape rates. Ultimately, together these three elements
exemplify how complicated the relationship between gender inequality and rape may be.
Specifically, it is possible that common indicators of gender disparity do not capture the
most important aspects of gender inequality. This could be one explanation for why some
measures of gender inequality were significantly related to rape and others were not. For
instance, inequality in educational attainment was not related to rape rates in any of the
Peterson and Bailey (1992) models. Through this study, it was also demonstrated that
some of the measures of gender inequality may be confounding other relationships. In
particular, Peterson and Bailey (1992) found that although in the main model there was
no significant relationship between gender inequality in occupation type and rape rates,
when looking at specific types of occupations the findings were mixed. Indeed, they
found support for backlash and amelioration when looking at specific types of
occupations. By thoroughly investigating the relationship between gender inequality and
rape, Peterson and Bailey (1992) illustrated how complicated the relationship may be and
demonstrated important ways to attempt to uncover more information.
Some scholars have noted that much of the research on gender inequality and
violence against women, including Peterson and Bailey (1992), have acted under the
axiom that amelioration and backlash are contradictory hypotheses (Bailey, 1999;
Whaley, 2001; Whaley et al., 2013). However, these hypotheses may be corollaries of one
another contingent either on time or the level of gender inequality. For instance, Bailey
(1999) and Whaley (2001) suggested and found some evidence that backlash occurs for a
period of time leading to amelioration. According to Bailey (1999), prior to his analysis
no studies had investigated the role of changes in gender inequality and changes in the
rape rate. Indeed, I was unable to locate any studies prior to Bailey (1999) that
investigated the relationship between changes in gender inequality and changes in any
form of violence against women. Bailey (1999) investigated the relationship between
gender inequality (using male and female differences in educational attainment,
occupation type, and income as indicators) and rape rates for central cities. He found
support for the backlash hypothesis between occupation type and rape rates for 1980 and
1990 (cross-sectional models), but found an ameliorative relationship between changes in
income inequality and changes in rape rates from 1980 to 1990 (change model). In the
end, Bailey suggested that although his analysis was an “improvement over previous
investigations this analysis does not fully resolve important feminist arguments regarding
the socioeconomic status of women and rape” (1999, p. 59).
Whaley (2001) presented itself as refining the temporal theoretical relationship
between change in gender inequality and violence against women. She hypothesized that
initially during times of higher gender inequality, women experience lower levels of rape
as a “reward, “but that in the long term, gender inequality prompts cultural changes that
build support for rape, prompting an increase. The reverse was hypothesized as well. To
use the hypotheses already introduced, Whaley (2001) hypothesized that backlash occurs
first and that after a sustained period of change amelioration can occur. Using 5-, 10-, and
20-year lagged indicators, Whaley (2001) was also unable to provide strong support for
her hypothesis; however, most of the significant relationships were in the proposed
direction. For example, cross-sectional analyses suggested that higher levels of inequality
in holding executive occupations were associated with lower levels of rape supporting the
backlash hypothesis. The model using a 5-year lag from 1970 to 1975 suggested backlash
also. Yet in accord with the hypothesis, the model using a 10-year lag between 1980 and
1990 found support for the ameliorative hypotheses. Specifically, decreases in gender
inequality in holding executive occupations were associated with decreases in rape.
Therefore, findings related to holding executive occupations appeared to support backlash
cross-sectionally and amelioration over a 10-year period. However, gender inequality in
executive occupations was not significantly related to rape when looking at change from
1980 to 1985, 1970 to 1980, and 1970 to 1990. Therefore, within this analysis, the results
reflect a complicated relationship.
Additionally, Whaley et al. (2013) suggested and found some evidence that the
nature of the relationship between gender inequality and violence against women may be
nonlinear and related to the level of gender inequality. Specifically, they investigated the
relationship between gender inequality (in terms of educational attainment, median
income, and employment status) and various rates of homicides for 1990 and 2000 from
the SHR (male-female, male-male, female-male, and female-female). Using negative
binomial regressions, Whaley et al. (2013) found that gender equality (represented by a
factor score of the three variables) was positively related to male offender – female victim
homicide rates and that the squared form of gender equality was negatively related in
1990. These significant findings suggest that the relationship between gender equality and
male-female homicide is nonlinear. Gender equality and its squared term were not
significantly related to any other homicide rates except female-offender and
femalevictim. For the female-female homicide rate, gender equality was significantly and
positively related in the model that did not include the robbery rate, which was included
as a “robust control for other factors related to criminal violence” (Whaley et al., 2013, p.
743). In 2000, gender equality and its squared form were significantly related to
malefemale homicide when the robbery rate was not included but only the squared form
was significantly related (negatively) when the robbery rate was included. The same was
true for male-male homicide rates. Just gender equality is significantly and positively
related to the female-female rate when the robbery rate was not included. In all other
instances, gender equality and its squared form were not significantly related to homicide
rates. Whaley et al. (2013) interpreted the effect of adding the robbery rate as suggesting
that unmeasured factors associated with violent crime may be mediating the relationship
or that there may be an issue with multicollinearity. As a whole, they found considerable
support for their hypothesis that backlash processes are likely to function at “low to
intermediate levels” of gender equality and that amelioration occurs when equality is high
for male-female homicide rates (p. 747).
Xie, Heimer, and Lauritsen (2012a) used longitudinal methods to assess whether
there was a relationship between non-fatal intimate partner violence and absolute/relative
indicators of women’s status. They created five five-year pooled periods using MSAlevel
data (which includes the 40 largest MSAs) from the NCVS for 1980 to 2004. To
thoroughly assess their hypotheses, they also disaggregated the non-fatal violence against
women data by victim-offender relationship using the following listed categories:
stranger, intimate partner, and other known. The four hypotheses included amelioration
(which was posited to function more for intimate partner violence than non-intimate
violence), backlash (which was also posited to function more for intimate partner
violence than non-intimate violence), routine activities (which suggests that increases in
absolute status also increases exposure and could lead to an increase in stranger and
known victimization), and protective (which suggests that an increase in absolute status
may increase the accessibility of protective mechanisms and decrease victimization).
Absolute women’s status was operationalized by the percent of women who were
employed, percent of voting age women who voted in the November presidential and
congressional elections, and an index of income-educational attainment, which included
median income and percent who completed four years of college or more. Relative status
was measured using the rate of women’s labor force participation minus that of men, the
level of income-educational attainment for women minus that of men, and the rate of
voter turnout for women minus that of men. Their results found support for each of the
hypotheses except for amelioration. They found a negative relationship between women’s
absolute economic and political status and intimate partner violence which supports the
protective hypothesis. They also found the negative relationship between inequality and
intimate partner victimization which supports the backlash hypothesis. The routine
activities hypothesis was supported through a positive relationship between absolute
indicators of labor force participation and victimization by strangers and non-intimate
persons known to the victim. This longitudinal analysis does not suggest that backlash,
amelioration, protective, nor routine activities hypotheses are corollary by explaining the
relationship between inequality, absolute status, and violence against women at different
periods. However, they found evidence that the nature of the relationship is contingent on
the victim-offender relationship.
The hypotheses and findings suggesting that backlash and amelioration are
corollary are particularly important considering that much of the empirical quantitative
literature on gender inequality has been cross-sectional and assumed a linear relationship.
By design much of the extant literature is unable to address these possibilities (Whaley et
al., 2013). The evidence supporting corollary processes indicate that future analyses
should continue to consider change over time and the structural form of the relationship
as the proposed dissertation will do. In addition, Xie et al. (2012a) presented evidence
that the victim-offender relationship is important to also consider in future analyses.
In addition, while not an inequality hypothesis, several scholars have investigated the role
of absolute status on trends of violence against women. Changes in absolute status are
increases or decreases in the status of a group with no consideration or comparison to
other groups. It is possible that there is a relationship between absolute status and
violence against women and that any findings suggesting a relationship
When race-disaggregated trends of violence against women are investigated,
major gaps in the crime decline, race, and gender literature are exposed. For example,
Lauritsen (2012) illustrated that although the rape and sexual assault trends for Latinas,
Black women, and white women have been similar for the last 40 years, the gaps between
them have varied occasionally. The variation in the gap could indicate different processes
at work. These types of findings are important in that they denote the importance of
attending to gender, race, and ethnicity but are often overlooked in later studies.
Ultimately, the crime decline literature has at times denoted the varying magnitude of the
decline by race and gender subgroups, but has not produced many intersectional
approaches to explaining the disparate decline (Heimer & Lauritsen, 2008; Lauritsen,
2012; Lauritsen & Heimer, 2010; Levitt, 2004; Parker, 2008). As a consequence, the
disparate declines and trends by race, ethnicity, and gender are not well understood. The
race and inequality literature tends to lack a gender-lens, a framework denoting the role
that gender plays in how things are enacted and perceived, while at the same time the
gender and inequality literature tends to lack a race-lens, a framework denoting the role
of race (McCall, 2005). This study argues that it is important to consider violence against
women from a gender x race-lens or an intersectional-lens.
HYPOTHESES AND EMPIRICAL SUPPORT
Before discussing the ways to apply an intersectionality perspective to the gender
inequality hypothesis, it is pertinent to review the various hypotheses that exist within the
literature. As noted in the introduction, the nature of the relationship between gender
inequality and violence against women has not been clearly defined within the empirical
literature (Whaley et al., 2013). In fact, evidence exists for multiple hypotheses. The
original hypothesis and the one many organizations work under suggests that as gender
inequality decreases and the genders become more equal violence against women will
decrease (Whaley et al., 2013). This is referred to as the amelioration hypothesis.
Peterson and Bailey (1992) found some support for the amelioration hypothesis. They
investigated the relationship between various forms of inequality (general, racial, and
gender) and rape rates for United States metropolitan areas in 1980. The rape rates were
obtained from the UCR in 1980. General inequality was measured using the GINI index
which represents income inequality in the total population. Racial inequality was
measured by the difference in median income for Black and white families. Gender
inequality was measured in terms of education using the ratio of men to women who
completed 4 years of high school, who completed 4 years of college, and who completed
5 or more years of college. It was also measured in terms of income using a ratio of men
to women’s median income for all people, people who completed 4 years of high school,
and people who completed 4 years or more of college. Gender inequality was also
measured by the percent of people employed in managerial and professional positions
who were women and by a ratio of men to women who live in poverty. Multiple models
were estimated with most containing one form of gender inequality and one containing a
representative of each type. The GINI index was found to be positively and significantly
related to rape rapes in each model. This finding suggests that places with greater general
income inequality exhibited higher rape rates. Racial inequality in income was significant
and positive in the models that included gender inequality by ratio of high school
graduates only and gender inequality in percentage of women in professional occupations
only. When all forms of gender inequality were included (income, high school graduation,
and profession occupations), gender inequality in median income was significantly and
positively related to rape rates. This finding suggests that where gender inequality in
income is greater, rape rates will also be higher and this is consistent with the
amelioration hypothesis.
Another hypothesis, the backlash hypothesis, suggests an opposite relationship.
Specifically, the backlash hypothesis suggests that as gender inequality decreases and the
genders become more equal, violence against women will increase (Whaley et al., 2013).
It is important to note that the term “backlash” is sometimes used to imply a specific
mechanism causing the increase of violence. For instance, sometimes backlash is defined
as a process by which men (or other dominant group) react violently to challenges to their
dominant status. In this study, as with many past studies, the mechanism of the
relationship between gender inequality and violence against women is not being
evaluated. Rather this study examines the relationship between gender inequality and
violence against women which therefore only allows for some speculation as to the
mechanism of the relationship. Backlash, herein, is used to represent a negative
relationship between changes in gender inequality and changes in violence against
women.
Within the same article described above, Peterson and Bailey (1992) also found
support for a backlash hypothesis when further investigating the relationship between
gender inequality in occupations and rape rates. Specifically, they separated professional
occupations into 1) executives, administrators, and managers 2) public officials and
administrators 3) engineers and natural scientists 4) health diagnostic occupations 5)
elementary and secondary teachers. They found that where the percent of executives,
administrators, and managers that are women were higher rape rates were also higher
suggesting backlash. However, they also found that where the percent of health diagnostic
occupations that are women were higher rape rates were lower. This was true of
elementary and secondary teachers as well. Since being an elementary or secondary
teacher is a women-typed occupation, the finding of a negative relationship could be
interpreted as backlash as well. The authors suggested that “the more that women remain
in traditional ‘female’ occupations, the greater their ‘reward’ – freedom from rape” (p.
172). However, the significant finding for health diagnostic occupations appears to
support amelioration, since health diagnostic occupations are traditionally men-typed
occupations.
The Peterson and Bailey (1992) study is illustrative for three primary reasons.
First, the analysis assesses the relationship between gender inequality and rape in
multiple ways. They use several forms of inequality including: education, income, and
occupation. In addition, they estimate models with each form individually as well as a
model with all of the forms. Second, this study is illustrative because it finds support for
both the ameliorative and backlash hypotheses. Due to these mixed findings, the authors
conclude that their “analysis by no means settles the issue of rape and the structural
position of women vis-à-vis men” (p. 174). Lastly, this study thoroughly investigated
different elements of inequality. For example, as discussed above, they investigated more
specific occupation categories to gain an understanding of the relationship between
gender parity in occupations and rape rates. Ultimately, together these three elements
exemplify how complicated the relationship between gender inequality and rape may be.
Specifically, it is possible that common indicators of gender disparity do not capture the
most important aspects of gender inequality. This could be one explanation for why some
measures of gender inequality were significantly related to rape and others were not. For
instance, inequality in educational attainment was not related to rape rates in any of the
Peterson and Bailey (1992) models. Through this study, it was also demonstrated that
some of the measures of gender inequality may be confounding other relationships. In
particular, Peterson and Bailey (1992) found that although in the main model there was
no significant relationship between gender inequality in occupation type and rape rates,
when looking at specific types of occupations the findings were mixed. Indeed, they
found support for backlash and amelioration when looking at specific types of
occupations. By thoroughly investigating the relationship between gender inequality and
rape, Peterson and Bailey (1992) illustrated how complicated the relationship may be and
demonstrated important ways to attempt to uncover more information.
Some scholars have noted that much of the research on gender inequality and
violence against women, including Peterson and Bailey (1992), have acted under the
axiom that amelioration and backlash are contradictory hypotheses (Bailey, 1999;
Whaley, 2001; Whaley et al., 2013). However, these hypotheses may be corollaries of one
another contingent either on time or the level of gender inequality. For instance, Bailey
(1999) and Whaley (2001) suggested and found some evidence that backlash occurs for a
period of time leading to amelioration. According to Bailey (1999), prior to his analysis
no studies had investigated the role of changes in gender inequality and changes in the
rape rate. Indeed, I was unable to locate any studies prior to Bailey (1999) that
investigated the relationship between changes in gender inequality and changes in any
form of violence against women. Bailey (1999) investigated the relationship between
gender inequality (using male and female differences in educational attainment,
occupation type, and income as indicators) and rape rates for central cities. He found
support for the backlash hypothesis between occupation type and rape rates for 1980 and
1990 (cross-sectional models), but found an ameliorative relationship between changes in
income inequality and changes in rape rates from 1980 to 1990 (change model). In the
end, Bailey suggested that although his analysis was an “improvement over previous
investigations this analysis does not fully resolve important feminist arguments regarding
the socioeconomic status of women and rape” (1999, p. 59).
Whaley (2001) presented itself as refining the temporal theoretical relationship
between change in gender inequality and violence against women. She hypothesized that
initially during times of higher gender inequality, women experience lower levels of rape
as a “reward, “but that in the long term, gender inequality prompts cultural changes that
build support for rape, prompting an increase. The reverse was hypothesized as well. To
use the hypotheses already introduced, Whaley (2001) hypothesized that backlash occurs
first and that after a sustained period of change amelioration can occur. Using 5-, 10-, and
20-year lagged indicators, Whaley (2001) was also unable to provide strong support for
her hypothesis; however, most of the significant relationships were in the proposed
direction. For example, cross-sectional analyses suggested that higher levels of inequality
in holding executive occupations were associated with lower levels of rape supporting the
backlash hypothesis. The model using a 5-year lag from 1970 to 1975 suggested backlash
also. Yet in accord with the hypothesis, the model using a 10-year lag between 1980 and
1990 found support for the ameliorative hypotheses. Specifically, decreases in gender
inequality in holding executive occupations were associated with decreases in rape.
Therefore, findings related to holding executive occupations appeared to support backlash
cross-sectionally and amelioration over a 10-year period. However, gender inequality in
executive occupations was not significantly related to rape when looking at change from
1980 to 1985, 1970 to 1980, and 1970 to 1990. Therefore, within this analysis, the results
reflect a complicated relationship.
Additionally, Whaley et al. (2013) suggested and found some evidence that the
nature of the relationship between gender inequality and violence against women may be
nonlinear and related to the level of gender inequality. Specifically, they investigated the
relationship between gender inequality (in terms of educational attainment, median
income, and employment status) and various rates of homicides for 1990 and 2000 from
the SHR (male-female, male-male, female-male, and female-female). Using negative
binomial regressions, Whaley et al. (2013) found that gender equality (represented by a
factor score of the three variables) was positively related to male offender – female victim
homicide rates and that the squared form of gender equality was negatively related in
1990. These significant findings suggest that the relationship between gender equality and
male-female homicide is nonlinear. Gender equality and its squared term were not
significantly related to any other homicide rates except female-offender and
femalevictim. For the female-female homicide rate, gender equality was significantly and
positively related in the model that did not include the robbery rate, which was included
as a “robust control for other factors related to criminal violence” (Whaley et al., 2013, p.
743). In 2000, gender equality and its squared form were significantly related to
malefemale homicide when the robbery rate was not included but only the squared form
was significantly related (negatively) when the robbery rate was included. The same was
true for male-male homicide rates. Just gender equality is significantly and positively
related to the female-female rate when the robbery rate was not included. In all other
instances, gender equality and its squared form were not significantly related to homicide
rates. Whaley et al. (2013) interpreted the effect of adding the robbery rate as suggesting
that unmeasured factors associated with violent crime may be mediating the relationship
or that there may be an issue with multicollinearity. As a whole, they found considerable
support for their hypothesis that backlash processes are likely to function at “low to
intermediate levels” of gender equality and that amelioration occurs when equality is high
for male-female homicide rates (p. 747).
Xie, Heimer, and Lauritsen (2012a) used longitudinal methods to assess whether
there was a relationship between non-fatal intimate partner violence and absolute/relative
indicators of women’s status. They created five five-year pooled periods using MSAlevel
data (which includes the 40 largest MSAs) from the NCVS for 1980 to 2004. To
thoroughly assess their hypotheses, they also disaggregated the non-fatal violence against
women data by victim-offender relationship using the following listed categories:
stranger, intimate partner, and other known. The four hypotheses included amelioration
(which was posited to function more for intimate partner violence than non-intimate
violence), backlash (which was also posited to function more for intimate partner
violence than non-intimate violence), routine activities (which suggests that increases in
absolute status also increases exposure and could lead to an increase in stranger and
known victimization), and protective (which suggests that an increase in absolute status
may increase the accessibility of protective mechanisms and decrease victimization).
Absolute women’s status was operationalized by the percent of women who were
employed, percent of voting age women who voted in the November presidential and
congressional elections, and an index of income-educational attainment, which included
median income and percent who completed four years of college or more. Relative status
was measured using the rate of women’s labor force participation minus that of men, the
level of income-educational attainment for women minus that of men, and the rate of
voter turnout for women minus that of men. Their results found support for each of the
hypotheses except for amelioration. They found a negative relationship between women’s
absolute economic and political status and intimate partner violence which supports the
protective hypothesis. They also found the negative relationship between inequality and
intimate partner victimization which supports the backlash hypothesis. The routine
activities hypothesis was supported through a positive relationship between absolute
indicators of labor force participation and victimization by strangers and non-intimate
persons known to the victim. This longitudinal analysis does not suggest that backlash,
amelioration, protective, nor routine activities hypotheses are corollary by explaining the
relationship between inequality, absolute status, and violence against women at different
periods. However, they found evidence that the nature of the relationship is contingent on
the victim-offender relationship.
The hypotheses and findings suggesting that backlash and amelioration are
corollary are particularly important considering that much of the empirical quantitative
literature on gender inequality has been cross-sectional and assumed a linear relationship.
By design much of the extant literature is unable to address these possibilities (Whaley et
al., 2013). The evidence supporting corollary processes indicate that future analyses
should continue to consider change over time and the structural form of the relationship
as the proposed dissertation will do. In addition, Xie et al. (2012a) presented evidence
that the victim-offender relationship is important to also consider in future analyses.
In addition, while not an inequality hypothesis, several scholars have investigated the role
of absolute status on trends of violence against women. Changes in absolute status are
increases or decreases in the status of a group with no consideration or comparison to
other groups. It is possible that there is a relationship between absolute status and
violence against women and that any findings suggesting a relationship
When race-disaggregated trends of violence against women are investigated,
major gaps in the crime decline, race, and gender literature are exposed. For example,
Lauritsen (2012) illustrated that although the rape and sexual assault trends for Latinas,
Black women, and white women have been similar for the last 40 years, the gaps between
them have varied occasionally. The variation in the gap could indicate different processes
at work. These types of findings are important in that they denote the importance of
attending to gender, race, and ethnicity but are often overlooked in later studies.
Ultimately, the crime decline literature has at times denoted the varying magnitude of the
decline by race and gender subgroups, but has not produced many intersectional
approaches to explaining the disparate decline (Heimer & Lauritsen, 2008; Lauritsen,
2012; Lauritsen & Heimer, 2010; Levitt, 2004; Parker, 2008). As a consequence, the
disparate declines and trends by race, ethnicity, and gender are not well understood. The
race and inequality literature tends to lack a gender-lens, a framework denoting the role
that gender plays in how things are enacted and perceived, while at the same time the
gender and inequality literature tends to lack a race-lens, a framework denoting the role
of race (McCall, 2005). This study argues that it is important to consider violence against
women from a gender x race-lens or an intersectional-lens.
HYPOTHESES AND EMPIRICAL SUPPORT
Before discussing the ways to apply an intersectionality perspective to the gender
inequality hypothesis, it is pertinent to review the various hypotheses that exist within the
literature. As noted in the introduction, the nature of the relationship between gender
inequality and violence against women has not been clearly defined within the empirical
literature (Whaley et al., 2013). In fact, evidence exists for multiple hypotheses. The
original hypothesis and the one many organizations work under suggests that as gender
inequality decreases and the genders become more equal violence against women will
decrease (Whaley et al., 2013). This is referred to as the amelioration hypothesis.
Peterson and Bailey (1992) found some support for the amelioration hypothesis. They
investigated the relationship between various forms of inequality (general, racial, and
gender) and rape rates for United States metropolitan areas in 1980. The rape rates were
obtained from the UCR in 1980. General inequality was measured using the GINI index
which represents income inequality in the total population. Racial inequality was
measured by the difference in median income for Black and white families. Gender
inequality was measured in terms of education using the ratio of men to women who
completed 4 years of high school, who completed 4 years of college, and who completed
5 or more years of college. It was also measured in terms of income using a ratio of men
to women’s median income for all people, people who completed 4 years of high school,
and people who completed 4 years or more of college. Gender inequality was also
measured by the percent of people employed in managerial and professional positions
who were women and by a ratio of men to women who live in poverty. Multiple models
were estimated with most containing one form of gender inequality and one containing a
representative of each type. The GINI index was found to be positively and significantly
related to rape rapes in each model. This finding suggests that places with greater general
income inequality exhibited higher rape rates. Racial inequality in income was significant
and positive in the models that included gender inequality by ratio of high school
graduates only and gender inequality in percentage of women in professional occupations
only. When all forms of gender inequality were included (income, high school graduation,
and profession occupations), gender inequality in median income was significantly and
positively related to rape rates. This finding suggests that where gender inequality in
income is greater, rape rates will also be higher and this is consistent with the
amelioration hypothesis.
Another hypothesis, the backlash hypothesis, suggests an opposite relationship.
Specifically, the backlash hypothesis suggests that as gender inequality decreases and the
genders become more equal, violence against women will increase (Whaley et al., 2013).
It is important to note that the term “backlash” is sometimes used to imply a specific
mechanism causing the increase of violence. For instance, sometimes backlash is defined
as a process by which men (or other dominant group) react violently to challenges to their
dominant status. In this study, as with many past studies, the mechanism of the
relationship between gender inequality and violence against women is not being
evaluated. Rather this study examines the relationship between gender inequality and
violence against women which therefore only allows for some speculation as to the
mechanism of the relationship. Backlash, herein, is used to represent a negative
relationship between changes in gender inequality and changes in violence against
women.
Within the same article described above, Peterson and Bailey (1992) also found
support for a backlash hypothesis when further investigating the relationship between
gender inequality in occupations and rape rates. Specifically, they separated professional
occupations into 1) executives, administrators, and managers 2) public officials and
administrators 3) engineers and natural scientists 4) health diagnostic occupations 5)
elementary and secondary teachers. They found that where the percent of executives,
administrators, and managers that are women were higher rape rates were also higher
suggesting backlash. However, they also found that where the percent of health diagnostic
occupations that are women were higher rape rates were lower. This was true of
elementary and secondary teachers as well. Since being an elementary or secondary
teacher is a women-typed occupation, the finding of a negative relationship could be
interpreted as backlash as well. The authors suggested that “the more that women remain
in traditional ‘female’ occupations, the greater their ‘reward’ – freedom from rape” (p.
172). However, the significant finding for health diagnostic occupations appears to
support amelioration, since health diagnostic occupations are traditionally men-typed
occupations.
The Peterson and Bailey (1992) study is illustrative for three primary reasons.
First, the analysis assesses the relationship between gender inequality and rape in
multiple ways. They use several forms of inequality including: education, income, and
occupation. In addition, they estimate models with each form individually as well as a
model with all of the forms. Second, this study is illustrative because it finds support for
both the ameliorative and backlash hypotheses. Due to these mixed findings, the authors
conclude that their “analysis by no means settles the issue of rape and the structural
position of women vis-à-vis men” (p. 174). Lastly, this study thoroughly investigated
different elements of inequality. For example, as discussed above, they investigated more
specific occupation categories to gain an understanding of the relationship between
gender parity in occupations and rape rates. Ultimately, together these three elements
exemplify how complicated the relationship between gender inequality and rape may be.
Specifically, it is possible that common indicators of gender disparity do not capture the
most important aspects of gender inequality. This could be one explanation for why some
measures of gender inequality were significantly related to rape and others were not. For
instance, inequality in educational attainment was not related to rape rates in any of the
Peterson and Bailey (1992) models. Through this study, it was also demonstrated that
some of the measures of gender inequality may be confounding other relationships. In
particular, Peterson and Bailey (1992) found that although in the main model there was
no significant relationship between gender inequality in occupation type and rape rates,
when looking at specific types of occupations the findings were mixed. Indeed, they
found support for backlash and amelioration when looking at specific types of
occupations. By thoroughly investigating the relationship between gender inequality and
rape, Peterson and Bailey (1992) illustrated how complicated the relationship may be and
demonstrated important ways to attempt to uncover more information.
Some scholars have noted that much of the research on gender inequality and
violence against women, including Peterson and Bailey (1992), have acted under the
axiom that amelioration and backlash are contradictory hypotheses (Bailey, 1999;
Whaley, 2001; Whaley et al., 2013). However, these hypotheses may be corollaries of one
another contingent either on time or the level of gender inequality. For instance, Bailey
(1999) and Whaley (2001) suggested and found some evidence that backlash occurs for a
period of time leading to amelioration. According to Bailey (1999), prior to his analysis
no studies had investigated the role of changes in gender inequality and changes in the
rape rate. Indeed, I was unable to locate any studies prior to Bailey (1999) that
investigated the relationship between changes in gender inequality and changes in any
form of violence against women. Bailey (1999) investigated the relationship between
gender inequality (using male and female differences in educational attainment,
occupation type, and income as indicators) and rape rates for central cities. He found
support for the backlash hypothesis between occupation type and rape rates for 1980 and
1990 (cross-sectional models), but found an ameliorative relationship between changes in
income inequality and changes in rape rates from 1980 to 1990 (change model). In the
end, Bailey suggested that although his analysis was an “improvement over previous
investigations this analysis does not fully resolve important feminist arguments regarding
the socioeconomic status of women and rape” (1999, p. 59).
Whaley (2001) presented itself as refining the temporal theoretical relationship
between change in gender inequality and violence against women. She hypothesized that
initially during times of higher gender inequality, women experience lower levels of rape
as a “reward, “but that in the long term, gender inequality prompts cultural changes that
build support for rape, prompting an increase. The reverse was hypothesized as well. To
use the hypotheses already introduced, Whaley (2001) hypothesized that backlash occurs
first and that after a sustained period of change amelioration can occur. Using 5-, 10-, and
20-year lagged indicators, Whaley (2001) was also unable to provide strong support for
her hypothesis; however, most of the significant relationships were in the proposed
direction. For example, cross-sectional analyses suggested that higher levels of inequality
in holding executive occupations were associated with lower levels of rape supporting the
backlash hypothesis. The model using a 5-year lag from 1970 to 1975 suggested backlash
also. Yet in accord with the hypothesis, the model using a 10-year lag between 1980 and
1990 found support for the ameliorative hypotheses. Specifically, decreases in gender
inequality in holding executive occupations were associated with decreases in rape.
Therefore, findings related to holding executive occupations appeared to support backlash
cross-sectionally and amelioration over a 10-year period. However, gender inequality in
executive occupations was not significantly related to rape when looking at change from
1980 to 1985, 1970 to 1980, and 1970 to 1990. Therefore, within this analysis, the results
reflect a complicated relationship.
Additionally, Whaley et al. (2013) suggested and found some evidence that the
nature of the relationship between gender inequality and violence against women may be
nonlinear and related to the level of gender inequality. Specifically, they investigated the
relationship between gender inequality (in terms of educational attainment, median
income, and employment status) and various rates of homicides for 1990 and 2000 from
the SHR (male-female, male-male, female-male, and female-female). Using negative
binomial regressions, Whaley et al. (2013) found that gender equality (represented by a
factor score of the three variables) was positively related to male offender – female victim
homicide rates and that the squared form of gender equality was negatively related in
1990. These significant findings suggest that the relationship between gender equality and
male-female homicide is nonlinear. Gender equality and its squared term were not
significantly related to any other homicide rates except female-offender and
femalevictim. For the female-female homicide rate, gender equality was significantly and
positively related in the model that did not include the robbery rate, which was included
as a “robust control for other factors related to criminal violence” (Whaley et al., 2013, p.
743). In 2000, gender equality and its squared form were significantly related to
malefemale homicide when the robbery rate was not included but only the squared form
was significantly related (negatively) when the robbery rate was included. The same was
true for male-male homicide rates. Just gender equality is significantly and positively
related to the female-female rate when the robbery rate was not included. In all other
instances, gender equality and its squared form were not significantly related to homicide
rates. Whaley et al. (2013) interpreted the effect of adding the robbery rate as suggesting
that unmeasured factors associated with violent crime may be mediating the relationship
or that there may be an issue with multicollinearity. As a whole, they found considerable
support for their hypothesis that backlash processes are likely to function at “low to
intermediate levels” of gender equality and that amelioration occurs when equality is high
for male-female homicide rates (p. 747).
Xie, Heimer, and Lauritsen (2012a) used longitudinal methods to assess whether
there was a relationship between non-fatal intimate partner violence and absolute/relative
indicators of women’s status. They created five five-year pooled periods using MSAlevel
data (which includes the 40 largest MSAs) from the NCVS for 1980 to 2004. To
thoroughly assess their hypotheses, they also disaggregated the non-fatal violence against
women data by victim-offender relationship using the following listed categories:
stranger, intimate partner, and other known. The four hypotheses included amelioration
(which was posited to function more for intimate partner violence than non-intimate
violence), backlash (which was also posited to function more for intimate partner
violence than non-intimate violence), routine activities (which suggests that increases in
absolute status also increases exposure and could lead to an increase in stranger and
known victimization), and protective (which suggests that an increase in absolute status
may increase the accessibility of protective mechanisms and decrease victimization).
Absolute women’s status was operationalized by the percent of women who were
employed, percent of voting age women who voted in the November presidential and
congressional elections, and an index of income-educational attainment, which included
median income and percent who completed four years of college or more. Relative status
was measured using the rate of women’s labor force participation minus that of men, the
level of income-educational attainment for women minus that of men, and the rate of
voter turnout for women minus that of men. Their results found support for each of the
hypotheses except for amelioration. They found a negative relationship between women’s
absolute economic and political status and intimate partner violence which supports the
protective hypothesis. They also found the negative relationship between inequality and
intimate partner victimization which supports the backlash hypothesis. The routine
activities hypothesis was supported through a positive relationship between absolute
indicators of labor force participation and victimization by strangers and non-intimate
persons known to the victim. This longitudinal analysis does not suggest that backlash,
amelioration, protective, nor routine activities hypotheses are corollary by explaining the
relationship between inequality, absolute status, and violence against women at different
periods. However, they found evidence that the nature of the relationship is contingent on
the victim-offender relationship.
The hypotheses and findings suggesting that backlash and amelioration are
corollary are particularly important considering that much of the empirical quantitative
literature on gender inequality has been cross-sectional and assumed a linear relationship.
By design much of the extant literature is unable to address these possibilities (Whaley et
al., 2013). The evidence supporting corollary processes indicate that future analyses
should continue to consider change over time and the structural form of the relationship
as the proposed dissertation will do. In addition, Xie et al. (2012a) presented evidence
that the victim-offender relationship is important to also consider in future analyses.
In addition, while not an inequality hypothesis, several scholars have investigated the role
of absolute status on trends of violence against women. Changes in absolute status are
increases or decreases in the status of a group with no consideration or comparison to
other groups. It is possible that there is a relationship between absolute status and
violence against women and that any findings suggesting a relationship
When race-disaggregated trends of violence against women are investigated,
major gaps in the crime decline, race, and gender literature are exposed. For example,
Lauritsen (2012) illustrated that although the rape and sexual assault trends for Latinas,
Black women, and white women have been similar for the last 40 years, the gaps between
them have varied occasionally. The variation in the gap could indicate different processes
at work. These types of findings are important in that they denote the importance of
attending to gender, race, and ethnicity but are often overlooked in later studies.
Ultimately, the crime decline literature has at times denoted the varying magnitude of the
decline by race and gender subgroups, but has not produced many intersectional
approaches to explaining the disparate decline (Heimer & Lauritsen, 2008; Lauritsen,
2012; Lauritsen & Heimer, 2010; Levitt, 2004; Parker, 2008). As a consequence, the
disparate declines and trends by race, ethnicity, and gender are not well understood. The
race and inequality literature tends to lack a gender-lens, a framework denoting the role
that gender plays in how things are enacted and perceived, while at the same time the
gender and inequality literature tends to lack a race-lens, a framework denoting the role
of race (McCall, 2005). This study argues that it is important to consider violence against
women from a gender x race-lens or an intersectional-lens.
HYPOTHESES AND EMPIRICAL SUPPORT
Before discussing the ways to apply an intersectionality perspective to the gender
inequality hypothesis, it is pertinent to review the various hypotheses that exist within the
literature. As noted in the introduction, the nature of the relationship between gender
inequality and violence against women has not been clearly defined within the empirical
literature (Whaley et al., 2013). In fact, evidence exists for multiple hypotheses. The
original hypothesis and the one many organizations work under suggests that as gender
inequality decreases and the genders become more equal violence against women will
decrease (Whaley et al., 2013). This is referred to as the amelioration hypothesis.
Peterson and Bailey (1992) found some support for the amelioration hypothesis. They
investigated the relationship between various forms of inequality (general, racial, and
gender) and rape rates for United States metropolitan areas in 1980. The rape rates were
obtained from the UCR in 1980. General inequality was measured using the GINI index
which represents income inequality in the total population. Racial inequality was
measured by the difference in median income for Black and white families. Gender
inequality was measured in terms of education using the ratio of men to women who
completed 4 years of high school, who completed 4 years of college, and who completed
5 or more years of college. It was also measured in terms of income using a ratio of men
to women’s median income for all people, people who completed 4 years of high school,
and people who completed 4 years or more of college. Gender inequality was also
measured by the percent of people employed in managerial and professional positions
who were women and by a ratio of men to women who live in poverty. Multiple models
were estimated with most containing one form of gender inequality and one containing a
representative of each type. The GINI index was found to be positively and significantly
related to rape rapes in each model. This finding suggests that places with greater general
income inequality exhibited higher rape rates. Racial inequality in income was significant
and positive in the models that included gender inequality by ratio of high school
graduates only and gender inequality in percentage of women in professional occupations
only. When all forms of gender inequality were included (income, high school graduation,
and profession occupations), gender inequality in median income was significantly and
positively related to rape rates. This finding suggests that where gender inequality in
income is greater, rape rates will also be higher and this is consistent with the
amelioration hypothesis.
Another hypothesis, the backlash hypothesis, suggests an opposite relationship.
Specifically, the backlash hypothesis suggests that as gender inequality decreases and the
genders become more equal, violence against women will increase (Whaley et al., 2013).
It is important to note that the term “backlash” is sometimes used to imply a specific
mechanism causing the increase of violence. For instance, sometimes backlash is defined
as a process by which men (or other dominant group) react violently to challenges to their
dominant status. In this study, as with many past studies, the mechanism of the
relationship between gender inequality and violence against women is not being
evaluated. Rather this study examines the relationship between gender inequality and
violence against women which therefore only allows for some speculation as to the
mechanism of the relationship. Backlash, herein, is used to represent a negative
relationship between changes in gender inequality and changes in violence against
women.
Within the same article described above, Peterson and Bailey (1992) also found
support for a backlash hypothesis when further investigating the relationship between
gender inequality in occupations and rape rates. Specifically, they separated professional
occupations into 1) executives, administrators, and managers 2) public officials and
administrators 3) engineers and natural scientists 4) health diagnostic occupations 5)
elementary and secondary teachers. They found that where the percent of executives,
administrators, and managers that are women were higher rape rates were also higher
suggesting backlash. However, they also found that where the percent of health diagnostic
occupations that are women were higher rape rates were lower. This was true of
elementary and secondary teachers as well. Since being an elementary or secondary
teacher is a women-typed occupation, the finding of a negative relationship could be
interpreted as backlash as well. The authors suggested that “the more that women remain
in traditional ‘female’ occupations, the greater their ‘reward’ – freedom from rape” (p.
172). However, the significant finding for health diagnostic occupations appears to
support amelioration, since health diagnostic occupations are traditionally men-typed
occupations.
The Peterson and Bailey (1992) study is illustrative for three primary reasons.
First, the analysis assesses the relationship between gender inequality and rape in
multiple ways. They use several forms of inequality including: education, income, and
occupation. In addition, they estimate models with each form individually as well as a
model with all of the forms. Second, this study is illustrative because it finds support for
both the ameliorative and backlash hypotheses. Due to these mixed findings, the authors
conclude that their “analysis by no means settles the issue of rape and the structural
position of women vis-à-vis men” (p. 174). Lastly, this study thoroughly investigated
different elements of inequality. For example, as discussed above, they investigated more
specific occupation categories to gain an understanding of the relationship between
gender parity in occupations and rape rates. Ultimately, together these three elements
exemplify how complicated the relationship between gender inequality and rape may be.
Specifically, it is possible that common indicators of gender disparity do not capture the
most important aspects of gender inequality. This could be one explanation for why some
measures of gender inequality were significantly related to rape and others were not. For
instance, inequality in educational attainment was not related to rape rates in any of the
Peterson and Bailey (1992) models. Through this study, it was also demonstrated that
some of the measures of gender inequality may be confounding other relationships. In
particular, Peterson and Bailey (1992) found that although in the main model there was
no significant relationship between gender inequality in occupation type and rape rates,
when looking at specific types of occupations the findings were mixed. Indeed, they
found support for backlash and amelioration when looking at specific types of
occupations. By thoroughly investigating the relationship between gender inequality and
rape, Peterson and Bailey (1992) illustrated how complicated the relationship may be and
demonstrated important ways to attempt to uncover more information.
Some scholars have noted that much of the research on gender inequality and
violence against women, including Peterson and Bailey (1992), have acted under the
axiom that amelioration and backlash are contradictory hypotheses (Bailey, 1999;
Whaley, 2001; Whaley et al., 2013). However, these hypotheses may be corollaries of one
another contingent either on time or the level of gender inequality. For instance, Bailey
(1999) and Whaley (2001) suggested and found some evidence that backlash occurs for a
period of time leading to amelioration. According to Bailey (1999), prior to his analysis
no studies had investigated the role of changes in gender inequality and changes in the
rape rate. Indeed, I was unable to locate any studies prior to Bailey (1999) that
investigated the relationship between changes in gender inequality and changes in any
form of violence against women. Bailey (1999) investigated the relationship between
gender inequality (using male and female differences in educational attainment,
occupation type, and income as indicators) and rape rates for central cities. He found
support for the backlash hypothesis between occupation type and rape rates for 1980 and
1990 (cross-sectional models), but found an ameliorative relationship between changes in
income inequality and changes in rape rates from 1980 to 1990 (change model). In the
end, Bailey suggested that although his analysis was an “improvement over previous
investigations this analysis does not fully resolve important feminist arguments regarding
the socioeconomic status of women and rape” (1999, p. 59).
Whaley (2001) presented itself as refining the temporal theoretical relationship
between change in gender inequality and violence against women. She hypothesized that
initially during times of higher gender inequality, women experience lower levels of rape
as a “reward, “but that in the long term, gender inequality prompts cultural changes that
build support for rape, prompting an increase. The reverse was hypothesized as well. To
use the hypotheses already introduced, Whaley (2001) hypothesized that backlash occurs
first and that after a sustained period of change amelioration can occur. Using 5-, 10-, and
20-year lagged indicators, Whaley (2001) was also unable to provide strong support for
her hypothesis; however, most of the significant relationships were in the proposed
direction. For example, cross-sectional analyses suggested that higher levels of inequality
in holding executive occupations were associated with lower levels of rape supporting the
backlash hypothesis. The model using a 5-year lag from 1970 to 1975 suggested backlash
also. Yet in accord with the hypothesis, the model using a 10-year lag between 1980 and
1990 found support for the ameliorative hypotheses. Specifically, decreases in gender
inequality in holding executive occupations were associated with decreases in rape.
Therefore, findings related to holding executive occupations appeared to support backlash
cross-sectionally and amelioration over a 10-year period. However, gender inequality in
executive occupations was not significantly related to rape when looking at change from
1980 to 1985, 1970 to 1980, and 1970 to 1990. Therefore, within this analysis, the results
reflect a complicated relationship.
Additionally, Whaley et al. (2013) suggested and found some evidence that the
nature of the relationship between gender inequality and violence against women may be
nonlinear and related to the level of gender inequality. Specifically, they investigated the
relationship between gender inequality (in terms of educational attainment, median
income, and employment status) and various rates of homicides for 1990 and 2000 from
the SHR (male-female, male-male, female-male, and female-female). Using negative
binomial regressions, Whaley et al. (2013) found that gender equality (represented by a
factor score of the three variables) was positively related to male offender – female victim
homicide rates and that the squared form of gender equality was negatively related in
1990. These significant findings suggest that the relationship between gender equality and
male-female homicide is nonlinear. Gender equality and its squared term were not
significantly related to any other homicide rates except female-offender and
femalevictim. For the female-female homicide rate, gender equality was significantly and
positively related in the model that did not include the robbery rate, which was included
as a “robust control for other factors related to criminal violence” (Whaley et al., 2013, p.
743). In 2000, gender equality and its squared form were significantly related to
malefemale homicide when the robbery rate was not included but only the squared form
was significantly related (negatively) when the robbery rate was included. The same was
true for male-male homicide rates. Just gender equality is significantly and positively
related to the female-female rate when the robbery rate was not included. In all other
instances, gender equality and its squared form were not significantly related to homicide
rates. Whaley et al. (2013) interpreted the effect of adding the robbery rate as suggesting
that unmeasured factors associated with violent crime may be mediating the relationship
or that there may be an issue with multicollinearity. As a whole, they found considerable
support for their hypothesis that backlash processes are likely to function at “low to
intermediate levels” of gender equality and that amelioration occurs when equality is high
for male-female homicide rates (p. 747).
Xie, Heimer, and Lauritsen (2012a) used longitudinal methods to assess whether
there was a relationship between non-fatal intimate partner violence and absolute/relative
indicators of women’s status. They created five five-year pooled periods using MSAlevel
data (which includes the 40 largest MSAs) from the NCVS for 1980 to 2004. To
thoroughly assess their hypotheses, they also disaggregated the non-fatal violence against
women data by victim-offender relationship using the following listed categories:
stranger, intimate partner, and other known. The four hypotheses included amelioration
(which was posited to function more for intimate partner violence than non-intimate
violence), backlash (which was also posited to function more for intimate partner
violence than non-intimate violence), routine activities (which suggests that increases in
absolute status also increases exposure and could lead to an increase in stranger and
known victimization), and protective (which suggests that an increase in absolute status
may increase the accessibility of protective mechanisms and decrease victimization).
Absolute women’s status was operationalized by the percent of women who were
employed, percent of voting age women who voted in the November presidential and
congressional elections, and an index of income-educational attainment, which included
median income and percent who completed four years of college or more. Relative status
was measured using the rate of women’s labor force participation minus that of men, the
level of income-educational attainment for women minus that of men, and the rate of
voter turnout for women minus that of men. Their results found support for each of the
hypotheses except for amelioration. They found a negative relationship between women’s
absolute economic and political status and intimate partner violence which supports the
protective hypothesis. They also found the negative relationship between inequality and
intimate partner victimization which supports the backlash hypothesis. The routine
activities hypothesis was supported through a positive relationship between absolute
indicators of labor force participation and victimization by strangers and non-intimate
persons known to the victim. This longitudinal analysis does not suggest that backlash,
amelioration, protective, nor routine activities hypotheses are corollary by explaining the
relationship between inequality, absolute status, and violence against women at different
periods. However, they found evidence that the nature of the relationship is contingent on
the victim-offender relationship.
The hypotheses and findings suggesting that backlash and amelioration are
corollary are particularly important considering that much of the empirical quantitative
literature on gender inequality has been cross-sectional and assumed a linear relationship.
By design much of the extant literature is unable to address these possibilities (Whaley et
al., 2013). The evidence supporting corollary processes indicate that future analyses
should continue to consider change over time and the structural form of the relationship
as the proposed dissertation will do. In addition, Xie et al. (2012a) presented evidence
that the victim-offender relationship is important to also consider in future analyses.
In addition, while not an inequality hypothesis, several scholars have investigated the role
of absolute status on trends of violence against women. Changes in absolute status are
increases or decreases in the status of a group with no consideration or comparison to
other groups. It is possible that there is a relationship between absolute status and
violence against women and that any findings suggesting a relationship
When race-disaggregated trends of violence against women are investigated,
major gaps in the crime decline, race, and gender literature are exposed. For example,
Lauritsen (2012) illustrated that although the rape and sexual assault trends for Latinas,
Black women, and white women have been similar for the last 40 years, the gaps between
them have varied occasionally. The variation in the gap could indicate different processes
at work. These types of findings are important in that they denote the importance of
attending to gender, race, and ethnicity but are often overlooked in later studies.
Ultimately, the crime decline literature has at times denoted the varying magnitude of the
decline by race and gender subgroups, but has not produced many intersectional
approaches to explaining the disparate decline (Heimer & Lauritsen, 2008; Lauritsen,
2012; Lauritsen & Heimer, 2010; Levitt, 2004; Parker, 2008). As a consequence, the
disparate declines and trends by race, ethnicity, and gender are not well understood. The
race and inequality literature tends to lack a gender-lens, a framework denoting the role
that gender plays in how things are enacted and perceived, while at the same time the
gender and inequality literature tends to lack a race-lens, a framework denoting the role
of race (McCall, 2005). This study argues that it is important to consider violence against
women from a gender x race-lens or an intersectional-lens.
HYPOTHESES AND EMPIRICAL SUPPORT
Before discussing the ways to apply an intersectionality perspective to the gender
inequality hypothesis, it is pertinent to review the various hypotheses that exist within the
literature. As noted in the introduction, the nature of the relationship between gender
inequality and violence against women has not been clearly defined within the empirical
literature (Whaley et al., 2013). In fact, evidence exists for multiple hypotheses. The
original hypothesis and the one many organizations work under suggests that as gender
inequality decreases and the genders become more equal violence against women will
decrease (Whaley et al., 2013). This is referred to as the amelioration hypothesis.
Peterson and Bailey (1992) found some support for the amelioration hypothesis. They
investigated the relationship between various forms of inequality (general, racial, and
gender) and rape rates for United States metropolitan areas in 1980. The rape rates were
obtained from the UCR in 1980. General inequality was measured using the GINI index
which represents income inequality in the total population. Racial inequality was
measured by the difference in median income for Black and white families. Gender
inequality was measured in terms of education using the ratio of men to women who
completed 4 years of high school, who completed 4 years of college, and who completed
5 or more years of college. It was also measured in terms of income using a ratio of men
to women’s median income for all people, people who completed 4 years of high school,
and people who completed 4 years or more of college. Gender inequality was also
measured by the percent of people employed in managerial and professional positions
who were women and by a ratio of men to women who live in poverty. Multiple models
were estimated with most containing one form of gender inequality and one containing a
representative of each type. The GINI index was found to be positively and significantly
related to rape rapes in each model. This finding suggests that places with greater general
income inequality exhibited higher rape rates. Racial inequality in income was significant
and positive in the models that included gender inequality by ratio of high school
graduates only and gender inequality in percentage of women in professional occupations
only. When all forms of gender inequality were included (income, high school graduation,
and profession occupations), gender inequality in median income was significantly and
positively related to rape rates. This finding suggests that where gender inequality in
income is greater, rape rates will also be higher and this is consistent with the
amelioration hypothesis.
Another hypothesis, the backlash hypothesis, suggests an opposite relationship.
Specifically, the backlash hypothesis suggests that as gender inequality decreases and the
genders become more equal, violence against women will increase (Whaley et al., 2013).
It is important to note that the term “backlash” is sometimes used to imply a specific
mechanism causing the increase of violence. For instance, sometimes backlash is defined
as a process by which men (or other dominant group) react violently to challenges to their
dominant status. In this study, as with many past studies, the mechanism of the
relationship between gender inequality and violence against women is not being
evaluated. Rather this study examines the relationship between gender inequality and
violence against women which therefore only allows for some speculation as to the
mechanism of the relationship. Backlash, herein, is used to represent a negative
relationship between changes in gender inequality and changes in violence against
women.
Within the same article described above, Peterson and Bailey (1992) also found
support for a backlash hypothesis when further investigating the relationship between
gender inequality in occupations and rape rates. Specifically, they separated professional
occupations into 1) executives, administrators, and managers 2) public officials and
administrators 3) engineers and natural scientists 4) health diagnostic occupations 5)
elementary and secondary teachers. They found that where the percent of executives,
administrators, and managers that are women were higher rape rates were also higher
suggesting backlash. However, they also found that where the percent of health diagnostic
occupations that are women were higher rape rates were lower. This was true of
elementary and secondary teachers as well. Since being an elementary or secondary
teacher is a women-typed occupation, the finding of a negative relationship could be
interpreted as backlash as well. The authors suggested that “the more that women remain
in traditional ‘female’ occupations, the greater their ‘reward’ – freedom from rape” (p.
172). However, the significant finding for health diagnostic occupations appears to
support amelioration, since health diagnostic occupations are traditionally men-typed
occupations.
The Peterson and Bailey (1992) study is illustrative for three primary reasons.
First, the analysis assesses the relationship between gender inequality and rape in
multiple ways. They use several forms of inequality including: education, income, and
occupation. In addition, they estimate models with each form individually as well as a
model with all of the forms. Second, this study is illustrative because it finds support for
both the ameliorative and backlash hypotheses. Due to these mixed findings, the authors
conclude that their “analysis by no means settles the issue of rape and the structural
position of women vis-à-vis men” (p. 174). Lastly, this study thoroughly investigated
different elements of inequality. For example, as discussed above, they investigated more
specific occupation categories to gain an understanding of the relationship between
gender parity in occupations and rape rates. Ultimately, together these three elements
exemplify how complicated the relationship between gender inequality and rape may be.
Specifically, it is possible that common indicators of gender disparity do not capture the
most important aspects of gender inequality. This could be one explanation for why some
measures of gender inequality were significantly related to rape and others were not. For
instance, inequality in educational attainment was not related to rape rates in any of the
Peterson and Bailey (1992) models. Through this study, it was also demonstrated that
some of the measures of gender inequality may be confounding other relationships. In
particular, Peterson and Bailey (1992) found that although in the main model there was
no significant relationship between gender inequality in occupation type and rape rates,
when looking at specific types of occupations the findings were mixed. Indeed, they
found support for backlash and amelioration when looking at specific types of
occupations. By thoroughly investigating the relationship between gender inequality and
rape, Peterson and Bailey (1992) illustrated how complicated the relationship may be and
demonstrated important ways to attempt to uncover more information.
Some scholars have noted that much of the research on gender inequality and
violence against women, including Peterson and Bailey (1992), have acted under the
axiom that amelioration and backlash are contradictory hypotheses (Bailey, 1999;
Whaley, 2001; Whaley et al., 2013). However, these hypotheses may be corollaries of one
another contingent either on time or the level of gender inequality. For instance, Bailey
(1999) and Whaley (2001) suggested and found some evidence that backlash occurs for a
period of time leading to amelioration. According to Bailey (1999), prior to his analysis
no studies had investigated the role of changes in gender inequality and changes in the
rape rate. Indeed, I was unable to locate any studies prior to Bailey (1999) that
investigated the relationship between changes in gender inequality and changes in any
form of violence against women. Bailey (1999) investigated the relationship between
gender inequality (using male and female differences in educational attainment,
occupation type, and income as indicators) and rape rates for central cities. He found
support for the backlash hypothesis between occupation type and rape rates for 1980 and
1990 (cross-sectional models), but found an ameliorative relationship between changes in
income inequality and changes in rape rates from 1980 to 1990 (change model). In the
end, Bailey suggested that although his analysis was an “improvement over previous
investigations this analysis does not fully resolve important feminist arguments regarding
the socioeconomic status of women and rape” (1999, p. 59).
Whaley (2001) presented itself as refining the temporal theoretical relationship
between change in gender inequality and violence against women. She hypothesized that
initially during times of higher gender inequality, women experience lower levels of rape
as a “reward, “but that in the long term, gender inequality prompts cultural changes that
build support for rape, prompting an increase. The reverse was hypothesized as well. To
use the hypotheses already introduced, Whaley (2001) hypothesized that backlash occurs
first and that after a sustained period of change amelioration can occur. Using 5-, 10-, and
20-year lagged indicators, Whaley (2001) was also unable to provide strong support for
her hypothesis; however, most of the significant relationships were in the proposed
direction. For example, cross-sectional analyses suggested that higher levels of inequality
in holding executive occupations were associated with lower levels of rape supporting the
backlash hypothesis. The model using a 5-year lag from 1970 to 1975 suggested backlash
also. Yet in accord with the hypothesis, the model using a 10-year lag between 1980 and
1990 found support for the ameliorative hypotheses. Specifically, decreases in gender
inequality in holding executive occupations were associated with decreases in rape.
Therefore, findings related to holding executive occupations appeared to support backlash
cross-sectionally and amelioration over a 10-year period. However, gender inequality in
executive occupations was not significantly related to rape when looking at change from
1980 to 1985, 1970 to 1980, and 1970 to 1990. Therefore, within this analysis, the results
reflect a complicated relationship.
Additionally, Whaley et al. (2013) suggested and found some evidence that the
nature of the relationship between gender inequality and violence against women may be
nonlinear and related to the level of gender inequality. Specifically, they investigated the
relationship between gender inequality (in terms of educational attainment, median
income, and employment status) and various rates of homicides for 1990 and 2000 from
the SHR (male-female, male-male, female-male, and female-female). Using negative
binomial regressions, Whaley et al. (2013) found that gender equality (represented by a
factor score of the three variables) was positively related to male offender – female victim
homicide rates and that the squared form of gender equality was negatively related in
1990. These significant findings suggest that the relationship between gender equality and
male-female homicide is nonlinear. Gender equality and its squared term were not
significantly related to any other homicide rates except female-offender and
femalevictim. For the female-female homicide rate, gender equality was significantly and
positively related in the model that did not include the robbery rate, which was included
as a “robust control for other factors related to criminal violence” (Whaley et al., 2013, p.
743). In 2000, gender equality and its squared form were significantly related to
malefemale homicide when the robbery rate was not included but only the squared form
was significantly related (negatively) when the robbery rate was included. The same was
true for male-male homicide rates. Just gender equality is significantly and positively
related to the female-female rate when the robbery rate was not included. In all other
instances, gender equality and its squared form were not significantly related to homicide
rates. Whaley et al. (2013) interpreted the effect of adding the robbery rate as suggesting
that unmeasured factors associated with violent crime may be mediating the relationship
or that there may be an issue with multicollinearity. As a whole, they found considerable
support for their hypothesis that backlash processes are likely to function at “low to
intermediate levels” of gender equality and that amelioration occurs when equality is high
for male-female homicide rates (p. 747).
Xie, Heimer, and Lauritsen (2012a) used longitudinal methods to assess whether
there was a relationship between non-fatal intimate partner violence and absolute/relative
indicators of women’s status. They created five five-year pooled periods using MSAlevel
data (which includes the 40 largest MSAs) from the NCVS for 1980 to 2004. To
thoroughly assess their hypotheses, they also disaggregated the non-fatal violence against
women data by victim-offender relationship using the following listed categories:
stranger, intimate partner, and other known. The four hypotheses included amelioration
(which was posited to function more for intimate partner violence than non-intimate
violence), backlash (which was also posited to function more for intimate partner
violence than non-intimate violence), routine activities (which suggests that increases in
absolute status also increases exposure and could lead to an increase in stranger and
known victimization), and protective (which suggests that an increase in absolute status
may increase the accessibility of protective mechanisms and decrease victimization).
Absolute women’s status was operationalized by the percent of women who were
employed, percent of voting age women who voted in the November presidential and
congressional elections, and an index of income-educational attainment, which included
median income and percent who completed four years of college or more. Relative status
was measured using the rate of women’s labor force participation minus that of men, the
level of income-educational attainment for women minus that of men, and the rate of
voter turnout for women minus that of men. Their results found support for each of the
hypotheses except for amelioration. They found a negative relationship between women’s
absolute economic and political status and intimate partner violence which supports the
protective hypothesis. They also found the negative relationship between inequality and
intimate partner victimization which supports the backlash hypothesis. The routine
activities hypothesis was supported through a positive relationship between absolute
indicators of labor force participation and victimization by strangers and non-intimate
persons known to the victim. This longitudinal analysis does not suggest that backlash,
amelioration, protective, nor routine activities hypotheses are corollary by explaining the
relationship between inequality, absolute status, and violence against women at different
periods. However, they found evidence that the nature of the relationship is contingent on
the victim-offender relationship.
The hypotheses and findings suggesting that backlash and amelioration are
corollary are particularly important considering that much of the empirical quantitative
literature on gender inequality has been cross-sectional and assumed a linear relationship.
By design much of the extant literature is unable to address these possibilities (Whaley et
al., 2013). The evidence supporting corollary processes indicate that future analyses
should continue to consider change over time and the structural form of the relationship
as the proposed dissertation will do. In addition, Xie et al. (2012a) presented evidence
that the victim-offender relationship is important to also consider in future analyses.
In addition, while not an inequality hypothesis, several scholars have investigated the role
of absolute status on trends of violence against women. Changes in absolute status are
increases or decreases in the status of a group with no consideration or comparison to
other groups. It is possible that there is a relationship between absolute status and
violence against women and that any findings suggesting a relationship
When race-disaggregated trends of violence against women are investigated,
major gaps in the crime decline, race, and gender literature are exposed. For example,
Lauritsen (2012) illustrated that although the rape and sexual assault trends for Latinas,
Black women, and white women have been similar for the last 40 years, the gaps between
them have varied occasionally. The variation in the gap could indicate different processes
at work. These types of findings are important in that they denote the importance of
attending to gender, race, and ethnicity but are often overlooked in later studies.
Ultimately, the crime decline literature has at times denoted the varying magnitude of the
decline by race and gender subgroups, but has not produced many intersectional
approaches to explaining the disparate decline (Heimer & Lauritsen, 2008; Lauritsen,
2012; Lauritsen & Heimer, 2010; Levitt, 2004; Parker, 2008). As a consequence, the
disparate declines and trends by race, ethnicity, and gender are not well understood. The
race and inequality literature tends to lack a gender-lens, a framework denoting the role
that gender plays in how things are enacted and perceived, while at the same time the
gender and inequality literature tends to lack a race-lens, a framework denoting the role
of race (McCall, 2005). This study argues that it is important to consider violence against
women from a gender x race-lens or an intersectional-lens.
HYPOTHESES AND EMPIRICAL SUPPORT
Before discussing the ways to apply an intersectionality perspective to the gender
inequality hypothesis, it is pertinent to review the various hypotheses that exist within the
literature. As noted in the introduction, the nature of the relationship between gender
inequality and violence against women has not been clearly defined within the empirical
literature (Whaley et al., 2013). In fact, evidence exists for multiple hypotheses. The
original hypothesis and the one many organizations work under suggests that as gender
inequality decreases and the genders become more equal violence against women will
decrease (Whaley et al., 2013). This is referred to as the amelioration hypothesis.
Peterson and Bailey (1992) found some support for the amelioration hypothesis. They
investigated the relationship between various forms of inequality (general, racial, and
gender) and rape rates for United States metropolitan areas in 1980. The rape rates were
obtained from the UCR in 1980. General inequality was measured using the GINI index
which represents income inequality in the total population. Racial inequality was
measured by the difference in median income for Black and white families. Gender
inequality was measured in terms of education using the ratio of men to women who
completed 4 years of high school, who completed 4 years of college, and who completed
5 or more years of college. It was also measured in terms of income using a ratio of men
to women’s median income for all people, people who completed 4 years of high school,
and people who completed 4 years or more of college. Gender inequality was also
measured by the percent of people employed in managerial and professional positions
who were women and by a ratio of men to women who live in poverty. Multiple models
were estimated with most containing one form of gender inequality and one containing a
representative of each type. The GINI index was found to be positively and significantly
related to rape rapes in each model. This finding suggests that places with greater general
income inequality exhibited higher rape rates. Racial inequality in income was significant
and positive in the models that included gender inequality by ratio of high school
graduates only and gender inequality in percentage of women in professional occupations
only. When all forms of gender inequality were included (income, high school graduation,
and profession occupations), gender inequality in median income was significantly and
positively related to rape rates. This finding suggests that where gender inequality in
income is greater, rape rates will also be higher and this is consistent with the
amelioration hypothesis.
Another hypothesis, the backlash hypothesis, suggests an opposite relationship.
Specifically, the backlash hypothesis suggests that as gender inequality decreases and the
genders become more equal, violence against women will increase (Whaley et al., 2013).
It is important to note that the term “backlash” is sometimes used to imply a specific
mechanism causing the increase of violence. For instance, sometimes backlash is defined
as a process by which men (or other dominant group) react violently to challenges to their
dominant status. In this study, as with many past studies, the mechanism of the
relationship between gender inequality and violence against women is not being
evaluated. Rather this study examines the relationship between gender inequality and
violence against women which therefore only allows for some speculation as to the
mechanism of the relationship. Backlash, herein, is used to represent a negative
relationship between changes in gender inequality and changes in violence against
women.
Within the same article described above, Peterson and Bailey (1992) also found
support for a backlash hypothesis when further investigating the relationship between
gender inequality in occupations and rape rates. Specifically, they separated professional
occupations into 1) executives, administrators, and managers 2) public officials and
administrators 3) engineers and natural scientists 4) health diagnostic occupations 5)
elementary and secondary teachers. They found that where the percent of executives,
administrators, and managers that are women were higher rape rates were also higher
suggesting backlash. However, they also found that where the percent of health diagnostic
occupations that are women were higher rape rates were lower. This was true of
elementary and secondary teachers as well. Since being an elementary or secondary
teacher is a women-typed occupation, the finding of a negative relationship could be
interpreted as backlash as well. The authors suggested that “the more that women remain
in traditional ‘female’ occupations, the greater their ‘reward’ – freedom from rape” (p.
172). However, the significant finding for health diagnostic occupations appears to
support amelioration, since health diagnostic occupations are traditionally men-typed
occupations.
The Peterson and Bailey (1992) study is illustrative for three primary reasons.
First, the analysis assesses the relationship between gender inequality and rape in
multiple ways. They use several forms of inequality including: education, income, and
occupation. In addition, they estimate models with each form individually as well as a
model with all of the forms. Second, this study is illustrative because it finds support for
both the ameliorative and backlash hypotheses. Due to these mixed findings, the authors
conclude that their “analysis by no means settles the issue of rape and the structural
position of women vis-à-vis men” (p. 174). Lastly, this study thoroughly investigated
different elements of inequality. For example, as discussed above, they investigated more
specific occupation categories to gain an understanding of the relationship between
gender parity in occupations and rape rates. Ultimately, together these three elements
exemplify how complicated the relationship between gender inequality and rape may be.
Specifically, it is possible that common indicators of gender disparity do not capture the
most important aspects of gender inequality. This could be one explanation for why some
measures of gender inequality were significantly related to rape and others were not. For
instance, inequality in educational attainment was not related to rape rates in any of the
Peterson and Bailey (1992) models. Through this study, it was also demonstrated that
some of the measures of gender inequality may be confounding other relationships. In
particular, Peterson and Bailey (1992) found that although in the main model there was
no significant relationship between gender inequality in occupation type and rape rates,
when looking at specific types of occupations the findings were mixed. Indeed, they
found support for backlash and amelioration when looking at specific types of
occupations. By thoroughly investigating the relationship between gender inequality and
rape, Peterson and Bailey (1992) illustrated how complicated the relationship may be and
demonstrated important ways to attempt to uncover more information.
Some scholars have noted that much of the research on gender inequality and
violence against women, including Peterson and Bailey (1992), have acted under the
axiom that amelioration and backlash are contradictory hypotheses (Bailey, 1999;
Whaley, 2001; Whaley et al., 2013). However, these hypotheses may be corollaries of one
another contingent either on time or the level of gender inequality. For instance, Bailey
(1999) and Whaley (2001) suggested and found some evidence that backlash occurs for a
period of time leading to amelioration. According to Bailey (1999), prior to his analysis
no studies had investigated the role of changes in gender inequality and changes in the
rape rate. Indeed, I was unable to locate any studies prior to Bailey (1999) that
investigated the relationship between changes in gender inequality and changes in any
form of violence against women. Bailey (1999) investigated the relationship between
gender inequality (using male and female differences in educational attainment,
occupation type, and income as indicators) and rape rates for central cities. He found
support for the backlash hypothesis between occupation type and rape rates for 1980 and
1990 (cross-sectional models), but found an ameliorative relationship between changes in
income inequality and changes in rape rates from 1980 to 1990 (change model). In the
end, Bailey suggested that although his analysis was an “improvement over previous
investigations this analysis does not fully resolve important feminist arguments regarding
the socioeconomic status of women and rape” (1999, p. 59).
Whaley (2001) presented itself as refining the temporal theoretical relationship
between change in gender inequality and violence against women. She hypothesized that
initially during times of higher gender inequality, women experience lower levels of rape
as a “reward, “but that in the long term, gender inequality prompts cultural changes that
build support for rape, prompting an increase. The reverse was hypothesized as well. To
use the hypotheses already introduced, Whaley (2001) hypothesized that backlash occurs
first and that after a sustained period of change amelioration can occur. Using 5-, 10-, and
20-year lagged indicators, Whaley (2001) was also unable to provide strong support for
her hypothesis; however, most of the significant relationships were in the proposed
direction. For example, cross-sectional analyses suggested that higher levels of inequality
in holding executive occupations were associated with lower levels of rape supporting the
backlash hypothesis. The model using a 5-year lag from 1970 to 1975 suggested backlash
also. Yet in accord with the hypothesis, the model using a 10-year lag between 1980 and
1990 found support for the ameliorative hypotheses. Specifically, decreases in gender
inequality in holding executive occupations were associated with decreases in rape.
Therefore, findings related to holding executive occupations appeared to support backlash
cross-sectionally and amelioration over a 10-year period. However, gender inequality in
executive occupations was not significantly related to rape when looking at change from
1980 to 1985, 1970 to 1980, and 1970 to 1990. Therefore, within this analysis, the results
reflect a complicated relationship.
Additionally, Whaley et al. (2013) suggested and found some evidence that the
nature of the relationship between gender inequality and violence against women may be
nonlinear and related to the level of gender inequality. Specifically, they investigated the
relationship between gender inequality (in terms of educational attainment, median
income, and employment status) and various rates of homicides for 1990 and 2000 from
the SHR (male-female, male-male, female-male, and female-female). Using negative
binomial regressions, Whaley et al. (2013) found that gender equality (represented by a
factor score of the three variables) was positively related to male offender – female victim
homicide rates and that the squared form of gender equality was negatively related in
1990. These significant findings suggest that the relationship between gender equality and
male-female homicide is nonlinear. Gender equality and its squared term were not
significantly related to any other homicide rates except female-offender and
femalevictim. For the female-female homicide rate, gender equality was significantly and
positively related in the model that did not include the robbery rate, which was included
as a “robust control for other factors related to criminal violence” (Whaley et al., 2013, p.
743). In 2000, gender equality and its squared form were significantly related to
malefemale homicide when the robbery rate was not included but only the squared form
was significantly related (negatively) when the robbery rate was included. The same was
true for male-male homicide rates. Just gender equality is significantly and positively
related to the female-female rate when the robbery rate was not included. In all other
instances, gender equality and its squared form were not significantly related to homicide
rates. Whaley et al. (2013) interpreted the effect of adding the robbery rate as suggesting
that unmeasured factors associated with violent crime may be mediating the relationship
or that there may be an issue with multicollinearity. As a whole, they found considerable
support for their hypothesis that backlash processes are likely to function at “low to
intermediate levels” of gender equality and that amelioration occurs when equality is high
for male-female homicide rates (p. 747).
Xie, Heimer, and Lauritsen (2012a) used longitudinal methods to assess whether
there was a relationship between non-fatal intimate partner violence and absolute/relative
indicators of women’s status. They created five five-year pooled periods using MSAlevel
data (which includes the 40 largest MSAs) from the NCVS for 1980 to 2004. To
thoroughly assess their hypotheses, they also disaggregated the non-fatal violence against
women data by victim-offender relationship using the following listed categories:
stranger, intimate partner, and other known. The four hypotheses included amelioration
(which was posited to function more for intimate partner violence than non-intimate
violence), backlash (which was also posited to function more for intimate partner
violence than non-intimate violence), routine activities (which suggests that increases in
absolute status also increases exposure and could lead to an increase in stranger and
known victimization), and protective (which suggests that an increase in absolute status
may increase the accessibility of protective mechanisms and decrease victimization).
Absolute women’s status was operationalized by the percent of women who were
employed, percent of voting age women who voted in the November presidential and
congressional elections, and an index of income-educational attainment, which included
median income and percent who completed four years of college or more. Relative status
was measured using the rate of women’s labor force participation minus that of men, the
level of income-educational attainment for women minus that of men, and the rate of
voter turnout for women minus that of men. Their results found support for each of the
hypotheses except for amelioration. They found a negative relationship between women’s
absolute economic and political status and intimate partner violence which supports the
protective hypothesis. They also found the negative relationship between inequality and
intimate partner victimization which supports the backlash hypothesis. The routine
activities hypothesis was supported through a positive relationship between absolute
indicators of labor force participation and victimization by strangers and non-intimate
persons known to the victim. This longitudinal analysis does not suggest that backlash,
amelioration, protective, nor routine activities hypotheses are corollary by explaining the
relationship between inequality, absolute status, and violence against women at different
periods. However, they found evidence that the nature of the relationship is contingent on
the victim-offender relationship.
The hypotheses and findings suggesting that backlash and amelioration are
corollary are particularly important considering that much of the empirical quantitative
literature on gender inequality has been cross-sectional and assumed a linear relationship.
By design much of the extant literature is unable to address these possibilities (Whaley et
al., 2013). The evidence supporting corollary processes indicate that future analyses
should continue to consider change over time and the structural form of the relationship
as the proposed dissertation will do. In addition, Xie et al. (2012a) presented evidence
that the victim-offender relationship is important to also consider in future analyses.
In addition, while not an inequality hypothesis, several scholars have investigated the role
of absolute status on trends of violence against women. Changes in absolute status are
increases or decreases in the status of a group with no consideration or comparison to
other groups. It is possible that there is a relationship between absolute status and
violence against women and that any findings suggesting a relationship
