Child Neglect Article Review

ORIGINAL ARTICLE

Gender Inequity Associated with Increased Child Physical Abuse and Neglect: a Cross-Country Analysis of Population-Based Surveys and Country-Level Statistics

Joanne Klevens1 & Katie A. Ports1

Published online: 27 July 2017 # US Government (outside the USA) 2017

Abstract Gender inequity is proposed as a societal-level risk factor for child maltreatment. However, most cross-national research examining this association is limited to developing countries and has used limited measures of gender inequity and child homicides as a proxy for child maltreatment. To ex- amine the relationship between gender inequity and child mal- treatment, we used caregivers’ reported use of severe physical punishment (proxy for physical abuse) and children under 5 left alone or under the care of another child younger than 10 years of age (supervisory neglect) and three indices of gender ineq- uity (the Social and Institutional Gender Index, the Gender Inequality Index, and the Gender Gap Index) from 57 countries, over half of which were developing countries. We found all three gender inequity indices to be significantly associated with physical abuse and two of the three to be significantly associ- ated with neglect, after controlling for country-level develop- ment. Based on these findings, efforts to prevent child abuse and neglect might benefit from reducing gender inequity.

Keywords Child abuse . Child neglect . Child maltreatment .

Etiology . Cross-national

Child abuse and neglect by parents and other caregivers hap- pens around the world at alarming rates. The World Health Organization (WHO) estimates that approximately a quarter of adults worldwide have experienced child physical abuse (WHO 2016a), but rates may vary across countries

(Stoltenborgh et al. 2013a). For example, studies have found that rates of child physical neglect vary by geographic location from 19% in North America to 7% in Europe (estimates from other regions are not available; Stoltenborgh et al. 2013b). Given these differences, it is important to ask: what is driving the geographic variation in rates and how can this inform prevention efforts?

The occurrence of child abuse and neglect, and the unequal distribution of exposure across geographic locations can be understood using the socioecological model, which proposes that factors at the individual-, family-, community-, and societal-level simultaneously operate and interact to increase or decrease the likelihood of child maltreatment (Garbarino 1977). Numerous individual- and family-level factors that in- crease the risk for both physical abuse and neglect have been identified (Stith et al. 2009). Fewer studies have identified community-level factors (Coulton et al. 2007; Freisthler et al. 2006), and research examining societal- or country- level factors of child maltreatment remains scarce. Modifying societal-level risk factors has great potential for achieving population-level reductions in rates of child mal- treatment (Putnam and Galea 2008).

Available research on societal-level factors of child maltreat- ment has predominantly focused on child homicides. These studies have found child homicides to be positively associated with the percent of women in the labor force (Briggs and Cutright 1994; Fiala and LaFree 1988; Gartner 1990, 1991; Hunnicutt and LaFree 2008), battle death rates (albeit only for homicides of children 0–4 years of age; Briggs and Cutright 1994; Gartner 1990, 1991), and rape rates (Briggs and Cutright 1994) and negatively associated with the ratio of females to males in tertiary education (Briggs and Cutright 1994; Fiala and LaFree 1988) or professional occupations (Fiala and LaFree 1988),. The link between child homicide and the ratio of female to males in tertiary education or

* Joanne Klevens [email protected]

1 Division of Violence Prevention, National Center for Injury Prevention and Control, Centers for Disease Control and Prevention (CDC), 4770 Buford Hwy, Mailstop F-63, Atlanta, GA 30341, USA

J Fam Viol (2017) 32:799–806 DOI 10.1007/s10896-017-9925-4

professional occupations are hypothesized to be related to the status of women (the higher women’s status, the lower child homicide rates would be). Battle death rates are based on the theory of Bcultural spillover^ which posits that legitimating of violence by the state (e.g., war or death penalty) may spillover to other types of violence (Baron and Straus 1987).

Inconsistent effects on child homicide rates have been found for income inequality (Briggs and Cutright 1994; Butchart and Engström 2002; Fiala and LaFree 1988; Gartner 1990; Hunnicutt and LaFree 2008), economic development (Fiala and LaFree 1988; Butchart and Engström 2002; Christoffel et al. 1981; Moniruzzaman and Andersson 2008), the divorce rate (Briggs and Cutright 1994; Fiala and LaFree 1988; Gartner 1990, 1991; Hunnicutt and LaFree 2008), female enrollment in tertiary education (Briggs and Cutright 1994; Fiala and LaFree 1988; Gartner 1990, 1991; Hunnicutt and LaFree 2008), coun- tries’ expenditures on social welfare programs ((Briggs and Cutright 1994; Fiala and LaFree 1988; Gartner 1990, 1991), overall homicide rates (Briggs and Cutright 1994; Christoffel et al. 1981; Fiala and LaFree 1988; Gartner 1991), teen births (Briggs and Cutright 1994; Gartner 1991), and the heterogeneity of the population (Briggs and Cutright 1994; Gartner 1990). Other societal-level factors that have not been found to be asso- ciated with child homicides include fertility and illegitimacy rates (Briggs and Cutright 1994; Gartner 1991); unemployment (Briggs and Cutright 1994; Hunnicutt and LaFree 2008), rapidity of development (Fiala and LaFree 1988); the death penalty (Gartner 1990); and physicians per 100,000 (Fiala and LaFree 1988; Hunnicutt and LaFree 2008). In one study, when the per- centage of women with access to paid jobs relative to men was considered, the association between child homicide rates and economic wealth (gross domestic product per capita) and income inequality among young girls was strengthened, whereby less economic wealth and greater income inequality were associated with higher child homicide rates (Butchart and Engström 2002).

While this cross-national research has highlighted several societal risk factors for child homicides, with the exception of three studies (Butchart and Engström 2002; Hunnicutt and LaFree 2008; Moniruzzaman and Andersson 2008), the bulk of the research has focused on developed or high income countries limiting the generalizability of their findings. In ad- dition, child homicides as an indicator of child maltreatment has several limitations. Previous research suggests that child homicide and child maltreatment are not similar constructs on a continuum, but rather, distinct forms of violence, and as such, examinations of child homicide may not be generalized to other forms of violence toward children, including child maltreatment (Gelles 1991). Moreover, death records and data tend to vary across countries and could lead to differential underestimates associated with country characteristics. For example, homicides could be coded erroneously as Bundetermined^ or Bunintentional^. Finally, deaths due to child neglect may be excluded from homicides in some

countries because they do not meet the definition of Bhomicide^ when defined as a deliberate act of commission.

Given the limitations of using child homicide as a measure of child maltreatment, surveys asking participants about perpe- tration or victimization may be a better way of assessing child maltreatment. Indeed, rates of child maltreatment based on self- reported or parent-reported incidents have found physical abuse to be 20 times greater than abuse reported to authorities (Finkelhor et al. 2014). To our knowledge, only three cross- national studies have examined a societal-level risk factor using child maltreatment data collected from surveys. Two studies used Multiple Indicator Cluster Survey (MICS) data from 25 middle and lower income countries and found that past-month caregiver reported child physical and psychological abuse was higher in countries where violence in families was more of a cultural norm (i.e., they reported greater acceptance of corporal punishment and intimate partner violence; Lansford et al. 2014) and in countries with lower levels of education (Lansford and Deater-Deckard 2012). The third study utilized data from 28 countries reporting childhood trauma questionnaire (CTQ) scores, which asks children 12 and over to rate the frequency that events related to child abuse and neglect occurred when they Bwere growing up^ (Viola et al. 2016). Findings revealed that countries with higher gross domestic products had lower childhood physical neglect estimates. It is likely that many other societal-level factors impact rates of child maltreatment, but such factors have yet to be explored.

The World Health Organization has suggested that gender inequity may be a societal risk factor for child maltreatment (Runyan et al. 2002). Gender inequity may lead to increased child abuse in at least two ways: (1) because limited opportu- nities among women may increase their stress and frustration with caring for children they may be more likely to abuse them (Fiala and LaFree 1988); and (2) disempowered women may be less able to protect their children from abuse (Gartner 1990). Although two studies mentioned above have examined proxies of gender inequity to be associated with child homi- cides [i.e., the ratio of females to males in tertiary education (Briggs and Cutright 1994) and professional occupations (Fiala and LaFree 1988)] we build on this research by exam- ining its associations with other, more robust indicators of child maltreatment by (a) focusing on caregiver-reported physical abuse and neglect; (b) examining child maltreat- ment’s relationship to three comprehensive indices of gender inequity generated by three different international agencies; and (c) expanding the number of countries and their regional and stage of development variability.

Method

This is a cross-sectional study based on country-level data as our unit of analysis. Our sample corresponds to all the

800 J Fam Viol (2017) 32:799–806

countries reporting either severe physical discipline or child neglect in a Multiple Indicator Cluster Survey (MICS; n = 49; United Nations International Children’s Emergency Fund [UNICEF] 2015) or a Demographic and Health Survey (DHS; n = 8; United States Agency for International Development [USAID] 2016) from 2011 to 2015. MICS and DHS are conducted in a large number of countries around the world using random sampling methods to obtain nationally representative samples and trained interviewers to administer a standardized questionnaire face-to-face to one adult in se- lected households. An index child between 1 or 2 and 14 years old is also selected randomly in the household for questions on child discipline. The 57 countries included in our analyses are listed by region in Table 1 with their corresponding level of development using the Human Development Index (HDI; United Nations Development Program 2015a) described in the measures section, data source (i.e., MICS or DHS), year data were collected, and age range of the children sampled.

Measures

Child Physical Abuse and Neglect Childhood physical abuse and neglect were the outcomes of interest. Childhood physical abuse is based on the MICS/DHS country reports of the percentage of caregivers who report that the index child experienced severe physical discipline in the past month. Severe physical discipline in MICS/DHS was defined as the index child being hit or slapped on the face, head or ear; beat/hit with something (e.g. an object); or hit over and over as hard as one could by anyone in the household in the past month in response to the child’s bad behavior. This question is prefaced by BAll adults use certain ways to teach or to address a behavior problem.^ Table 1 shows the percent of caregivers reporting this type of physical abuse for each country. Data for this variable were available for 51 countries.

Child neglect is based on another measure from the MICS/ DHS in which they report the percentage of children aged 0– 59 months left alone or under the care of another child youn- ger than 10 years of age for more than one hour at least once in the last week. This variable was available for 55 countries (reported in last column of Table 1). To correct for the skewness of the distributions for both variables, we used their natural log.

Gender Equity We used available data for the same year the MICS/DHS data were collected or the closest year available to that (± 1) for three measures of gender equity:

The Social and Institutional Gender Index (SIGI) The Social and Institutional Gender Index (SIGI) is a cross- country composite of 14 (in 2012) or 21 (in 2014) measures of discrimination against women in social institutions pub- lished by the Organisation for Economic Co-operation and Development (2016). It has five subindices: (a) discriminatory

family codes (e.g., equality in minimum age for marriage, in parental authority in marriage or divorce, and inheritance); (b) restricted physical integrity (e.g., laws on partner violence, rape, sexual harassment, frequency of partner violence and genital mutilation); (c) son bias (e.g., missing women and fertility preferences); (d) restricted resources and assets (e.g., access to land and non-land assets use, control and owner- ship); and (e) restricted civil liberties (e.g., access to public space and political representation). Each sub-index score is calculated by aggregating the different indicators it is com- posed of with a weighting scheme obtained through a polychoric principal component analysis. Each sub-index score is squared and multiplied by one-fifth. The five resulting scores are added. Scores vary from 0 to 1 with higher scores indicating greater discrimination against women.

The Gender Inequality Index (GII) The Gender Inequality Index (GII) is a composite of measures of across three dimen- sions: (a) reproductive health (i.e. maternal mortality ratio and adolescent birth rates); (b) empowerment (i.e., proportion of parliamentary seats occupied by females and proportion of adult females and males aged 25 years and older with at least some secondary education); and (c) economic status (i.e., la- bor force participation rate of female and male populations aged 15 years and older) representing women’s disadvantage published by the United Nations Development Program (2015b). To calculate the index for each country, a minimum value for each indicator is set at 0.1 and extreme high values are truncated. The geometric mean is then calculated separate- ly for women and men across each dimension. Finally, these means aggregated using a harmonic mean across genders. Scores vary from 0 to 1 with higher scores indicating women faring worse than men.

The Gender gap Index (GGI) The Gender Gap Index (GGI), published by the World Economic Forum (2014), measures gender-based gaps in access to resources and opportunities in countries rather than the actual levels of the available re- sources and opportunities in those countries to make the index independent from the countries’ levels of development. It is a composite of four constructs: (a) economic participation and opportunity, which captures three concepts: the participation gap (difference between women and men in labor force par- ticipation rates), the remuneration gap (ratio of estimated female-to-male earned income and wage equality for similar work) and the advancement gap (the ratio of women to men among legislators, senior officials and managers, and the ratio of women to men among technical and professional workers); (b) educational attainment, which assesses the gap between women’s and men’s current access to education (i.e., ratios of women to men in primary-, secondary- and tertiary-level ed- ucation and the ratio of the female literacy rate to the male literacy rate); (c) health and survival, which provides an

J Fam Viol (2017) 32:799–806 801

Table 1 Countries by region, Human Development Index, data source (MICS/DHS), year data collected, age range of children, and percent of children reported to have experienced physical abuse or neglect

REGION HDI MICS/DHS Year(s) collected Age range of children

% Physically abused past month

% Children <5 neglected past week

CENTRAL ASIA Kazakhstan 0.788 MICS 2015 1–14 1 5 Kyrgyzstan 0.655 MICS 2014 1–14 3 4.5 Mongolia 0.727 MICS 2013–14 1–14 4 10.2 Turkmenistan 0.688 MICS 2015 1–14 NI 0.8

EUROPE Belarus 0.796 MICS 2012 2–14 NI 3.6 Bosnia & Herzegovina 0.724 MICS 2011–12 2–14 4.5 1.6 Macedonia 0.738 MICS 2011 2–14 4.8 5 Montenegro 0.801 MICS 2013 1–14 2 2.6 Serbia 0.771 MICS 2014 1–14 1.0 3.6 Ukraine 0.743 MICS 2012 1–14 6.5 NI

LATIN AMERICA AND THE CARIBBEAN

Argentina 0.818 MICS 2011–2012 2–14 9.8 8.2 Barbados 0.793 MICS 2012 2–14 6.1 1.4 Belize 0.715 MICS 2015 1–14 6 12.9 Costa Rica 0.756 MICS 2011 2–14 3.1 4.0 Dominican Republic 0.715 MICS 2014 1–14 3 5.3 Guyana 0.636 MICS 2014 1–14 6 5.0 Haiti 0.479 DHS 2012 2–14 16.1 NI Honduras 0.612 DHS 2011–12 2–14 NI 4.4 Jamaica 0.727 MICS 2011 2–14 5.7 1.8 Mexico 0.756 MICS 2015 1–14 5.9 5.1 Panama 0.777 MICS 2013 1–14 2 2.6 St. Lucia 0.730 MICS 2012 2–14 5.9 4.7 Uruguay 0.790 MICS 2013 2–14 3 2.9

MIDDLE EAST & NORTH AFRICA

Algeria 0.732 MICS 2012–13 2–14 22.9 5.5 Egypt 0.690 DHS 2014 1–14 43.2 4.1 Iraq 0.648 MICS 2011 2–14 27.7 7.5 Jordan 0.746 MICS 2012 2–14 20.3 9.4 Oman 0.793 MICS 2014 1–14 NI 44.7 Palestine 0.793 MICS 2014 1–14 23 14.3 Qatar 0.844 MICS 2012 2–14 6.4 11.6 Tunisia 0.715 MICS 2011–12 2–14 31.9 13.2 Yemen 0.498 MICS 2013 2–14 42.1 NI

SOUTH AND EAST ASIA Afghanistan 0.456 MICS 2011 2–14 38.4 40.2 Bangladesh 0.563 MICS 2012–13 1–14 24.6 11.6 Cambodia 0.555 DHS 2014 2–14 NI 9.8 Lao PDR 0.552 MICS 2011–12 2–14 7.6 14 Nepal 0.548 MICS 2014 1–14 14 20.6 Thailand 0.723 MICS 2012–13 1–14 NI 4.6 Viet Nam 0.666 MICS 2014 1–14 2.1 7

SUBSAHARA AFRICA Benin 0.480 MICS 2014 1–14 23 34.1 Cameroun 0.512 MICS 2014 1–14 20 34.4 Chad 0.392 MICS 2014–15 1–14 15.3 NI Congo-Brazzaville 0.591 MICS 2014–15 1–14 28 41.6 Congo-Kinshasa 0.433 DHS 2013–14 1–14 28.3 49.1 Ghana 0.566 MICS 2011 2–14 14.4 20.7 Guinea Bissau 0.420 MICS 2014 1–14 18 30.6 Malawi 0.445 MICS 2013–14 1–14 6 37.1 Mali 0.419 MICS 2015 1–14 14 32.2 Mauritania 0.506 MICS 2015 1–14 27 34.4 Niger 0.342 DHS 2012 2–14 29.4 NI Nigeria 0.499 MICS 2011 2–14 34 39.9 Rwanda 0.483 DHS 2014–15 1–14 NI 35.0 Sâo Tomé e Príncipe 0.555 MICS 2014 1–14 10 15.5 Sudan 0.479 MICS 2014 1–14 14.0 NI Swaziland 0.531 MICS 2014 1–14 10 16.5 Togo 0.484 DHS 2013–14 1–14 11.4 29.1 Zimbabwe 0.509 MICS 2014 1–14 5 18.5

HDI Human Development Index <0.550 indicates low development, 0.550–0.699 for medium development, 0.700–0.799 high development, and ≥0.800 is very high development (United Nations Development Program 2015a). MICS Multiple Indicator Cluster Survey, DHS Demographic and Health Survey, NI no information

802 J Fam Viol (2017) 32:799–806

overview of the differences between women’s and men’s health (i.e., sex ratio at birth and the gap between women’s and men’s healthy life expectancy); (d) political empowerment, which measures the gap between men and women at the highest level of political decision-making (i.e., the ratio of women to men in minister-level positions, the ratio of women to men in parliamentary positions, and the ratio of women to men in terms of years in executive office (prime minister or president) for the last 50 years). The Index is con- structed using a four-step process. First, all indicators are con- verted to female-to-male ratios. Then, except for the two health indicators, these indicators are truncated at 1, meaning equal numbers of women and men. Next, indicators are nor- malized based on their standard deviations. The weighted av- erage of the indicators within each sub-index is calculated to create the sub-index scores. An un-weighted average of the sub-index scores is used to calculate the overall Gender Gap Index score. Scores vary from 0 to 1 with higher scores indi- cating greater gender equity. To correct for the negative skew- ness of these data, we reverse scored and converted these to their natural log.

Finally, level of development as a potential confounder was measured with the HDI. The HDI is a composite of indicators of life expectancy, education (i.e., adult literacy rate and com- bined gross enrollment in primary, secondary, and tertiary school), and Gross Domestic Product. Scores range from 0 to 1 with higher scores indicating higher levels of development.

Data Analyses

We estimated bivariate correlation coefficients, but report the squared correlation coefficient (or variance) for ease of inter- pretation as well as negative correlation coefficient signs, in parentheses, to inform the direction of the association. We used linear regression models to estimate the semi-partial cor- relation coefficient after adjusting for country-level of devel- opment using the Human Development Index.

Results

The 57 countries in our sample come from Central Asia (4), Europe (6), Latin America and the Caribbean (13), Middle East and Northern Africa (9), South and East Asia (7) and Sub-Saharan Africa (18). Nearly half of the countries (44%) are considered to have high or very high human development and a third are considered to have low human development. Child physical abuse rates in the past month varied from a range of 1% to 43% and child neglect rates varied from 0.8% to 49% (see Table 1).

Table 2 shows the squared correlation coefficients between our two outcomes and the three indices of gender inequity. The three measures of gender inequity were highly correlated

with each other (not shown) further demonstrating construct validity of the gender measures. We found statistically signif- icant and positive associations between the SIGI, the GII, and the reverse-scored GGI and both child physical abuse and child neglect. This means that higher scores indicating greater levels of discrimination against women on the SIGI, greater gender inequity on the GII, and lower scores on the GGI indicating greater gender gaps are associated with higher rates of child physical abuse and child neglect.

In separate regression models for child physical abuse and neglect, discrimination against women (SIGI) accounted for 31% of the variance in child physical abuse (14% after adjusting for the HDI measuring country-level development, p = .006) and 17% of the variance in child neglect (7% after adjusting for the HDI, p = .014); gender inequality (as assessed by the GII) accounted for 49% of the variance in child physical abuse (15% after adjusting for the HDI, p = .001) and 40% of the variance in child neglect (but 1% after adjusting for the HDI and no longer significant); and gender gaps (as assessed by the GGI) explain 53% of the variance in child physical abuse (20% after adjusting for the HDI, p = .001) and 15% of the variance in child neglect (8% after adjusting for the HDI, p = .019).

Discussion

We examined associations among three prominent and com- prehensive indicators of gender inequity (the SIGI, measuring discrimination against women, the GII measuring health, eco- nomic and power inequities, and the GGI, measuring econom- ic, education, health, and political power gender-based gaps), and two proxies of child maltreatment, severe child physical discipline and supervisory neglect. Rates of child maltreatment varied by geographic location, and our findings showed strong and statistically significant associations in the expected direc- tions with all three gender inequity indices. After controlling for country-level development, all three gender indices contin- ued to be significantly associated with physical abuse and two of the three (SIGI and GGI) continued to be significantly asso- ciated with neglect. These findings are consistent with previous research showing proxies of gender inequity to be associated with child homicides [i.e., the ratio of females to males in tertiary education (Briggs and Cutright 1994) and professional occupations (Fiala and LaFree 1988)]. The mechanisms for this relationship should be explored in future research.

Our study had several strengths including the use of nation- ally representative estimates of caregiver-reported abuse and neglect, robust indicators of gender inequity, and greatly ex- panded regional representation and developmental variability in our sample of countries. However, several limitations must be acknowledged before discussing the potential implications of these findings. First, our sample did not include countries in the

J Fam Viol (2017) 32:799–806 803

European Union or many of the most populated countries such as Brazil, China, India, Indonesia, Japan, Russia, or the USA. As such, our findings may not be generalizable to the countries not included. It is possible that comparable data are available from excluded countries but not published. Future studies might con- sider contacting Ministries of Health or Child Welfare in an effort to include a more diverse sample of countries from the European Union and other major economies.

Second, our measures of child maltreatment (severe physical discipline and supervisory neglect) do not capture the entire universe of maltreatment that children might ex- perience, such as physical abuse occurring outside disci- plinary interactions, emotional abuse and neglect, or sexual abuse. Nor do they provide any indication of other trau- matic experiences that can also impact the health and wellbeing of children across their life course. Severe phys- ical discipline and supervisory neglect are also limited by issues related to caregivers’ reports (i.e., recall bias, social desirability bias) and measurement problems (e.g., validity and reliability of one-item measures, short period of recall, including anyone in the household as a perpetrator). In addition, these measures may be culturally inappropriate for countries with different economic conditions, and con- cern over assessments that may be ethno-centric to Western post-industrial countries has been raised (Laird 2016).

Although we used three different indices of gender inequity that are composites of multiple indicators tapping many di- mensions of inequity such as political, institutional, education- al, economic, and health, these may not capture the universe of gender-based inequity (e.g., inequitable distribution of labor in the household). These composites are also constructed from government reports which may have variable reliability. Because data for some of the indices were available only for certain years, the exposure-outcome data are not optimally time-matched for all countries. However, examination of changes over time for those countries with various years of data revealed very small changes from year to year and there- fore this mismatch should not be of great concern.

A major limitation is that our findings are based on correla- tions from cross-sectional data. It is possible that gender ineq- uity and child maltreatment co-occur in a society due to some other, unexplored cause. Our small sample size (n = 57) limited the number of potential confounders we could control for. Multi-level data analyses with a larger sample could control for individual-level confounders and clustering. Future studies using natural experiments could provide some evidence of causal effects. Although we cannot claim causal relationships between variables, there are several potential mechanisms to explain how gender inequity might be related to child maltreat- ment. Women’s status has long been associated with several indicators of children’s health (Heaton 2015). When women are in better economic positions, financial stress which is asso- ciated with harsh parenting may decrease (Brooks-Gunn et al. 2013) and women can afford higher quality child care arrange- ments instead of leaving a child home alone or under the care of a sibling. High-quality child care is associated with fewer symptoms of maternal depression (Gordon et al. 2011), a risk factor for both child physical abuse and neglect (Stith et al. 2009). However, the relationship between gender inequality and child maltreatment should not only be considered in terms of women making more money, but also in terms of gender equality whereby the gap between men and women is reduced, and women’s economic position is equal to that of men. When women have greater political power, they tend to sponsor leg- islation and policy change that are supportive of children and families (UNICEF 2008). Policies that are supportive of fami- lies such as paid family leave (Klevens et al. 2016) or increased availability of child care are associated with decreased rates of child maltreatment (Klevens et al. 2015). Although the strong correlations of gender inequity and child abuse and neglect suggest new directions for understanding and addressing child maltreatment, their relevance should be replicated in future studies. Future research can take advantage of more robust measures of child maltreatment such as the Childhood Trauma Questionnaire that has been used in 28 countries from 6 continents (Viola et al. 2016). When policies or country levels

Table 2 Variance (R2) in physical abuse and neglect accounted for by the Social Institutional and Gender Index, the Gender Inequality Index, Gender Gap Index, and Human Development Index (negative correlations indicated) and variance after adjusting for Human Development (semipartial correlation squared)

Variable Ln Physical abuse Ln Neglect

R2 (p) Semipartiala R2 R2 (p) Semipartiala R2

(p) (p)

Social Institutional and Gender Index .31 (.000) .14 (.009) .17 (.008) .07 (.014)

Gender Inequality Index .49 (.000) .15 (.001) .40 (.005) .01 (.26)

Ln of reverse scored Gender Gap Index .53 (.000) .20 (.001) .15 (.014) .08 (.019)

Human Development Index (−).31 (.001) (−).51 (.000)

a adjusted for the Human Development Index

804 J Fam Viol (2017) 32:799–806

of gender inequity change, conducting longitudinal analyses would provide stronger evidence for causal relationships.

Child maltreatment, including physical abuse and neglect, is a significant public health issue that can be prevented. Extensive research on risk factors of child maltreatment at the individual- and family-level have resulted in several effec- tive interventions, including group-interventions that address harmful gender norms and gender-equity training (WHO 2016b). While child maltreatment interventions targeting individual- and family-level risk factors have led to declines in child physical abuse, prevention efforts that focus on societal-level risk factors like gender inequity would theoret- ically have a greater and more sustained impact on child mal- treatment rates. Thus, it is critical for researchers to continue to investigate societal-level factors associated with child mal- treatment so that interventions and prevention efforts can in- corporate strategies that have the greatest potential for population-level impact - assuring that all children and fami- lies have access to safe, stable, nurturing relationships and environments, the essentials for optimal health and development.

Acknowledgements We thank Ivan Ludlow, Chelsea Austin, and Jacqueline Hurd for their work in identifying data sources.

Author’s Contribution JK conceptualized the study, collaborated in data collection, conducted the statistical analyses, had full access to all the data in the study, and takes responsibility for the integrity of the data and the accuracy of the data analysis. KP contributed to the design of the study, collaborated in data collection, and helped interpret the data. JK and KP contributed to the drafting and revising of the manuscript and approved the final manuscript as submitted.

Compliance with Ethical Standards

Disclaimer The findings and conclusions in this report are those of the authors and do not necessarily represent the official position of the Centers for Disease Control and Prevention.

Funding No external funding.

Ethical Approval Secondary data analyses; IRB approval not required.

Declarations We declare no competing interests.

References

Baron, L., & Straus, M. A. (1987). Four theories of rape: A macrosociological analysis. Social Problems, 34(5), 467–489.

Briggs, C. M., & Cutright, P. (1994). Structural and cultural determinants of child homicide: A cross national analysis. Violence and Victims, 9(1), 3–16.

Brooks-Gunn, J., Schneider, W., & Waldfogel, J. (2013). The great reces- sion and the risk for child maltreatment. Child Abuse & Neglect, 37(10), 721–729. doi:10.1016/j.chiabu.2013.08.004.

Butchart, A., & Engström, K. (2002). Sex- and age-specific relations between economic development, economic inequality and homicide rates in people aged 0–24 years: A cross-sectional analysis. Bulletin of the World Health Organization, 80(10), 797–805.

Christoffel, K. K., Kiang, L., & Stamler, J. (1981). Epidemiology of fatal child abuse: International mortality data. Journal of Chronic Diseases, 34(2), 57–64. doi:10.1016/0021-9681(81)90051-5.

Coulton, C. J., Crampton, D. S., Irwin, M., Spilsbury, J. C., & Korbin, J. E. (2007). How neighborhoods influence child maltreatment: A re- view of the literature and alternative pathways. Child Abuse & Neglect, 31(11), 1117–1142. doi:10.1016/j.chiabu.2007.03.023.

Fiala, R., & LaFree, G. (1988). Cross-national determinants of child ho- micide. American Sociological Review, 53(3), 432–445. doi:10. 2307/2095650.

Finkelhor, D., Vanderminden, J., Turner, H., Hamby, S., & Shattuck, A. (2014). Child maltreatment rates assessed in a national household survey of caregivers and youth. Child Abuse & Neglect, 38(9), 1421–1435. doi:10.1016/j.chiabu.2014.05.005.

Freisthler, B., Merritt, D. H., & LaScala, E. A. (2006). Understanding the ecology of child maltreatment: A review of the literature and direc- tions for future research. Child Maltreatment, 11(3), 263–280. doi: 10.1177/1077559506289524.

Garbarino, J. (1977). The human ecology of child maltreatment: A con- ceptual model for research. Journal of Marriage and the Family, 39(4), 721–735. doi:10.2307/350477.

Gartner, R. (1990). The victims of homicide: A temporal and cross- national comparison. American Sociological Review, 55(1), 92– 106. doi:10.2307/2095705.

Gartner, R. (1991). Family structure, welfare, spending, and child homi- cide in developed democracies. Journal of Marriage and the Family, 53(1), 231–240. doi:10.2307/353147.

Gelles, R. J. (1991). Physical violence, child abuse, and child homicide: A continuum of violence, or distinct behaviors? Human Nature, 2(1), 59–72. doi:10.1007/BF02692181.

Gordon, R. A., Usdansky, M. L., Wang, X., & Guzman, A. (2011). Child care and mothers’ mental health: Is high-quality care associated with fewer depressive symptoms? Family Relations, 60(4), 446–460. doi: 10.1111/j.1741-3729.2011.00657.x.

Heaton, T. B. (2015). Are improvements in child health due to increasing status of women in developing nations? Biodemography and Social Biology, 61(3), 252–265. doi:10.1080/19485565.2015.1047487.

Hunnicutt, G., & LaFree, G. (2008). Reassessing the structural covariates of cross-national infant homicide victimization. Homicide Studies, 12(1), 46–66. doi:10.1177/1088767907311245.

Klevens, J., Barnett, S. B. L., Florence, C., & Moore, D. (2015). Exploring policies for the reduction of child physical abuse and neglect. Child Abuse & Neglect, 40, 1–11. doi:10.1016/j.chiabu. 2014.07.013.

Klevens, J., Luo, F., Xu, L., Peterson, C., & Latzman, N. E. (2016). Paid family leave's effect on hospital admissions for pediatric abusive head trauma. Injury Prevention, 22(6), 442–445. doi:10.1136/ injuryprev-2015-041702.

Laird, S. E. (2016). ‘If parents are punished for asking their children to feed goats’: Supervisory neglect in sub-Saharan Africa. Journal of Social Work, 16(3), 303–321. doi:10.1177/1468017315572037.

Lansford, J. E., & Deater-Deckard, K. (2012). Childrearing discipline and violence in developing countries. Child Development, 83(1), 62–75. doi:10.1111/j.1467-8624.2011.01676.x.

Lansford, J. E., Deater-Deckard, K., Bornstein, M. H., Putnick, D. L., & Bradley, R. H. (2014). Attitudes justifying domestic violence predict endorsement of corporal punishment and physical and psychologi- cal aggression towards children: A study in 25 low- and middle- income countries. Journal of Pediatrics, 164(5), 1208–1213. doi: 10.1016/j.jpeds.2013.11.060.

Moniruzzaman, S., & Andersson, R. (2008). Cross-national injury mor- tality differentials by income level: The possible role of age and

J Fam Viol (2017) 32:799–806 805

ageing. Scandinavian Journal of Public Health, 122(11), 1167– 1176. doi:10.1080/14034940510006021.

Organisation for Economic Co-operation and Development (2016). Social institutions and gender index (SIGI) [data set]. Retrieved from: http://stats.oecd.org/Index.aspx?DatasetCode=GID2.

Putnam, S., & Galea, S. (2008). Epidemiology and the macrosocial de- terminants of health. Journal Public Health Policy, 29(3), 275–289. doi:10.1057/jphp.2008.15.

Runyan, D., Wattam, C., Ikeda, R., Hassan, F., & Ramiro, L. (2002). Child abuse and neglect bny parents and other caregivers. In E. Krug, L. L. Dahlberg, J. A. Mercy, A. B. Zwi, & R. Lozano (Eds.), World report on violence and health (pp. 59–86). Geneva: World Health Organization.

Stith, S. M., Liu, T., Davies, C., Boykin, E. L., Alder, M. C., Harris, J. M., ... &, Dees, J. E. (2009). Risk factors in child maltreatment: A meta- analytic review of the literature. Aggression and Violent Behavior, 14(1), 13–29. doi:10.1016/j.avb.2006.03.0.

Stoltenborgh, M., Bakermans-Kranenburg, M. J., IJzendoorn, M. H., & Alink, L. R. (2013a). Cultural–geographical differences in the oc- currence of child physical abuse? A meta-analysis of global preva- lence. International Journal of Psychology, 48(2), 81–94. doi:10. 1080/00207594.2012.697165.

Stoltenborgh, M., Bakermans-Kranenburg, M. J., & van IJzendoorn, M. H. (2013b). The neglect of child neglect: A meta-analytic review of the prevalence of neglect. Social Psychiatry Psychiatric Epidemiology, 48(3), 345–355. doi:10.1007/s00127-012-0549-y.

UNICEF. (2008). The state of the world’s children 2007: Women and children: The double dividend of gender equality. New York, NY: United Nations Children's Fund Retrieved from: http://www.unicef. org/sowc07/.

UNICEF (2015). Multiple indicator cluster surveys. New York: United Nations Children's Fund. Retrieved from: http://mics. unicef.org/surveys.

United Nations Development Program. (2015a). Human development report statistical annex, table 2. Human development index trends. New York, NY: United Nations Retrieved from: http://hdr.undp.org/ sites/default/files/hdr_2015_statistical_annex.

United Nations Development Program (2015b). Gender inequality index [data set]. \ Retrieved from: http://hdr.undp.org/en/ composite/GII.

USAID. Demographic and Health Surveys Program (2016). Demographic and health surveys. Rockville: USAID, The DHS Program. Retrieved from: http://www.dhsprogram.com/.

Viola, T. W., Salum, G. A., Kluwe-Schiavon, B., Sanvicente-Vieira, B., Levandowski, M. L., & Grassi-Oliveira, R. (2016). The influence of geographical and economic factors in estimates of childhood abuse and neglect using the childhood trauma questionnaire: A worldwide meta-regression analysis. Child Abuse & Neglect, 51(1), 1–11.

WHO (2016a). Child Maltreatment Fact Sheet. Retrieved from http:// www.who.int/mediacentre/factsheets/fs150/en/.

WHO (2016b). INSPIRE: Seven strategies for ending violence against children. Geneva: Author. Retrieved from: http://www.who.int/ violence_injury_prevention/violence/inspire/en/.

World Economic Forum (2014). Global Gender Gap Report, 2014: Gender Gap Index [data set]. Retrieved from: http://reports. weforum.org/global-gender-gap-report-2014/wp-content/blogs.dir/ 60/mp/files/pages/files/gggr-2014-table-b1.pdf.

806 J Fam Viol (2017) 32:799–806

Journal of Family Violence is a copyright of Springer, 2017. All Rights Reserved.