Introduction The sex ratio at birth

The sex ratio at birth, generally defined as the number of male births divided by the number

of female births, is relatively stable for human beings over time and across space.

Historically, consistent sex ratio imbalances have rarely been observed except for some

crises, such as wars and epidemics. The rise in the ratios of belligerent countries during and

after wars has been observed in France, England, and Wales during World Wars. A similar

rise has also been acknowledged in the white population of the United States at the time of

World War II (Chahnazarian, 1988). As far as the disease, the Hepatitis B virus (HBV) has

been recognized as one possible cause to the high sex ratio because an HBVinfected mother

is 1.5 times more likely to give birth to a male (Chahnazarian, Blumberg, & London, 1988).

In addition to these reasons, a more striking imbalance in the sex ratio at birth has emerged

since the wide technological application for fetal sex inspection, especially being coupled

with the cultural preference for one sex over the other and lowered fertility rates. In turn,

such an imbalance could constitute an alteration to social norms and values, and lead to

behavior modifications. Previous research has widely explored its influence on the

marriage market but neglected the other implications of the sex ratio imbalance in a broader

and remoter social sphere.

This dissertation connects the increasing proportion of female crime in China with

the imbalanced sex composition caused by the historically unique one-child policy.

Specifically, the dissertation explores the mechanisms of gendered crime and consequently

the gender gap in offending from an external policy force treated as a natural experiment.

Its results show that women’s share of documented crime dramatically increased after the

mid-1990s when the first one-child generation reached the age of legal responsibility. This

change reflects an interplay of behavioral changes and the net widening. The increasing

criminality of the one-child generation is attributable to the gap between similar gender

expectations at the individual level, which has been reshaped by the unique socialization

practices under the influence of the policy, and a stubbornly unequal gender hierarchy in

the society. As a result, the one-child-generation women who disproportionately experience

the resulting social strain are more likely to become involved in property and occupational

crime as the alternative means to fulfill their aspirations for economic success. Additionally,

the policy affects not only the individual gender roles of the only children but also their

peers who have siblings through the intermediary of a culture shift. Therefore, the policy

has changed the behavior of a whole generation through the process of socialization and

the lag in the structural change. The net-widening effect, moreover, is also rooted in the

unequal gender structure and traditional gender ideology. Moral panic associated with the

emergence of diverse forms of female offenses led to an inordinate degree of adverse

attention focused upon the one-child-generation women by criminal justice professionals.

The increased criminalization of sexuality brought an increasing number of the one-child-

generation women into conflict with the law, usually for prostituting themselves for easy

money to fulfill their material satisfaction. Consequently, the one-child-generation female

offenders are perceived as “doubly deviant” by the law and the socially prescribed ideology

of gender and are, therefore, punished more harshly than before by the criminal justice

system.

This dissertation not only explores an understudied country in criminological

research but also applies the findings to a broad sphere to explain the increasing female

crime that has been observed worldwide. It disentangles the theoretical controversy in

explaining the increase in the share of crime committed by women in general by embedding

the argument in a multidimensional gender role repertoire. The dissertation is organized

around three objectives.

Objective 1: Identifying the relationship between the sex composition and the share

of women’s offending, and the influence of the one-child policy on the relationship.

Demographic characteristics have always been used to explain variation in

criminality. Specifically, measures of poverty, racial composition, region, income

inequality, age distribution, and family disruption are well represented in most such studies.

However, the gender composition of the population has not been researched as frequently,

although gender is one of the most potent and substantial correlates of criminality (e.g.,

Smith & Visher, 1980; Steffensmeier & Cobb, 1981; Zimmerman & Messner, 2010;

Messner & Sampson, 1991). One reason for this phenomenon is the relatively stable gender

distribution of human beings, which limits the effects of variation in crime rates.

Guttentag and Secord (1983) proposed a dedicated framework to explain the

relationship between the gender composition of a population and gender roles and norms

in society. A society with more men than women (i.e., a higher sex ratio of the male

population divided by female population) would tend to impose more constraints on

women. When in shorter supply than, women are highly valued as marriageable partners,

mothers, and homemakers (Angrist, 2002; Scott J. South & Messner, 1987). This is a result

of the counteracting forces of dyadic power in interpersonal relationships and structural

power in societal gender separation (Guttentag & Secord, 1983). Although women’s dyadic

power is likely to be enhanced in a society with an imbalanced sex ratio where women are

relatively scarce, the overwhelming structural power owned by men limits the ways in

which women’s potential dyadic power can be exercised. As a result, women in societies

with higher sex ratios tend to marry at a younger age and, in cultural contexts, invest less

in education and career development. It is more common for women to stay at home and

work as full-time homemakers with constrained extrafamilial goals in such societies than

it is in lower-sex-ratio societies. According to this theoretical paradigm, therefore, we

expect to see more traditional gender norms and roles in societies with higher sex ratios

and, therefore, less female criminality than is observed in societies with lower sex ratios.

This theory connects gender structure and gender roles at a macro level and has had

considerable empirical support (Barber, 2001; Barber, 2000a; Messner & Sampson, 1991;

Pedersen, 1991; Porter, 2007; South, 1988; Trent & South, 1989). Some research that has

applied this theory to explaining criminality found a negative causal relationship between

sex ratio and female crime (South & Messner, 1987) or a positive relationship for male

crime (Barber, 2000b; Edlund, Li, Yi, & Zhang, 2013). However, this theory oversimplifies

the endogeneity of the gender composition of society by setting it as the cause or at least

as one crucial causal factor of gender norms. Gender composition as a demographic feature

of a society is also a product of its cultural and social structures. Almost always, an

imbalanced sex ratio is accompanied by some inherent culture and structure causing it. For

example, a society with a sex ratio higher than normal would probably have a culture

favoring sons and a structure empowering men (Becker & Posner, 2009; Das Gupta et al.,

2003). The culture and structure of society are also the determinants of gender norms and

roles. Therefore, the relationship between the demographic sex ratios of a population and

its gender norms as well its gendered behaviors is just like “two apples on one tree”—they

are both outcomes of a society’s culture and structure. The direct relationship between the

sex ratios and gender norms or gendered behaviors tends to be spurious unless we find

causal logic supporting it or the “tree trunk” generating them both. However, since they are

both products of culture and structure, it is hard to find the exogeneity of either.

Because those limitations veil the real relationship between the sex ratios and

gendered crime, a clearer understanding of how imbalanced gender structure drive crime

is needed to help us understand the underlying mechanism to criminality. China’s onechild

policy provides such a window into the social effects of sex ratio imbalance and therefore

the causal relationship between the sex ratio of a population and gendered criminality.

Objective 2: Evaluate the extent to which the one-child policy has changed the

socialization practices across generations and the effect of the potential inter-generational

differences on the gender gap in criminality.

The criminal world is an extremely masculine world. According to the statistics, up

to 84% of crimes were committed by men, and 88% of prisoners were men worldwide

(UNODC, 2015). Therefore, the gender gap in criminal offending is one of the most

acknowledged phenomena in criminology (Britton, 2000; Leonard, 1982; Renzetti, 2013).

The gender gap in offending is viewed as a product of biological and sociological

differences between men and women. The former emphasizes the natural differences

between genders, including physical and psychological aspects, whereas the latter regards

socialization processes and practices as the critical factors explaining the gender gap. After

all, gender socialization is the first and most important socialization process that a person

receives as soon as he or she is born. Almost all socialization institutions teach men and

women to behave differently. These differences in socialization eventually create different

gender roles and norms and, thereby, different gendered behaviors, including deviant ones.

However, the causal mechanisms behind the relationship between the gender gap in

offending and the socialization process are still controversial and obscure.

China’s one-child policy has externally changed the socialization practices on the

new Chinese generations. Although girls are less valued than boys in China under the

influence of cultural preference for sons, the imbalanced sex composition in favor of boys

may benefit girls who survived because (a) the average parental treatment of girls improves

because the ones who are not really wanted by their parents have been aborted, (b) there is

more investment in the single child's education and human capital development, regardless

of gender, in families with fewer children (Becker & Posner, 2009), and (c) for those

families with only a girl, they put their hopes in her, since no family wants to give up their

dream just because they only have a girl. In addition, extensive propaganda during the

implementation of the one-child policy, designed to educate people that girls are as valuable

and promising as boys, plays an important part in changing gender socialization and gender

roles, although it has little effect in eradicating prenatal sex selection as expected.

Therefore, the gender socialization process in China for the one-child generation is

remarkably different from traditional practices. Women and men are treated more similarly

by parents, teachers, relatives, and friends. The one-child-generation women are expected

to be ambitious, competitive, and goal-oriented—just like their male peers. As a result,

women are less relation-concerned or family-centered than before. Instead, similar to men,

they are achievement-oriented (Edlund et al., 2013). Consequently, economic success and

status achievement is no longer solely a masculine goal. It is shared by both genders in the

new generations in China.

Therefore, the one-child policy artificially changes the socialization of girls and

boys and, therefore, it also affects the outcomes of socialization, such as the criminality of

men and women. An exploration of the policy’s effects on socialization and deviance helps

to clarify the etiological mechanisms of socialization facilitating crime.

Objective 3: Evaluate the influence of the one-child policy on the treatment of

offenders by the criminal justice system and thereby affect the gender gap in crime.

Some criminologists have proposed that the convergence may not be due to actual

changes in men’s and women’s criminality but is due to changes in crime control and

punishment within the criminal justice system. In contrast to a behavior change hypothesis,

then, these criminologists posit the net-widening hypothesis. The net-widening hypothesis

suggests that women’s documented crime rates may be increased in two ways. First,

societal attitudes toward women offenders change from being sympathetic to punitive,

which leads to greater willingness by citizens to report women suspects, a higher

probability that police will arrest them, and more severe punishment from the courts

(Steffensmeier, 1980). Second, the increasing bureaucratization and routinization of the

criminal justice system decreases individual discretion and leads to non-discriminatory

standards and outcomes (Chesney-Lind & Shelden, 2014; Feld, 2009).

If a change in crime is captured by documented official data, it is necessary to

differentiate whether this change reflects a real change in behavior or is an artifact of the

responses from enforcement authorities to clarify the influence path of the one-child policy.

1.2. Methodology

Addressing these objectives requires a combination of empirical research methodologies

based on an integration of theories. This study uses a mixed-method approach to studying

the effects of the one-child policy on the gender gap in crime in China and draws heavily

from my fieldwork in China.

The quantitative research involves three kinds of methods. First, time-series

analysis methods are employed to capture the general relationship between the sex ratios

and female share of crime from the trends of these variables. I collected data from archived

official records in China, from 1972 to 2014. The policy’s effect is measured by a proxy

variable—the sex ratios of a cohort aged 0 to 4—to show its external force. The Student’s

Test method is used for the inter-generational comparisons of socialization practices

influenced by the policy. In addition to the generation category, I also compared gender

differences to show the gender gap in socialization. I collected data from 2,006 responses

to an online survey of Chinese adult citizens from 31 provinces of China as well as from

Hong Kong and Taiwan. Following with the comparisons, I employed structural equation

modeling (SEM) to evaluate the direct and indirect paths of socialization to criminality.

Corresponding to the comparative categories by genders and over generations, I also

estimated and compared four models.

The qualitative research in this dissertation focuses on the different treatment of

female and male offenders by the Chinese criminal justice system and professionals. To

examine the potential effect of the one-child policy, the analyses center around

differentiation by genders (men and women) and over generations (pre-1980s and

post1980s, i.e., pre-one-child and one-child generations). Comparisons are made between

these two sets of groups to capture any inter-generational change in the gender gap in crime.

I adopted interview methods for this study. The research participants were professionals in

the criminal justice system in China.

1.3. Structure

The dissertation is structured as three papers. The findings of the first study set a backdrop

for the other two studies. The second study explains the change found by the first study

from a socialization perspective. The third study investigates the question from the

perspective of justice system perceptions but additionally verifies the findings in the first

and second papers (see Figure 1.1 for the theoretical framework).

The first paper, Chapter 2, pursues the first objective of identifying the relationship

between the sex composition and the share of women’s offending under the influence of

the one-child policy. It provides a quantitative examination on how the one-child policy

influences the female offending in China. This article uses time-series analysis methods to

trace the effects of that policy and the accompanying imbalanced sex ratios at birth on the

percent of female offenders as of the one-child cohort reaching the age of criminal

responsibility. It explores the relationship between sex composition and female crime at a

macro level and provides a premise on the relationship between the artificial sex ratio

imbalance and the gender roles for the following research. This paper has been submitted

to The Sociological Quarterly for peer review.

Chapter 3, the second paper, addressing the second objective of the dissertation,

compares the socialization practices between the one-child and pre-one-child generations

and, thereby, evaluates the influence path of the socialization to criminality. It explains the

change in the gender gap in crime from the behavioral perspective and ascribes the change

to the socialization process in family, school, peers, and workplace. The findings of this

paper support those of the first study with an exploration of the causal pathways.

Chapter 4, the third paper, addresses all three objectives of the dissertation from the

perspective of Chinese criminal justice professionals. It answers the trend in the gender gap

in crime and, moreover, clarifies relative gender contribution to the change and explains

what factors drive the change. Both changes in individual criminal behaviors and responses

to those behaviors from the criminal justice system are examined with consideration of the

effect of the one-child policy. This study not only supports the findings of the other two

papers but explores the net-widening effect on the documented crime.

The fifth and last chapter concludes by summarizing the implications, limitations,

and extensions for future research.

1.4. Figures

Figure 1.1. Dissertation theoretical framework.

Chapter 2. Sex Ratios and Female Crime

2.1. Introduction

Demographic characteristics have always been used to explain variation in criminal

offending. Specifically, most of these studies represent measures of poverty, racial

composition, region, income inequality, age structure, and family disruption. However, the

gender composition of the population has not been commonly researched, although gender

is one of the most potent and substantial correlates of delinquency (e.g., Messner &

Sampson, 1991; Smith & Visher, 1980; Steffensmeier & Cobb, 1981; Zimmerman &

Messner, 2010). One reason for this phenomenon is the relatively stable gender composition

of human beings, which hinders its effect on variation in delinquency rates to be revealed.

Fortunately, we can take advantage of a unique social experiment in China, which caused a

significant shift in the gender composition and, thereby, gives us a “window” to explore its

actual effect on crime.

Guttentag and Secord (1983) proposed a dedicated framework to explain the

relationship between the gender composition of a population and gender roles and norms

in society. A society with more men than women (i.e., a higher sex ratio) would tend to

impose more traditional constraints on women. When in short supply, women are highly

valued as marriageable partners and after that as mothers and homemakers (Angrist, 2002;

South & Messner, 1987). This is a result of the counteracting forces of dyadic power in

interpersonal relationships and structural power in societal relationships between genders

(Guttentag & Secord, 1983). Although women’s dyadic power is likely to be enhanced in a

society with an imbalanced sex ratio where they are scarce, the overwhelming structural

power owned by men limits the ways in which women’s potential dyadic power can be

exercised. As a result, women in societies with higher sex ratios tend to marry at a younger

age and, therefore, invest less in education and career development. It is more common for

women to stay at home and work as full-time homemakers with constrained extrafamilial

goals in those societies than in the lower-sex-ratio societies. According to this theoretical

paradigm, therefore, we expect to see more traditional gender norms and roles in societies

with higher sex ratios and, therefore, less female offending and vice versa.

This theory connects gender structure and gender roles at a macro level and has had

considerable empirical support (Barber, 2001; Barber, 2000a; Messner & Sampson, 1991;

Pedersen, 1991; Porter, 2007; South, 1988; Trent & South, 1989). Some research has

applied this theory in explaining criminal offending and found a negative causal

relationship between sex ratio and female crime (Anne Campbell et al., 2001; Scott J. South

& Messner, 1987) or positive for male crime (Barber, 2000b; Edlund, Li, Yi, & Zhang,

2013). However, this theory oversimplifies the endogeneity of the gender composition by

setting it as the cause or at least as one crucial causal factor of gender norms. Gender

composition as a demographic feature of a society is also a product of its cultural and social

structures. Usually, an imbalanced sex ratio is accompanied by some concomitant cultural

and structural specific. For example, a society with a sex ratio higher than normal would

probably have the culture favoring sons and the structure empowering men (Becker &

Posner, 2009; Das Gupta et al., 2003). Meanwhile, the culture and structure of society are

also the determinants of gender norms and roles. Therefore, the relationship between the

demographic sex ratio of a population and gender norms as well as gendered behaviors is

just like “two apples on one tree”—they are both outcomes of a society’s culture and

structure. The direct relationship between the sex ratio and gender norms or gendered

behaviors tends to be spurious unless we find causal logic supporting it or the “tree trunk”

generating them both. However, since they are both products of culture and structure, it is

hard to find exogeneity of either.

In this paper, I take advantage of a radical social experiment—the one-child policy

in China—as a window into the social effects of sex ratio imbalance to assess the causal

relationship between the sex ratio of a population and gendered offending. By setting the

sex ratio as an exogenous variable, I aim to provide a counter-example from a society with

an artificially imbalanced sex ratio to challenge the current theory.

2.2. China’s One-Child Policy

its one-child policy and has since regulated the birth and life of its people. The original and

most important intention of this policy was to control the fertility rate and curb the then

fast-growing population, thereby concentrating limited resources on enhancing population

quality and eventually eliminating poverty in China. The content of the policy varied from

province to province, but the basic guideline was similar. For the Han ethnic group

(comprising over 90% of the Chinese population), every urban couple could only have one

child regardless of its sex; every rural couple could have two children but only if the first

was a girl, or only one child if the first was a boy. This policy put fewer restrictions on

minority ethnicities. To assist the process, contraception (including distributing free

condoms and sterilization surgeries for either men or women) and abortion were widely

used to achieve the goal of controlling fertility at such a low rate.

However, the one-child policy went against the cultural preference for sons. Chinese

people valued boys more than girls because of the prevalent desire to raise sons to provide

for one’s old age (Chen & Silverstein, 2000). Sons were perceived as the future of a family

and the guarantees of their parents’ security in Chinese culture. In contrast, daughters were

regarded as a debt or an economic burden since they were expected to join another family

upon marriage, ceasing to take responsibility for care for their parents when they were ill

or elderly (Hesketh & Xing, 2006).

As a response to this conflict between culture and policy, some practical methods

were used to circumvent or reduce the policy’s influence on the traditional preference of a

child’s sex. Among all such methods, prenatal sex selection was the most popular one.

Especially after the introduction of ultrasound for prenatal inspection in the early of 1980s,

fetal sex identification followed by gender-specific induced abortion was pervasive.

According to a medical project covering eight provinces and cities in 1986, the sex ratios

among 500 and 1,226 aborted sex-identifiable fetuses in rural and urban areas were 94.6

and 96.8, respectively (Zeng et al., 1993). These ratios can be compared with the real sex

ratio of living birth that year (107.68) as evidence of the impact of offspring sex selection

on gender structure.

Along with a stunning drop in the fertility rate, the gender composition of the new

Chinese generation eventually changed dramatically as a side effect of 35 years of the most

radical and largest social experiment in history (see Figure 2.1). Hesketh and Zhu (2006)

estimated that there were up to 34 to 41 million “missing females” in China by the end of

2000. This number is even higher according to Amartya Sen's (1990) estimate that there

were 50 million “missing women” in China alone by the end of the 1980s. These staggering

numbers of missing women eventually brought about 32 million surplus men than women

among the one-child generation by 2005 (Zhu, Lu, & Hesketh, 2009) and a deficit of one

million marriageable women in the first-marriage market per year after 2010 (Tuljapurkar,

Li, & Feldman, 1995). Men occupied about 97% of all unmarried persons aged 28 to 49 in

China (Hudson & Den Boer, 2002). Although the policy has been changed, its influence is

lasting and striking in every aspect of the society.

A large proportion of unmarried young men in a society could be an unstable factor

leading to conflicts, social disorders, and crimes. A substantial amount of literature links

high delinquency rates with a large population of young, unmarried, and competitive men

through psychological, biological, financial, and sociological processes (Barber, 2003;

Cameron, Meng, & Zhang, 2016; Cheatwood & Block, 1990; Mesquida & Wiener, 1996;

Wilson & Daly, 1985). Accordingly, most criminological forecasts predicted the male

percentage of offending in mainland China would go up when numerous single men could

not find wives (e.g., Den Boer & Hudson, 2004; Edlund, 2005; Edlund, Li, Yi, & Zhang,

2010, 2013, Hudson & Den Boer, 2002, 2004; Tuljapurkar, Li, & Feldman, 1995). In

contrast, the female crime rate unexpectedly outpaced the male rate and had led to a

substantial increase in the percent of female criminal offending (calculated as the number

of female offenders/the total number of offenders) in recent decades (see Figure 2.2). This

surprising outcome has raised questions regarding the relationship between the sex ratio

and gendered patterns of offending.

2.3. Theory and Literature

2.3.1. Sex Ratio, Cohort Size, & Gendered Delinquency

Guttentag and Secord (1983) explored the relationship between sex ratios and sex roles

historically and internationally and proposed a power-matrix paradigm for explaining the

relationship between the sex ratios and gender roles. There are two types of power in the

society; one is the micro-dyadic power attached to interpersonal relationships, and the other

is the macro-structural power that is related to the social hierarchy of genders. Compared

to the latter, dyadic power is more flexible according to the relative dependence between

two parties. Therefore, one gender gains the dyadic power when it is in relative

undersupply. However, the implementation of the dyadic power must be subject to the

structural power. Societies with higher sex ratios have more traditional social norms with

respect to female roles because the structural power owned by males is strengthened by the

patriarchal structure. Under these circumstances, women are highly valued as homemakers

and mothers (Angrist, 2002; Scott J. South & Messner, 1987) and, so, tend to marry at a

younger age and have fewer career aspirations than their counterparts in low-sex-ratio

societies. As such, girls and women tend to stay at home, work as full-time homemakers,

and invest less in education and career development with constrained extrafamilial goals,

although they may have a subjective sense that they are powerful in the relationship and be

able to require a substantial economic investment from wooers or partners, such as a “quasi-

wage” or intra-marriage transfers of income (Grossbard, 1984), or a lifetime commitment

to the marriage as an economic and reproductive union (Guttentag & Secord, 1983; South

& Messner, 1987). Moreover, chastity is often prized in high-sex-ratio societies, with

women usually controlled strictly by their fathers before marriage or by their husbands after

marriage (Guttentag & Secord, 1983). In contrast, women in low-sex-ratio societies are

relatively free from traditional role constraints, although they may have less dyadic power

in hand. Since the benefits of marriage are relatively low, women aspire for more economic

and political independence and pursue career success on a par with men. Because of their

success in extrafamilial activities and advantage in numbers, women can ask for more

structural power as a remedy for the lack of the dyadic power.

Building on this theory, South and Messner (1987, 2000) posited a negative

association between the sex ratio of the population and female involvement in criminal

offending. Because women are likely to be constrained more in traditional gender roles in

a society with a high sex ratio, their activities primarily center around family and domestic

issues. Therefore, they have less ability and opportunity to commit crimes (Adler, 1975;

Simon, 1975). Conversely, women in the lower-sex-ratio societies tend to be more

delinquent due to more extrafamilial activities providing them with opportunities to become

involved in crime. Using the female share of property crime as the dependent variable and

the sex ratio for ages 15 to 49 as the key explanatory variable, this hypothesis was supported

by a cross-national empirical study covering 60 countries (South & Messner, 1987).

Campbell, Muncer, and Bibel (2001) linked the sex ratio with female violent crimes to an

evolutionary perspective. In their model, a low sex ratio in a region tends to strain male

partner resources, which is likely to induce more female crime, especially intrafemale

assaults, for competing for male partners and future paternity and paternal investment

(Buss, 1994). This proposition has also been applied to explain increasing black female

crime in the United States. Due to race-specific mass incarceration and the increasing

economic marginality of black men, women in the black community have to face the

shortage of suitable male partners who are expected to hold a higher socioeconomic status

than themselves (Becker & Posner, 2009). The regional low sex ratio leads to a significant

increase in the proportion of female-headed households and female crimes (Taylor,

Chatters, Tucker, & Lewis, 1990).

O'Brien (1991) studied the effect of the sex ratio on rapes testing the thesis of

Guttentag and Secord (1983) and found a negative relationship between the two. In other

words, rape incidents are less likely in societies with higher sex ratios and more likely when

the sex ratios are lower. O'Brien ascribed this association to the degree of men’s protection

over women and women’s extrafamilial activities producing opportunities for perpetrators.

Moreover, in a low-sex-ratio society, rape can be seen as a response to the threat to men

posed by women engaging in nontraditional gender roles and a restoration of the challenged

structural power (O'Brien, 1991).

Studies directly exploring the effect of the sex ratio on overall crime rates have

mixed results. Barber (2000) found a negative association between the sex ratio and violent

crime through the mediating effect of lower sex ratios inducing more family conflict

(Guttentag & Secord, 1983), which would, in turn, lead to a higher level of delinquency.

However, most research suggests a positive association between the sex ratios and crimes

(Dreze & Khera, 2000; Edlund et al., 2013; Messner & Sampson, 1991; Oldenburg, 1992).

A society with more men, especially the young and unmarried, is likely to have more

conflicts and delinquency due to more motivated offenders, fiercer competition for a wife,

and less social control from marriage (Edlund et al., 2013). Messner and Sampson (1991)

ascribed this inconsistent or “null” effect of the sex ratio to the countervailing direct and

indirect effects and argued that the negative indirect effect suppressed the positive direct

effect of sex ratios on criminal offending via family disruption. However, these empirical

studies did not disaggregate the crime rate by the sex of the offenders, so they were

relatively ineffective in explaining the effect of sex composition on gender roles or

gendered offending. In other words, it is hard to say which sex contributes more to the

marginal change in the crime without considering sex-specific offending rates.

Sex ratios affect gender roles and behaviors through the “marriage squeeze”

(Caldwell, Reddy, & Caldwell, 1983; Hvistendahl, 2011; Trent & South, 2011). In a

highsex-ratio society, there is likely to be a marriage squeeze for men, which means they

would face a shortage of suitable female marriage partners; on the contrary, the marriage

squeeze is experienced by women when marriageable men are undersupplied. Moreover,

birth rates and corresponding cohort sizes may catalyze or buffer the marriage squeeze

effect of imbalanced sex ratios on gender roles and the gender gap in offending (Guttentag

& Secord, 1983; Hvistendahl, 2011). Because a traditional age gap exists between spouses,

such that men are on average older than women, the influence of the sex ratio imbalance is

mediated by the corresponding age cohort size. In a society with a surplus of men but a

deficit in women, an expanding cohort size weakens the marriage squeeze effect of the sex

ratio imbalance, and a shrinking cohort size intensifies the effect. Conversely, in a society

with a surplus of women but a deficit in men, an expanding cohort size intensifies the effect

of the sex ratio imbalance, and a shrinking cohort size weakens the effect. Research about

the relationship between the sex ratio and cohort size has focused on the causal effect of a

shrinking cohort size on the sex ratio imbalance (Das Gupta & Li, 1999; Hvistendahl,

2011), but rarely has studied the mediating effect of cohort size on the marriage squeeze

pressure caused by the imbalanced sex ratio.

2.3.2. Economic Development & Gendered Delinquency

Economic development has effects on gender roles and gendered offending in different

ways. Straightforwardly, the economic situation contemporaneously influences the

occupational opportunities faced by men and women. In addition, the economic

development may also have an effect on gendered socialization through its influence on the

relative positions between parents and, thereby, the gender norms and roles of offspring.

The latter reflects an inter-generational effect of the economic development on gendered

offending, whereas, the former can be categorized as an intra-generational effect

correspondingly, which is confined within the contemporary generation.

2.3.2.1 Intra-generational Effect

The process of economic development plays a role in influencing the levels of criminal

activities of both genders. Along with economic development, there is more property

available to be stolen (Nettler, 1982) and more extrafamilial activities available for both

genders to participate in, especially for women who are traditionally constrained by

domestic responsibilities (Shelley, 1981). As women become less comfortable with

traditional gender roles and more ambitious about social roles outside of the home, they

would strive to behave more like men in a variety of ways, including delinquent behaviors,

thereby increasing the female involvement in criminal activities (Adler, 1975). Moreover,

along with more participation in business and economic affairs, women enjoy more

opportunities, including educational, occupational, and criminal ones, due to the liberation

from traditional gender roles (Simon, 1975). The emancipation theory generated a

significant amount of research about the relationship between equality and the gender gap

in offending (e.g., Chamlin & Cochran, 1997; Engelhardt, Rocheteau, & Rupert, 2008;

Hunnicutt & Broidy, 2004; Messner & Golden, 1992; Ruppanner, 2010; Steffensmeier &

Streifel, 1992). The results regarding the relationship between gender equality and

gendered offending rates have been mixed, in part because of the variety of ways that

gender equality has been operationalized. Emancipation theory focuses on the “bright side”

of economic development by connecting it with gender role liberation and predicts a

positive association between the economic development and female involvement in the

crime.

2.3.2.2 Inter-generational Effect

Hagan (1990) and Hagan, Simpson, and Gillis (1987) proposed power-control theory in

explaining the inter-generational influence of gender power dynamics on the delinquency

of man and women. Wives have much less power than their husbands in patriarchal families

in which husbands are employed in an authority position while wives work as fulltime

homemakers; correspondingly, daughters enjoy much less freedom than their brothers in

the same family. The former imbalance explains the latter one, whereas the latter imbalance

explains the reason why girls traditionally commit less crime than boys because (a) the

gender division in domestic work imposes more social control on daughters and (b) the

resulting attitudes toward risk make daughters more risk-averse. Therefore, patriarchal

families tend to reproduce daughters to focus their futures around domestic labor and

consumption following the model of their mothers, whereas they reproduce sons to prepare

for participation in direct production following the model of their fathers (Hagan et al.,

1987; McCarthy, Hagan, & Woodward, 1999). However, with more women joining the

labor force, more women are increasing their power at home. This results in the formation

of more egalitarian families, in which the consumption and production spheres are

undivided by gender. Having relatively equal power at home results in the similar treatment

of sons and daughters. Boys are encouraged to reject the traditional masculine gender norm

supporting risk-taking activities and delinquency, whereas girls are encouraged to be

independent and ambitious. Therefore, the gender gap in delinquency is narrower in the

egalitarian families than it is in the patriarchal families because of the decrease in male

offending and the increase in female offending (Hagan et al., 1987; McCarthy et al., 1999).

Therefore, family pattern, either patriarchal or egalitarian type, can reproduce itself from

generation to generation and keep the gender inequality or equality relatively stable.

Most empirical research on power-control theory provided at least moderate support

for its central argument (Blackwell & Piquero, 2005; Grasmick, Hagan, Blackwell, &

Arneklev, 1996; Hagan, 1990; Hagan, Gillis, & Simpson, 1985; Jensen, 1993; Leiber &

Wacker, 1997; McCarthy et al., 1999; Morash & Chesney-Lind, 1991). Women experience

changes in the workplace, which brings about more control over resources and power

outside and at home. This change increases women’s chances to exercise their power in

family relations. Consequently, child-rearing practices for sons and daughters are also

transformed (McCarthy et al., 1999; Morash & Chesney-Lind, 1991). As egalitarian

families become more common and prevalent, daughters become freer and less constrained,

thereby increasing preference for risk among women, thus reducing gender difference in

delinquency in the society (Blackwell & Piquero, 2005; McCarthy et al., 1999). Empirical

tests of power-control theory suggested that gendered delinquency changes depending on

the level of patriarchy and corresponding control at home.

Furthermore, the power dynamics between parents may also influence the offspring

sex ratio. Cultural son preference is rooted in a functionalist perspective that men are better

able to provide their parents with economic security than women since their higher capacity

in wage earning. As such, if women make more or as much as money as their male

counterparts, their parents should adjust the pro-male bias and balance back their preference

of offspring sex (Hudson & Den Boer, 2002; Zeng et al., 1993). Therefore, if the power

dynamics are balanced between a couple, the son preference should be weakened, which

correspondingly leads the sex ratio to be balanced. Another link between sex ratio and

economic development is evidenced by the positive association between parents’ education

and occupation and offspring sex ratio (Banister, 2004; Gu & Roy, 1995; Zeng et al., 1993).

Education increases the awareness of potential technological options for offspring sex

selection, whereas good occupations make those options affordable (Hvistendahl, 2011).

These theories connect the sex ratio, economic development, and gendered

delinquency from different perspectives, but all three have rarely been tested in the same

analysis. Moreover, the stable and endogenous nature of gender composition limits the

explanatory power of the sex ratio on gendered roles and behaviors in previous research.

This study aims to address these gaps in the literature based on an experiment that changed

the population sex ratio dramatically in a short time and provide a counter-example from a

rarely researched society with substantively different culture from the “mainstream”

Western countries to challenge the current theory.

2.4. Research Questions and Hypotheses

This study aims to identify the means by which the sex ratio and economic development

influence gendered offending and to test if Guttentag and Secord’s (1983) thesis holds when

the sex ratio imbalance is artificially and exogenously caused or (at least) strengthened by

a policy. As such, there are three primary questions I seek to address with the present study.

First, in discussing the primary effect of the exogenous shock caused by the onechild

policy, I need to ask whether and to what extent the shock changed the relationship between

the sex ratios and gendered offending. If the gendered behaviors are influenced by sex ratios

as predicted by Guttentag and Secord (1983), the shock is supposed to be irrelevant to this

causal relationship. In contrast, if the relationship is changeable to the exogenous shock,

there should be a mechanism underlying the relationship that was impacted by the shock.

To answer this question, this study adopts the policy as the structural breaker to check if

the effect of the sex ratios of the marriageable population on the gendered offending has

changed after years of policy enactment.

Hypothesis 1: The effect of the sex ratio of the marriageable population on the

gendered offending is expected to be structurally broken by the one-child policy after a

generation of its enactment.

By controlling for the effect of the shock via its proxy variable, the second objective

of this study is to assess the real effect of sex ratio on gendered offending. On the one hand,

I believe there is a relationship between the sex ratio and gendered offending as most

research has assumed to be the case; on the other hand, I suspect this relationship is only

the appearance of an underlying causality in the culture and structure leading to concurrent

changes in the sex ratio and gendered offending. The effect of sex ratio on gendered

offending may not be same as predicted after controlling the exogenous influence, which

means the policy has changed the underlying factor and, therefore, the relationship could

not keep up. As such, this study proposes the second hypothesis.

Hypothesis 2: After controlling for the exogenous shock from the one-child policy,

the effect of the sex ratios of the marriageable population on the gendered offending is

expected to be non-negative and, thereby, different from the prediction of Guttentag and

Secord (1983).

Last, this study also tests the relationships between the economic development and

the gendered offending from intra- and inter-generational perspectives. According to the

literature, the third hypothesis tested in this study is as follows.

Hypothesis 3: Economy is expected to be positively related to the percent of female

offending contemporaneously based on emancipation theory, and positively related to the

one-generation-lagged percent of female offending based on power-control theory.

2.5. Data and Methods

2.5.1. Data

I adopted the percentage of female offenders from 1972 to 2014 as the dependent variable

to show the trend in gendered offending. However, because gendered offending data in

China were not always available, I constructed a proxy measure, combining two relevant

variables in time series to make up the missing values in certain years. There are two official

databases in China reporting the gendered offending rates but in different measures. One is

the China Statistics Yearbooks reporting the annual number of prison inmates and female

inmates; the other is the Chinese Yearbooks of the Court of Justice reporting the female

share of convicts. Nevertheless, the availability of the time-series data for each variable

separately was too low for our study as it involved tests of one policy’s accumulative effect

on structural and behavioral changes across generations. The official data for the female

share of inmates are only available from 2003 to 2012, whereas the data for the female

share of convicts is only available from 1994 to 2007 (c.f. Lin, 2012). Other secondary data

sources for the female share of inmates include the Department of Correction in China for

data from 1972 to 1985 (c.f. Tong, 1995), the Supreme Court in

China from 1986 to 1992 (c.f. Tong, 1995), and a BBC news report for the variable in 2014

(Hatton, 2015). Therefore, the time-series from 1992 to 2002 are missing for the female

share of inmates, which is supplemented by the female share of convicts. Although the

dependent variable is hybrid, the time-series trends of these two types of gendered

offending rates are highly consistent (see Figure 2.3). To capture the influence of the data

source change on the results, I generated a dummy variable controlling the data source

effect in the following regressions. The dependent variable is calculated as [the number of

female offenders/the total number of offenders*(1+ the sex ratio of the population)]. This

method ensures the dependent variable is controlled for gender composition and, therefore,

reflects the female percentage of recorded offenders in the population filtering out noise

from sex ratios.

The independent variables were obtained from publicly available data from national

agencies, including population censuses (1964, 1982, 1990, 2000, 2005, 2010), the Chinese

Population & Employment Statistics Yearbooks (named as the Chinese Population

Statistics Yearbook before 2000; 1987-2015), and the China Statistics Yearbooks

(20002015). There are three variables related to our research questions. To test Hypothesis

1 and 2, this study specifies the marriageable cohort as population aging from 15 to 49

because this cohort is the primary population experiencing marriage squeeze that is

influenced by the sex ratio imbalance according to Guttentag and Secord's (1983) thesis

(South & Trent, 1988; South, 1988). Two variables are derived from this cohort: sex ratio

and cohort size variation. The former is calculated as the ratio of male population per 100

female population in the cohort, and the latter calculated as (cohort size at year t / cohort

size at year t-1). Therefore, if the rate is higher than 1, the cohort size is expanding; if less

than 1, the cohort size is shrinking. To answer the second question, I used the sex ratios of

the cohort aged 0 to 4 as the proxy variable to reflect the policy effect. Rather than the sex

ratio at birth, I chose the sex ratio of a larger cohort because I want to include not only the

influence of the prenatal sex selections but also postnatal sex selections against females in

our model so to manifest the influences of the one-child policy more comprehensively.

I expect to see a higher sex ratio of young children in years when the one-child policy was

implemented more strictly.

Based on theoretical predictions about the relationship between economic

development and gendered offending, I adopted GDP per capita and its 27-year lagged form

as another two independent variables. Higher GDP per capita indicates higher economic

development, and, therefore, there should be more occupational and extrafamilial activities

in the economy available for females, with which the percent of female offenders is

expected to be higher. The variable at the same period controls for the intra-generational

effect of economic development on gendered offending; whereas the lagged variable

controls for the inter-generational effect.

I adopted the linear interpolation method for missing values for all variables.

According to the t-statistics of the augmented Dickey-Fuller (ADF) test, all variables except

the marriageable cohort size variation are non-stationary at level but stationary at their first

difference. Therefore, I included their first difference forms in our models (See Table 2.1

and 2.2 for descriptive data statistics and zero-order correlations).

2.5.2. Methods

Although the time for the one-child policy announcement is known, its effect on the

structure is not apparent because the exact time for implementation was different from

province to province and the effect would be manifested after certain years of accumulation.

To test Hypothesis 1, therefore, we adopted the Chow test (Chow, 1960) to detect any

possible structural break changing the effect of sex ratio of the marriageable cohort on

female offending during the period from 1973 to 2014.

The time of breakpoint is meaningful to our model specification, from which we

can determine the number of lagged years for the proxy variable of the policy (sex ratio of

cohort aged 0 to 4). Building on the structural breakpoint, we used time-series and robust

OLS methods to test our model to see whether and how much the policy changed the effects

of sex ratio and cohort size on the gendered offending as well as the degree to which the

economic development can explain the percent of female offending.

2.6. Results

Tables 2.3 and 2.4 show the results of the Chow breakpoint test, from which we can identify

that year 1997 is a structural breakpoint for the effect of sex ratio on gendered offending.

Before 1997, the effect of sex ratio on the percent of female offenders is consistent with

Guttentag and Secord’s (1983) thesis, although it is not statistically significant; however,

the coefficient signal was changed to be significantly positive by the policy after 17 years

of its performance. After 1997, every 0.01-unit increase in the sex ratio would bring about

8.25% increase in the female percentage of offenders, which is contrary to the theoretical

prediction. Therefore, the exogenous policy shock structurally changed the relationship

between the sex ratio and gendered offending, which supports Hypothesis 1. Moreover, the

dummy variable is not significant in the breakpoint test, which means the break is caused

by the policy rather than the data source change.

According to the Chinese criminal law, the age for full penal responsibility is

sixteen; therefore, the structural breakpoint has its practical meaning. It reflects that the

change in gender composition caused by the policy had a lagged effect on gendered

offending through socialization and that the effect showed up when the influenced

generation reached the age of legal responsibility.

Table 2.5 shows the results of the specified model guided by Hypothesis 2 and

Hypothesis 3. The effect of the policy is manifested by the coefficient of the proxy variable

in model 3, which is positively related to the percent of female offenders. In other words,

the offspring selection against girls would bring in more female offenders after the

generation grew up. Different from the theoretical predictions (Guttentag & Secord, 1983;

South & Messner, 1987), the sex ratio of the marriageable cohort has no significant effect

on the percentage of female offenders although the coefficient signal conforms to the

expected relationship (model 3). However, the marriageable population variation has a

positive effect on the percent of female offenders, and the counteracting effect of these two

is also significant. Therefore, the relationship between sex ratio and gendered offending

does not stand. Moreover, differences between model 2 and model 3 manifest the effect of

the proxy variable on the sex ratio and size variation of the marriageable cohort. The

coefficients of the three variables related to sex ratio and cohort size are more salient in

model 3 than those in model 2, which means the real effects of these variables are

counteracted by the effect of the policy to some extent. Therefore, the inclusion of the

policy effect is necessary to show the net effects of sex ratio and cohort size variation on

gendered offending. Moreover, there is no structural breakpoint found in the specified

model, which means the proxy variable has explained the effect of the policy.

With regard to the economic development, the lagged variable for the

intergenerational effect of economic development is significantly positive; however, the

variable for the intra-generational effect of the economic development is not significant nor

consistent with theoretical predictions. Therefore, the results support power-control theory,

but not emancipation theory. The power dynamics between parents influence the gender

roles of their offspring, which shows a profound influence of socialization on gendered

offending. Therefore, Hypothesis 3 gets partial support.

Model 4 includes the dummy variable controlling for the data source. The

coefficient of the dummy variable is not significant; moreover, the differences between

model 3 and model 4 are not salient either on the statistical significances or the coefficients,

which means the data source change does not influence the results. Therefore, the results

are not an artifact of the different data sources.

These results prove to be robust to the replacement of other variables reflecting the

economic development and marriageable cohort in the robustness tests.

2.7. Discussion

This study sought to challenge Guttentag and Secord’s (1983) theory in consideration of

China’s purely exogenous shock to the sex ratio and its effect on gendered delinquency.

This artificially imbalanced sex ratio is expected to have a distinctive effect and mechanism

on gender roles and norms from those of naturally imbalanced sex ratios because of the

gender socialization process. Rather than a natural derivation from culture and structure,

the socialization practices experienced by the one-child generation in China changed

dramatically because of the policy and, thereby, led to the changes in the relationship

between the sex ratio and female offending. Therefore, we suspect the underlying factor

influencing both sex ratio and gendered offending is the process of gender socialization.

The results provide support for this argument with manifestation as to when and

how the exogenous shock took effect. After 17 years of enactment, the lagged effect of the

one-child policy appeared when the first one-child generation grew up to the age of criminal

responsibility. The artificially and exogenously imbalanced sex ratio produced more female

offending outpacing male offending, which is an unexpected outcome. Combined with the

positive inter-generational effect of economic development on the percentage of female

offending, we speculate this is because of the unique socialization process accompanying

the sex ratio imbalance. The propaganda designed for educating people that girls are as

valuable and promising as boys in the process of implementing the one-child policy did not

eradicate prenatal sex selection as expected, but played a part in

marriageable cohort age range to 20 to 49, considering the legal age for marriage in China

is 20 and 22 for women and men respectively. The significances and signals of all

independent variables keep the same, so our model is robust. Results are available upon

request.

changing gender socialization (Das Gupta et al., 2003). Every family put its hopes on the

only child, girl or boy, and treated them more similarly. No family wanted to give up its

dream just because they only had a girl. This socialization process was remarkably different

from the traditional practices, such that girls were raised to be ambitious, competitive, and

goal-oriented just like their male counterparts. As a result, Chinese women are less

relationconcerned than before, but similarly as men, are more achievement-oriented

(Edlund et al., 2013). This unique gender socialization process is distinctively accompanied

with and caused by the artificial sex ratio imbalance and low fertility rate. As a result,

economic success is no longer solely a masculine goal. It is shared by both genders in the

new generation in China. In contrast with this equality goal, however, the distribution of

the means to achieve it is still unequal in a patriarchal society with structural power

dominated by men. Compared with men, women have much fewer opportunities, especially

legitimate ones to fulfill their goals. According to the Global Gender Gap Report 2016, the

gender gap in earned income in China was 1: 0.676 although the gender parity has improved

significantly during the past decades. To conquer this strain (Merton, 1938) faced

specifically by young females in China, illegal means to success became an alternative

option for some.

After controlling the policy influence, the effects of the sex ratio and size variation

of the marriageable population are not completely consistent with the theoretical

predictions (South & Messner, 1987). An expanding marriageable cohort relieves the

pressure of a marriage squeeze where females would be less constrained to the traditional

roles by the structural power owned by males and could be involved in more extrafamilial

activities, including illegal ones. Conversely, a shrinking marriageable cohort boosts the

marriage squeeze pressure, so females were more constrained and less in pursuit of goals

out of the home. However, the sex ratio of the marriageable cohort is not significant in our

results, although it plays a counteracting role to the effect of cohort size change on gendered

offending.

Finally, the inter-generational effect of economic development plays a part in

determining the percentage of female offending in the offspring generation. A higher

position held by mothers or a more balanced power structure between parents is influential

to the socialization of more neutral gender roles of offspring. However, the

intragenerational effect of economic development is not consistent with theoretical

expectation. This may be due to the countervailing effects produced by more criminal

opportunities with occupations available for females and less economic temptation to

commit crime with a legal and stable income.

2.8. Conclusion

The one-child policy not only controlled the Chinese population but also changed the social

structure dramatically. It increased the percent of female offending after the first one-child

generation grew up. Moreover, its effect on the percent of female offending was intensified

over time along with more single children growing up and further changes happening

within the social structure. The one-child policy structurally broke and increased the

percentage of female offending in China after 1997. By influencing the sex ratio of a very

young generation, the one-child policy unexpectedly contributed to gender role

convergence in China as one of its side-effects. In turn, the policy enhanced the percentage

of female offenders because there tends to be more similarity between gender roles in the

one-child generations and more structural strains specifically experienced by young

females in China.

Cross-national offending data from self-reports or arrest data from official records

are not available for many countries, particularly broken down by gender and/or offense

type (Stamatel, 2006). That kind of detailed data would be ideal to better understand the

gendered dynamics of the criminal offending outside of Western nations. Nonetheless, a

non-mainstream country study is meaningful because it uniquely explores the relationship

between demography and gendered behaviors within a less-researched culture with an

experimental social break. The findings unveiled the factor that undergirds a relationship

between the sex ratio and gendered patterns of offending that has been ignored in most

research based on Western samples. This is inspirational not only to the country that was

studied but also to criminological research on this topic.

2.9. Tables and Figures

Table 2.1. Descriptive Statistics for All Continuous Variables

Variable Mean

Std. Dev.

Min

Max

Obs

∆ percent of femaleoffenders 0.08

∆ sex ratio of cohort aging 0 to 4 2.71e-3

∆ sex ratio of marriageable -2.47e-4

population

change of marriageable 1.01 population

∆ GDP per capita 804.03

∆ 27-yr-lagged GDP per capita 35.31

0.22

0.02

0.01

0.03

1345.65

57.95

-0.61

-0.14

-0.03

0.96

-33

0.56

0.03

1.09

5527

255

Table 2.2. Zero-order Correlations Between Va

riables

1. ∆ percent of female

offenders

1.00

2. ∆ 17-yr-lagged sex ratio of

cohort aging 0 to 4

0.14

1.00

3. ∆ sex ratio of marriageable

population

0.17

0.26

1.00

4. change of marriageable

population

0.37**

0.05

0.62***

1.00

5. ∆ GDP per capita

0.21

0.10

0.31*

0.23

1.00

6. ∆ 27-yr-lagged GDP per

capita

0.31*

0.03

0.15

0.04

0.74***

1.00

*p<0.1; **p<0.05; ***p<0.01

Table 2.3. Chow Breakpoint Tests (F-statistics) on the Percent of Female Offenders (First

Difference)

Breakpoint

1997

∆ sex ratio of marriageable population

6.90***

5.63**

Dummy

2.88

*p<0.1; **p<0.05; ***p<0.01

Table 2.4. Structural Break on Sex Ratio Effect (Standard Errors) on the Percent of Female

Offenders (First Difference)

Breakpoint

1997

time span

1973-1996

1997-2014

1973-2014

∆ sex ratio of marriageable population

-6.15

(4.16)

8.25**

(3.57)

2.32

(2.89)

*p<0.1; **p<0.05; ***p<0.01

Table 2.5. Regression Coefficients (Robust Standard Errors) for the Percent of Female

Offenders (First Difference)

Expected

relationship

Model 1

Model 2

Model 3

Model 4

∆ 17-yrs lagged sex ratio

0-4

1.53**

(0.72)

1.44*

(0.75)

∆ sex ratio of

marriageable cohort

variation of

marriageable cohort size

∆ sex ratio * change of

marriageable population

-2.15

(3.68)

3.98e-3**

(1.70e-3)

-0.11**

(0.05)

-3.24

(4.12)

4.33e-3**

(1.68e-3)

-0.12**

(0.05)

-3.22

(4.25)

4.06e-3**

(1.77e-3)

-0.11**

(0.05)

∆ GDP per capita

∆ 27-yrs lagged GDP per

capita

-6.32e-6

(2.88e-5)

1.80e-3*

(9.53e-4)

-1.52e-5

(3.14e-5)

2.17**

(9.39e-4)

-1.64e-5

(3.11e-5)

2.22**

(9.62e-4)

-1.60e-5

(3.16e-5)

2.26e-3**

(9.81e-4)

dummy

none

-0.03

(0.09)

0.10

0.29

0.32

Obs

*p<0.1; **p<0.05; ***p<0.01

Figure 2.1. Sex Ratio at Birth in China (1953-2015). Data source: The Chinese population

censuses (1953, 1964, 1982, 1990, 2000, 2005, 2010), the Chinese Population &

Employment Statistics Yearbooks (named as the Chinese Population Statistics Yearbook

before 2000) (1987-2015), and the China Statistics Yearbooks (2000-2015).

Figure 2.2. The Percent of Female Offenders in China (1972-2014). Note: The percent of

female offenders equals the percentage of female prison inmates from 1972 to 1992 and

from 2003 to 2014 and equals the percentage of female convicts from 1994 to 2001. Data

source: the Department of Correction in China (1981-1985), the Supreme Court in China

1970

1980

1990

2000

2010

2020

year

(1986-1992), the Chinese Yearbooks of Court of Justice (1994-2001), the China Statistics

Yearbooks (2003-2012), and a BBC news report (Hatton, 2015).

Figure 2.3. Data Sources for the Percent of Female Offenders in China. Note: The percent

of female offenders equals the percent of female inmates from 1972 to 1992 and from 2003

to 2014 and equals the percent of female convicts from 1994 to 2001. Therefore, 1992 and

2003 are two change points of the data source.

Chapter 3. How Does the One-child Policy Influence Gendered Deviance in China?

3.1. Introduction

Gender is one of the strongest predictors of delinquent and deviant behaviors across space

and over time. Men are inarguably more involved in crime than women (Agnew, 2009;

Vaske, Boisvert, & Wright, 2015). As a result, the gender gap in deviance is prominent.

However, scholars have debated since the 1970s, after the proposition of emancipation

theory (Adler, 1975; Simon, 1975), whether the gender gap has changed and, if there has

been the change, what has contributed to this change. Since the 1970s, the convergence of

the gender gap in deviance and crime has been observed worldwide (Carrington, 2006;

Estrada, Backman, & Nilsson, 2016; Heimer, Johnson, Lang, Rengifo, & Stemen, 2012;

Lauritsen, Heimer, & Lynch, 2009; Savolainen et al., 2017), yet the causes of the change

are still controversial. Etiological research on the gender convergence in crime has been

based on the gender role convergence from the emancipatory perspective (Adler, 1975;

Simon, 1975), the economic marginalization perspective ( Heimer, 2000; Steffensmeier,

1978, 1980), and the civilizing perspective (Goodkind, Wallace, Shook, Bachman, &

O’Malley, 2009; Haynie & Armstrong, 2006). One problem shared by the previous

research, however, is the oversimplification of gender roles. Most research, either the

theory explaining the increase in women’s offending or the theory explaining the decrease

in men’s offending, has operationalized gender role change as a result of the alteration in

women’s labor force participation and ignored the multidimensional aspects of gender roles

in other domains, such as family, school, and interpersonal relationships (Giordano &

Cernkovich, 1979). Moreover, femininity and masculinity are set as the ends of the gender

role continuum, in which femininity deters the delinquency and masculinity facilitates the

delinquency. The convergence, therefore, is a result of the masculinization of women, or

femininization of men, or the neutralization of gender roles in general.

Another limitation to this line of research is the endogeneity embedded in the

affected variables and the possible spuriousness of the correlations between them. The

overwhelming majority of empirical research has been based on observations from

advanced Western countries where the social order is relatively stable in the contemporary

era. Even the emancipation theorists who set their focus on the effects of the women’s

liberation movement cannot clarify whether the movement has also been an outcome of the

change in other fundamental cultural and social structure and, therefore, the relationship

between the movement and the increase in women’s crime might be illusory (Giordano,

1978; Smart, 1979). Similarly, the other two theories, one based on the gender-specific

structural change that occurred in the labor market and the other based on the cultural

change of gender ideologies, also cannot avoid the same question: What is the initiative of

the change?

To subdue these two problems in research, the current study (a) employs Giordano

and Cernkovich's (1979) multidimensional gender-role frameworks to reflect the gendered

pathology of deviance comprehensively and (b) applies this framework to a rarely

researched country with a unique natural experiment on a massive population. China’s

onechild policy provides an experimental setting to explore the underlying mechanism of

gendered delinquency. The findings of the current study present the ways in which the

policy has transformed the gendered pathways to deviance with an integrated theoretical

model.

3.2. The Context of the Current Study

onechild policy was implemented and has since regulated the birth and life of one in five

people in the world. For the Han ethnic group (comprising over 90% of the Chinese

population), every urban couple could only have one child regardless of its sex; every rural

couple could have two children only if the first was a girl, or only one child if the first was

a boy. However, the one-child policy went against the cultural preference for sons. Chinese

people valued boys more than girls because of the prevalent desire to “raise sons to provide

for one’s old age.” (Chen & Silverstein, 2000). As a result, offspring sex selection was

pervasive as a method to circumvent the conflict between the policy and the culture. To

reduce its occurrence, the nationwide propaganda promoted gender equality, and the value

of women crusaded to change the attitudes toward gender at a macro level. Although the

propaganda did not eradicate the cultural preference for sons, it played a part in the

offspring socialization practices with the effect of the low fertility rate directly produced

by the policy.

Although girls are less valued than boys in China under the influence of

Confucianism, the imbalanced sex composition in favor of boys may benefit girls who

survived because (a) the average parental treatment of girls improves because the ones who

were not really wanted by their parents had been aborted (Sen, 1990), (b) there is more

investment in the single child's education and human capital development, regardless

whether the child is a girl or a boy, in families with fewer children (Becker & Posner, 2009),

and (c) for those families with only a girl, families pin hopes on her, since no family wants

to give up their dream just because they only have a girl. This new socialization trend

accompanied with economic reform and marketization incubates the materialism and

money fetishism in China among the young generations (Bao & Hass, 2009; Bao, Hass, &

Tao, 2017). As a result, the gender role ideologies have been shifted to a more balanced

repertoire that accommodates economic roles in place of family responsibilities for both

women and men.

Therefore, the gender socialization process in China for the one-child generations

is remarkably different from traditional practices (Bao et al., 2017; Curran & Cook, 1993).

Women and men are treated more similarly by parents, teachers, relatives, and friends. The

one-child-generation women are expected to be ambitious, competitive, and goal-oriented

just like their male peers. As a result, women are less family-centered than before and are,

instead, similar to men in their achievement orientation (Edlund et al., 2013). Consequently,

economic success and status achievement are no longer solely masculine goals but are

shared by both genders among the new generations in China. The unique socialization of

the one-child generations is an outcome of a natural experiment imposing an external force

on a nation’s demographic structure, which is impossible to find in other nations even in

those with naturally imbalanced gender structure.

3.3. Gendered Pathways to Deviance

Before the rise of feminist criminological theories, the development of theories of deviance

and delinquency primarily concerned male behaviors. Women were treated as an

unimportant subset that was compatible with or neglected by the “masculine” theories

because of the marginal position of women in the criminal world. Feminist criminology,

however, has brought gender on the agenda and treats gender as a crucial structure to

deviance and delinquency, rather than a demographic category, and accordingly modifies

the traditional criminogenic pathology with gendered pathways to criminal offending.

Moreover, such study has advocated a gender-centered introspection of the traditional

theories and, therefore, adds a new dimension to reactivate the vitality of those frequently

researched theories. This study absorbs the idea of feminist criminology and tests the

explanatory power of variables extracted from three established criminological theories to

assess how the one-child policy has affected the pathways to deviance for both women and

men.

3.3.1. Multidimensional Gender Roles

Gender has always been simplified as a binary identity in most criminological research and

treated as an add-on (or a control) variable. Correspondingly, gender roles have also been

perceived as a unidimensional concept and simply connected with the delinquency change.

One criminogenic interpretation of the traditionally greater involvement of men in

delinquency argues that there is simply a correlation between traditional masculine

behavior and deviant behavior; traditional feminine behavior is perceived as the antithesis

of deviance and criminality (Messerschmidt, 2000; Shover, Norland, & Thornton, 1979).

In the vein of this theory, gender roles are viewed as behavioral expectations that people

hold for themselves, expectations about such matters as appropriate conduct or plans for

the future (Shover et al., 1979). Therefore, the link between gender roles and criminal

involvement is straightforward and direct.

Following such an assumption, changes in gender roles have been connected to

changes in the gendered deviance. Adler explains the narrowing gender gap by claiming

that women imitate more masculine than before (Adler, 1975). In the vein of emancipation

theory, many studies have perceived masculinized women or reduced femininity as the

reason responsible for the change in the gender gap (e.g., Applin & Messner, 2015;

Savolainen et al., 2017).

Feminist researchers insist that we should elaborate gender in the model and

acknowledge the multidimensional complex of gender from how it is shaped to how it

influences other social factors. Relying on this perspective, this study proposes a

gendercentered analytic framework based on Giordano and Cernkovich's (1979) model and,

accordingly absorbs explanatory variables from current criminological theories to evaluate

the influence of the one-child policy on gendered crime in China. Giordano and Cernkovich

conceived a multidimensional gender role framework to explain the complexity of gender

and its cultural and structural influence on individuals’ behaviors. This framework has been

adopted and extended to understand the relationship between gender and crime from a

feminist criminological perspective (e.g., Heimer & De Coster, 1999; Savolainen et al.,

2017). Also deriving from this model, the current study adds another structural dimension

to the societal-level gender role repertoire as the settings of specific behaviors of actors.

Correspondingly, three assumptions are made about the concept of gender roles, each of

which has important implications for understanding the relationship between such variables

and delinquent activities: (a) The concept “gender roles” can be layered as a general social

level and a specific individual level; (b) in each layer, gender roles can be decomposed into

attitudes toward gender and actual action corresponding to gender; and (c) concrete gender

roles are also related to the specific domains of the actor’s role repertoire. Table 3.1 presents

these conceptual distinctions schematically. Gender roles are a set of concepts composed

of the general and individual attitudes toward gender, gender structure, and personal

gendered behavior within a variety of social institutions. Moreover, the attitude and actual

structure or behavior do not always match each other. The generalized individual gender

roles reflect the societal-level gender roles, but individual traits may fluctuate widely from

those of the societal level. Moreover, the individual-level gender roles are a result of

socialization under the influence of the societal-level gender roles.

This system of gender roles is relatively stable and reproducible by itself. However,

this system is also dynamic and changeable. A substantial change in either individuals’

gender role expectations or behaviors will initiate the change in societal-level gender roles,

which, in turn, produces more individual-level change and then ripples out.

Guided by this multidimensional gender role framework, the current study abstracts

variables from three most researched criminogenic theories and categorizes them according

to their possible influence on specific dimensions of the gender role repertoire.

3.3.2. Gendered Decomposition of Criminological Theories

3.3.2.1. Control theories

Delinquency, according to control theories, requires no particular skills. It is frequently the

easiest or most expedient way for individuals to satisfy their needs and desires. What

requires explanation is the conformity. Control theorists argue that people conform because

of the controls to which they are subjected (Gottfredson & Hirschi, 1990; Hirschi, 1969;

Kornhauser, 1978; Sampson & Laub, 1995). Therefore, control theories emphasize the

mechanism of socialization on the offending behaviors. These controls can be categorized

into three types: (a) external control or social control (Hirschi, 1969), (b) stake in

conformity (Briar & Piliavin, 1965; Toby, 1957), and (c3) internal control or self-control

(Gottfredson & Hirschi, 1990). In addition, power-control theory is another strand that

specifies the relationship between family control and offspring offending from an

intergenerational perspective.

External control largely manifests as the fear of sanctions from others for delinquent

behaviors. The “others” include family, friends, school officials, employers, colleagues,

neighbors, and police. External control can be enhanced if the others set clear rules

forbidding delinquency and related behaviors, monitor the individual’s behavior, and

consistently sanction rule violators in a meaningful manner (Robert Agnew, 2009;

Gottfredson & Hirschi, 1990). Stake in conformity can be found in some individuals who

are less likely to engage in delinquency because they have a lot to lose through delinquent

behaviors. In particular, they have strong emotional bonds with conventional others, like

family and friends, that may be jeopardized by delinquency (Sampson & Laub, 1995).

Investment in conventional activities, such as getting an education, can increase the stake

in conformity. For example, adolescents who are doing well in school, devoting more time

to homework, and anticipating obtaining a good education and job are less likely to engage

in delinquency because it might jeopardize their accomplishments and life plans (Giordano,

Cernkovich, & Rudolph, 2002; Smith & Paternoster, 1987). Another restraint against

delinquency is individuals’ abilities and beliefs inhibiting themselves from responding to

temptations and provocations with delinquency, which is defined as internal control or

selfcontrol. Some individuals have been taught to condemn delinquency and, therefore, are

less likely to engage in delinquency as a result. Therefore, self-control is partly a result of

absorption of the external control and, therefore, internalization into individuals’ traits

(Gottfredson & Hirschi, 1990).

Gottfredson and Hirschi (1990) explained why men have higher rates of offending

than women by arguing that men are subject to fewer controls than are women. Men are

less strictly supervised by parents and less strongly tied to the household than women;

moreover, men are less likely to be sanctioned by family, peers, and others for aggressive

and deviant behaviors than women. Compared with women, men are less likely to condemn

crime and more likely to internalize pro-criminal definitions (Burton, Cullen, Evans, &

Dunaway, 1994; Warr & Stafford, 1993). Moreover, committing crime, to some extent, is a

way to demonstrate masculinity (Messerschmidt, 1993, 2000). These gender-specific

differences in crime participation are products of different socialization experiences of

women and men that result in the gender stratification of controls (Lagrange & Silverman,

1999). “Parents tend to socialize children to accept society’s values and thereby develop

their conscience” (Kempf, 1993, p. 145). However, social values and beliefs are different

across genders. Correspondingly, women are socialized differently from their male

counterparts since their day of birth and are expected to be more submissive than men in

patriarchal societies. Besides its impact on internal control, the family also functions

fundamentally in shaping the stratification of external control. Women tend to be more

closely monitored and controlled from their childhood through adulthood than their male

counterparts (Steffensmeier & Allan, 1996), which limits opportunities for involvement in

delinquency through family socialization (Lagrange & Silverman, 1999). These

genderspecific differences survive in other institutions of socialization including school,

workplace, and friends. Experiencing a distinctively higher amount of controls, women are,

therefore, much less likely to be involved in crime than their male counterparts. Empirical

findings substantially support control theories with two additional extensions. First, the

interpersonal differences in controls are determined primarily as early as in childhood from

family socialization. Better parenting is associated with higher control levels among

children and thereby less offending committed by them after they grow up (Rebellon &

Anskat, 2018; Rebellon, Straus, & Medeiros, 2008). Second, control level experienced by

individuals not only explains the variation in criminal offending, but also works well with

other analogous deviant behaviors, such as smoking, fighting, speeding driving, excessive

drinking, and gambling (Pratt & Cullen, 2000; Rebellon & Waldman, 2003; Widom, 2014).

These behaviors are very similar to criminal offending in the way that they can also achieve

immediate gratification in an expedient way (Gottfredson & Hirschi, 1990).

In addition to traditional control theories, Hagan, Simpson, and Gillis (1987)

proposed power-control theory to explain the inter-generational influence of gender power

dynamics on offspring’s delinquency. Wives have much less power than their husbands in

patriarchal families in which husbands are employed in an authority position while wives

work as full-time homemakers; correspondingly, daughters enjoy much less freedom than

their brothers in this type of family. The former imbalance explains the latter; whereas the

latter imbalance explains the reason why girls traditionally commit less crime than boys:

(a) the gender division of domestic work imposes more social control on daughters, and (b)

the resulting attitude toward risk makes daughters more risk-averse. Therefore, patriarchal

families tend to reproduce daughters socially to focus their futures around domestic labor

and consumption following the model of their mothers, while reproducing sons to prepare

for participation in direct production following the model of their fathers (Hagan et al.,

1987; McCarthy et al., 1999).

However, with women joining the labor force, the family power structure has been

rebalancing. This change results in the formation of more egalitarian families, in which the

consumption and production spheres are undivided by gender. Relatively equal power

between parents leads to the similar treatment of sons and daughters. Boys are encouraged

to reject the traditional masculine gender norms supporting risk-taking activities and

delinquency, and girls are encouraged to be independent and ambitious. Therefore, the

gender gap in delinquency is narrower in egalitarian families than in patriarchal families

because of the expected decrease in men’s offending and increase in women’s offending

(Hagan et al., 1987; McCarthy et al., 1999). Family structure, either patriarchal or

egalitarian, can reproduce itself from generation to generation and keep gender inequality

or equality relatively stable.

Most empirical research on power-control theory has provided at least moderate

support for the argument (Blackwell & Piquero, 2005; Grasmick, Hagan, Blackwell, &

Arneklev, 1996; Hagan, 1990; Jensen, 1993; Leiber & Wacker, 1997; McCarthy et al., 1999;

Morash & Chesney-Lind, 1991). Women experience the change in the workplace, which

brings about more control over resources and power outside and at home. This change

increases women’s chance to exercise their power in family relations. Consequently, child-

rearing practices for sons and daughters are also transformed (McCarthy et al., 1999;

Morash & Chesney-Lind, 1991). As egalitarian families become more common and

prevalent, daughters become freer and less constrained, thereby reducing the gender

difference of delinquency (Blackwell & Piquero, 2005; McCarthy et al., 1999). Empirical

tests of power-control theory suggest that the gender gap in offending is correlated to the

level of patriarchy and corresponding control at home.

Gender has been embedded into control theories, which has theoretically explained

the substantial gender gap in delinquency and deviance within their analytic paradigms,

respectively. However, the pivotal variables are independently studied, and sometimes

researchers would find contradictory or competitive explanations. For example, in the

egalitarian families where mothers spend less time on supervising children, the external

control experienced by either boys or girls tends to be less intense than in the traditional

patriarchal families. The lessened family control should increase offspring’s deviance level,

which should derive a contradictory prediction on boys’ delinquency. This concurrence of

several explanatory variables hides and mixes the real effect of each variable on deviance.

Therefore, an integrated model including those competing variables is crucial for

differentiating and comparing effects between them. Corresponding to the

multidimensional gender role repertoire, we extract four factors from control theories

influencing gendered deviance: (a) external control that is affected by the societal gender

ideologies, such as family control, (b) stake in conformity that is shaped by the individual

behaviors, such as school performance, (c) internal control related to the individual-level

gender expectations, such as legal conscience, and (d) parents’ power dynamics

experienced in the original family, which corresponds to the societal-level gender structure.

3.3.2.2. Anomie/strain theory

Different from control theory, strain theory explains why people offend. Emile Durkheim

(1897/1951) ascribed crime and deviance to the imbalance between the norms or values

that are previously held and the ones that are newly evolving. Durkheim called this

imbalance anomie, which gestated people’s criminality in searching for a more stable

environment. Building on Durkheim’s anomie theory, Robert Merton (1938) developed a

more specific theory for explaining the origin of crime. He refined the imbalance that

Durkheim focused on, theorizing it to be the mismatch between socially accepted goals and

the means to achieve those goals, which may produce frustration and other negative

feelings. People may deal with strain through delinquency so that their goals can be

fulfilled, or they just turn to delinquency to as an outlet for negative feelings. Messner and

Rosenfeld (1997) applied this theory in explaining why the United States suffers higher

crime rates than other nations. They ascribed the root to the prevalence of “the American

dream” overemphasizing economic success but with limited means to fulfill this dream,

especially for the underclass. Meanwhile, there is also an institutional anomie intensifying

the mismatch under the societal emphasis on economic success. Or, in other words, the

economy dominates over other social institutions, such as family and school, that should

otherwise produce constraints over individuals’ deviant behaviors. Therefore, those who

lack legal means to achieve those cultural goals and necessary social control from

institutions would be likely to turn to illegal means to fulfill the goals. This theory is

successful in explaining the extremely high property crime rate among the underclass.

Therefore, anomie/strain theory successfully connects individual experience with the

institutional structure to explain motivations for delinquency, regardless of gender.

Broidy and Agnew (1997) extended general strain theory to explain the gender

difference in crime. Women are traditionally perceived to have much less strain than their

male counterparts because of the different requirements of gender roles. Traditionally,

women are perceived to be insulated from the pressures of public life, especially the

pressure of economic success and high-class status (Applin & Messner, 2015). The ideal

image of women is often “sugar and spice,” and women’s roles are assumed to be less

demanding than men’s, which is assumed to produce less strain and therefore less

delinquency (Naffin, 1985). Moreover, differences in the types of strain experienced exist

between genders, which mediate the influence of strain on behavior and emotions reacting

to it (Moon & Morash, 2017). Compared to men, women tend to suffer strain more from

relational conflict than their male counterparts, who tend to suffer strain from struggling in

pursuing monetary success and extrinsic achievement (Broidy & Agnew, 1997). Moreover,

gender discrimination is another distinct source of strain for women (Chesney-Lind &

Sheldon, 1992). The connecting mechanism from strain to deviance is also different

between genders. More concern with relational issues causes women to be constrained

more by various social controls from family, friends, and colleagues and to have more social

supports as well, all of which reduce the possibility of delinquency, especially serious and

aggressive offenses. On the contrary, men always have fewer social controls over their

aggressive behaviors (Broidy & Agnew, 1997). Therefore, the different amount and type of

strain experienced by men and women, various emotional reactions to strain, and distinctive

conditioning factors between strain and responding behaviors explain the gender gap in

deviance.

Corresponding to the multidimensional gender role repertoire, strain theory explains

the gap between the individual gender ideology and its actual performance as affected by

the societal-level gender structure. To include strain theory into the integrative model, three

sets of variables are extracted: (a) the family hope burdened by different genders, which is

a part of the societal-level ideologies of gender, (b) the life goals of either genders, such as

the aspiration for monetary success, as a reflection of the individual gender expectations,

and (c) the generalized gender structure facilitating the fulfillment of the life goals for

women and men.

3.3.2.3. Learning Theory of Deviance

Peer influence is another correlate of deviance that has frequently been used to explain the

gender difference of deviance (Jensen & Akers, 2017). During interaction with intimate

personal groups, definitions favorable or unfavorable to delinquency are learned with the

assimilation of motives, drives, rationalizations, and attitudes toward whether viewing legal

codes as rules to be observed or broken (Matsueda, 1982; Sutherland, 1947). This

differential association effect has been widely found in the interactions of peers with

friends, which has been developed to explain the gender-specific effect of peers on criminal

involvement (Alarid, Burton, & Cullen, 2000; Covington, 1985).

As one gender-oriented derivation of learning theory, Giordano and her colleagues

affirmed that girls would be more deviant and delinquent with a mixed-gender peer group,

while boys would be so when they were typically accompanied by same-gender

companions (Giordano, 2009; Giordano & Cernkovich, 1979; Giordano et al., 2002). These

mixed-gender peer groups play a role in taking the learning channel into effect, from which

women can learn about and carry out delinquent activities with the maximum opportunity

to interact with men and masculinity. As a result, women’s delinquency and deviance are

“group influenced in a way that is similar to the pattern of male delinquency.” (Cernkovich

& Giordano, 1979, p. 474).

In contrast, men are less aggressive when they are around women, which has been

supported by the civilizing perspective (Lauritsen & Heimer, 2008). The increasing

presence of women is expected to be associated with a civilizing process restraining men’s

deviance to some extent. In this way, women act as men’s “moral guardians” or at least as

some form of informal control over men’s delinquent and criminal behaviors (Lauritsen et

al., 2009; Rosenfeld, 2000).

Both gender-specific contentions are derivations of learning theory and are

consistent. By setting men as the origin of deviance learning but women as the other side,

interaction with men increases the involvement in crime and occurrence of deviance, while

interaction with women decreases these in contrast. Therefore, the gendered perception of

cross-sex friendship connects the gender roles and learning theory of deviance (Cernkovich

& Giordano, 1979), which reflects the individual-level gender ideologies corresponding to

the gender role framework. To include the peers’ effect on deviance, our model also

integrates this variable.

3.3.3. Reformulate Gendered Pathways to Deviance

Most criminological research treats the criminogenic factors and deviance flatly without

differentiating the structural positions held by explanatory variables. However, every social

and cultural factor does not exist independently, because they are intricately related to other

factors and, thereby, compose a complex effect influencing people’s decision, including the

decision of criminal involvement. To reflect our society more accurately, therefore, we must

distinguish the effects of each variable on deviance as well as on other potentially relevant

variables. Guided by the multidimensional gender role assumptions, this study reformulates

the extracted variables from the three theories and configures a framework explaining the

gendered pathways to deviance (see Figure 3.1).

All variables are categorized into three groups according to their initiative features

and onset stage. The first category corresponds to the societal-level ideologies of gender,

including family control, family hope, parents’ interaction dynamics, interaction with

family members, peer interaction, and school life, which represents the socialization

processes within three primary institutions: family, peers, and schools. The second category,

reflecting the individual-level gender role repertoire, includes self-control, life goal, and

gender role perceptions, which reflect cognitive traits. The last category of variables

corresponds to the societal-level structure of gender roles and mainly reflects the gendered

means of fulfilling the individual expectations. Compared to the third category, the

cognitive traits are more individual and likely to fluctuate from person to person. However,

they are affected by the societal settings of gender ideologies that are beyond the

individual’s control and classified as socialization practices. Moreover, the effects of

socialization practices are intermediated by individuals’ cognitive traits; in other words,

individuals, even though having grown to maturity in similar environments, make different

decisions on criminal engagement. In this process, personal cognition plays as a receptor

and translator of external forces to behavior (Giordano et al., 2002). Those three categories

of variables influence the deviance (individual behavior) directly and indirectly. Therefore,

we assume that the layered gender role repertoire affects gendered deviance and,

correspondingly, that the alteration occurring in the former also drives the latter to change.

Following the socialization change, a sequence of shifts would happen in the gender

role repertoire, conductive from the individual layer to the societal layer, and then to the

individual layer again. As a result, this substantial change of gender role repertoire

influences people’s behaviors in the other realms, including in deviance and delinquency.

3.4. Hypotheses

The present study examines the effect of the one-child policy on gendered pathways to

deviance by comparing the differences between generations before and after the enactment

of the one-child policy with an integrated model. It also explores the change in

multidimensional gender roles and its effect on deviance. Considering the possible unique

socialization process experienced by the one-child generations, we expect to see:

Hypothesis 1a: Gender socialization in family, school, and peers is significantly

different between the one-child and pre-one-child generations.

Hypothesis 2a: The gender socialization differences transform the one-child

generation people’s cognitive traits from their older counterparts.

Hypothesis 3a: The socialization, cognitive traits, and gender structure variables

influence subsequent deviant behavior, but the pathways are different between genders and

generations.

Hypotheses 1a and 2a are tested by the differences in specific measurements between

the one-child and pre-one-child generations. Whereas, a series of path-equation regressions

is adopted for the test of hypothesis 3a, from which the gendered pathways to deviance and

their inter-generational dynamics are revealed.

3.5. Methods

3.5.1. Sample

The sample included 2,006 adult Chinese citizens (ages 18 and above) who were recruited

online. The reason for employing adult subjects rather than adolescents, as much research

has disproportionally done (e.g., Agnew & Brezina, 1997; Hoffmann & Su, 1997; Jennings,

Piquero, Gover, & Perez, 2009; Kaufman, 2009; Moon & Morash, 2017), is out of fidelity

to the original version of strain theory, which focuses on financial goals (Merton, 1938).

Adolescent subjects may be a good choice when considering sampling convenience but are

only feasible for the analysis of educational strain, which is insufficient to capture the direct

motive driving delinquency even if there is a positive connection between those two kinds

of strain.

The geographic distribution of the respondents was wide, covering 31 Chinese

provinces and municipalities as well as Hong Kong, Taiwan (97.1%), and some foreign

countries (25 respondents were residing out of China), as identified by the respondents’ IP

addresses. The age and gender distributions are highly representative for the one-

childpolicy-oriented study. There were 908 respondents born after 1980, the year for the

announcement of the policy, and 1,098 born before that year. Women were the majority

(63.5%) of the respondents across generations (see Table 2).

3.5.2. Data Collection

The survey with 46 multi- and single-choice questions and seven self-rating scales (see

appendix for the survey instrument) was carried out in the summer of 2017 in China. The

questionnaire was designed at the Wenjuanxing Survey Service Website (www.wjx.cn) and

then posted and distributed on the WeChat platform (the most popular social networking

software in China, akin to Facebook) to recruit respondents. The intent of the research was

briefly introduced in the advertisement for the survey. The respondents who finished the

survey with “good quality” (taking more than 180 seconds to finish) would get ￥1 Chinese

Yuan (approximates $0.16 US dollar) cash reward, which was also clarified in the

introduction of the survey as an incentive. The survey was anonymous, which meant there

was no identifiable information collected. Because of the voluntary nature of the online

survey, participation per se represented consent. Moreover, the respondents were free to

end the survey at any time in the process, and all information would be dropped by the

survey system automatically.

3.5.3. Measures and Comparisons

To reveal the gendered influence of the one-child policy, all respondents were categorized

into four groups according to their ages (born before or after 1980) and genders (female or

male). They were one-child-generation men (OCG men), one-child-generation women

(OCG women), pre-one-child-generation men (POCG men) and pre-one-child-generation

women (POCG women). Answers were compared between generations and genders with

the Student’s Test method to capture the inter-generational and gender differences (see

Table 3.3). Control theories are measured by sibling care, parents’ power dynamics, family

hope, parents’ control, legal conscience, and school performance. Learning theory is

measured by peer effect. Strain theory is measured by life goal and societal gender

structures in terms of means to success. In addition, individual ideologies of gender are also

included as cognitive traits. I reformulated those variables according to the path framework

in Figure 3.1, including (a) the socialization from family, school, and peers, (b) personal

cognitive variables such as individual attitudes toward gender roles, life goal, and

conscience, and (c) gender differences experienced in multiple institutions as the gender

structure at the generalized level (see Table 3 for the descriptive and comparative statistics).

The latter two directly affected deviance; whereas the socialization factors were expected

to influence the deviance directly and indirectly intermediated by the cognitive traits.

Among these variables, the measurement of family hope and definitions of

masculinity/femininity were measured by multiple-item scales. Inclusion in a scale required

items to reach a minimum loading of 0.5 in the rotated matrix of the factor analysis (Bao et

al., 2017). Moreover, the Cronbach’s alpha for each scale had to be higher than 0.7 to ensure

the internal consistency and measurement reliability statistically.

3.5.3.1. Socialization process

Sibling care is measured by two questions. One concerns the number of siblings, and the

other is about the time spent in caring for their siblings for the participants who were not

single children. The responsibility of taking care of siblings cultivates the conventional

bonding within the family and the orientation toward personal relations; thus, it is expected

to decrease the deviance (Hirschi, 1969). Combining the responses, the sibling care variable

is binary (1: need to care; 0: no need). The inter-generational difference is significant in

that there was a higher proportion of older-generation people responsible for their siblings’

care when they were young (mean = 0.55) than the OCG counterparts (mean = 0.39).

However, this responsibility has been skewed to women over time. More young women

shouldered the responsibility of sibling care than their male counterparts (mean = 0.41 for

women and 0.34 for men). In contrast, fewer POCG women had to take the responsibility

than did their male counterparts (mean = 0.52 for women and 0.59 for men). This gender-

specific change might be a direct reflection of the policy’s intervention. Families that had

the “luck” of having a boy as the first child would stop childbearing and comply with the

policy; however, families without such “luck” would not be satisfied with the single child

and would tend to have a larger family, in which the first child turned to be the elder sister

and likely the babysitter of her younger sibling(s). As a result, the one-child generations

have been largely freed from the sibling-care responsibility, though this change is not equal

between genders.

The parents’ power dynamic is measured by parents’ relative position at home.

Respondents were asked who had the final say at their homes (1 = mother, 2 = mother for

the major things and father for the less important, 3 = the same, 4 = father for the major

things and mother for the less important, 5 = father). Therefore, the higher the number is,

the more patriarchal the family was in which the respondent grew up. The mean is 3.24 (SD

= 1.34). This variable is not different between generations nor genders.

Family hope combines the scales for father’s and mother’s hope (0 = no, 1 =

somewhat, 2 = yes for either) and, therefore, ranges from 0 to 4. A higher score in family

hope means the respondent burdened more by parents with the family dream. The mean is

2.84 (SD = 1.47), and Cronbach’s alpha is 0.81. The gender difference is significant, where

men (mean = 3.00) were burdened with more family hope than women (mean = 2.75).

Moreover, the gender gap has shrunk (the gender difference is 0.31 for the POCG with p <

0.01, whereas the difference is 0.23 for the OCG with p<0.01).

Family control is captured by the question whether the respondent would receive

sanctions from the family if they were deviant or intractable when they were young (1=no,

2=not always, 3=yes). The mean is 2.38 (SD = 0.73) and increases across generations.

Compared to the older-generation counterparts (mean = 2.35), the OCG people (mean =

2.42) generally received more sanction from the family against their defiant behaviors

(t=2.08, p < 0.04). Moreover, there is no significant gender difference among the young

generations. In contrast, older-generation males were subject to more disciplinary control

than were their female counterparts.

Cross-gender friendship measures the learning effects on deviant behaviors

(Giordano, 2009; Lei, Simons, Simons, & Edmond, 2014). The variable is measured by the

response to the question “is friend’s gender a concern when you are making friends?” (0 =

no, 1 = somewhat, 2 = yes.) The mean is 0.75 (SD = 0.92). Compared to men, women were

concerned more about their friend’s gender in general (mean = 0.80 for women, and 0.66

for men, with t = 3.32 and p < 0.01); however, this gender difference only existed among

the POC generations, which reflected a more permissive trend in making friends for

Chinese young women.

School effect is captured by the measure of the self-reported school performance (1

= bad, 2 = below average, 3 = average, 4 = above average, and 5 = good). The mean is 3.56

(SD = 1.09). There is no significant difference between generations nor genders.

In sum, either boys or girls of the one-child generations have been more at the center

of the family (Bao et al., 2017). They have been freed from sibling care responsibility and

pinned with more hope and control from the family because of their parents’ low fertility

rate. However, the inter-generational change is not the same between genders. On the one

hand, the gender gap has been narrowing in terms of socialization. Women have received

relatively more family hope and control and been more open to making male friends than

they were before. On the other hand, however, some family chores, such as sibling care, is

skewed to women’s shoulders. Overall speaking, our results support hypothesis 1. The

socialization practices have been significantly changed over generations because of the

one-child policy; moreover, the change has been gender-specific.

3.5.3.2. Cognitive Traits

Legal conscience reflects the level of self-control and the internalization of the external

controls. This variable is captured by the responses to the question whether the respondent

agreed on unconditional compliance with laws and regulations (1 = no; 2 = yes for most

except the unpunishable; 3 = yes for most except the unreasonable; and 4 = yes). The mean

is 3.32 (SD = 0.81) with no difference between genders nor generations.

Masculinity and femininity reflect the result of gender socialization on individual

gender roles. They are measured in two ways: first, as the interaction between the definition

and the self-rating scale by men and women, respectively and second, as the comparison

between the respondents’ same-gender parent and themselves. Both definitions are gauged

with 16 items, similar to those in previous studies (e.g., Kaufman & Taniguchi, 2006; Lei

et al., 2014). Masculinity was defined by male respondents based on a multi-response

question, including being independent, enterprising, aggressive, physically strong,

emotionally repressed, controlling, pleasure seeking, economically successful, muscular,

sporty, tough, adventurous, sagacious, powerful, assertive, and knowledgeable. Based on

the constructed definition, the male respondents were asked to rate their masculine level

from 1 (not masculine at all) to 10 (very masculine). Correspondingly, femininity was

defined by female respondents with options including being elegant, affectionate, delicate,

sociable, submissive, family-centered, dependent, chaste, kind-hearted, quiet, cautious,

sensitive, considerate, cultured, neat, and pure. After constructing the femininity definition,

the respondents were asked to rate their feminine score from 1 (not feminine at all) to 10

(very feminine). Therefore, the scales of masculinity and femininity range from 0 to 160

and are positively related to persons’ perceptions of their gender roles. Both definitions’

internal reliability is high (Cronbach’s alpha is 0.78 for masculinity and 0.79 for

femininity). Moreover, the self-rated femininity significantly decreases over time (mean =

7.05 for the POC generation, and 6.24 for the OC), but there is no significant

intergenerational change among men. Correspondingly, the interaction is significantly

higher for the OCG females than for their older counterparts but are not different between

male groups. Another measurement is the parent-child comparison of masculinity and

femininity (1 = I am more masculine/feminine, 2 = the same, 3 = my father/mother is more

masculine/feminine). The mean is 1.99 (SD = 0.80) for the masculine difference and 1.95

(SD = 0.80) for the feminine difference, which means both genders regard their parents as

being more masculine or feminine, and this perception is consistent across generations.

Life goal is composed of two parts. One is the economic-oriented definition of

success, and the other is the aspiration to achieve economic success. The respondents were

asked to construct their definition of success from ten diverse items—being economically

abundant, politically powerful, reputable, popular, knowledgeable, healthy, to having many

friends, a happy family, outstanding children, and making the world a better place.

Compared to the older generations, who had a diverse spectrum of success definition, more

of the OCG people, both women and men, disproportionately defined success as economic

abundance (1 = yes, no = 0; mean = 0.49 for the POC generations and 0.66 for the OC; see

also Dutton, 2005; Shen, 2015). Moreover, the gender difference has shrunk across

generations (no difference among the new generations but a significant gap among the older

generations with 0.52 and 0.46 for men and women, respectively). This pattern is consistent

with the trend of aspiration for economic success (0 = no, 1= somewhat, 2 = yes). A higher

proportion of the OCG people rated high in their desire to achieve success economically

than their older counterparts (mean = 1.59 for the older generations, and 1.74 for the

younger, with t = 4.91 and p < 0.01). In particular, a higher proportion of the single children

among the new generations defined success economically (0.75 versus 0.62 with p < 0.01)

and aspiring for economic success (1.81 versus 1.70 with p < 0.01) than their peers who

had siblings.

In sum, a higher proportion of the OCG people defined success as monetary

abundance and aspired for such a success than the older generations. Moreover, femininity

reduced over time, but masculinity saw no change. Therefore, these findings support

hypothesis 2. The individual gender role expectations are different across generations.

3.5.3.3. Gender Structure

Gender structures are measured in four ways as the gender differences experienced by the

respondents in family, school, workplace, and from the general income level to reflect the

societal level of gender structure.

Family inequality is measured by the responses to the question about whether their

parents told or implied that “It is harder for women to succeed than for men” (0 = no, 1 =

somewhat, 2 = yes). The mean is 0.63 (SD = 0.82). The gender difference was less

emphasized by the parents of the OCG people than for their older counterparts (mean =

0.68 for the POC generations, and 0.56 for the young ones, with t = 3.40 and p < 0.01).

Compared to women’s parents (mean = 0.60), men’s parents (mean = 0.68) were more

likely to transfer the chauvinist message to their sons to promote the sons’ confidence in

their gender (t = 1.94 and p < 0.05). Moreover, this difference was from the gender gap

among the older generations, which was not noted among the one-child generations.

School inequality is measured similarly to family inequality, except the gender

discriminatory message senders were changed to teachers (0 = no, 1 = somewhat, 2 = yes).

The mean is 0.57 (SD = 0.79). Gender discrimination has been strengthened over

generations. A significantly higher proportion of young women (mean = 0.61) experienced

such discrimination at schools than did their male counterparts (mean = 0.50) and older

counterparts (mean = 0.54).

Job inequality is captured by the promotion prospect for women and men.

Respondents were asked to predict the outcome of a scenario in which “a man and a woman

are working in the same company at the same position. One day, the boss has a mind to

promote one of them. Who do you think is more likely to get this promotion (1 = the

woman, 2 = the same, 3 = the man)?” The mean is 2.22 (SD = 0.78). Fewer young women

agreed with the man getting the opportunity than did their male counterparts (mean = 2.18

for women and 2.31 for men, with p < 0.02). However, the gender gap was reversed for the

POC generations, among whom more women predicted the man as the lucky person than

their male counterparts (mean =2.28 for women and 2.13 for men, with p < 0.01).

Income inequality is measured by the perception of who had a higher income on

average (1 = women, 2 = the same, 3 = men). The mean is 2.46 (SD = 0.73), which means

more people regarded men as the winner of income. The unequal perception has been

reinforced over generations. More OCG people believed men’s average income is higher

than women’s than did POCG people. Like job inequality, however, younger generation

women were more optimistic than their older counterparts in the gender equality of earning,

which reflects the dynamics of individual expectation of gender roles.

In general, regardless of the equalized socialization practices and individual gender

role perceptions, the societal gender structure turns out to be even more unequal than

before, especially in school and in the general income level. Compared to other

socialization agents, the family has less gender discrimination among the one-child

generations, and, correspondingly, women who have been socialized with such equality

expect similarly equal treatment in other societal institutions. However, those institutions

keep or even strengthened gender disparity, and fail the expectation of the OCG women.

3.5.3.4. Deviance

Deviance is measured by a self-rated scale of respondents’ deviant levels regarding deviant

behaviors (from 1, not deviant at all, to 10, very deviant). Examples were offered, including

smoking or drinking before 18, fighting at school or on the street, getting something by

force, shoplifting, deliberately damaging private or public property, and disobeying the

authorities (Bao et al., 2017). The mean is 4.59 (SD = 2.44). The gender gap in deviance is

significant across generations. On average, men are more deviant than women. The reason

why deviance was used as the dependent variable rather than more serious criminal

behaviors is in consideration of the data quality. There are two concerns if we explicitly

collect information on delinquency: (a) the reliability and validity may be impaired

considering the respondents’ willingness to reveal their criminal activities (Thornberry &

Krohn, 2000) and (b) the data tend to be highly skewed with an inflation of zero considering

the general low crime rates in China (Curran & Cook, 1993). As an alternative, the measure

of deviant behavior, which is milder than criminal offending, avoids these problems to a

large extent; and more importantly, they are analogous to crime in the nature of fulfilling

instant gratification and therefore widely used as the sound prediction for criminal

involvement in research (Gottfredson & Hirschi, 1990; Perrone, Sullivan, Pratt, &

Margaryan, 2004; Widom, 2014).

3.6. Results

3.6.1. Bivariate Correlations

Table 3.4a and 3.4b present Pearson correlations for men’s and women’s groups of

respondents, respectively. Deviance has significant correlations with cognitive traits and

moderate correlations with structural inequalities. Moreover, the connections between

cognition and socialization variables are salient, which reflects the effect of socialization

on deviance via its intermediary effect on shaping cognitive traits. These associations are

consistent with both theoretical propositions and empirical findings based on Western

samples (e.g., Burton, Cullen, Alarid, & Dunaway, 1998; Lauritsen et al., 2009; Lei et al.,

2014). Comparing the Table 4a and 4b, all structural variables have a consistent correlation

with deviance between genders, which is also true for the cognition-related variables except

for masculine/feminine ones. However, some socialization-related variables have a

reversed effect on women’s and men’s deviance, such as parents’ power dynamics and

cross-sex friendship.

Three demographic control variables are measured with the single-choice questions.

In addition to the gender and age variables, childhood residence was also collected and

coded as 0 = non-country and 1 = country, with 60 - and 40 - percent distributions,

respectively. From the IP addresses, the overwhelming responses were from industrial or

economically advanced cities. However, the information revealed by the IP address is vague

at the city level and impossible to capture more residential details after childhood.

Combining with the fact of massive rural-to-urban migration in China during past decades

(Lo, Cheng, Bohm, & Zhong, 2018; Solinger, 1999) and the relatively inconvenient

infrastructural condition in rural areas, there is a high probability that the Internet-based

survey respondents who spent childhood in the countryside have moved out after they grew

up.

3.6.2. Path Model - General

Structural equation modeling (SEM) is employed to evaluate direct and indirect effect paths

between variables in the model, which integrated elements of socialization, cognition, and

inequality. Collinearity diagnostics did not show multicollinearity problems between

parenting variables, with all the VIF values under 1.3. The result for the general model with

all responses without differentiation over gender nor generation is shown in Table 3.5.

Socialization is operationalized by six variables ranging from the realm of family, school

to peers. Cognitive traits reflect through two variables, and gender structure includes four

variables. The influences of these three parts on deviance are estimated with the

intermediary effects of socialization on cognitive traits. Three demographic variables were

used as controls to exclude possible artifacts of group differences on the measures.

The results in Table 3.5 show that cognitive variables are highly related to deviance.

In general, high conscience is a good predictor of low deviance level; whereas the aspiration

for economic success increases deviance. These two variables are results of socialization.

Except for the sibling care responsibility, all other five variables have significantly positive

effects on one or both cognition variables. Among all socialization institutions, family plays

the most important role in shaping self-control and life goal through the external control,

hope anchoring, and power structure between parents. Nevertheless, the direct effects of

family socialization are largely diminished after controlling its indirect effects on deviance.

In contrast, the effect of school socialization is salient on deviance, either in the direct or

indirect way. The acceptance of cross-gender friends only connects to the self-control but

does not affect the other two response variables. Structural gender inequality contributes to

the increase in deviance through discrimination in family and school. Among the control

variables, gender has the only direct effect on deviance that men have a higher deviance

level than women on average. However, generation and rural residence during childhood

connect to the shaping of cognition and therefore indirectly influence deviance. The one-

child generations are more eager for economic success than their older counterparts.

Growing up in rural areas plays a role in reducing success aspiration and self-control, which

can be ascribed to the unequal distribution of education resources and loosened parent-to-

children relationships due to the massive labor force migrations from rural to urban areas.

After including all the variables into the model, the effect of generation is not

significant, which means the inter-generational difference has been explained. Although the

direct pathway model to deviance is significant, its R2 is relatively low. This may be because

the aggregate data not fit this designated path model well. Therefore, a gendered and

generational disaggregation is necessary to explore the pathways to deviance.

3.6.3. Path Model – Gendered and Generational

Whether current criminogenic explanations are universal and gender-neutral has been in

debate for a long time (Mears, Ploeger, & Warr, 1998; Smith & Paternoster, 1987).

Moreover, in line with the gender-specific effects of the one-child policy, we disentangle

the data over gender and generation so to capture the policy’s influence on the

intergenerational trend of the gender gap in deviance. Categorized by gender and

generation, therefore, four sets of path models are estimated separately representing each

group. Facilitated by the specification in gender, two gender role measures are additionally

included as cognitive variables. Corresponding to the data specification, gender and

generation are removed from the control variables. The complete results of the grouped

path analyses are presented in Table 3.6a, 3.6b, 3.6c, and 3.6d. To make it easy to read and

compare, Figure 3.2 only contains the paths that are statistically significant (at 90%

confidence level), while controlling for the effect of childhood residence variable. Table

3.7 decomposes the direct and indirect effects of each variable on deviance and compares

the differences between groups.

The results show salient differences between generations and genders in the

etiological paths to deviance. Among the socialization variables, patriarchal family, family

control and school performance are the only three variables influential on deviance for all

groups with consistent effects. A more patriarchal family, where the mother has a lower

position than the father, decreases the deviance level of its daughters but increases that of

sons on the contrary. This finding supports power-control theory. In addition to the

genderspecific direct effect, girls from a more patriarchal family have a higher level of self-

control, which deters deviant behaviors; whereas, boys tend to have a higher level of

masculinity, which promotes the sense of toughness and eventually, deviance, indirectly.

Family control plays a role in reducing deviance for four groups through the indirect paths.

For male groups, family control significantly promotes the self-control level and therefore

reduces the level of deviance. In contrast, family control reduces the deviance level of

POCG women by increasing their sense of femininity. Moreover, both mechanisms can be

found among the OCG women, whose experience of strict family control increases their

femininity (comparing to the level of their mother) and legal conscience. School

performance has a strong effect on deviance prevention. Except for the POCG women

whose school performance decreasing their deviance via the increase in femininity, good

school performance directly reduces the deviance for the other three groups. This finding

is consistent with the conformity stake hypothesis that deviance is more “costly” for the

better school performers. The exclusiveness of same-gender friend network shows the

gender-specific effects on deviance. Women’s deviance is likely to be higher if they are

willing to make friends with men, which is consistent across generations. However, the

effect is reversed for the OCG men’s deviance. The deviance of a man of the younger

generation is higher if he tends to make friends with only men and correspondingly lower

if with women. This finding supports learning theory including the mixed-gender

association effect on women’s deviance (Giordano & Cernkovich, 1979) and the civilizing

effect on men’s (Lauritsen et al., 2009). Being burdened with family hope has no effect on

the POCG women but plays a part in affecting the OCG women’s and all generation men’s

deviance via indirect paths. For the younger generation people, family hope increases

conscience and therefore reduces the deviance level. The family hope placed on young

women particularly increases their aspiration for economic success, so it has an extra

positive effect on their deviance, which however is buffered by the effect of increased

conscience. For the POCG men, the family hope enhances not only the conscience but also

their masculinity. After counteracting, the deviance level of this group is positively related

to the family hope. The responsibility of sibling care, although has been decreased

strikingly as a benefit to the OCG people, has no significant effect on their deviance. The

only influenced group is the POCG men, whose deviance would be increased by the sibling

care responsibility.

Among the cognition variables, the parent-child difference in femininity or

masculinity is the only factor significantly affecting the deviance for all four groups.

Consistent with the predictions that masculinity encourages toughness and aggression

while femininity to the opposite (Connell & Messerschmidt, 2005; Lei et al., 2014;

Messerschmidt, 1993), decreased femininity promotes the deviance of younger women, but

decreased masculinity reduces the deviance of men in contrast. Therefore, the opposite

changes in individual gender ideologies narrow the gender gap in deviance that men tend

to be less defiant but women more defiant. Another measure of gender roles, self-rated

femininity/masculinity, is significant for the POCG women and both male groups.

Consistent with the finding in the inter-generational changes of gender roles, a lower score

of femininity or a higher score of masculinity increases the deviance level for women and

men, respectively. Conscience consistently decreases deviance among the OCG women and

both male groups, which supports self-control theory (Burton et al., 1998; Gibbs et al.,

1998; Perrone et al., 2004). In particular, only the OCG women’s deviance is specifically

increased by the aspiration for economic success. If a young Chinese woman yearns for

being economically successful as her life goal, her deviance level is likely to be higher than

her counterparts who have no such a goal.

With regard to the gender structure, inequality variables only influence women’s

deviance across generations, which is sensible for the gender inequalities are experienced

disproportionally by women in a patriarchal society. If a POCG woman experiences gender

discrimination at family, she is likely to be more deviant than the ones who are protected

from such experience by parents. Whereas for the OCG women, the income inequality

plays a more important role in increasing their deviance level. These findings empirically

support strain theory with the positive effect of the success aspiration on deviance. The

young women bearing with the dream to be monetary successful will find themselves in

short of legitimate means of fulfilling their dream when they enter the patriarchal society

with stratified gender structure. Even if they have a paid job, they are still more likely to

confront the disadvantages of the “double shift” (H. Zhang, 2014a) and less chance to

promotion than their male counterparts (Applin & Messner, 2015), which exposed them

further to the anomic pressures associated with the economically successful goal.

Growing up in rural areas is positively related to deviance as a control variable for

all groups except for the OCG men. It increases deviance via the indirect paths by

decreasing the legal conscience. Also, living in the countryside decreases the success

aspiration for the younger generation women and femininity for the older-generation

women, respectively. The former plays a role in reducing deviance, but that is too weak to

offset the positive effect of the decreased conscience.

Compared to the aggregate model (Table 3.5), the regression R2 has been enhanced

in each model, especially for male groups. The lowest R2 is for the POCG female group,

which means their deviance relates more to other variables not included in our model. The

R2 for the OCG female model is higher and between that of their older and male

counterparts. Therefore, the model fits the pathways to deviance better for men and OCG

women.

The findings for the one-child generations, either women or men, are stable after

being tested with data disaggregated by single or non-single children status, although the

success-inequality strain effects are more significant among the only-children group. The

perception of income inequality has a positive effect on OCG women’s deviance including

the only children and the ones who have siblings, but not on their male counterparts’. For

the single male children, more skewed income distribution toward men, in contrast,

decreases their deviance level because they can succeed through legitimate channels in a

more patriarchal society. Thus, the one-child policy has changed not only the socialization

for the only children and therefore their gender roles and gendered pathways to deviance,

but also the other people who belong to the very special generations. This effect process

functions through the mutual influences between the layered gender role repertoire. The

only children, as the ones directly influenced by the policy, initially change the generalized

attitudes toward gender, which in turn affects the individual-level gender ideologies of more

peer children in society. Despite the difference of the definition of success and aspiration

between single and non-single children, we can expect the difference will gradually fade

out when the policy effect ripples out over time.

3.7. Discussion

The pathways to deviance are distinctive for every group representing each combination of

the gender and the generation. Compared to the POCG groups’, some influential variables

are replaced or added in the pathways to the one-child generations’ deviance. Most

explanatory factors are the same between the groups of men, except for the stronger peers’

and weaker family interaction’s effects on the deviance of the OCG men than on that of

their older counterparts. For groups of women, the inter-generational differences are more

salient. The pathways to the OCG women’s deviance reflect the features of both deviant

pathways for their older counterparts and male counterparts. Or in other words, the OCG

women are in the middle between the POCG women and the OCG men in terms of the

pathways to deviance. Similar with men, family hope and legal conscience are affecting

Chinese younger-generation women’s deviance that are not significant in the

oldergeneration women’s pathways. Moreover, school performance, which only has an

indirect effect on the POCG women’s deviance, plays a role in decreasing deviance directly

in men’s and OCG women’s pathways. Like the POCG women’s model, moreover, the

younger-generation women’s deviance is also deterred by the patriarchal family structure

and the same-gender friendship. In addition to the generational differences, the distinctions

between genders are also prominent. Compared to men, women are subjected to the gender

inequality embedded in the social structure, which increases their deviance across

generations. For the POCG women, gender discrimination from family is substantial in

promoting their deviance level. However, for the OCG women’s deviance, income

inequality between genders pushes deviance up, especially when they have the aspiration

for economic success. This finding is consistent with strain theory. Although the younger

generation women have been socialized more equally with their male counterparts at home

by their parents, they are still suffering from the gender discrimination and inequality in

other social institutions. Because of such a lag in the attitude toward gender roles and actual

gender structure in society, Chinese younger-generation women are subjected to the

specific strain between their newly instilled dream and the relentless reality that the gender

structure is still hierarchical, and therefore the opportunities to fulfill their life goal are

unequal and meager.

In addition to testing the applicability of these theories, we also examine the relative

explanatory power of variations derived from them. For all four groups, control theories

get the most support, especially social control and stake in conformity theories. Learning

theory is also functional in the pathways to deviance for all respondents except for the

POCG men. Whereas, strain theory only explains the deviant pathways for the OCG women

because of the faster change in gender ideologies than that in structure, which has been

externally driven by the demographic policy.

Another extensive finding is in line with learning theory. For the OCG men’s group,

the same-gender friendship network demonstrates a positive connection with high

masculinity and thereby a high level of deviance. However, the effect is reversed for

women’s deviance that is increased by the opposite-gender friendship. Despite the reversed

direction in the gendered pathways, the effect of cross-gender friendship is consistent on

deviance. Male friends play as an aggressive model in peer interaction and a socialization

agent for definitions and motives favorable to delinquency. Moreover, such a peer effect on

male and female deviance can be enlarged by the imbalanced sex ratios among the onechild

generations. It is easier for women making friends with men than before, but harder for

men making friends with women when we take the huge deficit of women among the one-

child generations into consideration, which was caused by the conflict between the policy

and cultural son preference. Thus, the increase in women’s deviance may be partially a

consequence of increased involvement in mixed-gender peer groups.

As a result, the gender gap in deviance that was rated higher than five has been

narrowed from 4.7% for the pre-one-child generations to 3.9% for the one-child generations

(calculated as 50%- [women’s ratio/ women’s + men’s ratios] *100%).

Moreover, the convergence is primarily a result of the increase in women’s deviance. There

are few differences between male groups inter-generationally. Therefore, unique

socialization practices experienced by the one-child generations updated the individual

attitudes toward gender roles, which has equalized the life goal of the younger generation

people between men and women. Nevertheless, the hierarchical gender structure is

stubborn and untouched, which limits women’s means to achieve their life goal and

therefore creates the strain specifically faced by Chinese young women.

3.8. Conclusion and Implication

The one-child policy had been strictly enacted for over 35 years and affected the birth and

life of one-fifth people in the world. As a result, it has reshaped a society’s culture and

social structure externally and therefore provides an extraordinary chance to explore the

underlying mechanisms and effect paths to gendered delinquency. Contrary to the prevalent

predictions, however, the imbalanced sex ratios and scarce of women do not endow women

more value, at least no more value as workers or producers. The gender gap in deviance

narrows, but is not because of the increase in gender equality as numerous research

indicates (Lauritsen & Heimer, 2008; Lauritsen et al., 2009).

In general, two mechanisms function in narrowing the gender gap in deviance. First,

women are less feminine than before; meanwhile, masculinity also decreases

intergenerationally, which caused a convergence in the gender difference of deviance.

However, this change is different from the “masculinization” of women or “feminization”

of men because the pathways to deviance continue being distinctive between genders. If it

is a simple neutralization of gender roles or shortening of the gender continuum, the gender

differences in other realms should also fade, and the deviant pathways should be

“degendered.” However, the findings still indicate strong differences between genders in

almost all domains, which, in return, support the multidimensionality of gender role

repertoire. The other mechanism to the converged gender gap in deviance is the

equalization lag of the societal gender structure to gender ideologies happened in gender

role repertoire. The one-child policy drives the socialization, especially family socialization

of gender, to be changed and therefore alters the definition of what is normative femininity

but maintains the existing societal gender structure (Seidman, 2010). Even as stereotypes

are modified, however, the cultural shift may still leave structural inequalities in place

(Applin & Messner, 2015). This mismatch produces the genderspecific strain for Chinese

young women, especially those who have been instilled in gender equality and pinned with

the family dream by their parents. Therefore, the new generation women have been less

“immune from the criminogenic institutional forces” (Applin & Messner, 2015) but turned

to be as likely as their male counterparts to exhibit support for the values associated with

economic achievement (Steven F. Messner & Rosenfeld, 1994). Eventually, the deviance

functions as an outlet and a reaction after their dream is struck by the unequal means in

society.

Although this study is based on Chinese data, the findings are revelatory to the

criminogenic research in general. The change in socialization practice was a side outcome

of the demographic policy curbing the population growth, but the transformation has been

experienced by all people belonging to the one-child generations, regardless if they are the

only children or not.

Emancipation theory connects the liberated gender roles with the increase in

women’s crime, which has been widely criticized from the mismatch between the explained

subjects – the women who have been liberated are not the ones who are brought into the

justice system (Giordano & Cernkovich, 1979; Renzetti, 2013). The latter, in contrast, are

always found in the underclass, who are far from being liberated but trapped deep with

financial burdens by the economic marginalization (Holtfreter, Reisig, & Morash, 2004;

Reckdenwald & Parker, 2008; Steffensmeier & Allan, 1995).

However, the liberated women and the underclass women do not live in a vacuum.

In fact, they share the same world and dramatically influence each other. When the middle-

and upper-class women enjoy liberation and go out of home to compete with men in school

and workplace, the other groups of women are also influenced. They may not be directly

influenced by the liberalization, but indirectly through the changed societal gender roles.

For example, the liberated life of the middle-class women is advertised as the standard and

normal style in the modern time by mass media. Influenced by such ideologies, other groups

of women are also desiring for such a liberated life, even if they have not been emancipated

in real. Bearing with such a dream but having to face the crude reality that they are

marginalized in the society, and therefore the means to fulfill their dream is bleak. They, as

a result, are subjected to the strain produced by the structural inequality not only inter-

gender but also intra-gender. Mismatched by the societal structure from their ideologies of

gender roles, those groups of women are more likely to involve in delinquency to

alternatively fulfill their dream or simply avoid the frustrating feelings. Therefore,

emancipation and economic marginalization theories tell us the same story within a

dynamic system of gender roles. Some groups of women have achieved numeric gains in

society as the result of emancipation and therefore influenced the societal attitudes toward

gender, which in turn incubates the equalized individual ideologies in a broad scope but

does not change the unequal structure fundamentally. This equalization lag happening in

the gender role repertoire leads to the strain and therefore an increase in delinquency for

the ones experiencing the mismatch most. Therefore, there is still a long way to go to reach

the real equality of gender for the whole society.

3.9. Tables and Figures

Table 3.1. A Schematic Representation of Gender Role Repertoire

Layer

Societal

Individual

Generalized

Attitudes toward

Gender in Society

Generalized

Structure of

Gender in

Society

Specific

Expectation

within an actor’s

frame of

reference

Specific

Behaviors of

actor

Family

The desirability of

gendered familial

behavior (e.g.,

Women should be

family-centered)

The general

family position

held by genders

(e.g., men are

boss at home)

Actor’s familial

goals (e.g.,

Whether and

when to marry

and how many

children)

Actor’s behavior

within the

family (e.g.,

performing

“feminine”

chores)

School

The desirability of

gendered behavior

at school (e.g.,

girls should be

good at humanities

while boys at

sciences)

The

equal/unequal

treatment of

genders by

teachers

Actor’s schooling

goal

(e.g., achieve

good school

performance)

Actor’s behavior

at school (e.g.,

overboard in

math)

Peers

The desirability of

relationship with

gendered peers

(e.g., a woman

should be with

women)

The friendship

network

structure (e.g.,

the proportion of

cross-sex

friends)

Actor’s

desirability of

gendered

relationships

(e.g., I want to

hang out with

guys)

Actor’s

friendship

network

structure (e.g.,

how many

cross-sex friends

in the network)

Work

and

income

The gendered

desirability of

work (e.g., a

woman should

work and earn

income)

The general

gender structure

at the workplace

(e.g., women

occupy low-paid

jobs)

Actor’s job and

income goal

(e.g., a good job

with decent pay)

Actor’s actual

employment

(e.g., an

assembly girl)

Note: This table is modified based on Giordano and Cernkovich’s (1979) work.

Table 3.2. The Composition of Survey Respondents

Pre-one-child

generations

One-child

generations

Total

Women

638

636

1,274

Men

460

272

732

Total

1,098

908

2,006

Table 3.3. Comparisons over Generations and Genders (T-test Statistics with Unequal Variances)

Category

Variable

Mean

(Std.

Dev.)

Intergenerational

Change = the

OCa – the POCb

Gender

Difference =

Women – Men

Gender Gap

among the

POC =

Women - Men

Gender Gap

among the OC

= Women -

Men

Dependent

Variable

Deviance

4.59

(2.44)

-0.04

(0.38)

-0.47***

(4.22)

-0.44***

(2.98)

-0.53***

(2.94)

Socialization

Sibling Care Responsibility (0 = no, 1

= yes)

0.48

(0.50)

-0.16*** (7.11)

-0.04

(1.60)

-0.08**

(2.52)

0.07*

(1.86)

Parents’ Power Dynamic (1 =

matriarchal, 2 = mild matriarchal, 3 =

egalitarian, 4 = mild patriarchal, 5 =

patriarchal)

3.24

(1.34)

0.01

(0.23)

1.90e-3

(0.03)

0.06

(0.74)

-0.09

(0.93)

Family Hope (0 – 4: the bigger the

number, the more hope expressed)

2.84

(1.47)

0.05

(0.71)

-0.25***

(3.73)

-0.31***

(3.01)

-0.23**

(2.53)

Family Control (1 = no, 2 = not

always, 3 = yes)

2.38

(0.73)

0.07** (2.08)

-0.05

(1.37)

-0.08*

(1.80)

-0.02

(0.37)

Cross-gender Friends’ Effect (0 = no,

1 = somewhat, 2 = yes)

0.75

(0.92)

-0.03

(0.82)

0.14***

(3.32)

0.18***

(3.23)

0.10

(1.51)

School Performance ( 1 = bad, 2 =

below average, 3= average, 3 =

above average, 4 = good)

3.56

(1.09)

0.01

(0.18)

0.03

(0.58)

0.03

(0.41)

0.03

(0.38)

Cognition

Legal Conscience (1 = no, 2 = except

for the unpunishable, 3 = except for

the unreasonable, 4 = yes)

3.32

(0.81)

-3.74e-3 (0.10)

0.01

(0.35)

5.38e-4

(0.01)

0.03

(0.56)

Masculinity/Femininity (0 – 160: the

bigger the number, the more

masculine/feminine)

28.86

(24.70)/

26.65

(23.53)

-0.57 (0.43)/-

9.53*** (4.90)

Table 3.3 (Continued) Comparisons over Generations and Genders (T-test Statistics with Unequal Variances)

Category

Variable

Mean

(Std.

Dev.)

Intergenerational

Change = the

OCa – the POCb

Gender

Difference =

Women - Men

Gender Gap

among the

POC =

Women - Men

Gender Gap

among the OC

= Women –

Men

Cogn

ition

Parent-child difference of

Masculinity/Femininity (1 = I am

more masculine/feminine, 2 = the

same, 3 = my parent is more

masculine/feminine)

1.99

(0.80)/

1.95

(0.80)

-0.07

(1.11)/

0.03

(0.62)

Economic Success Definition (0 =

no, 1 = yes)

0.57

(0.50)

0.18*** (8.13)

-0.02

(0.91)

-0.06*

(1.82)

-0.03

(0.83)

Success Aspiration (0 = no, 1 =

somewhat, 2 = yes)

1.66

(0.69)

0.15*** (4.91)

0.01

(0.27)

0.03

(0.72)

-0.07*

(1.73)

Inequality

Family Gender Discrimination (0 =

no, 1 = somewhat, 2 = yes)

0.63

(0.82)

-0.12*** (3.40)

-0.07*

(1.94)

-0.09*

(1.82)

-0.01

(0.22)

School Gender Discrimination (0 =

no, 1 = somewhat, 2 = yes)

0.57

(0.79)

0.02

(0.53)

0.02

(0.52)

-0.05

(0.94)

0.10*

(1.85)

Workplace Gender Discrimination (1

= women easier being promoted, 2 =

the same, 3 = men easier being

promoted)

2.22

(0.78)

-2.15e-3 (0.06)

0.03

(0.86)

0.15***

(3.23)

-0.13**

(2.39)

Income Gender Inequality (1 =

women have higher income, 2 = the

same, 3 = men have higher income)

2.46

(0.73)

0.06*

(1.83)

-0.05

(1.60)

0.05

(1.18)

-0.22***

(4.62)

*p<0.1; **p<0.05; ***p<0.01 (two-tailed test); a OC = One-child Generation; b POC = Pre-one-child Generation

Table 3.4a. Correlation Matrix for the Study Variables (Men, n = 732)

1 2 3 4 5 6 7 8 9 10 11

1.Deviance 1.00

2.Sibling -0.02 1.00

3.patriarchal 0.09** -0.02 1.00

4.Hope 4.30e-3 0.02 0.23*** 1.00

5.Control -0.05 0.05 0.08** 0.23*** 1.00

6.Friends 0.06* 0.06 -0.02 0.01 -0.02 1.00

7.School -0.10*** 0.04 0.02 0.19*** 0.05 0.09** 1.00

8.Cons. -0.13*** 0.01 0.06 0.28*** 0.26*** 0.01 0.15*** 1.00

9.Fem. 0.30*** 0.09** 0.12*** 0.21*** 0.05 -0.02 0.13*** 0.14*** 1.00

10.Fem. Dif. -0.12*** 0.04 0.05 -0.06 -0.09** -0.06 -0.03 -0.06 -0.11*** 1.00

11.Asp. 0.04 -0.01 0.18*** 0.43*** 0.26*** -0.06 0.13*** 0.31*** 0.22*** -0.12*** 1.00

12.Fam. Dis. 0.07* 0.18*** 0.01 0.03 0.02 0.17*** 0.09** -0.03 0.06* -0.07** -0.05

13.Sch. Dis. 0.07* 0.07* 0.02 -0.03 -0.02 0.17*** 0.08** 0.04 0.03 -0.06* -0.01

1.00

0.32***

1.00

0.06

0.03

1.00

0.36***

1.00

14.Job Dis. 0.01 -0.06* 0.17*** 0.18*** 0.11*** 0.03 0.06* 0.14*** 0.05 -0.07* 0.19*** -0.02

15.Inc. Ine. 0.01 -0.11*** 0.16*** 0.29*** 0.22*** -0.02 0.05 0.18*** 0.16*** -0.13*** 0.28*** -0.09** *p<0.1; **p<0.05;

***p<0.01

Table 3.5. Standardized Regression Coefficients for the Path Model (n=2,006)

Explanatory Variables

Response Variables

7. Legal Conscience

8. Success Aspiration

Deviance

Socialization

1. Sibling Care

2. Patriarchal Family

3. Family Hope

4. Family Control

5. Same-sex Friend Exclusiveness

6. School Performance

0.03

0.03**

0.09***

0.22***

0.04**

0.07***

-0.01

0.04***

0.12***

0.17***

-3.33e-4

0.06***

0.02

-0.11***

-0.02

-0.16**

-0.11*

-0.18***

Cognitive Traits

7. Legal Conscience

8. Success Aspiration

-0.35***

0.17**

Table 3.4b. Correlation Matrix for the Study Variables (Women, n = 1,274)

1 2 3 4 5 6 7 8 9 10 11

1.Deviance 1.00

2.Sibling -0.02 1.00

3.Patriarchal -0.14*** 0.10*** 1.00

4.Hope -0.06** 0.10*** 0.11*** 1.00

5.Control -0.07*** 0.13*** 0.07** 0.15*** 1.00

6.Friends -0.09*** 0.11*** 0.04 0.04 0.09*** 1.00

7.School -0.08*** 0.07*** 0.03 0.21*** 0.06** 0.05* 1.00

8.Cons. -0.12*** 0.07** 0.11*** 0.20*** 0.23*** 0.09*** 0.15*** 1.00

9.Fem. -0.13*** 0.07** 4.50e-3 0.07** 0.02 0.01 0.07** 0.08*** 1.00

10.Fem. Dif. 0.20*** -0.04 -0.06** -0.06** -0.18*** -0.09*** -0.09*** -0.17*** -0.15*** 1.00

11.Asp. -0.04 0.03 0.10*** 0.26*** 0.22*** 0.06** 0.17*** 0.27*** 0.03 -0.14*** 1.00

12.Fam. Dis. 0.03 0.08*** 0.10*** 0.08*** 0.02 0.14*** -0.02 0.03 0.05* -0.08*** -0.05*

1.00

0.02

1.00

0.35***

1.00

13.Sch. Dis. 0.04 0.02 0.02 0.04 0.05* 0.11*** 0.07** 0.01 -0.01 -0.01 -0.03 0.28***

14.Job Dis. 0.02 -0.03 -0.01 0.08*** 0.07** 0.07** 0.11*** 0.15*** -0.02 -0.10*** 0.18*** -0.04

15.Inc. Ine. -0.01 0.03 0.08*** 0.19*** 0.09*** 0.11*** 0.14*** 0.17*** 0.03 -0.12*** 0.25*** -0.08*** *p<0.1; **p<0.05;

***p<0.01

Gender Structure

9. Family Discrimination

10. School Discrimination

0.14**

0.15**

11. Job Discrimination

0.10

12. Income Inequality

0.12

Control

13. Gender-Woman

14. Generation-One Child

15. Childhood Rural Residence R2

0.05

-0.03

-0.16***

0.12

0.03

0.12***

-0.08**

0.16

-0.45***

0.02

0.19*

0.05

*p<0.1; **p<0.05; ***p<0.01

Table 3.6a. Standardized Regression Coefficients for the Path Model (Pre-one-child Generation Women, n=638)

Explanatory Variables

Response Variables

7. Legal

Conscience

8. Success

Aspiration

9. Femininity

10. Decreased

Fem.

Deviance

Socialization

1. Sibling Care

2. Patriarchal Family

3. Family Hope

4. Family Control

5. Same-sex Friend Exclusiveness

6. School Performance

0.01

0.04*

0.08***

0.17***

0.08**

0.10***

0.01

0.03

0.12***

0.18***

0.03

0.06**

0.08

0.09

0.05

0.22*

-0.07

0.19**

4.69e-3

-0.03

-0.21***

-0.04

-0.06**

-0.12

-0.17**

-0.08

-0.09

-0.25**

-0.02

Cognitive Traits

7. Legal Conscience

-0.10

8. Success Aspiration

9. Femininity

10. Decreased Fem.

-0.03

-0.17***

0.30**

Gender Structure

11. Family Discrimination

0.23*

12. School Discrimination

0.21

13. Job Discrimination

0.10

14. Income Inequality

0.09

Control

15. Childhood Rural Residence

-0.21***

0.11

-0.14***

0.15

-0.23

0.03

0.18***

0.08

0.22

0.08

*p<0.1; **p<0.05; ***p<0.01

Table 3.6b. Standardized Regression Coefficients for the Path Model (One-child Generation Women, n=636)

Explanatory Variables

Response Variables

7. Legal

Conscience

8. Success

Aspiration

9. Femininity

10. Decreased

Fem.

Deviance

Socialization

1. Sibling Care

2. Patriarchal Family

3. Family Hope

4. Family Control

5. Same-sex Friend Exclusiveness

6. School Performance

0.10

0.06**

0.08***

0.24***

0.04

-0.05

0.05***

0.06***

0.14***

0.02

0.08***

0.36*

-0.13*

0.15**

-0.18

0.06

0.04

-0.03

0.01

-0.13***

-0.07**

-0.06*

0.28

-0.32***

-0.02

-3.37e-4

-0.23**

-0.21**

Cognitive Traits

7. Legal Conscience

8. Success Aspiration

-0.34***

0.32**

9. Femininity

10. Decreased Fem.

-0.05

0.65***

Gender Structure

11. Family Discrimination

0.17

12. School Discrimination

0.04

13. Job Discrimination

14. Income Inequality

0.08

0.34***

Control

15. Childhood Rural Residence

-0.21***

0.13

-0.13**

0.12

-0.24

0.02

0.05

0.03

0.14

*p<0.1; **p<0.05; ***p<0.01

Table 3.6c. Standardized Regression Coefficients for the Path Model (Pre-one-child Generation Men, n=460)

Explanatory Variables

Response Variables

7. Legal

Conscience

8. Success

Aspiration

9. Masculinity

10. Decreased

Mas.

Deviance

Socialization

1. Sibling Care

2. Patriarchal Family

3. Family Hope

4. Family Control

-0.01

0.01

0.14***

0.27***

0.01

0.04*

0.22***

0.15***

0.67*** 0.15*

0.40***

0.03

0.08

0.02

-0.03

-0.08

-0.18

0.02

0.03

3.19e-3

5. Same-sex Friend Exclusiveness

6. School Performance

-0.01

0.03

-0.02

0.02

0.03

0.14

-0.05

2.97e-3

0.14

-0.21**

Cognitive Traits

7. Legal Conscience

-0.43***

8. Success Aspiration

9. Masculinity

10. Decreased Mas.

0.04

0.35***

-0.28**

Gender Structure

11. Family Discrimination

0.17

12. School Discrimination

0.14

13. Job Discrimination

0.09

14. Income Inequality

-0.13

Control

15. Childhood Rural Residence

-0.14*

0.16

0.02

0.25

0.32

0.11

-4.59e-4

0.01

0.14

0.18

*p<0.1; **p<0.05; ***p<0.01

Table 3.6d. Standardized Regression Coefficients for the Path Model (One-child Generation Men, n=272)

Explanatory Variables

Response Variables

7. Legal

Conscience

8. Success

Aspiration

9. Masculinity

10. Decreased

Mas.

Deviance

Socialization

1. Sibling Care

2. Patriarchal Family

3. Family Hope

4. Family Control

5. Same-sex Friend Exclusiveness

6. School Performance

-0.07

-0.03

0.09**

0.22*** 0.04

0.11***

-0.01

0.04*

0.07***

0.22***

-0.09***

0.05**

0.02

0.09

0.11

-0.09

-0.25*

0.25**

0.10

0.08**

-0.02

-0.12

-0.07

-0.02

0.16

0.31***

-0.13

-0.29

0.29*

-0.37***

Cognitive Traits

7. Legal Conscience

-0.54***

8. Success Aspiration

9. Masculinity

10. Decreased Mas.

0.38

0.30***

-0.31*

Gender Structure

11. Family Discrimination

-0.11

12. School Discrimination

0.12

13. Job Discrimination

-0.06

14. Income Inequality

-0.21

Control

15. Childhood Rural Residence

0.03

0.12

0.13***

0.21

-0.28

0.04

-0.04

0.03

-0.10

0.17

*p<0.1; **p<0.05; ***p<0.01

Table 3.7. Effect Decomposition for the Path Models

DirectNOC Women Indirect Total Direct OC WomenIndirect Total Direct NOCIndirect Men

Total Direct OC MenIndirect Total

Socialization

1.Sibling 0.23*** 0.23***

2.Patriarchal -0.17** -0.17** -0.32*** -4.40e-3** -0.32** 0.05* 0.05* 0.31*** -0.02* 0.29*

3.Hope -0.01** -0.01** 0.08*** 0.08*** -0.05** -0.05**

4.Control -0.10* -0.10* -0.12** -0.12** -0.12*** -0.12*** -0.12*** -0.12***

5.Friend -0.25** -0.25** -0.23** -0.05** -0.28** 0.29* -0.08* 0.21*

6.School -0.02** -0.02** -0.21** -0.01* -0.22* -0.21** -0.21** -0.37*** 0.02** -0.35**

Cognitive Traits

7.Cons. -0.54***

8.Aspiration

9.Fem./Mas. -0.17*** 0.30***

10.Dec. Fem./Mas. 0.30** - 0.31*

Gender Structure

11.Family Dis. 0.23*

0.23*

-0.17***

0.30**

-0.34***

0.32**

0.65***

-0.34*** -0.43***

0.32**

0.35***

0.65*** -0.28**

-0.43*** -0.54***

0.35*** 0.30***

-0.28** -0.31*

12.School Dis.

13.Job Dis.

14.Inc. Ine.

0.34***

15.Residence

0.05**

Control

0.03** 0.03**

0.02*

*p<0.1; **p<0.05; ***p<0.01; Only significant effects are displayed. The rest is represented in Table 3.6a, 3.6b, 3.6c, and 3.6d.

Figure 3.1. Theoretical Re-composition of the Gendered Pathways to Deviance

Women Men

POC

Figure 3.2. Grouped Structural Equation Models Detecting Influence Paths to Deviance. Note: Only paths significant are displayed. The

rest is represented in Table 3.6a, 3.6b, 3.6c, and 3.6d. Childhood residence is controlled but not presented here.

Chapter 4. Who Is Like Whom: A Qualitative Research on the Convergence of the

Gender Gap in Offending Among the One-child Generations in China

4.1. Introduction

The trend toward the convergence of the gender gap in offending has been widely

acknowledged and quantitatively researched worldwide (Estrada, Backman, & Nilsson,

2016a, 2016b; Lauritsen, Heimer, & Lynch, 2009; Savolainen et al., 2017; Schwartz, 2013a;

Schwartz, Steffensmeier, Zhong, & Ackerman, 2009). The difference in men’s and women’s

offending has been shrinking as indicated by official measures of crime. Quantitative

research verifies this change as well but has failed both to clarify who contributes most to

the convergence—men or women—and to explain what factors drive the convergence.

Moreover, most research, limited by the accessibility of the quantitative data, has been

disproportionally based on the “mainstream” sample (i.e., data from Western countries,

particularly the United States; Schwartz, 2013b). However, considerable endogeneity and

complexity exist in the association between the patriarchal social order and other

theoretically salient macrosocial characteristics because of their relatively stable social

structure in the contemporary era (Savolainen et al., 2017). For example, one of the most

frequent critiques on emancipation theory is that women’s liberation is an outcome of

changes in other fundamental culture and social structure. Thus, the assumed relationship

between emancipation and the increase in women’s offending may be spurious.

Consequently, the stability in social structure makes it difficult to detect the real causal

relationship between variables due to the endogeneity of the variations in those variables.

In contrast to highly developed countries, some nations are still in transitional stages

of economic modernization. Therefore, they may experience tremendous changes in their

culture and social structure within a short time brought about by external social forces.

Studies of possible variations caused by some unusual external force are valuable in

uncovering the relationships between variables and contribute to our understanding of the

underlying mechanisms of those relationships. China’s one-child policy is the source of

one of those external forces, which has reshaped the culture and social structure of a nation

within 35 years. This policy, then, provides us with scholarly leverage for exploring the

change in the gender gap in offending and the causes of that change from the perspective

of a natural experiment. Moreover, China has rarely been studied, even though one-fifth of

the world’s population lives there. A study of crime in China, therefore, offers a unique

opportunity to test current criminological theories regarding the gender gap in offending.

It is widely recognized that China’s one-child policy has dramatically imbalanced

the gender composition of the Chinese population (Hesketh & Zhu, 2006; Hudson & Den

Boer, 2002; Zhu, Lu, & Hesketh, 2009). The gender composition imbalance, moreover, has

been affecting every aspect of the culture and social structure of Chinese society. As a

result, it has profoundly influenced gender norms and gendered behaviors, including

criminal activities (Barber, 2000b; Campbell et al., 2001; Edlund et al., 2013; Guttentag &

Secord, 1983; South & Messner, 1987, 2000). Statistics indicated that the share of crime

contributed by women has doubled within 20 years in China since the first one-child

generation reached the age of criminal responsibility (see Figure 4.1). This change has

contributed to the convergence of the gender gap in offending, which is consistent with the

global trend. However, the policy adds an extra dimension to China’s case that enables us

to probe the causes of the convergence by controlling the endogeneity in the variations of

social variables through an exogenous experiment.

4.2. Literature Review

4.2.1. Convergence

Although the narrowing of the gender gap in offending has been observed across countries

(Carrington, 2006; Estrada et al., 2016b; Heimer et al., 2012; Heimer et al., 2009; Lauritsen

et al., 2009; Savolainen et al., 2017; Schwartz et al., 2009; Steffensmeier et al., 2005, 2006),

the pattern of the trend is still controversial: Does it reflect a change in men’s behavior over

time or is it due to a change in women’s behavior? Influenced by the works of Adler (1975)

and Simon (1975), most research assumes male behavior as the “norm” in the criminal

world and female behavior as the “mimic,” thus hypothesizing that the convergence is a

result of increases in women’s involvement in crime. Moreover, this female-to-male pattern

has been supported by a significant body of empirical studies, especially those based on

data from the United States (e.g., Heimer et al., 2012; Schwartz et al., 2009; Steffensmeier

et al., 2005, 2006) and other highly developed nations (e.g., Carrington, 2006; Estrada et

al., 2016b, 2017; Schwartz, 2013b). The opposite pattern, however, has been supported by

research in Sweden, where Estrada and colleagues found the convergence due to decreases

in men’s offending (Estrada et al., 2016a, 2017).

In addition to the longitudinal studies, most cross-national research also confirms a

narrowing gender gap in offending that is caused by increased offending by women

(Hunnicutt & Broidy, 2004; Schwartz, 2013a). Using a larger sample of countries, recent

research has indicated a mix of both patterns, i.e., increased women’s offending and

decreased men’s offending, contributing simultaneously to the narrowing of the gap

(Savolainen et al., 2017). However, in both single country and cross-national research, the

100

data are drawn primarily from Western and economically advanced nations. Therefore, the

generality of the converging pattern is limited by this consideration.

4.2.2. Causes of Convergence

Some criminologists have proposed that the convergence may not be due to actual changes

in men’s and/or women’s criminality but is due to changes in crime control and punishment

within the criminal justice system. In contrast to a behavior change hypothesis, then, these

criminologists posit the net-widening hypothesis. Of course, it is possible that convergence

of official crime rates reflects both changes in individual criminal behaviors and responses

to those behaviors from the criminal justice system; that is, official crime rates reflect the

interplay of these two effects.

The behavior change hypothesis assumes that the change in the gender gap reflects

the real propensities to offending. Emancipation theory originally specified the behavioral

and predisposing changes. Enhanced freedom and equality between genders (Adler, 1975)

and augmented opportunities to commit crime brought about by women’s increasing

engagement in extrafamilial activities (Simon, 1975) promoted criminal involvement by

women. This hypothesis sees greater gender equality as “masculinizing” women’s

behavior, which results in an increase in women’s motivations for offending (Levin & Fox,

2001). In addition, “conflict between what is supposed to be possible and what is actually

available for women” (Schwartz, 2013b, p. 791) may cause strain and frustration, which

may provoke criminal responses (Agnew, 1992; Rosenfeld, 1989; Schwartz, 2013b).

Economic marginalization theory refutes the liberation hypothesis; in contrast, it

ascribes the increasing women’s offending to their worsening economic position in society

and the “feminization” of poverty (Steffensmeier, Emilie Allan, & Cathy Streifel, 1989).

101

Gains in gender equality do not necessarily translate into economic independence or parity

of women (Hartnagel & Mizanuddin, 1986; Hunnicutt & Broidy, 2004; Renzetti, 2013).

While the women’s movement may have allowed some upper- or middle-class women to

make inroads into formerly men’s professions, most women have become increasingly

economically marginalized. They are more likely to be unemployed or, if employed, they

are more likely to be in less secure, lower paid, less or unskilled part-time jobs in which

career prospects are minimal (Chesney-Lind & Shelden, 2014; Holtfreter, Reisig, &

Morash, 2004). Consequently, the economic pressures lead women to be more involved in

crime, especially non-violent, petty, property offending (Steffensmeier, 1978).

The net-widening hypothesis is also rooted in emancipation theory but as a

derivation of the original theory, which centers on the reactions of citizens and criminal

justice system professionals toward the alleged “liberated” women and new gender roles

and norms. This hypothesis emphasizes changes in the treatment of women involved in

crime. It postulates that the convergence of the gender gap in offending was brought about

by harsher treatment and stricter punishment of women offenders, rather than by actual

behavioral change on the part of women (Merlo & Pollock, 2015). Moreover, another

gender-specific, net-widening effect relates to changes in consumption patterns and

development of technology. More women are involved in crime due to the shifting market

consumption patterns (e.g., more self-help markets and heavier credit card usage) that

enable certain kinds of property crimes. Moreover, advanced technologies (e.g.,

surveillance cameras) have been applied prevalently in preventing consumer crimes that

are mainly perpetrated by women (Scott J. South & Messner, 1987). The net-widening

hypothesis suggests that women’s documented crime rates may be increased in two ways:

102

(a) societal attitudes toward women offenders change from being sympathetic to punitive,

which leads to more willingness of citizens to report women suspects, a higher probability

that police will arrest them, and more severe judgment from courts (Steffensmeier, 1980)

and (b) the increasing bureaucratization and routinization of the criminal justice system

decreases individual discretion and leads to non-discriminatory standards and outcomes

(Chesney-Lind & Shelden, 2014; Feld, 2009; Hagan, 1977).

Compared to the behavior change hypothesis, the net-widening effect has received

a greater amount of empirical support (e.g., Chesney-Lind & Paramore, 2001; Estrada,

2006; Estrada et al., 2016b; Schwartz, Steffensmeier, & Feldmeyer, 2009; Steffensmeier et

al., 2005; Witterbrood & Junger, 2002). By employing diverse operationalizations of net-

widening indicators and comparisons between official crime rates and self-reported crime

or victim-survey data, the convergence of the gender gap in offending is argued to be

largely an artifact of changes in social control by net-widening theorists. The

criminalization of less serious forms of deviant behaviors brings more women into the

criminal justice system because women’s offending is more likely than that of men to be

petty and minor (Steffensmeier & Allan, 1996). Although the “War on Drugs” since the

1980s “has translated into a war on women” (Chesney-Lind, 1991, p. 57) and is more likely

to bring women into justice system than before, women still play a marginal role in drug

organizations and are likely to receive lesser sentences for a much smaller amount of drugs

(Sokoloff, 2005) or just for possession of drugs rather than trafficking in them

(ChesneyLind, 1991). Compared to more stringent enforcement of serious offenses, one

might expect more latitude or discretion by the criminal justice system professionals for

103

the less serious offenses, if such discretion is legally permissible (Schwartz, 2013b;

Steffensmeier et al., 2005, 2006)

Existing empirical research into this topic is disproportionately quantitative and

varies significantly in sample selection and variable operationalization, which lowers the

reliability of the results (Schwartz, 2013b; Schwartz, Steffensmeier, & Feldmeyer, 2009).

Moreover, the quantitative methodology may be inappropriate for testing the behavior

change hypothesis because the behavioral change is latent and, therefore, not easily

detected by quantified variables. In addition, there has been no single study that answers

both questions about who is converging to whom and what is causing the convergence.

Another shortcoming of the existing research is that the possible endogeneity of variations

of diverse social variables may blur and impede the discovery of the causal relationship

and underlying mechanism. Finally, both single-country and cross-national research has

been based on data from a limited number of advanced Western countries; thus, the results

are relatively circumscribed.

4.3. Current Study

To overcome some of the limitations of quantitative research for understanding the

convergence of the gender gap in crime, this study adopts qualitative methods to explore

changes in the gender gap in offending and to explore potential reasons for such changes

from the perspective of criminal justice professionals. Moreover, an understudied country,

China, is the site of data collection. Besides being one of the world’s most populous

countries, China’s one-child policy provides a unique “window” for exploring various

causes of the convergence in the gender gap in offending. Different from the Western

experience, China saw no explicit women’s movement in the 1960s; nor was it influenced

104

by the Western movement because of its closed regime at that time. However, the onechild

policy, although not directly on gender order, sharply changed the structure of the Chinese

population. Therefore, such a different country sample provides a unique way to explore

the mechanism by filtering the influence of the emancipation movement that has been

argued about for decades.

To examine the potential effect of the one-child policy, the analyses here center

around differentiation by gender and by generation (pre-1980s and post-1980s, i.e., preone-

child and one-child generations). Comparisons are made between these two sets of groups

to capture any inter-generational change in the gender gap in offending.

4.4. Method

4.4.1. Qualitative Approach

I use the interview method for this study. The research participants were professionals in

the criminal justice system in China who were recruited through personal networks rather

than a formal institutional channel to alleviate any pressure from the system they might

feel and to encourage them to express their honest opinions. Most interviews were

conducted in casual settings. Some interviews were conducted through WeChat (a popular

message and social networking platform in China), in which the researcher did not meet

the interviewee in person, but rather conducted the interview through audio and typed

messages. To increase rapport and put the participants at ease, the interviews were

conducted in the style of a casual chat (Warren & Karner, 2010).

4.4.2. Participants

Fifteen professionals currently working in the Chinese criminal justice system participated

in the interviews. Participants were recruited using snowball sampling methods. The

105

researcher began with her personal network, contacting persons who were criminal justice

professionals or who might have friends or relatives working in the system. After each

interview, the interviewee was asked to introduce the researcher to colleagues or friends

who worked in the system, and whom they thought would be interested in and appropriate

for this research. Recruitment continued until 15 interviews were completed, which was

the number approved by the university’s institutional review board.

The participants represented a wide range of professions in the Chinese criminal

justice system, including one professor at a police college, several police officers,

prosecutors, court clerks, judges, and one sheriff (see Table 4.1). Nine of the 15

interviewees were men. Eight of the 15 were born before 1980, the year the one-child

policy was announced, while seven were born after 1980 and belonged to the one-child

generation (OCG for short, correspondingly POCG as the acronym of pre-one-child

generation). Five of them had less than five years of experience in the justice system, but

four had more than 20 years of experience in the system. Among the 15, the shortest tenure

was 2.5 years, and the longest was 36 years.

4.4.3. Data Collection

Interviews were conducted between June and August 2017 in China. Twelve interviewees

were professionals in the criminal justice system in a capital city of an eastern coastal

industrial province. Three interviewees were criminal justice professionals in another city

in the same province. The economy of this province played a pilot role in the process of

economic reform and opening up in China, which incubated a commerce-oriented

environment and massive private and small enterprises. Correspondingly, the per-capita

income of this region was relatively high in China, and therefore, attracted an immense

106

amount of migrating population from the rural areas of China. The duration of interviews

depended on the conversation location and environment. The in-person interviews lasted

approximately 1-1.5 hours; those conducted via WeChat were approximately 15-30

minutes long.

Prior to each interview, the researcher introduced herself and this study to the

interviewee including the aim of the research (academic research on women’s crime in

China) and the sample question of the interview (“What is the trend of women’s

offending?”). Based on the potential participants’ initial understanding of the purpose and

content of the interview, they decided whether to participate in the research. If individuals

consented to participate, the researcher then asked them to decide when and where they

would like the interview to take place. As a result, most interviews were conducted at

locations that were private and convenient for the interviewees, such as a restaurant,

teahouse, coffee shop, or the interviewee’s home. Some interviews were conducted via the

WeChat Message system. The criminal justice professionals identified the evenings or

weekends as the most convenient times for the interview. Most in-person interviews were

conducted while the researcher was dining with the interviewees, which was defined as a

casual environment for personal conversation in Chinese culture (Shen, 2015). Because the

interviews occurred in a casual and informal setting, most participants who attended in

person refused to allow the use of a recording device after being asked for permission to

do so by the researcher; nor was it appropriate in this context for the interviewer to take

notes. In addition, many participants mentioned political sensitivity as a concern. Five

explicitly refused permission to use a voice recorder for the interview because of their

worries that the record would be used as evidence detrimental to their career. Only one

107

participant agreed to record the interview. However, the responses of that participant

seemed to reflect hesitation (perhaps holding back); for example, she gave “unsure” or “no

feeling” to questions regarding trend changes but went no further when the interviewer

asked for examples supporting her conclusion than to offer the passive response, “I have

never done statistics.” Except for the one recorded interview, all transcripts were generated

by the researcher right after the interviews when her memory was freshest. The transcripts

were word-processed and simultaneously translated into English after all interviews were

completed in 2017.

The researcher began the interview with questions regarding the interviewee’s

demographic information—age, occupation, and years of experience with the justice

system. The interviewee’s gender was determined by interviewer observation. After that,

the interviewer asked the questions following the guide. During each in-person interview,

the interviewer used pause and probe techniques with verbal and nonverbal responses to

the participant. During each WeChat interview, the interviewer adjusted and supplemented

questions according to the participant’s responses.

4.4.4. Interview Questions

A semi-structured interview guide was developed. Questions focused on both

interviewees’ perception of the trend in the gender gap in offending and on the treatment

of men and women offenders. About the gender gap in offending, two main questions were

asked to stimulate discussion:

1. Do you think women’s share of crime has changed in the post-1980s’ generation

compared to the pre-1980s’ generation?

108

2. Have any differences emerged in criminal offending patterns between

generations?

The first question sought to gauge the interviewee’s perception of the change in the

share of crime committed by women, whereas the second question was intended to

determine if interviewees perceived a change in crime patterns across generations.

Two major questions also set the stage for the discussion of differences in the

treatment of male and female offenders:

1. Is there any difference in the treatment of men and women suspects or offenders

in the criminal justice system? For example, are women more likely than men

to get a lenient sentence?

2. Do you treat men and women who are involved in crime differently in your

work?

The first question was intended to gauge the participants’ perception of whether the

criminal justice system treats male and female offenders differently, while the second was

intended to measure the professionals’ sense of their own differential treatment of male and

female offenders.

Guided by these four primary questions, additional questions were asked during the

interview depending on the responses, identity, and status of the interviewees to probe for

additional and more specific information.

4.4.5. Data Analysis

Data were imported into NVivo12 (QSR International Pty Ltd., 2018). Using an

interpretive method, the researcher read, identified, and categorized units of meaning or

text units. Themes emerged after reflection on these categories. Guided by the four primary

109

questions, three themes emerged from the interviews: (a) the convergence of the gender

gap in offending, (b) the net-widening effect on the convergence, and (c) the socialization

effect on the convergence. The first two themes were explicitly revealed by the responses

to the guide questions, whereas the third theme emerged during the interviews. New codes

were added, and definitions of codes were revised as the analysis process continued. An

iterative process was followed as the researcher returned to the data to make certain the

interpretation as represented by categories and themes was consistent with the data.

4.5. Results

Three themes emerged from the data (see Figure 4.2). The first theme centers on the

convergence between men’s and women’s crime in the one-child generations compared to

their older counterparts. This theme has two sub-themes corresponding to the first two

interview questions: (a) changes in terms of the crime rate, and (b) changes in crime patterns.

The second theme relating to the treatment of male and female offenders by the criminal

justice system has three sub-themes: (a) systematic treatment, (b) the professionals’

individual treatment, and (c) the inter-generational transformation in leniency. A final theme

emerged around the roles of socialization, success aspiration, and gender inequality

experienced by offenders and the effect of these factors on behavioral change. To protect

the participants, all specific information (e.g., the name of the employer or the name of a

relevant place) was coded and anonymized

4.5.1. Theme 1: Convergence of the Gender Gap in Offending

Most professionals stated that the gender gap in offending has been narrowing

across generations in China. The convergence is reflected in two ways: as an increase in

110

the share of crime committed by women and as a(2) similarity in men’s and women’s

offending patterns. 4.5.1.1. Crime Rate

Twelve professionals (80.0%) stated they felt men’s crime was growing, but

women’s crime was growing faster, which contributed to an increase in women’s share of

criminal offending. Two interviewees (13.3%) differentiated the types of crime and stated

there was no change in violent crime, but that there was a sharp increase in property and

occupational crime perpetrated by women. Another participant (6.7%). whose interview

was audio-recorded, said she “never made statistics and had no feeling about the change.”

This participant was also the one who had the least experience (2.5 years) in the criminal

justice system (see Table 4.2).

The conclusion that the share of crime committed by men and women was

converging was based on the observation that there were more young women involved in

crime now compared to those of the pre-1980s generations. One professional with rich

experience in prison guard training gave some numerical examples.

The female crime rates have increased a lot. The rates are very high nowadays.

Although the general crime rates have been increasing, the women’s part grows

more rapidly. Look, there is another women’s prison newly built (in our region).

Why? [That is] just because the number of women inmates was over the maximum

capacity of the old one … [answering the supplemental question about the

capacities of both facilities] The old one is at C [a district name]. It was designed

to accommodate 2,000 inmates, but in fact, contained 4,000. However, it was still

not enough, so another one has been built. [I am] not sure about its capacity;

however, it must be above 4,000. (Respondent A)

The convergence is significant but may be sensitive to the type of offenses. Two

male professionals noted that the convergence happened inconsistently across offenses.

There was no obvious change for violent crimes, but they observed significant increases in

property and occupational crimes.

111

It depends on types of crime. There is little change that happened in violent crimes.

Compared to men, women only occupy a tiny share in all (violent) criminal cases.

However, women’s share in crime has increased a lot in property crimes, especially

the occupational crimes in which the women’s share has been on par with men’s

and close to 50 percent. The occupational crimes were not common in women’s

crime in the past, mainly because female heads were rare at that time. (Respondent

Respondent L noted, “I have no particular feeling of the ascent, maybe only a

little.…There is no change in violent crime, but the proportion of women offending has

been increasing in property-, drug- and job-related crimes.”

Because the inherent difference in physical strength between genders is constant

across generations, violent crime is still a male-dominated arena in a country with strict

gun control. However, changes have appeared in other crimes that demand less physical

power by the offender. Some professionals ascribed these changes to emancipatory gender

roles: women now have more occupational options and extrafamilial activities than in the

past, which affords them more opportunities to engage in some traditionally

maledominated offenses, such as corruption, embezzlement, drug trafficking, and other

property crimes.

4.5.1.2. Crime Pattern

Consistent with the convergence of the gender gap in crime rates, most professionals noted

that the patterns and forms of crime committed by men and women are also converging

(see Table 4.3). Nevertheless, four professionals believed that major differences in men’s

and women’s crime patterns continue to exist and are significant across generations. Two

OCG professionals noted that the similar trend in new types of crime was the drive for the

convergence.

112

Eight professionals noted there was a convergence in the crime pattern between

genders and that changes in women’s offending were the causes of the convergence/

Women were converging toward men in terms of offending types and tactics. Many

professionals mentioned the diverse types of crime committed by young women, but they

observed that occupational and drug offenses were those in which women’s offending had

become most similar to that of men. One professional (Respondent E) connected this

phenomenon with women’s emancipation in recent generations, nevertheless he also

indicated there was still a difference in terms of severity because of the gender difference

in extrafamilial activities and physical characteristics:

Women’s crime has been increasing in some types, such as embezzlement and

larceny. The primary cause, in my opinion, probably is the expanding range of

women’s occupations….(answering the supplemental question about the severity

compared to men’s crime) That is still a bit different. For example, the amount of

money involved in larceny perpetrated by men is on average higher than that by

women. Women are common to see in petty theft. Moreover, there are usually more

techniques or strengths involved in men’s crime. Women, because of their lack of

technical skills or physical strength, can only steal some easy things. Therefore, the

sums involved in their cases are less. Women’s preys are always those things that

are easy to steal, or they have opportunities to encounter in their work or daily life.

(Respondent E)

This view is consistent with research findings based on quantitative studies in the

United States (Archer, 2000; O’Brien, 1999; Steffensmeier, 1980; Steffensmeier,

Steffensmeier, & Rosenthal, 1979). Regardless of the increase in women’s offending, the

consequences and dangerousness caused by their criminal behavior are still not as serious

as those of men. Therefore, even if the convergence has happened in terms of crime rate

and pattern, the severity is typically still significantly different between men and women.

Despite that the convergence in the gender gap in the crime rate and crime patterns

was noted by most professionals, crime was still viewed as a traditionally masculine

113

symbol in Chinese culture. Some professionals believed the increase in women’s offending

was a result of the masculinization of gender roles. Respondent M and Respondent F

opined, “I would say women converge toward men. Women’s offending is becoming

increasingly masculine, including the offense types or the offending methods” and “The

crime types and tactics are converging, indeed. You can find women criminals commit

almost every crime type that was traditionally committed by men,” respectively.

Those professionals believed that crime was a way of showing masculinity;

correspondingly, they equalized women who committed traditionally male-dominated

crime to men, because the women were believed to be doing something “exclusively”

reserved to men, thereby breaking the traditional ideology of femininity.

Not only in those traditional offenses but the convergence also emerges in new types

of crime along with the technological development. Professionals indicated that Internet-

based crime is an emerging domain contributing to the narrowing of the gender gap.

Per Respondent H, “In my opinion, there is some difference (between generations),

but not (women) converging to men in offending. The whole criminal world is turning from

the traditional ‘offline’ to the newly emerging ‘online’ field, such as gambling. Another

trend is more women involved in drug-related crime. According to Respondent I,

The share of the new types of crime in women’s offending is increasing to some

extent, especially the internet fraud and pyramid schemes. For those offenders, men

and women are converging in the means of crime commission and culpability. I

believe that is because the division of labor is elaborate in those new types of

offenses. Moreover, they are not the violent crime and are demanding more in

intelligence capital input, therefore more similar for both men and women

physically. (Respondent I)

Compared to the traditional offenses, these new types of crime fit women’s

characteristics better, which the professionals perceived to be a reason for the narrowing of

114

the gender gap in offending. Some professionals, however, believed there was no

significant change in the gendered pattern of offending across generations. For women’s

crime, the non-violent offenses were still the dominant types because of either the physical

insufficiency or “high morality” of women. Per Respondent B,

I do not think (there is any change). Men commit much more violent crimes than

women, and women’s share of violent crime is always meager. After all, violent

crimes are demanding force and strength. The most common types of women’s

crime are fraud, larceny, prostitution, and drug-related offenses. Even some

women who are involved in some violent crimes, such as robbery, are mainly

aiding and abetting. It is rare to see a mix of property and personal offense types

in women’s offending. It is either property or personal crime, because of

(women’s) moral value. It is rare to see a mixture of both, like killing or hurting

someone for money or other extremely vicious cases (perpetrated by women).

While (for women), the most common reason for the personal crime is emotional

disputes. (Respondent B)

The marginal position held by women in serious and violent crimes was also

observed in the quantitative research based on the Western nations’ data (Schwartz, 2013a),

although the reason has rarely been revealed in previous research. The Chinese

professionals attributed the difference to feminine characteristics, both physical and

psychological, which restrict women’s participation in violent offending. Moreover,

consistent with the findings from Western nations, women’s violent crime typically

happens in intimate and private settings (Archer, 2000) and involves personal relationships.

From this view, the rate and type of violent crime committed by women are different from

that of men; men commit more violent crimes that result in more injuries, and they commit

these crimes against strangers as well as people they know (Campbell, 1993; Miller, 2001;

Miller, 2005).

This view of the gender difference in seriousness also extends to non-violent crimes.

Women are more likely to be involved in property-related crime, especially the petty

115

offenses, which is responsible for the marginal increase in women’s offending. Respondent

K proposed, “There is an increase in women’s share of offending. However, there is no

particular change in the crime pattern or form. The property infringement still dominates

women's crime.” Respondent J believed,

Differences between the men’s and women’s crime continue to exist, although the

women’s offending is catching up in terms of crime rates. Women’s and men’s

crime have distinctive characteristics, respectively. Women are increasing as the

offenders in some types of crime, such as larceny, drug abuse, and traffic accidents.

(Respondent J)

A professional summarized the observations of women’s crime patterns from the

perspective of prisoners, in which violent offenses were the primary type of offending.

Respondent A proposed,

Violent crimes are the primary (in prison). Because most property crimes are petty

offenses, the imprisonment term is short. If the remaining term is shorter than one

year after deducting the time spent in the detention house, the inmates will not be

transferred to prison. After the Penal Correction (VIII) (2011), if (the remaining

term is) shorter than three months, the inmates will not be transferred and stay in

the detention house for the remaining sentence. Therefore, only the inmates who

have long imprisonment term will be transferred to prison. Consequently, the

violent crime is the primary offense type (in prison). Also, one-third of the women

inmates are for drug-related crimes. In general, (they) take the drug at first, then

commit drug trafficking. However, they all are at the bottom (of the drug trafficking

chain). None of them have had connections with the higher levels. They even don't

know the persons who are at the higher levels. Only one alleged “big sister” among

the inmates I contacted in recent decades is the exception. Her partner was the boss

(of drug trafficking), who could be said: “a foxy rabbit with three burrows.” He had

many crack dens all over the country. Eventually, her partner was sentenced to

death, and she got a suspended death sentence. Later, her sentence abated to life

imprisonment and then to fixed-term imprisonment. When I saw her, she has served

for over ten years. Also, there are many criminals of fraud, especially in XX

Province (where we are). However, fraudsters are not very common among the

young generations because it (committing fraud) needs experience after all. The

next most common crime is accommodating prostitution. (Respondent A)

Even within the field of violent and other serious offenses, which would result in

long-term imprisonment for the offenders, the inferior position held by women was still

remarkable and consistent, as this interviewee’s comments indicate. The only high-level

116

drug criminal known by this professional was the partner of and dependent on a real male

boss. Most women drug dealers were at the bottom of the trafficking chain with little, if

any, connection to the important roles in the business. Because they played primarily as the

“retailer” of the drug trade, they were highly vulnerable to being detected and arrested.

Therefore, if there is a net-widening tendency of the criminal justice system, women’s

arrests may increase without any corresponding change in offending behaviors simply

because there is a great share of women in the pool of offenders who could be easily

targeted and arrested (Blumstein & Wallman, 2000; Steffensmeier et al., 2005).

The minor-offense pattern of female offenders is consistent with the findings of

other research (e.g., Shen, 2015; Steffensmeier, 1983). Even being charged for serious

crimes, women are more likely to play a minor role in the process, such as compliances

(Giordano et al., 2002; Leonard, 1982). The gender stratification in the criminal world

mirrors that in the whole society in China. Even the women offenders, who were perceived

as masculine by professionals, are still largely dominated by men and observe the

patriarchal gender order consistent with the socially prescribed ideologies.

4.5.2. Theme 2: Net-widening Effect on Convergence

Most interviewed professionals noted that the system’s and their own individual treatment

of men and women offenders is unbiased (see Table 4.4), except for some special situations

regulated by law, such as pregnancy and lactation. However, many of them also mentioned

that women offenders’ penitent attitude and their victimization experiences that would also

be considered for leniency. Nevertheless, there is a trend of diminishing lenient conditions

for the one-child generation of women offenders.

117

4.5.2.1. System Treatment

Most professionals mentioned the tenet that “all people are equal before the law.” One

professional also described the structure of the Chinese criminal justice system designed

to ensure the unbiased treatment of all offenders. The antagonistic roles played by the court

and the prosecutor reduce the possibility of discretion in judgment and sentencing. He gave

an example to illustrate the power balance in the Chinese system:

The relationship between the courts and the prosecutors is antagonistic to balance

the power of each other, which is the foundation of Chinese judicatory system.

Therefore, even if the court sympathizes with this woman, the prosecutor is

responsible for supervising (whether the law is indiscriminately carried out and the

sentence is reasonable). In contrast, if the prosecutor sympathizes with someone,

then the court has the responsibility for supervising (the process and decision). (For

example,) There was a case detected by me years ago: a woman poured

concentrated sulfuric acid caused another person’s severe injury, but the trial court

only sentenced her to four-year imprisonment considering her prior experience of

victimization. This was a very light sentence for a case with serious injuries. The

prosecutor, however, filed a protest against this sentence afterward, which

eventually pressured the court to change the sentence to ten-year

imprisonment…(Moreover,) the court would form a collegiate bench for each case

before 2017, which was composed of seven members. The verdict would not be

submitted or valid unless four of the seven agreed. Therefore, the outcome was

unlikely to mix with personal emotions. However, judges are in charge of simple

cases independently after 2017, so the personal feelings may affect the result after

that. (Respondent B)

Regardless of such a mechanism to avoid biased treatment, however, the law regulates

some specific situations in which women can get lenient treatment for humanitarian

considerations during sentencing and imprisonment, including pregnancy and lactation.

In addition, some “soft” conditions would also influence the judgment and sentencing as

well, including the attitude of contrition, the reason for the crime, and subjective

culpability. The law also considers those conditions but gives more latitude of discretion

for professionals rendering judgment. According to Respondent A,

A lot of women criminals are for retaliatory crimes. Therefore, the judges will

consider their specific victimization and lighten penalty as appropriate. Besides,

118

women, in general, have a better attitude to a guilty plea, which is also considered

in the sentence and detention.

Per Respondent G,

[We] will consider whether to give her lenient sentencing for some extenuating

circumstances… such as the former experience of victimization, the coercion

experienced during the criminal commission, and the fact whether she surrendered

on her own.

In contrast, male offenders are much less likely to be afforded extenuating

conditions. Respondent G observed, “[M]en are usually the principal offenders in most

serious cases, whereas women are often accomplices or coerced accomplices. Moreover,

women’s social status determines that they experience more victimization than men.”

Therefore, there are several “hard” (pregnancy and lactation) and “soft” conditions

according to Chinese Law that will possibly give female offenders some leniency relative

to their male counterparts. Nevertheless, lenient treatment of female offenders by the

individual professionals interviewed for this study appears less likely.

4.5.2.2. Individual Treatment

Most professionals (80.0%) stated that they were blind about gender or other demographic

characteristics of offenders in their work. They would only be in line with the law and do

the job with impassive professionalism. Only three interviewees mentioned the slight

difference in their treatment of men and women offenders. Their reasons were very similar

to the extenuating circumstances regulated by law, but with some personal sympathy

toward those women who were involved in crime involuntarily.

Thus, Respondent I proffered, “I treat women more genial with the consideration of their

emotional sensitivity and the fact that many of them have had some experiences of

victimization more or less before their crimes. Some of them were forced to become

119

involved in crime.” Respondent J similarly responded, “I usually treat men harsher than

women, because it is believed that male criminals are mostly subjectively intentional

malice, while women (criminals) are often the victims of being coerced or incited.”

Beliefs in gender differences are persistent. Compared to men, women are

perceived as emotional, passive, and relational, which have been alleged as the reasons that

deter women from committing crime or serious crime (Miller & Mullins, 2006). Even if

women catch up to men in terms of crime share, the perceived differences between men

and women offenders are still prominent.

Regardless of these chivalrous statements about women offenders, most

interviewed professionals declared they were only doing their job and treated everyone the

same. As stated by the net-widening hypothesis, in addition to the willingness of police and

the courts to arrest and punish women offenders, the bureaucratization and routinization of

the Chinese criminal justice system may also play a role in decreasing personal discretion

in processing.

4.5.2.3. Inter-generational transformation

It is worth noting that the extenuating circumstances mentioned by many professionals

and implying disproportionate consideration for women offenders have been declining

among the post-1980s generations. The most remarkable change has been in the attitudes

of young women involved in crime. Two professionals gave examples from their

observations of prison inmates and suspects, respectively.

Nowadays many young girls who have been convicted have bad attitudes when they

are serving their sentence, which can be said very resistant. Well, she is always

impenitent and refuses to change no matter how you communicate with or educate

her; she even does not regret at all. However, the previous generation inmates,

without being educated much, regretted very much once they entered in (the prison).

(Respondent A)

120

The former women criminal suspects had a better attitude of confession (than those

today). Generally speaking, they would confess their offending during interrogation

actively, which is very different from the new-generation girl offenders. Nowadays,

those young women suspects are very resistant and uneasy to surrender, even with

a variety of strategies (used for the interrogation). I remember that when I started

working, I met a girl who was involved in a robbery case. In fact, she was just an

accomplice and did not participate in the looting. The principal criminals were her

boyfriend and their friends. She was only standing beside and looking around.

However, despite that, the accomplice was also a criminal. From the moment she

was arrested, she was very regretful and scared. She was trembling and crying all

the time until swollen her eyes. Eventually, she got a not heavy sentence taking

consideration of her small role in the criminal process, good attitude of pleading

guilty and meritorious contribution by reporting the facts of other offenders. In

contrast, there are more shrewd girls today. They declare that “they have the right

to keep silent” or “they will only say when their lawyers come” as soon as they

enter (the interrogation room), which they learned from some Hong Kong and

foreign films or TV series. Some young girls even threaten us by asserting their

connections with some powerful people who will protect them. (Respondent N)

The change in the attitude among the younger generation reduces the possibility

that leniency regulated by law will be applied to their cases. Moreover, it may also lower

the level of sympathy that criminal justice professionals feel toward them. Their

uncompromising and antagonistic manner may play a role in reducing the chivalrous

treatment of women offenders among the one-child generations.

Another change is the decrease in the victimization or coercion experienced by

younger-generation female offenders. However, the motives for their offending are

seen by the professionals as more like those of men and less pardonable than was

seen in the past for female offending, on average. Respondent I offered, “[The

victimization was] more common among the pre-80s generations. The post-80s are

still very young after all so many offenses are out of impulsivity, and [therefore the

motivations are] relatively similar between men and women.”

121

One professional summarized both changes in attitude and motivation based on his

observations of women’s offending at different times, implying that those changes would

influence punishment and possible leniency of the treatment by the criminal justice system.

Women offending [in the past] was not as usual as it today. On the one hand, many

[women)]offenders were the victims of previous violent crimes. The victimization

experience would be taken into consideration while sentencing and result in a

lenient treatment if possible. On the other hand, [I] have to say that the women

offenders or suspects in the previous time had better attitudes than the new

generation. They would confess or surrender their criminal facts actively and

truthfully, which would also be considered for the leniency. (Respondent L)

The lenient treatment of female offenders is a responsive outcome of traditional

Chinese gender roles, which regard women as being vulnerable and obedient. This

stereotype is also applicable to women offenders. Perceived by the interviewed

professionals, female suspects “should” be nonresistant and female criminals “should” be

repentant as the “right” way to behave and, subsequently, to earn the sympathy and leniency

that may be awarded. If they fail the socially prescribed roles, such as being “bad in

attitude” or unrepentant, they are punishable pursuant to the male model. Moreover, women

offenders who break the stereotype are subject to a backlash and tend to be punished more

harshly because of their being “doubly deviant”(Lloyd, 1995). They are blamed for

jeopardizing the social safety and, simultaneously, for challenging the traditional moral

code of femininity (Shen, 2015). Consequently, the increasingly punitive response to

female offenders among the new generation is in line with “equality with a vengeance”

(Chesney-Lind & Pollock, 2006). Compared to the pre-1980s generation, therefore, we can

expect an increase in the harsh treatment of female offenders among the one-child

generations, who are more likely to break with traditional “womanhood” (Chesney-Lind,

122

1991). The responses from the criminal justice system thus contribute, at least in part, to

the narrowing of the gender gap in offending.

4.5.3. Theme 3: Socialization and Criminality

Some professionals expressed opinions on why the gender gap in offending was narrowing

among the one-child generations. The reasons can be divided into four categories: (a) the

socialization of the younger generation of women; (b) the strain caused by the gap between

economic aspiration and the means to fulfill the aspiration that was specifically

experienced by Chinese young women; (c) a permissive change that has occurred in gender

norms; and (d) the loosening social control structure in modern times.

A professional connected the unique process of socialization with the strain

experienced by young Chinese women together as a reason why there had been such a

sharp increase in women’s offending in contemporary China. Her observation is based on

her experience with women prison inmates.

[It is] very common [to see the single child in women criminals]. Most of them

were taken care of by their grandparents and lived in their hometowns when their

parents migrated to urban areas for work….Their grandparents were unable to

restrain their behaviors when they grew up. Moreover, most of them made a poor

performance at school, so they had to continue their education in some vocational

schools in urban or suburban areas. These were not regular high schools, so their

education background was not good, which hindered their career development,

especially many of them dreamed of high-status and highly paid jobs. Moreover,

those vocational schools had an atmosphere in which students did not study but

cliquing, brawling and fighting every day. Influenced by such an environment,

those girls could commit any crime you can imagine. (Respondent A)

The city-centered economic development style in China spawns a huge flow of

migrants to provide cheap labor in urban areas; however, constrained by the hukou

(household registration system), their statuses are impossible to change to align with their

migration. As a result, those rural-urban migrants cannot be registered into the welfare

123

system in their residential areas where they make striking contributions. Among the variety

of social benefits affiliated with the welfare system, children’s education is one of them,

which limits the possibility of taking their children with them to the place where they work.

For the higher incomes in those industrial cities, they have to leave their children at home

in villages with their grandparents or other relatives; hence, these children are called the

“left-behind” children (Chang, Dong, & MacPhail, 2011; Wen & Lin, 2012; Ye, Murray, &

Wang, 2010). The left-behind experience influences girls more profoundly. On the one hand,

girls are traditionally more controlled and supervised than boys, which is believed to be the

reason for their higher rates of conformity (Chesney-Lind & Sheldon, 1992) . Therefore,

the loosened family control should have a greater effect on their behavior than on their male

counterparts (Lagrange & Silverman, 1999; Savolainen et al., 2017; Schwartz, 2013a). In

other words, the inter-generational change that happened in men’s socialization was not as

significant as that in women’s socialization. On the other hand, the patriarchal nature of

Chinese society structurally limits women’s legitimate opportunities to pursue success on

par with men.

The low fertility rates as a direct result of the one-child policy, together with

propaganda promoting gender equality that accompanied the implementation of the policy,

changed traditional socialization practices, especially for girls. Parents placed their hopes

on their only child regardless of the child’s gender and, accordingly, treated boys and girls

more similarly than ever before. Compounded with the influence of economic

marketization in China, monetary abundance is emphasized and uni-defined as the symbol

of success by the society (Dutton, 2005; Shen, 2015). Girls were also infused with the

dream of being as successful and as independent as boys. This change in the gender

124

socialization practices has not only influenced the single children, but also their peers who

had siblings. Therefore, the gender roles and norms of the whole one-child generations

have been changed. Such a unique socialization process, which is remarkably different

from traditional practices, has created new generations of Chinese women who are more

likely to be ambitious, competitive, and goal-oriented than were their older counterparts

(Edlund et al., 2013). However, they are also more likely to experience strain than their

male counterparts upon realizing the inequality of means to fulfill their goals when they

are out in the society (Merton, 1938). A nationwide survey shows that sex discrimination

is common in Chinese job markets with 86.6 percent of female college-graduate

respondents having experienced it and 95 percent of all female respondents acknowledging

the pressure of “double shift” (Xinhua News, 2016; H. Zhang, 2014b). Female migrants,

moreover, are even more vulnerable to the involvement of crime because of extra

exploitation of their labor (Harvey, 2005) and marginalization of their social position in an

economy-driven society (Bakken, 2005), which results in greater strain.

The change, either in the way of social control or in the way of strain, would

positively contribute to criminal offending among the one-child generations, especially for

women (Broidy & Agnew, 1997; Hirschi, 1969; Steven F. Messner & Rosenfeld, 1994;

Sampson & Laub, 1995). Therefore, we can expect women’s criminal offending to catch

up to men’s and, therefore, create the convergence in the gender gap of offending.

Consistent with this mechanism, one professional illuminated another situation for young

women offenders by focusing on the strain resulting from their goals or dreams of material

success and their realistic chances of fulfilling the goal on their own.

Along with the “Beat the Tigers” [anti-corruption] campaign going, many women

have also been charged with corruption and bribery as men do. Some of them were

125

government officials themselves, but some were wives or mistresses of male

governors and accomplices of those occupational crimes. Generally speaking, [for

the second situation,] most of them were mistresses. They were not government

employees but used their sexuality to hook the ones who were holding power and

sold the “guanxi” [personal connection] to persons who needed it. [They], in this

situation, are convicted of the accessories. Most of them were young women

pursuing material enjoyment and vanity satisfaction. However, they might have a

bleak prospect on the job market and didn’t want to toil. (Respondent B)

The strain between the life goal of economic success and the ways to realize the

goal is demonstrated in Respondent B’s statement and is believed to drive some women to

take advantage of their sexuality as a shortcut in fulfilling the goal. The shortcut is not

illegal per se; however, it is criminalized with an emphasis on the sexualization of women

in general. Therefore, the professional was of the opinion that women have a greater

tendency to derive status from male partners without expending too much effort (see also

Hall, 2013), which echoes the traditional sexist perception toward women by many of the

Chinese professionals interviewed. Combined with the expected “womanhood” from

female offenders by the professionals, the culturally ingrained sexism is vivid. Although

these perceptions came from legal professionals, they reflect the unequal societal-level

expectations on gender roles in China and specific constraints imposed on Chinese women.

However, challenges against the traditional gender roles have emerged in the new

generations under the influence of economic and social changes. A professional analyzed

the increase in women’s crime from an emancipatory perspective:

From a social perspective, women have more and more job options and

opportunities. They are as active as men in every field, including the criminal world.

From the perspective of gender perception, we no longer think that women are weak,

but that women are equal to men, including women themselves. [Asking the female

interviewer], do you think that your ability is the same as that of men? [Do you think

that] you can do the same as men? …[After getting the positive response from the

interviewer, he continued], Right! Another aspect [of the new perception] is there

are more tolerance and freedom rendered to women by the present society. There

are no longer the shackles of the former “three obedience and four virtues.”

126

Women’s thoughts and behaviors are more open and permissive than before.

(Respondent J)

This statement connects the behaviors of women with the perception of gender.

Consistent with both hypotheses of emancipation theory, the behavioral change and

netwidening effect are simultaneously influencing the gender gap in offending. More

specifically, the emancipation in women’s activities has resulted in a perception of gender

equality. Another professional confirmed such a change in the gender norms and

perceptions from a more extreme example, prostitution.

[T]he perception of prostitution has also changed. The girls who were arrested for

prostitution in the past all felt embarrassed and asked us to keep them confidential.

The girls nowadays seem to be quite indifferent. They plan to make some money

by prostitution in the urban areas for several years and then return to their

hometowns where there will be no one knowing their story and what they have

done. With this idea, there are few moral constraints. The effect of the sole legal

sanctions is not enough [to restrict prostitution]. (Respondent N)

One side effect caused by the one-child policy is the prevalence of offspring sex

selection against girls, because of the conflict between the policy and the son-preference

culture, and eventually a dramatic sex composition imbalance of the young generations. As

estimated, there were about 32 million “surplus” men among the one-child generations by

2005 (Zhu et al., 2009) and a deficit of 1 million marriageable women in the firstmarriage

market per year after 2010 (Tuljapurkar et al., 1995). Men occupied 97 percent of the all

unmarried population aged 28 to 49 in China (Hudson & Den Boer, 2002). Derived from

the sex ratio imbalance in the marriageable cohort, the demand for prostitution soars. At

the same time, the more sexually permissive gender norms and higher population mobility

than ever before promoted the supply-end to cater to the huge demand. As a result,

prostitution has become the fastest growing type of female crime (Chan, 2006). As this

statement illustrates, in addition to more permissive attitudes toward prostitution, the rural-

127

to-urban migrating status was another reason for an increase in the offense. The loosened

social network accompanying the migrant flows anonymized this group of people.

Therefore, the high population mobility has brought about another side effect—lessened

social control (Wang & Shi, 2018). This structural change reduces the moral scruples of

the migrants and also increases the difficulty of preventing recidivism.

Female suspects were not as common as nowadays when I just worked as a

policeman 36 years ago. There are many migrants [in our city] currently. Many

suspects are migrants, and it is difficult to manage them. Female recidivists and

habitual criminals are also more common than before, especially for the prostitution.

[It is frequent to see] some women caught in another place just days after their

release. The primary reason is the loosened social management system. In the past,

the migration flow was not so significant [as today], and most criminals were locals.

Therefore, just a call to their employers or communities, they would be tightly

supervised and less likely to recidivate. Particularly for women, in addition to the

physical supervision, they were also subjected to the pressure from public opinion.

Now those constraints are gone. (Respondent O)

Sen’s (1990) estimate that there were 50 million “missing women” in China alone by the

end of the 1980s.

Because of the difficulty of constraining recidivism rooted in the structural and

cultural changes, the criminal justice system would respond with harsher punishment for

the new-generation offenders, especially women, who are the primary group affected by

the inter-generational changes. Therefore, socialization not only has played an important

role in the convergence of the gender gap in offending, but it also influenced perceptions

and treatment of women offenders by criminal justice professionals.

4.6. Discussion

This study empirically confirms (a) the converging trend of the gender gap in offending in

China, (b) the convergence as an outcome of the rising women’s offending, and supports

128

connection between the two causal effects (see Figure 4.3).

The inter-generational change, either in crime share or in crime pattern, confirms a

trend of close resemblance between genders. Believed by most professionals, this

convergence reflects women’s greater participation in offending, especially in non-violent

crime. Women’s shares of property, drug, and job-related offenses outpace those of men

and, therefore, have contributed to a narrowing of the gender gap. Nevertheless, gender

roles in the criminal world are still distinct. Compared to men, women offenders play an

inferior and marginal role in criminal enterprises, which places them in a vulnerable

position for being caught. Also, women’s offending is more likely to be petty and minor in

consequences compared with that of men, which leaves a large latitude of discretion in

criminal justice processing.

Because of the gendered characteristics of criminal offending, the justice system

and professionals also play a part in narrowing the gender gap in offending. The law has

created some “hard” extenuating circumstances exclusively applicable to women, such as

pregnancy and lactation, and some “soft conditions” disproportionately enjoyed by women,

such as good attitude and victimization experience. Both categories are eligible for lenient

consideration in Chinese legal settings. Therefore, women offenders are systematically

chivalrously treated in China. However, this gender-specific leniency has been fading

away, along with the vanishing of the “soft conditions,” especially for the one-child-

generation women offenders. Moral panics associated with the emergence of diverse forms

of female offending lead to an inordinate degree of adverse attention focused upon the one-

childgeneration women by criminal justice professionals. Defiance against the traditional

129

ideology of women places the young female offenders in the form of “social double

jeopardy” and causes them to suffer repugnant treatment by professionals. As a result,

professionals’ belief in the growing resemblance of women’s and men’s criminal motives

and the perceived attitude of resistance by women offenders both contribute to a

convergence in the treatment of men and women offenders by both the system and

individual professionals.

Another contributor to the convergence of the gendered offending gap is the real

change in the criminal behavior of younger generation women. As one unintended outcome

produced by the one-child policy, there have been more women brought into the criminal

justice system under the influence of the unique socialization practices of the special

generation. The one-child policy has provoked a gender-neutral approach to socialization

in the family but has failed to promote gender equality simultaneously in other social

institutions, such as the work world. The inconsistency in the equalization pace has created

strain between aspirations and the means to fulfill those aspirations, specifically for women.

Moreover, the rural-to-urban migration generated by the interplay of regional inequalities

and marketization in China has loosened social control imposed on the migrants and their

children as well. Consequently, gender roles and norms have become more permissive.

These shifts in socialization are gender-neutral but have gender-specific effects on

behavioral change, especially on the criminal behavior of women.

Finally, changes in the socialization process have not only influenced behavior but

have also broken down gender stereotypes that might portray women as pitiful, dependent,

and in need of care. To some extent, therefore, these changes also influence the perceptions

and treatment of female offenders by the criminal justice system. Women offenders, who

130

are against the traditional gender norms, are more likely to be punished more harshly than

their submissive counterparts because they have threatened not only social safety but also

the cultural ideology of gender. From this perspective, the change in socialization brought

about by the one-child policy, directly and indirectly, narrows the gender gap in offending.

4.7. Conclusion

This study was limited by the small sample size as well as by the use of a convenience

sampling method. If we could have more diverse participants, such as justice professionals

from rural areas, the discovery would be more profound. For example, in addition to the

inflowing migrants’ influence on women’s crime in cities, the effect of the outflowing

migrants from the countryside could also be explored, which would increase the

comprehension of causes of the gap convergence from a comparative perspective.

Moreover, the interview method also limited the study, especially for a topic that was

politically sensitive and related to professional integrity. Other qualitative methods, such

as ethnographic methods that involve direct observations, might have yielded deeper

findings.

Despite these limitations, this study makes an important contribution to knowledge

about the gender convergence in crime in a rarely researched, but socially and economically

important country. Moreover, the results also promote the understanding of the causal

effects of the convergence, which have been widely debated for some time. A narrowing

gender gap is confirmed both by the crime rate and by the crime pattern. Nevertheless, the

gender order is still significant, both in the criminal world and the society.

Gender equality has been enhanced to some extent in China by the one-child policy, but

China is still far from achieving gender parity. The increase in women’s offending could

131

be seen less as a manifestation of gender emancipation and more as a responsive vengeance

from the stubborn hierarchical structure against the progressive cultural shift in gender

ideology.

4.8. Tables and Figures

Table 4.1. Interviewee Characteristics

Gender & Generation

Position POCG Men WomenPOCG OCG Men WomenOCG Total

Professor 0 1 0 0 1

Police Officer 2 0 2 0 4

Prosecutor 1 0 0 2 3

Court Clerk 0 0 0 2 2

Judge 2 1 1 0 4

Sheriff 1 0 0 0 1

Total 6 2 3 4 15

Table 4.2. Statements about the convergence of the gender gap in the crime rate

Gender & Generation

Category

POCG POCG OCG OCG Men Women

Men Women

Total

Increasing women’s 4 2 3 3 share of crime

Depending on crime 2 0 0 0 type

No change 0 0 0 1

12 (80.0%)

2 (13.3%)

1 (6.7%)

Table 4.3. Statements about the convergence of gender gap in crime pattern

Gender & Generation

Category

POCG POCG OCG OCG Men Women

Men Women

Total

Women converging 4 1 1 2

to men

8 (53.3%)

Similar on the new 0 0 1 1

crime

2 (13.3%)

132

No change 2 1 1 0

4 (26.7%)

Not sure 0 0 0 1

1 (6.7%)

Table 4.4. Statements about the treatment of men and women offenders

Unbiased Lenient toward Women

System 8 (53.3%) 7 (46.7%)

Individual 12 (80.0%) 3 (20.0%)

Figure 4.1. The Percent of Female Offenders in China (1972-2014). Note: The percent of

female offenders equals the percentage of female prison inmates from 1972 to 1992 and

from 2003 to 2014, and equals the percentage of female convicts from 1994 to 2001. Data

Source: the Department of Correction in China (1981-1985), the Supreme Court in China

(1986-1992), the Chinese Yearbooks of Court of Justice (1994-2001), the China Statistics

Yearbooks (2003-2012), and a BBC news report (Hatton, 2015).

1970

1980

1990

2000

2010

2020

year

133

Figure 4.2. The structural map of themes and subthemes.

Figure 4.3. The Causes of the gendered offending convergence and their relationships.

134

Chapter 5. Conclusion

Profoundly influenced by Confucian values, Chinese people endorse the relational

hierarchy (Y. B. Zhang, Lin, Nonaka, & Beom, 2005). Gender is the very realm manifesting

the strict hierarchical relation between men and women, which has ingrained the patriarchal

gender orders in every Chinese social institution and also shaped traditional gender

ideologies in China. The Confucian perspective emphasizes the differences between

genders and justifies them by its metaphysical and normative account (Wang, 2011).

Although Confucianism had been criticized during the Cultural Revolution and diluted in

the modern era, its influences are still alive in every cell of Chinese culture and social

structure.

This dissertation explored the etiological mechanisms of the increasing proportion

of female crime in China and eventually anchors on structural gender inequality as its

essential cause. The results show that the women’s crime is converging to men’s in China,

which has narrowed the gender gap in crime in China after the first one-child generation

reached the age of legal responsibility. The one-child policy has changed the socialization

practices on the new generations with gender-equalizing propaganda and low fertility rate.

As a result, the gender role repertoire has changed both at the individual and societal levels.

From the individual gender role perspective, discrimination has diminished in the

family because of the low fertility rate imposed by the policy. Consequently, the individual

gender expectation, especially that of one-child generation women, has become more

egalitarian. Young Chinese women are more likely to aspire for economic success on par

with men than their older counterparts. In contrast, the one-child policy keeps the gender

inequality and gender order intact in other social institutions. Chinese society is still a

135

patriarchal society where the prescribed social ideology constrains women; moreover,

driven by the imbalanced sex ratios of the population, the degree of patriarchy has been

strengthened with women are more valued as marital partners and after that as mothers and

homemakers but less as employees or productive workers. Accompanied by this change,

gender inequality and women’s discrimination are exacerbated in the work world. This

imbalance in the progressiveness of gender equality creates strain for the younger

generation of women from the mismatch between their life goals and the means to fulfill

such goals in reality.

Another factor influencing crime in the new generations is social control

experienced by both genders driven by the economic reform in China. The massive ruralto-

urban migration generated by the interplay of regional inequality and marketization in

China has loosened social control imposed on the migrants and their children as well.

Compared to men, this change has a more prominent influence on women, who used to be

closely monitored by and bonded with parents reducing their exposure to definitions

favorable to deviance. Loosened social networks anonymize migrants and result in

decreasing moral constraints. The loosened control combined with the extra strain

produced by the labor exploitation and social marinization of migrants leads to a sharp

increase in crime among migrants, particularly female migrants. Moreover, increased

exposure to male peers, because of the extremely imbalanced sex ratios among the onechild

generations, drives women to absorb the pro-deviance definitions and therefore increases

their propensity for crime than before. In contrast, the sex ratio imbalance limits younger

men’s chance to group with women, which also increases the one-child-generation men’s

deviance. This change brought about by the one-child policy has influenced not only the

136

only children but also their peers who have siblings; therefore, the policy reshaped the

individual gender role repertoire in the whole one-child generations. As a result, both

genders’ crime is growing, but women’s is growing faster, which narrows the gender gap

in crime.

From the perspective of societal gender role ideologies, the gender difference in

crime commission is consistent across generations, which are hard to be changed by the

policy. Gender hierarchy is stubborn in most social institutions regardless of the changing

socialization practice in the family. The permissive transformation in gender norms,

moreover, brings about “vengeful” criminalization of female sexuality by criminal justice

professionals who largely represent the societal perception of gender standard. This gender

order keeps the stereotype of genders, in which women are expected to be passive,

submissive, and vulnerable. If they fail such an expectation, social punishment will be

imposed, or chivalrous treatment will be revoked. When the ideology failure happens in

female offenders, harsher treatment is conceivable due to their “double deviance” from the

law and the prescribed social ideologies of women. Because of the ingrained gender order,

gender differences do not diminish but are still vivid in every field in China. Even in the

perceived “masculine” criminal world, women still play a marginal (e.g., petty non-violent

offenses) and inferior (e.g., accomplices in serious offenses) role dominated by men, and

have a gendered pathway to deviance, which mirrors the gender stratification in the whole

society in China. Therefore, the progressive imbalance in gender equality reflects that the

gender roles are multidimensional.

Therefore, the increase in the female share of crime and thereby the convergence in

the gender gap is an outcome reflecting the behavioral change and the responses of the

137

criminal justice system to that change. The obstinate gender inequality disappoints the

“emancipated” individual gender prospects and as a result, leads to the increase in female

crime.

5.1. Implications for Strain Theory

The findings of this dissertation strongly support Merton’s strain theory (1938) and imply

the applications of his typology of deviance in explaining the influence of sex composition

on female crime. The case study is based on a sample with artificially imbalanced sex ratios

driven by policy. The exogenous force has emancipated the gender ideologies of

individuals and infused the equalized life goals into the one-child-generation women.

However, the force keeps the gender hierarchy untouched in most social agents, which

determines that the institutional means of fulfilling those goals are unequal or even

inaccessible for women when compared with men. This goal-means mismatch fits one type

of deviance categorized by Merton including the acceptance of the cultural goals but the

rejection of the legitimate means of attaining those goals. As a result, illegal channels turns

out to be attractive as an innovative way to fulfill the goal. Therefore, the one-child policy,

which has changed the socialization practices and, thereby, the cultural goals of women,

has initiated the mismatch and eventually triggered the increase in female offending.

As a comparison, countries with naturally imbalanced high sex ratios are typically

low in female crime rates (Scott J. South & Messner, 1987), because women, in those

countries, are repressed by the traditional gender ideologies (Guttentag & Secord, 1983).

Specifically, women are valued more as marriageable partners and, thereafter, as

homemakers and mothers but less as workers and employees in the high-sex-ratio societies.

Without any external force, the process of socialization will reproduce culture and social

138

structure through the generations. Therefore, natural sex ratio imbalance will not have the

same force as does its artificial counterpart. It will not create the mismatch between the

goals and the means. As a result, women in those countries are more likely to have the

cultural goals of being home-stay mothers and around family, which are compatible with

the means provided by the social institutions. This type of deviance (or non-deviance) fits

the conformity category that involves the acceptance of both the cultural goals and means

of attaining those goals.

As a derivation, a way to curb the increasing female crime in the society with

artificially imbalanced sex ratios, according to Merton’s typology, should be equalizing the

institutional means for women to fulfill their already-emancipated life goals. Once the

goalmeans mismatch disappears, the strain, and thereby, the crime rates will reduce

correspondingly. Thus, the real gender parity in both cultural ideologies and social

structures can adjust the deviant innovation back to conformity.

5.2. Implications for Feminist Criminology

This dissertation has important implications for explaining the increase in female crime at

a global level. The trend of convergence of the gender gap in crime has been widely

acknowledged and observed worldwide (e.g., Estrada, Backman, & Nilsson, 2016; Heimer,

Lauritsen, & Lynch, 2009; Savolainen et al., 2017). However, the cause of the trend has

been controversial for decades due to the theoretical debate around whether it is because

of the enhanced gender equality or worsened economic position of women, which are seen

as opposite and rivalry.

Using a policy experiment, this dissertation illustrates that the two theories are not

contradictory; in contrast, they explain the causal effects of the transformation in gender

139

role repertoire on the increasing female crime at different levels. Emancipation theory

connects the liberated gender roles with the increase in women’s crime, which has been

widely criticized from the mismatch between the explained subjects – the women who have

been liberated are not the ones who are brought into the justice system (Giordano &

Cernkovich, 1979; Renzetti, 2013). The latter, in contrast, are always found in the

underclass, who are far from being liberated but trapped deep with financial burdens by the

economic marginalization (Holtfreter, Reisig, & Morash, 2004; Reckdenwald & Parker,

2008; Steffensmeier & Allan, 1995).

However, the liberated women and the underclass women do not live in a vacuum.

In fact, they share the same world and dramatically influence each other. When the middle-

and upper-class women enjoy liberation and go out of home to compete with men in school

and workplace, the other groups of women are also influenced. They may not be directly

liberalized, but indirectly influenced by the changed societal gender roles. For example,

the liberated life of the middle-class women is advertised as the standard and normal style

in the modern time by mass media. Influenced by such ideologies, other groups of women

are also desiring for such a liberated life, even if they have not been emancipated in real.

Bearing with such a dream but having to face the crude reality where the means to fulfill

their dream is bleak, they, as a result, are subjected to the strain produced by the structural

inequality not only inter-genders but also intra-gender. Mismatched by the social structure

from their ideologies of gender roles, those groups of women are more likely to become

involved in delinquency to fulfill their dream alternatively or to avoid the frustrating

feelings. Therefore, emancipation and economic marginalization theories tell us the same

story within a dynamic system of gender roles. Some groups of women have achieved

140

numeric gains in society as the result of emancipation and therefore influenced the societal

attitudes toward gender, which in turn incubates the equalized individual ideologies in a

broad scope but does not change the unequal structure fundamentally. This equalization lag

happened in the gender role repertoire leads to the strain and therefore an increase in

delinquency for the ones experiencing the mismatch most. Therefore, there is still a long

way to go to reach the real equality of gender for the whole society, and when that day

comes, we can expect a drop in crime for both genders.

5.3. Limitations and Future Research

This dissertation is a starting point for research on China’s one-child policy and gendered

crime and therefore has a number of limitations. This section reviews some limitations and

points to future research for an enhancement. First, inaccessibility to disaggregate crime

rates over crime types and lack of victimization data have impeded the macro-level

quantitative research to reveal which categories of crime contribute most in the increasing

women’s crime. An enhancement in data quality in future studies, therefore, is expected to

illustrate further the underlying factor beneath the relationship between the sex ratio and

gendered patterns of offending that has been ignored in most research based on Western

samples.

Convenience sample for survey data might bias the findings because the

respondents were the ones who could and would participate in such an Internet survey. This

sampling method filtered out many kinds of respondents, such as rural residents who were

restrained in internet access, and undereducated or aged people who were unfamiliar with

the online content. Under such an influence, therefore, the results might only show a partial

141

image of Chinese people’s perceptions on gender, success, and deviance, especially when

we take the enormous rural-urban disparity in China into consideration.

Moreover, political sensitivity largely restricts both research methods and sample

size. Other qualitative methods, such as ethnographic methods that involve direct

observations might have yielded more in-depth findings. Snowball sampling method,

although convenient, might also bias the scope of the sample and therefore, the research

findings. The sampled participants, because of this method, were highly homogeneous in

terms of their work and resident contexts. All of them were from urban areas with massive

inflows of migrants from rural areas. if I could have more diverse participants, such as

justice professionals from rural areas, the discovery would be more profound. For example,

in addition to the inflow migrants’ influence on women’s crime in cities, the effect of the

outflow migrants in the countryside could also be explored, which would increase the

comprehension of causes to the gap convergence from a comparative perspective. Also, the

results would be more revealing if I could include criminal convicts in the research and

compare their socialization, gender role definition and reasons for offending with the

noncriminal sample.

Students also viewed