|USFDC Home | USF Electronic Theses and Dissertations||| RSS|
This item is only available as the following downloads:
xml version 1.0 encoding UTF-8 standalone no
record xmlns http:www.loc.govMARC21slim xmlns:xsi http:www.w3.org2001XMLSchema-instance xsi:schemaLocation http:www.loc.govstandardsmarcxmlschemaMARC21slim.xsd
leader nam 2200385Ka 4500
controlfield tag 001 002069503
007 cr bnu|||uuuuu
008 100422s2009 flu s 000 0 eng d
datafield ind1 8 ind2 024
subfield code a E14-SFE0003286
Mitchell, Jessica Nicole.
Power-control theory :
b an examination of private and public patriarchy
h [electronic resource] /
by Jessica Nicole Mitchell.
[Tampa, Fla] :
University of South Florida,
Title from PDF of title page.
Document formatted into pages; contains 57 pages.
Thesis (M.A.)--University of South Florida, 2009.
Includes bibliographical references.
Text (Electronic thesis) in PDF format.
ABSTRACT: The gender difference in crime is indisputable. In an attempt to explain gender differences in adolescents' involvement in crime, secondary data analysis of middle and high school students and their neighborhoods will be examined. Feminists have identified the concept of patriarchy as the root of gender differences in all behavior and particularly in criminal behavior. Hagan's Power-Control Theory incorporates the concept of patriarchy through measures within home to examine how differences in occupational authority between parents affects the gender difference in delinquency through differential controls placed on sons and daughters. However, it has been suggested that the measure of patriarchy be extended into the public sphere (Walby, 1990). Specifically, this study compares a traditional private patriarchy model to a public patriarchy model in order to determine which approach better explains the gender gap in crime. Patterns of findings were not substantively different between private and public patriarchy models; however, a number of theoretical implications point to the fact that alternate measures of patriarchy could lend support for power-control theory that it currently lacks.
Mode of access: World Wide Web.
System requirements: World Wide Web browser and PDF reader.
Advisor: Christine S. Sellers, Ph.D.
t USF Electronic Theses and Dissertations.
Power Control Theory: An Examination of Private and Public Patriarchy by Jessica Nicole Mitchell A thesis submitted in partial fulfillment of the requirements for the degree of Master of Arts Department of Criminology College of Behavioral and Community Sciences University of South Florida Major Professor: Christine S. Sellers Ph.D. Christopher J. Sullivan, Ph.D. Jennifer J. Wareham, Ph.D. Date of Approval : November 20, 2009 Keywords: gender, delinquency, neighborhood characteristics, feminism, family structure Copyright 2009 Jessica Nicole Mitchell
Dedication I dedicate the following to my family and friends who provided me with unwavering support and encouragement throughout the process.
Acknowledgements To begin, I would like to acknowledge and express my sincere appreciation to my major professor, Dr. Christine Sellers, who provided me with the data from the Pinellas study of middle and high school students. Additionally, I would like to thank Dr. Sellers for her continuous support, guidance, and patience. Likewise, Dr. Christopher Sullivan served a vit al role, offering me his expert opinion and assistance. Furthermore, Dr. Jennifer Wareham made it possible to utilize and understand the census data and geocode my sample. Without the help of these individuals, through countless emails, phone calls, and meetings, this thesis would not have been achievable. Finally, I would like to recognize and give special thanks to my family and friends for always being there and listening even though they may not have understood what I was going through.
i Ta ble of Contents List of Tables ...................................................................................................................... ii List of Figures .................................................................................................................... iii Abstract .............................................................................................................................. iv Chapter 1 Introduction .........................................................................................................1 Chapter 2 Patriarchy, Power Control Theory, and the Gender Difference in Crime ..........5 Gender Differences in Crime ...................................................................................5 Patriarchy: The root of gender differences ..............................................................7 Forms of patriarchy in society: private and public ....................................10 Hagans Power Control Theory .............................................................................12 Research on Power Control Theory .......................................................................15 The Present Study ..................................................................................................23 Chapter 3 Method ..............................................................................................................26 Sample and Procedure ............................................................................................26 Measurement of Variables .....................................................................................29 Analytic Plan ..........................................................................................................32 Chapter 4 Resul ts ...............................................................................................................33 Private Patriarchy Model .......................................................................................33 Public Patriarchy Model ........................................................................................38 Chapte r 5 Discussion and Conclusion ...............................................................................43 References ..........................................................................................................................50 Appendices .........................................................................................................................56 Appendix A: Factor Loadings for Latent Variables ..............................................57
ii List of Tables Table 1: Description of the Sample ...................................................................................28 Table 2: Co rrelations of Variables Included in Final Private Patriarchy Model Less Patriarchal Sample (N=330) ................................................................................34 Table 3: Correlations of Variables Included in Final Private Patriarchy Model More Patriarchal Sample (N=155) ................................................................................35 Table 4: Binary Logistic Re gression Predicting Probabilities for Power Control Theory Latent Constructs/Gender and Delinquency in Private Patriarchy Model ...........37 Table 5: Correlations of Variables Included in Final Public Patriarchy Model Less Patriarchal Sample (N =309) ................................................................................39 Table 6: Correlations of Variables Included in Final Public Patriarchy Model More Patriarchal Sample (N=176) ................................................................................39 Table 7 : Binary Logistic Regression Predicting Probabilities for Power Control Theory Latent Constructs/Gender and Delinque ncy in Public Patriarchy Model ............41
iii List of Figures Figure 1: Private Patriarchy Conceptual Model .................................................................23 Figure 2: Public Patriarchy Conceptual Model ..................................................................25
iv Power Control Theory : An Examination of Private and Public Patriarchy Jessica Nicole Mitchell ABSTRACT The gender difference in crime is indisputable. In an attempt to explain gender differences in adolescents involvement in crime, secondary data analysis of middle and high school students and their neighborhoods will be examined. Femini sts have identified the concept of patriarchy as the root of gender differences in all behavior and particularly in criminal behavior. Hagan s PowerControl The ory incorporates the concept of patriarchy through measures within home to examine how differences in occupational authority between parents affect s the gender difference in delinquency through differential controls placed on sons and daughters However, it has been suggested that the measure of patriarchy be extended into the public sphere (Walby, 1990) Specifically, this study compares a traditional private patriarchy model to a public patriarchy model in order to determine which approach better explains the gender gap in crime. Patterns of findings were not substantively different between priva te and public patriarchy models; however, a number of theoretical implications point to the fact that alternate measures of patriarchy could lend support for power control theory that it currently lacks.
1 Chapter 1 Introduction An indisputable f inding in criminological research is that men commit substantially more crimes than women (Heimer, 2000). This difference in crime rates for men and women is often referred to as the gender gap in crime. Despite the persistent evidence of a gender gap in crime, mainstream criminologists have often failed to acknowledge gender in the development of their theories and empirical studies (Parker & Reckdenwald, 2008). Feminist criminologists have recognized that inadequate attention has been given to this iss ue. Feminists argue that gender must be looked at as a product of power relations between males and females and not simply the enactment of roles and formation of masculine or feminine attitudes (Daly & Chesney Lind, 1988, p. 511; Heimer, 2000, p. 444). This feminist perspective, along with feminist thought in general, focuses on patriarchy as the root of gender differences in all behavior and particularly in criminal behavior. Patriarchy is a social construct in which men and masculine qualities are v alued more highly than women and feminine qualities (Belknap, 2007). A patriarchal society thus creates gender inequities, oppressing women. As an abstract concept, a variety of definitions exist for patriarchy. Sylvia Walby defines patriarchy as a syst em of social
2 structures and practices in which men dominate, oppress and exploit women (1990, p. 20). In examining and assessing various theoretical approaches to patriarchy, she offers a perspective on theorizing patriarchy in a manner that articulates two distinct forms of patriarchy, which she labels private patriarchy and public patriarchy, that are evident in a number of societal structures. Private patriarchy is the oppression of women by limiting them to unpaid household labor and keeping them from the public sphere. Alternatively, public patriarchy gives women access to both private and public spheres, but this access is seen as inferior compared with that of their male counterparts, and women are still collectively subordinated by societal constructions. Patriarchy is a central idea and necessary in understanding gender inequities, especially in the study of gender and crime. One attempt to explain the gender gap in crime that utilizes the concept of patriarchy was offered by criminol ogist John Hagan and his colleagues, who developed power control theory (1985). This theory explains the gender difference in crime through power and control relations between mothers and fathers. Family relations are differentiated as more or less patri archal by the amount of authority the mother and father have in their workplace relative to one another. In more patriarchal families the fathers workplace authority exceeds that of the mother, especially when the mother does not work outside the home. Within these families, sons and daughters are socialized to take on gender roles similar to those of their parents; this is accomplished through stronger controls on daughters and weaker controls on sons. In more patriarchal families, the gender gap in de linquency is expected, therefore, to be fairly substantial. In contrast, less patriarchal families are characterized by a balance of
3 power and occupational authority between mothers and fathers. In these types of families, both sons and daughters are exposed to fewer controls and encouraged to take risks. It is this similar lack of control and similar ideations of risk between sons and daughters that narrows the gender gap in delinquency in less patriarchal families. R esearch on power control theory has demonstrated limited support for the theory. Hagan and his associates have been able to find consistent support for his model ( Hagan, 1990; Hagan, Simpson, & Gillis, 1987; Hagan & Wheaten, 1993; Hagan et al., 2002; Grasmick et al., 1996; McCarthy et al., 1999). However, a number of studies that attempt to replicate Hagans findings (Hill & Atkinson, 1988; Singer & Levine, 1988; Jensen & Thompson, 1990; Morash & Chesney Li nd, 1991; Avakame, 1997; Lieber & Wacker, 1997; Uggen, 2000; Blackwell, 2000; Blackw ell et al., 2002) have failed to find much support for the theory. A common feature in all of these studies, however, is the persistent conceptualization of patriarchy as strictly a family le vel or private phenomenon. Yet Walby (1990) theorizes patriar chy as a phenomenon reaching beyond the family into the public sphere. Might the support for power control theory be strengthened or clarified if a measure of public patriarchy were included in the power control model? It is conceivable that measures of p ublic patriarchy that illustrate gender inequalities occurring at the macro level may provide a test of power control theory more closely aligned with Hagans conception of the theory. The current study compares a typical private patriarchy power control model with a public patriarchy power control model in order to determine which approach better explains the gender gap in crime.
4 In the next chapter, a closer look is taken at Walbys theory of patriarchy and its applicability to Hagans power control theory of gende r differences in delinquency. A n alternative conceptualization of power control theory is then presented that offers a public patriarchy, rather than strictly a private patriarchy, explanation of the gender gap in delinquency. Chapter three de lineates an empirical test of the original micro level power control model, the private patriarchy model, as well as the alternative multilevel power control model, the public patriarchy model In that chapter, the methods are described for each mode l tested and compared. Chapter four presents the results of these empirical tests, and the implications of those findings are discussed in Chapter five.
5 Chapter 2 Patriarchy, Power Control Theory, and the Gender Difference in Crim e Gender Differences in Crime Gender and its link to crime and delinquency can not be ignored. According to the FBI Uniform Crime Report there were an estimated 2.6 million females arr ested, accounting for about 24.5% of all arrests in 2008. Additiona lly, the Bureau of Justice Statistics indicates that women make up about 12.9% of the jail population (Sabol & Couture, 2008) and 7.2% of the state or federal prison inmate population (Sabol & Couture, 2008). In a lifetime, men have a greater chance of going to prison than women, with 11.3% of all men and 1.8% of all women incarcerated in prisons. Nine out of every 10 inmates in the jail system are male and of the 3.2% of all adults in the United States who are under some type of correctional supervision, men make up 77% compared with 23% for females in 2007 (Sabol & Couture ). Men also exceed women in other types of antisocia l and deviant behavior ( Toro, Urberg, & Heinze, 2004). For example, in a national sample, women found suicide less acceptable than men (Stack et al., 1994; Stack & Wasserman, 1992, 1995); men report greater drinking and drug use, as well as heavy smoking compared to females (Rodham et al., 2005), and men are also more likely to initiate aggressive or violent behaviors such as intimate partner violence (Tjaden &
6 Thoennes, 2000). Specifically, females account for only about 14% of all violent offenders. Despite the persistent gender gap in crime, delinquency, and other deviant acts, efforts to account for this discrepancy have been som ewhat limited. Historically, the gender gap has been ignored by criminological theory. Most attempts to account for delinquency have based theoretical explanations on the c ircumstances of male offending exclusively. More recently, as feminist criminolog ists have called for increased attention to the criminal involvement of females, mainstream criminologists have routinely controlled for gender in empirical tests of theories of delinquency. However, merely including girls in samples and controlling for g ender does little to explain the gender gap in delinquency. Such statistical controls may identify that gender differences exist but fail to explore why those differences exist. Even feminist criminologists, who have alternatively promoted gender specific theories of delinquency involvement, and who have uncovered more completely the circumstances of female offending, neglect to account for why males and females differ so greatly in their offending behavior. Recent developments in feminist perspectives on crime are beginning to address this oversight. Chapple and colleagues assert that Researchers are increasingly suggesting that gender as structure is created, maintained, and differentially experienced within families leading to gender differences in boys and girls delinquency (2005, p. 358). In particular, they argue that social controls have a different level of importance and effect for boys and girls, therefore demanding gender specific theories of social control (Chapple et al., 2005). Feminist t hought has introduced many new approaches to
7 various fields of study, emphasizing the need to include feminist views and to account for females in research. One such development emphasized early on in feminism, and later in feminist criminology, is the con cept of patriarchy. These advances have increased the awareness of including females in criminological research, yet a majority of studies concentrate on female victimization, overlooking the influence of patriarchy on female offending (Yllo, 1983, 1984; Yllo & Straus, 1984; Parker & Reckdenwald, 2008). While t here are many dimensions of the concept of patriarchy and the feminist perspective, the focus remains constant in attempting to explain the gender differences in crime and its link to the structure of society. Patriarchy: The root of gender differences In general, patriarchy is a promising explanation of gender differences and inequalities. Much has been written about this concept, but Sylvia Walby (1990) is one of few that have theorized patriarchy. According to Walby (1990, p. 1), the concept of patriarchy is indispensible for an analysis of gender inequality. Patriarchy may thus serve as a key in unlocking the question: Why do women have a significantly lower rate of criminality th an men? As a building block of the feminist perspective, patriarchy reflects the subordination and disadvantaged position of women in social institutions. In general, patriarchy is the structuring of society by family units where fathers are the head of the household. Within these family units, fathers hold the responsibility of the welfare of the family. This position in the family in turn gives fathers authority over their families
8 (Parker & Reckdenwald, 2008). Patriarchy describes males as having m ore power and control over females, dominating them both culturally and socially. Walby (1990) identifies several approaches to understanding the role played by patriarchy in accounting for gender differences. These theoretical perspectives include radi cal feminism, Marxist feminism, liberal feminism, and dualsystems theory. Radical feminism asserts that the cause of gender inequality arises from men as a group controlling women as a group. This form of patriarchy is seen as an independent social cons truct that oppresses women in any way to benefit mens needs and desires. Some radical feminists have gone as far as arguing that modern reproductive technologies (i.e. birth control and fertility drugs) are, far from liberating women, a way for men to control womens sexuality and reproduction. Unlike radical feminists, Marxist feminism views gender inequality as a product of capitalism. Class relations are emphasized as the primary cause of gender relations. Women are dominated by men as a result of economic exploitation, which forces women to remain in the household as unpaid laborers. While both radical and Marxist feminists analyze womens oppression as a deep rooted social structure, liberal feminism accounts for all the practical inequalities wom en encounter. For example, this approach focuses on womens access to education, paid employment, and general equal treatment between men and women. Finally, dual systems theory combines both Marxist and radical feminist theory. This perspective explains gender inequality as a result of both patriarchy and capitalism, either interacting in society together or each as a separate system that functions independently at different levels. These four main feminist arguments all provide insight into the concept of patriarchy.
9 Walby (1990) further distinguishes areas or settings within which patriarchal relations occur: the household, paid employment, the state, sexuality, male onfemale violence, and cultural institutions. Beginning with the household, this private setting of womens oppression focuses on womens roles in the household and division of labor. Under a patriarchal household with the husband considered the breadwinner, the wife is the homemaker whose full time job is reproduction and production within the home. Marxist theorists view patriarchy in the household as class relations using a materialistic approach. Materialism explains the subordination of women by the higher value placed on mens surplus production in a capitalist society compar ed to the lower value of women reproducing and rearing children. Drawing on reproduction as an oppressive force, the main focus of radical feminism in regard to the household are the biological dangers of reproduction. Although both liberal feminists and dual systems theorists hold to similar arguments, they give more attention to the division of labor by gender. They argue that the core of patriarchal exploitation is the expropriation of domestic work required of women. This domestic labor is not value d or considered work because it is unpaid. The increase of paid employment of women in contemporary society would naturally be considered a move away from a patriarchal society. However, feminists explain that paid employment has also been influenced by patriarchy. These effects can be seen through the lower pay rates, occupational positions, and amount of occupational authority held by women compared to men. National statistics in the U.S. have consistently expressed a gender bias in regard to hourly wa ges and salary, as well as the opportunity for overtime and longer shifts (Walby, 1990, p. 25). Over time, women have
10 gained access to employment outside of the home, yet remain exploited as cheaper labor than men. The lack of power held by women in society is also seen as a form of patriarchy exuded by the state as a whole. While distinct theoretical approaches explain the states political influence on various structures in society differently, the main focus is on the patriarchal effort to maintain gender inequality. Similarly, culture sexuality and male onfemale violence are viewed as constructions in society built to preserve mens power and subordination of women. By looking at these various units of society, Walby is able to explain in detail the different feminist perspectives on how patriarchy is embedded in society. She then critically assesses them relative to one another to develop a contemporary model or theory of patriarchy. An important aspect of Walbys contribution to literature on patriarchy is her distinction between private and public forms of patriarchy, into which she fits the above mentioned structures of oppressive social relations that exist in contemporary society. Forms of patriarchy in society: private and public. Patriarchy, as it contributes to gender inequality, exists in both the private and public spheres (Walby, 1986; Parker & Reckdenwald, 2008). Womens oppression through inequalities within the home and family constitutes private patriarchy In this private sphere, the family is structured in such a way that the male is labeled the head of the household solely based on the fact that the male works more hours outside of the home and has a greater occupational authority outside of the hom e. Through private
11 patriarchy, women are oppressed and controlled individually (Walby, 1990, p. 24). Walby identifies household production as the dominating patriarchal structure in the private sphere, with employment, the state, sexuality, violence, and culture as secondary patriarchal influences. By means of patriarchal socialization and gender role preferences, female responsibilities are restricted to the family, which is not recognized in this ideology as work because it is unpaid labor. Consequently, because women are relegated to the household, they undergo socialization and experiences that restrict their activities, whereas men maintain their power and dominance over economic resources and hold a position of power over women (Parker & Reckdenw ald, 2008, p. 9). In contrast to Walbys concept of private patriarchy, public patriarchy includes institutional structures such as employment, schools, churches, and the government (Walby, 1990; Atwell, 2002). These institutions preserve patriarchy in society by perpetuating the inequality of men and women in their position, power, and controls within them. The dominating forms of patriarchy in the public sphere are employment and the state, with household production and the other patriarchal structure s mentioned by Walby having less of an impact on gender relations. Unlike conditions under private patriarchy, where women were isolated within the home, women under public patriarchy are allowed roles in the public sphere, but remain oppressed by the gender inequalities in paid employment, education, economic conditions, and positions of power held in society. Although women are involved in the public sphere, patriarchy is still evident through measures of female poverty, part time work, income, and educ ational disparities
12 (Heimer, 2000). Accordingly, even when private patriarchy is not observed, public patriarchy may still be controlling and limiting power to females relative to males. While Walby does not apply her theoretical model of patriarchy to t he gender difference in crime, one criminological theory that has used the concept of patriarchy, at least in a limited way, to account for the gender gap in crime is power control theory. Hagans power control theory Hagan and associates (1985) used both macro level and micro level concepts to explain the link between gender and delinquency, integrating traditional theories of class (power) and family function (control). In his initial version of power control theory, Hagan centered the theory on Marxi st ideas of class relations and positions of authority held by the head of the household in families. He argued that class affects the gender difference in delinquency. Specifically, higher classes with greater occupational authority held by the head of the household (i.e. owners, managers) present a greater difference in delinquency between genders. In families of lower classes (i.e. workers, unemployed surplus population) the gender disparity in delinquency narrows. According to Hagan, males are freer to deviate from social norms than females and they are freest to deviate in the higher classes. However, Hagans initial (1985) version of power control theory did not directly address gender relations in the workplace and at home; instead, the gender po wer imbalance was assumed by virtue of the class position of the male head of household. In response to counterarguments on this point (Coser & Coser, 1974; Coser, 1985), Hagan further elaborated on the power control model in 1987. By looking at the re lative difference in authority in the workplace between the wife and the husband,
13 instead of simply the absolute authority of the male head of the household, the gender power relations within the family unit could be observed. In this elaborated model (1987), Hagan and his colleagues introduced the basic elements of power control theory as it is commonly known today. The main idea of the theory is that the amount of authority parents have in the workplace, relative to one another, is reproduced within the two parent household. The difference in power between parents is then believed to differentially influence the social control of daughters and sons, in turn, having an effect on gender differences in the childs preference for risky behavior, as well as assessments of risk. At last, it is these differences in control and access to risk that explain the gender difference in delinquency. In particular, the aim of power control theory is not to predict the probability of females being involved in delinque nt behavior; instead, it seeks to clarify the differences between males and females in delinquency and how these differences may be greater in certain types of households compared with others (Blackwell, 2000). Hagan et al. (1987) distinguish households a s being more patriarchal or less patriarchal on the basis of whether individuals have positions of authority in the workplace relative to their spouse. For example, a household would typically be described as more patriarchal if the father holds an authority position, while the mother does not have authority in her job or is unemployed. Additionally, even if the father has no authority in his job outside the home, but the mother is unemployed, the household is considered more patriarchal. On the ot her hand, a less patriarchal family would exist when both parents hold jobs of equal authority, creating a more even balance of power between the two spouses.
14 Basically, the theory looks at the relational rather than absolute measures of authority of th e father and mother. In more patriarchal families, where the father has greater power and occupational authority outside the home as opposed to the mother, the child socialization in these unbalanced families generally is delegated to mothers (Leiber & Wacker, 1997). Hagan et al. (1985) state that mothers more than fathers are likely to be the instruments or means of control over the children and that daughters more than sons are likely to be the objects of that control. This instrument object relati onship is stronger in more patriarchal families, whereas it is less prominent in less patriarchal (egalitarian) families. More patriarchal families socialize their daughters to take on feminine roles similar to that of their mothers, while sons are expect ed to take on occupational positions like those of their father (Leiber & Wacker, 1997). These reproductions of order allow for sons to have greater freedom and to be encouraged to take risks, while daughters are controlled more and discouraged from risky behavior. As a result of the differences in socialization and therefore differences in risk preferences and perceptions, sons are more likely to be involved in delinquent activities compared to daughters (Blackwell, 2000). Alternatively, less patriarc hal families, where mothers and fathers have similar levels of occupational authority and power in the family, do not exhibit such pronounced differences between sons and daughters since they are controlled more similarly than in unbalanced families (Hagan et al., 1987). It is the similarity in the controls and ideation of risk that narrows the gender gap in delinquency within less patriarchal families in comparison to more patriarchal families. Power control theory is not merely a theory of
15 the gender pa ttern of delinquency; rather, it is a theory of the gender patterning of risk preferences more generally, which could potentially lead to a wide variety of risktaking behaviors (Grasmick et al., 1996, p.181). Hagans first attempt to advance power c ontrol theory reveals his awareness of the impact of private patriarchy on the gender difference in crime. In homes where wives do not work outside the home but husbands are fully employed, private patriarchy prevails. It is in these more patriarchal h ouseholds that Hagan hypothesizes a larger gender difference in delinquency. Alternatively, in families where both husband and wife are employed and enjoy equal levels of authority in their employment, private patriarchy is diminished; likewise, the gende r gap in delinquency in these less patriarchal households is expected to be reduced. Although Hagan makes no explicit distinction between private and public patriarchy in his development of power control theory, it is clear that the theoretical statemen t singles out the influence of private patriarchy specifically on the gender gap in delinquency. This emphasis on private patriarchy is further evident in empirical tests of power control theory. Research on power control theory While power control theor y has been praised as a fundamental development in criminological theory, empirical support for the theory is mixed (Grasmick et al., 1996; Hagan et al., 1979, 1987; Hill & Atkinson, 1988; Singer & Levine, 1988; Jensen & Thompson, 1990, Morash & Chesney Li nd, 1991; Avakame, 1997; Leiber & Wacker, 1997; Uggen, 2000; Blackwell, 2000; Blackwell et al., 2002). The inconsistencies in support of power control theory have resulted in criticisms that attack Hagan and his
16 associates from several sides, but most con troversy has centered on the measures used to test power control theory. A weakness feeding this debate is that while Hagans Canadian data show strong support for the theory, attempts to replicate these findings with other data sets have often failed. H agan has attempted to answer these criticisms and lack of support by replying that their contradictory findings are due to poor measurement of the latent variables in power control theory. Of central concern is the attempt to measure the effects of patriarchy. Hagans own empirical tests of the theory have evolved and changed from the initial version (1985) of power control theory. In fact, in the first test of the theory, Hagan et al. (1985) had no measure of patriarchy at all. Instead, they tested the hypothesis that movement upward in social position or class increases the gender difference in delinquency. Hagan assumed that the freedom to deviate is directly related to ones class position, explicitly asserting that males are freer to deviate tha n females, and that males are freest to deviate in the highest classes. To test their hypothesis, Hagan et al. (1985) compared the effect of gender on delinquency controlling for parental controls and risk among four neoMarxian class categories: employer managerial, working, and surplus c lasses (Wright 1978; Wright et al., 1982). Four questions were asked about only the head of the household to determine the class category in which the respondent belongs. Employers were those working for themselves and had others working for or paid by them. Managers were full time workers who worked for someone else and supervised others. Workers were those working full time but did not work for themselves, did not supervise others, and did not have others working for them. Finally,
17 the surplus category comprised those not working full time. In this study, patriarchy was assumed to be present in the employer and managerial classes and absent in the working and surplus classes, but patriarchy itself was never measured. Hagan et al.s (1985) finding in this initial study was that the gender effect in the employer class differed most from the gender effect in other classes. A possible counterargument in the literature (Coser & Coser, 1974; Coser, 1985), however, emphas ized the need to look at the relative difference between the occupational power of both the father and mother within a household. Hagans second test of power control theory (Hagan et al., 1987) incorporated some of these counterarguments and introduced a more deliberate measure of patriarchy. In this second test, Hagan measured fathers occupational authority relative to mothers occupational authority in order to determine empirically whether a family was balanced or unbalanced. Similar to the measures used in the first (1985) study, Hagan et al. (1987) asked questions about whether the individual supervised, or was supervised by, others in the workplace. In the second study, however, these questions were extended to both husband and wife rather than limited to only the head of household. Responses to these questions produced what Hagan et al. (1987) referred to as a Dahrendorfian model of family class relations. The relative occupational authority measure, or occupational patriarchy, separated famil ies into six categories, which were subsequently designated as balanced or unbalanced. Balanced families included: upper command class (both wife and husband have occupational authority), upper obey class (both wife and husband both have no authority), an d lower obey class (wife not employed and husband has no occupational authority).
18 Female headed households (wife has occupational authority and husband has no authority) and male headed households (wife has no occupational authority or is unemployed and husband commands) were considered unbalanced or patriarchal. Hagan et al. (1987) found that the gender effect on delinquency, controlling for parental controls and risks, was larger in patriarchal families than in balanced families. In other words, male s from patriarchal families were involved in significantly more delinquency than females within patriarchal families, while males and females in balanced families were not significantly different in their involvement in delinquency. Most of Hagans subse quent tests of power control theory have operationalized the concept of occupational patriarchy using measures identical to or conceptually indistinct from those used in the 1987 test of the theory (Hagan, 1990; Hagan et al., 1988; Hagan et al., 2004; Haga n and Kay, 1990; McCarthy et al., 1999). Moreover, a handful of studies conducted by other researchers have also adopted measures identical to those used in the Hagan et al. (1987) study (Singer and Levine, 1988; Grasmick et al., 1996; Blackwell, 2000; Bla ckwell et al., 2002; Blackwell and Piquero, 2005). Other researchers attempting to test power control theory have not always operationalized patriarchy in a manner consistent with Hagans own practices. In one early test of the theory, Hill and Atkinson (1988) analyzed gender differences in the effects of parents instrumental and relational controls but neglected to examine these relationships within the context of patriarchy or even economic class. Leiber and Wacker (1997), in a test of power control theory that focused on delinquency among children of single mothers, used an index of mothers education and employment status combined
19 with neighborhood socioeconomic status as a proxy for mothers power. Bates et al. (2003) substituted a measure of attitudinal patriarchy, assessing the degree of adherence to traditional sex role beliefs, in the absence of a measure of occupational patriarchy1Additional studies that test power control theory offer closer approximations of th e occupational patriarchy measure advocated by Hagan et al. (1987) but still fall somewhat short (Jensen and Thompson, 1990; Morash and Chesney Lind, 1991; Avakame, 1997; Finckenauer et al., 1998; Uggen, 2000; Blackwell and Reed, 2003; Hadjar et al., 2007) Rather than questioning respondents on whether or not they supervise or are supervised by others in the workplace, several studies employ a survey item that simply elicits the respondents occupation, which is then classified into one of several categori es consistent with the neoMarxist framework used by Hagan et al. (1985). Occupations in the employer and manager categories are assumed to be with authority and those in the worker and surplus categories are assumed to be without authority. Occupatio nal patriarchy is then estimated on the basis of similarity or difference in the categorization of the occupations of husbands relative to wives (but see Uggen, 2000 for an exception). Although these studies make a greater effort to assess relative workpla ce authority for husbands and wives, the means by which authority is Even Hagan (Hagan et al., 1988; Hagan et al., 1993) himself has deviated from his own established measure of occupatio nal patriarchy in some of his research, substituting instead a measure of marital power based on a question about who in the family decides whether or not the wife works. 1 Grasmick et al. (1996) also incorporated attitudinal patriarchy into their power co ntrol model but retained the occupational patriarchy measure used by Hagan et al. (1987).
20 determined leaves open the possibility of error. Occupational titles can be deceiving, and an occupation that one researcher designates as with authority another may designate as witho ut authority. Given the variability with which the concept of patriarchy has been operationalized in tests of power control theory, it is not surprising that the results of these tests have been quite mixed. Although Hagans own research has produced con siderable support for the theory over the years, the results of other tests of the theory have not been as successful, even among those using measures of occupational patriarchy identical to those advocated by Hagan. For example, Singer and Levine (1988) f ound that gender differences in controls, risk preferences, and delinquency are all greater, rather than reduced as the theory predicts, in balanced households. Moreover, studies by Blackwell (2000; Blackwell et al., 2002; Blackwell and Piquero, 2005), usi ng different samples, failed to find any statistically significant differences between more and less patriarchal households in the gender effect on delinquency when power control variables were estimated in the models tested. It is important to note that the measure of patriarchy used in all of these tests of power control theory focuses solely on conventional ideas of household production and paid employment in order to label a family as more or less patriarchal. Under a more patriarchal family, th e breadwinning father and homemaker mother are intended to illustrate an imbalance of power, while less patriarchal families are described as more balanced in power as a result of equal occupational authority between the mother and father. Walby (1990) id entifies the setting for this type of gender relations as private
21 patriarchy. However, Walby also acknowledges the existence of public patriarchy. Public patriarchy expresses the perpetuation of gender bias though institutional structures at the macro le vel such as education, employment, and the government. Differing from private patriarchy, this type of oppression allows women to join the public sphere, but maintains power and control over females through various practices, such as subordination and seg regation. Moreover, Walby (1990) makes a persuasive argument that contemporary society has been shifting from private patriarchy to public patriarchy. Unfortunately, none of the extant tests of power control theory attempt to measure patriarchy in its pu blic form. In Hagans more recent work, he acknowledges that private patriarchy is now fairly rare in Western families, and uses that argument to account for a persistent gender difference in delinquency even within less patriarchal families. Nevertheles s, he does not speculate on whether private patriarchy has been replaced by public patriarchy (Hagan et al., 2004). As a result, he never conceptualizes patriarchy beyond the family setting in power control theory, and thus misses the opportunity to incor porate measures of public patriarchy at a macro level of analysis. Public patriarchy, parallel to private patriarchy, influences the controls and risks placed on individuals differently for males and females. P ublic patriarchy does not oppress women by keeping them in the household as w ith forms of private patriarchy; access is given to the public sphere yet is still oppressive through societal constructions (i.e. unequal pay in employment or limited access to higher education). It is the inequities fo und in the public sphere that carry into the gender roles placed on adolescents in the home. Perhaps the weak findings in previous studies (Singer &
22 Levine, 1988; Blackwell, 2000; Blackwell et al., 2002; Blackwell & Piquero, 2005) may be a failure to reco gnize the role that public patriarchy may play in the raising of children. Power control theory currently lacks any measurement of public patriarchy and could be improved by including such measures. Walby (1990) emphasizes public patriarchy as an integ ral part of the conceptualization of patriarchy, which is the foundation of power control theory. Outside of tests of power control theory, other studies of gender and crime (Parker & Reckdenwald, 2008; Atwell, 2002; Heimer; 2000; Steffensmeier & Haynie, 2000; DeWees & Parker, 2003) have used measures that distinguish between private and public patriarchy. For example, private patriarchy has been measured at the macro level in these studies using the Traditional Family Index (percentage of families where the husband works, while the wife does not; percentage of married families with children) and Family Unpaid Work Index (percentage of women working full time with no income; percentage of employed females who are unpaid family workers) (Parker & Reckdenwal d, 2008). In contrast, public patriarchy has been measured at the macro level in these studies as the Income Inequality Index (ratio of men to women with bachelors or higher degree; ratio of men to women in professional occupations), the percentage of women working part time, and the percentage of women below the poverty line. Measures such as these have been found to have influence on female crime rates, even when controlling for measures of private patriarchy (Parker & Reckdenwald, 2008)
23 The Present S tudy The review of the literature on power control theory has revealed some weaknesses in the measurement of patriarchy and the role it plays in accounting for gender differences in delinquency. Although patriarchy has been measured in a variety of ways i n the extant research, all of these measures have the commonality that they assume that only private patriarchy, the oppression of women within the family, has an impact on the way sons and daughters are raised. None has considered the impact of public pat riarchy, at the community level, on gender differences in childrearing and delinquency. The present study investigates the extent to which public patriarchy plays a role in explaining gender differences in delinquency. Specifically, the study tests two alt ernative models of power control theory. The first model (see Figure 1) operationalizes private patriarchy at the family level in a manner consistent with extant tests of the theory. This model is then compared with the second model (see Figure 2), a publ ic patriarchy model that operationalizes patriarchy at the community level and links it to childrearing processes and delinquency at the family level. It is hypothesized that the public patriarchy model of power control theory will provide a more robust ex planation of gender differences in delinquency than the private patriarchy model alone. Figure 1. Private Patriarchy Conceptual Model Private Patriarchy Maternal Controls Gender Differences in Delinquency Risk
24 Figure 1 represents the basic elements of Hagans power control theory using measures of private patriarch y. It illustrates that occupational patriarchy at the family level (private patriarchy) affects the gender difference in delinquency both directly and indirectly through risk a nd maternal control factors. As illustrated, patriarchy measured through occupational authority between the mother and father impacts the level of controls placed on sons and daughters. In more patriarchal families, greater controls are put on daughters and they are raised to take fewer risks. It is this difference in childrearin g that leads to a greater gender gap in delinquency. In less patriarchal families or more balanced families, similar controls are placed on sons and daughters. They also have more equal risk preferences and risk perception leading to a narrower gender ga p in delinquency. Alternatively, Figure 2 demonstrates a new model that includes a community level measure of public patriarchy. In this conceptual model, patriarchy measured at the census tract level will, similar to the private patriarchy model, directly and indirectly affect the gender difference in delinquency through maternal controls and risk factors. A daughter raised in a more patriarchal community will be tau ght to take fewer risks and given fewer opportunities for delinquency with g reater contr ols placed upon her; alternatively, her male counterpart will be raised very different ly Sons and daughters raised in less patriarchal communities will be more equally controlled and given more equal opportunities to offend.
25 Figure 2. Public Patriarchy Conceptual Model -----------------------------------------------------------------------------------------------------------Based on the work of Walby (1990), including community characteristics that reflect public patriarchy in the power control model is crucial when looking at gender inequities. Power control theory must take a more holistic approach to the fact that patriarchy operates at not only the individual level but at the community level as well. It may be that the lack of support in previous tests of the theory is due to the need for measures of public patriarchy. For that reason, this study compares a private patriarchy model of power control with a public patriarchy model of power control to further explore the w ays in which patriarchy might account for gender differences in delinquency. Public Patriarchy Maternal Controls Gender Differences in Delinquency Risk
26 Chapter 3 Method Sample and Procedure The analyses in this study are based on data gathered for a cross sectional study designed to examine delinquency among students in a m iddle school and high school in Largo, Florida in 1998. In the middle school, the survey was administered in all social studies classes, a required course for all students in the school. During the administration of the survey, one researcher read the ques tions aloud, and a second researcher assisted as needed while the students marked their responses on the questionnaire. A total of 1,043 surveys were completed by these middle school students, representing 81% of the students enrolled in the school. Nonre sponse was due mainly to absences on the day of survey administration; very few parents actively withheld consent for their children to participate and few students declined to participate after passive parental consent was secured. In the high school, the survey was self administered in 30 randomly selected third period classrooms. The third period was chosen because more classes were taught during this period, affording a greater chance of representation of the entire student body. A total of 621 surveys were completed by the high school students, representing 79% of
27 those enrolled in the randomly selected classrooms. Again, nonresponse was due largely to absences on the day of survey administration. The analyses in the present study utilize a subsample from the original data collection effort. The private patriarchy model requires the sample to be reduced to those students living in two parent (mother/father, mother/stepfather, father/stepmother) households. This restriction is necessary in order to asse ss the occupational authority of the father relative to that of the mother. This subsample, comprising 1,058 middle and high school students, is further subdivided into those who can be classified as more patriarchal and those who can be classified as l ess patriarchal households. Missing data on variables used to determine the private patriarchy classification further eliminated cases from the final sample. Additionally, the public patriarchy model required another criterion for inclusion in the sample. Data from individual students had to be linked to macro level data in the census tracts in which these students lived. In the self report survey, students were asked to report the names of the two streets that formed the intersection closest to their home Of the 1,674 students in the original sample, 1,188 responded with information that allowed them to be geocoded. However, in order to compare the private and public patriarchy models using the same subjects, the final sample had to include only those g eo coded individuals residing in twoparent families that could be classified into more or less patriarchal families. Once missing data on variables in the analysis were taken into consideration, 485 cases were usable for both the full private and public patriarchy models. Table 1 provides descriptive statistics of the sample.
28 Table 1. Description of the Sample Variable Percentage Gender Male Female 48.7% 51.3% Type of School Middle School High School 62.7% 36.3% Race White Black Hispanic Asian Other 77.7% 11.6% 4.1% 3.2% 3.9% Avg. Age 13.72 yrs. Delinquen cy Total Vandalism Stole a backpack Petty theft Grand theft Hit someone intentionally 59.8% 25.4% 11.9% 17.5% 6.4% 38.7% With the large drop in sample size, it is dually important to mention some highlights of the sample that were omitted. While there was still more middle school kids overall in the sample omitted, a number of high school students were part of this sample. Additionally, a significant number of the students eliminated from the sample were delinquent. Although losing a large number of the students who are delinquent could be
29 problematic by losing variation and/ or biasing the re sults, more than half of the sample was still labeled delinquent lending sufficient variation in the sample Measurement of Variables Measurements of the variables are described below in detail. Gender is coded as male = 0 or female = 1, and is referred to as female in the models. T he sample was divided into m ore and less patriarchal groups based on two different ways of measuring patriarchy First, the sample was split into more and less patriarchal groups using a measure of private patriarchy an d second using a measure of public patriarchy. To operationalize private patriarchy an initial survey question asked students to describe separately the main jobs of their fathers and mothers in terms of whether or not they supervised others, had others s upervising them, or were not employed. To be classified as residing in a more patriarchal family, the father had to (a) have an authority position over others, with the mother either not having authority or no job at all, or (b) have a job with no authorit y, with the mother having no job. To be classified as residing in a less patriarchal family, both parents had to exhibit equal levels of: (a) authority, (b) lack of authority, or (c) unemployment. To classify the sample as residing in a more or less patr iarchal community using the public patriarchy concept, data were collected from the Pinellas County Census (2000). Geo coding was utilized to match the sample to census tracts, with respondents in the final sample residing in a total of 71 different tracts. Three measures were used: 1) the percentage of females living below the poverty level (reverse coded), 2) the r atio of men to women aged 25 or older with a bachelors or higher degree, and 3) the ratio of
30 men to women in management and professional occ upations (se e Parker & Reckdenwald, 2008). Factor analysis was used to create a scale and the factor scores were saved (see Appendix A) To distinguish students residing i n the more patriarchal tracts (coded as 1) from those residi ng in less patriarchal tracts (coded as 0), the median on each of the three variables was used as a cut point separating the two groups. Maternal control 2Risk preference is operationalized by four items that measure impulsivity and attraction to risk: sometimes I will take a risk just for the fun of it, I test myself every is a scale that incorporates aspects of both relational and instrumental control (see Appendix A). Relational controls by mother were assessed through a semantic differential technique that asked respondents to rate, on a scale of 1 to 6, the degree to which they could talk to their mother, were trusted by their mother, could ask their mother for advice, received praise fr om their mother, and wanted to be like their mother. Instrumental controls by mother were evaluated by eliciting the degree to which the respondent reported that their mother knows both their whereabouts and with whom they sp end time (both reverse coded). Items comprising the maternal control scale were standardized prior to scaling and factors were saved. Appendix A reports the results of exploratory factor analyses on the construction of the maternal control scale. All of the variables within the scale loaded highly, ranging from .537 .816, well above standard acceptable loadings. Furthermore, this scale had an eigenvalue of 3.48, with Cronbachs Higher scores on the maternal control va riable indicate greater cont rol 2 A scale for paternal control was created similar to maternal control; however, it was omitted from the final model due to multicollinearity with the maternal contr ol variable.
31 now and then by doing something risky, I act on the spur of the moment w ithout stopping to think, and I often do whatever brings me pleasure here and now. Responses to these four items were combined into an additive scale. The results of an exploratory factor analysis of this scale are reported in Appendix A. The risk prefer ence scale met all acceptable criterion, with variables load scores ranging from .594 .786 (Eigenvalue = Higher scores on the risk preference va riable indicate greater perception of risk. Risk perception is measured as the degree to whic h one believes he/she would get caught by the police if they committed certain delinquent acts: skipping school, stealing, hitting someone intentionally, and using marijuana. Responses to these four items were combined into an additive scale. The results of an exploratory factor analysis of this scale are reported in Appendix A. The risk perception scale met all acceptable criterion, with variables load scores ranging from .742 .833 (Eigenvalue = 2.49; Higher scores on the risk perception va riable indicate greater preference for risk. Delinquency is measured by five observed variables in the past twelve months : stolen a students backpack/lunch money, stolen other things worth $50 or less, st olen other things worth more than $50, damaged or destroyed property, and hit someone with the intention of hurting them. These measures of common delinquency (Hagan et al., 1985) are theoretically consistent with the initial test of power control theor y. Responses to these five items were then converted to a dichotomous variable, coded as 1 = having committed any number of delinquent acts specified and 0 = having committed none of
32 these delinquent acts. This process was used to accommodate and adjust f or several outliers that were skewing the data. Analytic Plan Latent constructs for the final model consisted of gender, maternal control, risk perception, and risk preference. Using the latent constructs, logistic regression was employed for e ach family and community type (more patriarchal and less patriarchal), first using a measure of private patriarchy and then using a measure of public patriarchy to account for gender differences in delinquency. In order to determine when the model estimates were significantly different, a z test was performed between estimates for gender as well as for all latent power control measures ( Knoke, Bohrnstedt, & Mee, 2002) Coefficients from the less and more patriarchal models were examined through these calculations to determine significant differences. From these two models, comparisons can be made to conclude whether or not a public patriarchy model better explains the gender difference in crime, as opposed to the traditional private patriarchy model.
33 C hapter 4 Results Private Patriarchy Model Of the 485 cases in the final sample, 330 of the youths were categorized as belonging to a less patriarchal family and 155 resided in a more patriarchal family when private patriarchy was used as the measure of patriarchy. Within each group, there were nearly equal numbers of females and males. Also, a majority of both groups consisted of middle school students with an average age of 13.9 years old in the less patriarchal sample and 13.7 years old in the more patriarchal sample. Descriptive and bivariate statistics given in Table 2 and 3 illustrate the differences in the more and less patriarchal groups. The less patriarchal group indicates higher mean levels of delinquency (mean = .620) than the more patriarc hal group (mean = .571) but this difference is not statistically significant Moreover, contrary to the predictions of power control theory, gender is significantly correlated with delinquency in less patriarchal families (r = .252) but not in more patri archal families (r = .118). In fact, there are no significant gender differences in the maternal control, risk perception, and risk preference variables among those in the more patriarchal group, whereas in the less patriarchal group, males are more likel y than females to report weaker maternal controls, lower perceptions of risk, and greater
34 preference for risk. However, when gender is not held constant, delinquency is greater among those who indicate higher preference for risk and fewer perceptions of ri sk within both the less and more patriarchal groups. Additionally, lower maternal control is associated with higher likelihood of delinquency, but only for those in less patriarchal families. Maternal control is also significantly related to risk preferen ce and risk perception in both the more and less patriarchal samples. These findings indicate that regardless the type of patriarchal influence, the stronger maternal control, the more the individual has a negative perception of risk and the less likely he/she will be involved in risky behavior. Table 2 Correlations of Variables Included in Final Private Patriarchy Model Less Patriarchal Sample (N=330) Variables 1 2 3 4 5 1. Gender --2. Delinquency .252 ** --3. Maternal Control .108* .275 ** --4. Risk Perception .160* .353 ** .274** --5. Risk Preference .137 .367 ** .392** .388** --Mean .501 .620 .022 .023 .007 S.D. .500 .486 .986 1.022 .963 *p<.05, **p<.01
35 Table 3 Correlations of Variables Included in Final Private Patriarchy Model More Patriarchal Sample (N=155) Variables 1 2 3 4 5 1. Gender --2. Delinquency .118 --3. Maternal Control .014 .137 --4. Risk Perception .004 .279 ** .327** --5. Risk Preference .072 .320 ** .199* .350 ** --Mean .484 .571 .041 .031 .294 S.D. .501 .496 .915 .925 .457 *p<.05, **p<.01 In an effort to more fully understand these relationships, binary logistic regression was used to examine the extent to which the control and risk variables from power cont rol theory mediated the effect of gender on delinquency. If power control theory is a valid explanation of gender differences in delinquency, then the gender effec t on delinquency will be stronger in more than in less patriarchal families. Further, it is e xpected that the gender effect will be reduced for both more and less patriarchal families when the control and risk variables are introduced into the model; however, the reduction will be more pronounced in the less patriarchal than in the more patriarcha l group. As illustrated in Table 4, Equation 1 is the base model that includes only the variable female using measures of private patriarchy as the selection variable. Consistent with the bivariate analysis in Tables 2 and 3, only the less patriarchal sample shows a significant impact of gender on delinquency, with females having 1.027 lower logodds of being delinquent than males. Contrary to power control theorys predictions, males and females are equally likely to be delinquent in more patriarchal families. A z test for difference in the log odds of females being delinquent between more and less patriarchal families (z = 1.644) indicates that the difference between the two groups in the effect of
36 gender on delinquency just misses significance at the .05 level. As a result, when patriarchy is measured in terms of relative occupational authority at the family level, gender differences in delinquency are greater rather than smaller in less patriarchal families, and the impact of gender is somewhat gr eater (although not significantly so) in less than in more patriarchal families. Despite these unexpected findings, it remains instructive to examine the effect of adding control and risk variables to these models, as demonstrated in Equation 2 in Table 4. In the less patriarchal model, the log odds of females being delinquent is reduced from 1.027 to .897, indicating that maternal control, risk perception, and risk preference partially mediate the gender effect on delinquency. Further, lower mat ernal controls, lower perceptions of risk, and higher preferences for risk all contribute to the probability of being delinquent when gender is held constant. Nevertheless, the gender effect on the probability of being delinquent is not completely eliminated when control and risk variables are included in the model, indicating that power control theory cannot completely explain gender differences in delinquency even in less patriarchal families. Although there was no difference between males and females in the probability of being delinquent in the more patriarchal sample to account for, the model is presented for heuristic purposes. When maternal control and the risk variables are added to the model with gender, gender remains a nonsignificant predictor of delinquency, as does maternal control. However, lower perceptions of risk and higher preference for risk are both associated with a higher probability of delinquency within the more patriarchal sample, as power control theory would predict.
37 Table 4. Bi nary Logistic Regression Predicting Probabilities for Power Control Theory Latent Constructs/Gender and Delinquency in Private Patriarchy Model Less Patriarchal More Patriarchal Variable Log Odds SE Odds Ratio Log Odds SE Odds Ratio Z Score Equation 1 Female 1.027*** .221 .358 .397 .313 .672 1.644 NS Constant .848 .187 2.334 .535 .245 1.707 N 375 170 2LL 477.566 229.421 Pseudo R 2 .089 .013 Equation 2 Female .897** .262 .408 .464 .357 .629 .97 7 NS Mat. Cont. .316* .151 .729 .076 .191 .927 .986 NS Risk Perc. .562*** .148 .570 .509* .235 .601 .191 NS Risk Pref. .636*** .172 1.889 .689** .237 1.991 .181 NS Constant .838 .225 2.312 .702 .287 2.018 N 330 155 2LL 355.902 184.694 Pseudo R 2 .317 .199 *p<.05, **p<.01, ***p<.001
38 Public Patriarchy Model Using the public patriarchy measure, the final sample of 485 students consisted of 309 adolescents residing in communities categorized as less patriarchal and 176 ado lescents residing in communities categorized as more patriarchal, both having nearly equal numbers of males and females. Both samples consisted of more middle school students with an average age of 13.8 years old in the less patriarchal sample and 13.7 ye ars old in the more patriarchal sample. Interestingly, and contrary to the findings utilizing the private patriarchy selection criterion, as shown in Tables 5 and 6, those residing in more patriarchal communities show greater mean levels of delinquency (M ean = .629) than those residing in less patriarchal communities (Mean = .588) however, this difference is not statistically significant Nonetheless, individuals living in both less and more patriarchal communities show significant correlations between g ender and delinquency (r= .218 and r= .193, respectively,). Specifically, they both are in the expected direction, such that females are less likely to be delinquent compared to their male counterparts. Females in less patriarchal communities report a stronger perception of getting caught if participating in risky behaviors and less preference for risky behavior. Moreover, the females residing in the less patriarchal communities indicate higher perception of risk than males, while females in more patri archal communities are not statistically significantly different from males in their perception of risk. Furthermore, delinquent adolescents (without regard to gender) from both community types report fewer maternal controls, lower perception of risk, and greater risk preference than non delinquents. Maternal controls are also significantly related to both risk preference and
39 risk perception. Similar to the fin dings in the private patriarchy model, regardless of the type of patriarchal influence, strong maternal control predicts lower risk preference and higher perception of risky behaviors as negative or having a negative outcome Table 5 Correlations of Variables Included in Final Public Patriarchy Model Less Patriarchal Sample (N=309) Varia bles 1 2 3 4 5 1. Gender --2. Delinquency .218 ** --3. Maternal Control .085 .265 ** --4. Risk Perception .193* .349 ** .308** --5. Risk Preference .141 .406 .317** .378** --Mean .515 .588 .033 .030 .015 S.D. .500 .493 .996 .969 .960 *p<.05, **p<.01 Table 6 Correlations of Variables Included in Final Public Patriarchy Model More Patriarchal Sample (N=176) Variables 1 2 3 4 5 1. Gender --2. Delinquency .193** --3. Maternal Control .035 .156* --4. Risk Perception .025 .296 ** .251** --5. Risk Preference .070 .240* .353** .372 ** --Mean .496 .629 .085 .048 .016 S.D. .501 .484 1.016 1.003 .969 *p<.05, **p<.01 Since males are significantly more likely to be delinquent than females in both types of communities, it is appropriate to explore the degree to which variables from power control theory mediate those gender differences using a multivariate logistic regression technique. Again, it is expected that the gender effect on delinquency will decr ease when the maternal control, risk perception, and risk preference variables are
40 added to the basic model. Equation 1 in Table 7 in which delinquency is regressed on only the gender variable, demonstrates that females living in less patriarchal neighbor hoods have .885 lower log odds of being delinquent than males, while females residing in more patriarchal neighborhoods have .705 lower log odds of being delinquent than males. Contrary to theoretical expectations, though, the gender gap in delinquency is wider in the less than in the more patriarchal communities. Equation 2, in which the control and risk variables are incorporated, demonstrates that the gender effect on delinquency is indeed mediated by the power control variables, but only in the less pa triarchal communities, where the log odds of females being delinquent decreases from .885 to .692. On the other hand, in the more patriarchal communities, once maternal controls, risk perception, and risk preference are added to the model, the log odds of females being delinquent actually increases from .705 to .918. In the more patriarchal model, then, the power control variables demonstrate a suppression effect rather than a mediating effect on the gender delinquency relationship. Despite the diff erence between the less and more patriarchal samples in the way the power control variables operate on the gender delinquency relationship, the difference in the gender effect across the two models is not statistically significant (z = .511).
41 Table 7. Binary Logistic Regression Predicting Probabilities for Power Control Theory Latent Constructs/Gender and Delinquency in Public Patriarchy Model Less Patriarchal More Patriarchal Variable Log Odds SE Odds Ratio (O.R.) Log Odds SE Odds Ratio Z Score Equation 1 Female .885*** .223 .413 .705* .303 .494 .478 NS Constant .767*** .186 2.153 1.018 .239 2.767 N 348 197 2LL 458.436 251.019 Pseudo R 2 .061 .045 Equation 2 Female .692** .268 .501 .918** .352 .339 .511 NS M at. Cont. .273 .146 .761 .141 .203 .868 .528 NS Risk Perc. .508*** .159 .602 .635** .205 .530 .490 NS Risk Pref .839*** .178 2.314 .360 .225 1.434 1.67 NS Constant .743 .227 2.102 1.196 .291 3.306 N 309 176 2LL 338.089 200.707 Pseudo R 2 .319 .208 *p<.05, **p<.01, ***p<.001 Examination of both models shows more support for power control theory when patriarchy is measured at the public or community level rather than at the private or
42 family level; however, this support is limited. Mater nal controls are not significant in either the more or less patriarchal sample and moreover, none of the z tests indicate significant differences between the more and less patriarchal communities. Further discussion of these results is presented in the ne xt chapter.
43 Chapter 5 Discussion and Conclusion Gende r differences in delinquency are evident across a wide spectrum of criminological research. Hagans power control theory was designed specifically to account for this gender difference. H ow ever, the evi dence in support of the power control model has been modest at best. At t he foundation of the power control explanation of gender differences in delinquency is the concept of patriarchy Specifically, children reared in homes where the fathe r holds a position of occupational authority, while the mother does not is considered a more patriarchal family environment. This difference in occupational authority is translated into varying parental controls and risk ideations for sons and daughters. In this family type, power control theory predicts a greater gap in delinquency between sons and daughters due to differential controls, risk perception, and risk preference Contrarily, families with parents who have equal authority in their jobs are considered less patriarchal, producing reduced gender differences in controls and risky behavior. Whereas Hagan s own research has demonstrated support for power control theory, others who have attempted to replicate his studies have failed. This study suggests that patriarchy, as the pillar of power control theory, is found not only at the family level, but also at the community level. Sylvia Walby ( 1990)
44 conceptualizes these two forms of patriarchy as private and public. Private patriarchy is describ ed as oppression of females at the individual level within the family. On the other hand, public patriarchy is considered to be oppression at the community level through unequal employment opportunities or state institutions. Power control theory does ta ke into consideration private patriarchy but does not consider public patriarchy in its model. Conceivably, the lack of support in previous studies could be a result of the exclusion of public patriarchy measures. The current study compare d a tradit ional, private patriarchal model of power control theory to a model which used public patriarchal measures to delineate be tween more and less patriarchal influences on gender differences in crime among adolescents. Although no significant gender differenc es were found between more and less patriarchal influences in either the private or public patriarchal model, there were a number of significant and insignificant paths within each model and at various levels of analysis worth discussing. Within the priva te patriarchy model, b ivariate analysis illustrated a strong correlation between gender and delinquency in less patriarchal families ; however, the more patriarchal sample showed an insignificant correlation between gender and delinquency. There was also a lack of correlation between delinquency and risk/control variables in the more patriarchal sample. Multivariate analysis shows similar results; in the private patriarchy model, there is only a significant relationship between gender and delinquency in the less patriarchal family types, while an insignificant relationship was
45 found in the more patriarchal sample. Furthermore, there was no significant difference between the two samples gender coefficients using measures of private patriarchy. A primary reason for these findings could be a result of the final sample size. Once the measures were determined and missing variables were accounted for in the analysis, the sample size was reduced from 1,674 cases to 485 cases total. After implementation of the more and less patriarchal standards, further reduction in the sample sizes could have affected the outcomes especially given the disparity in the distribution of cases in regards to more and less patriarchal samples Specifically, the sample size for th e more patriarchal sample i s drastically smaller than t he less patriarchal sample size. Walby (1990) explains that there is a shift in modern society from private to public patriarchy. Therefore, relatively low numbers of students being categorized as more patriarchal in the private model is expected, but could be detrimental to the analysis in terms of finding significant results. Furthermore, with only about a fourth of the sample being used in the models, there are issues with generalizability. It is possible that those who answered the questions to the surveys that were in the final model are unique or have something in common. Also, with the survey being administered within schools, adolescents who may be absent from school for a number of reasons are not being represented. For instance, all of the kids who may be the most delinquent, such as students who are truant, those who may have dropped out of school, or even those children who are in juvenile detention centers are missing from the sample. Therefore, it is difficult to generalize these findings to all adolescents.
46 Unlike the private patriarchy model gender differences in delinquency were present in both more and less patriarchal communities when patriarchy was measured at the public level However, the correlation was stronger in less patriarchal communities than i n more patriarchal communities. While this finding is contrary to the power control theory, it may be explained by the fact that males naturally commit more crimes and in less patriarchal communities where both parents could be working outside the home, ther e are even fewer controls placed on the children overall, leading to more delinquency. Also, within communities where more equal opportunities are given to women, daughters may be raised to aspire to reach certain goals and in turn are controlled more and have less risk preference. Another key finding in the model was that maternal control was not a significant predictor of delinquency in both less and more patriar chal communities Perhaps the explanation for these findings is that when using measures of public patriarchy, maternal control is no longer an appropriate measure when looking at the community level influence on adolescent behavior The maternal control scale is based at the individual level. If Walby (1990) is correct, then with the shift from private to public patriarchy, maternal control is no longer as influential on gender differences given the oppressive effects of patriarchy would be mostly found in co mmunity level variables. Multivariate analysis did show a significant relationship between gender and delinquency in both more and less patriarchal communities. Unfortunately, using z score calculations, there was no significant difference found betw een the two community types. Again, the power control theory model operates as expected in less patriarchal
47 communities, but does not in the more patriarchal communities. In fact, the gender difference increases rather than decreases when control/risk va riables are incorporated, indicating a suppressor effect. S uppression usually occurs when a n independent variable is related to the dependent variable in one direction, but relates to the other independent variable in the opposite direction. In this part icular model, females are less likely to be delinquent (negative relationship), but females have a positive relationship with the risk preference variable. Moreover, females have a negative relationship between the other control and risk variables. S e veral of the limitations in this study related to the measures and models in addition to the previously mentioned issue of sample size The measure of delinquency as a dichotomous variable limited the type of analyses that could be performed as well as t he variation. Creating a dichotomous variable was necessary with the data given the low numbers of delinquent adolescents as defined by the scale adapted from Hagans measure of delinquency. Despite the weakness of the variable, it was important to adher e to Hagans measure as closely as possible to test power control theory appropriately; especially given that the focus of the study was to compare the traditional model with another measure of patriarchy. However, future studies could either use larger datasets or find a more appropriate scale to measure common delinquency among adolescents. Additionally, there was no measure of paternal control in the model. Hagans model of power control theory includes this particular measure. A scale for paternal control was created similar to maternal control; however, it was eliminated due to multicolli nearity with the maternal control variable. Furthermore, inclusion of the
48 paternal control measure in the model had no impact on the results compared to the model excluding paternal control. Also, a combined parental control variable was created, but did not meet sufficient eigen values to be utilized nor was it suitable for finding key relationships in power control theory. Specifically, it was most important to measure the difference in maternal control to understand the relationship sought in power control theory. Maternal control is the variable that illustrates the patriarchal influence in the home and truly uncovers sons and daughters difference in controls and risk. Nevertheless, exclusion of paternal control in the model limits the study in that it does not follow the traditional power control theory. Given the results, there are a number of theoretical implications of this study. Although gender is a significant factor in some of the analyses, ultimately there was no gender difference found between individuals or neighborhoods characterized as having more or less patriarchal influences. Even though no gender differences were found in the final model, there is still support for the public patriarchy measure. The models using the public patriarchy measure show more significant paths as predicted in power control theory, while the private patriarchy models do not. Regardless of the public or private mea sures of patriarchy, the study failed to uncover any statistically significant relationships between less and more patriarchal contexts in the gender pattern of delinquency. The gender difference did not vary between more and less patriarchal samples, con sistent with other studies (Singer & Levine, 1988; Blackwell, 2000; Blackwell et al., 2002; Blackwell & Piquero, 2005). Patterns of findings were not substantively different between private and public
49 patriarchy models. Development of a multilevel model could be a more appropriate measure of the influence that public patriarchy has on individual level behavior. Additionally, with the changing of family dynamics, power control theory must evolve to not only include measures of public patriarchy but also t he prevalence of singleparent families as well as alternative family structures While a few studies ( Hagan et al., 1987; Jensen & Thompson, 1990; Leiber & Wacker, 1997; Morash & Chesney Lind, 1991) have indeed investigated single parent households, the ir findings remain limited to mainly single mother families, used alternative measures to those in traditional power control theory and remained inconclusive Another limitation to the study is not identifying and using siblings to test the models. Usin g siblings in future studies would be ideal in measuring wh ether or not the family type or even the community type influenced sons and daughters differently. Specifically, by using siblings in studies, gender differences in delinquency between siblings w ithin families can be examined to truly test power control theory as it is conceptualized (Blackwell & Reed, 2003). Future studies should take into consideration the fact that the central tenet of power control theory is the idea of patriarchy and its effect on gender differences in delinquency. Patriarchy influences exist in more aspects of society than simply at the family or individual level. In order to truly assess the effect of patriarchy on gender differences in crime, it should be measured at the community level as well.
50 References Atwell, M. W. (2002). Equal Protection of the Law? Gender and Justice in the United States New York: McGrawHill. Avakame, E. F. (1997). Modeling the Patriarchal Factor in Juvenile Delinquency: Is There Room for Peers, Church, and Television? Criminal Justice and Behavior 24, 477494. Bates K. A., Bader, C. D., & Mencken, F. C. (2003). Family Structure, Power Control Theory, and Deviance: Extending Power Control Theory to Include Alternate Family Forms Western Criminology Review 4, 170190. Belknap, J. (2007). The Invisible Woman: Gender, Crime, and Justice (3rd ed.). Belmont, CA: Wadsworth. Blackwell, B. S. (2000). Perceived Sanction Threats, Gender, and Crime: A Test and Elaboration of Power Contr ol Theory. Criminology 38, 439488. Blackwell, B. S. & Piquero, A R. (2005). On the Relationships Between Gender, Power Control, Self Control, and Crime. Journal of Criminal Justice 33, 117. Blackwell, B. S. & Reed, M. D. (2003). Power Control as a Between and WithinFamily Model: Reconsidering the Unit of Analysis. Journal of Youth and Adolescence 32, 385399.
51 Blackwell, B. S., Sellers, C. S., & Schlaupitz, S. M. (2002). A Power Control Theory of Vulnerability to Cr ime and Adolescent Role Exits --Revisited Canadian Review of Sociology and Anthropology 39, 199218. Chapple, C. L., McQuillan, J. A., & Berdahl, T. A. (2005). Gender, Social Bonds, and Delinquency: A Comparison of Boys and Girls Models. Social Science Research 34, 357383. Co ser, R. (1985). Power Lost and Status Gained: The American Middle Class Husband. Paper presented at the meetings of the American Sociological Association, Washington, D.C. Coser, L. & Coser, R. (1974). Greedy Institutions: Patterns of Undivided Commitment New York: Free Press. Daly, K. & Chesney Lind, M. (1988). Feminism and Criminology. Justice Quarterly 5, 497538. DeWees M. A., & Parker, K. F. (2003). Women, Region, and Types of Homicide: Are there Regional Differences in the Structural Status of Women and Homicide Offending? Homicide Studies 7, 368393. Grasmick, H. G., Hagan, J., Blackwell, B., & Arneklev, B. J. (1996). Risk Preferences and Patriarchy: Extending Power Control Theory. Social Forces 75, 177199. Hadjar, A., Baier, D., Boehnke, K ., & Hagan, J. (2007). Juvenile Delinquency and Gender Revisited. European Journal of Criminology 4, 3358. Hagan, J. (1990). Destiny and Drift: Subcultural Preferences, Status Attainments, and the Risks and Rewards of Youth. American Sociological Revi ew 55, 567582.
52 Hagan, J., Gillis, A. R., & Simpson, J. (1985). The Class Structure of Gender and Delinquency: Toward a Power Control Theory of Common Delinquent Behavior. American Journal of Sociology 90, 11511178. Hagan, J. & Kay, F. (1990). Gender a nd Delinquency in White Collar Families: A Power Control Perspective. Crime and Delinquency 36, 391407. Hagan, J., Simpson, J., & Gillis, A. R. (1987). Class in the Household: A Power Control Theory of Delinquency. American Journal of Sociology 92, 788816. Hagan, J. & Wheaten, B. (1993). The Search for Adolescent Role Exits and the Transition to Adulthood. Social Forces 71, 955980. Heimer, K. (2000). The Nature of Crime: Continuity and Change. Changes in the Gender Gap in Crime and Womens Economic Marginalization. Criminal Justice 1, 427483. Hill, G. D. & Atkinson, M. P. (1988). Gender, Familial Control, and Delinquency. Criminology 26, 127149. Jensen, G. F. & Thompson, K. (1990). Whats Class Got to Do with It? A Further Examination of Power Control Theory. The American Journal of Sociology 95, 10091023. Knoke, D., Bohrnstedt, G. W., & Mee, A. P. (2002). Statistics for Social Scientists (4th ed.). Belmont, CA: Wadsworth. Lieber, M. J. & Wacker, M. E. (1997). A Theoretical and Empir ical Assessment of Power Control Theory and Single Mother Families. Youth & Society 28, 317350.
53 McCarthy, B., Hagan, J. & Woodward, T. S. (1999). In the Company of Women: Structure and Agency in a Revised Power Control Theory of Gender and Delinquency Criminology 37, 761788. Morash, M. & Chesney Lind, M. (1991). A Reformulation and Partial Test of Power Control Theory of Delinquency. Justice Quarterly, 8, 347377. Rodham, K. Hawton, K., Evans, E., & Weatherall, R. (2005). Ethnic and Gender Differ ences in Drinking, Smoking, and Drug Taking Among Adolescents in England: A Self Report School Based Survey of 15 and 16 Year Olds. Journal of Adolescence, 28, 6373. Parker, K. & Reckdenwald, A. (2008). Women and Crime in Context: Examining the Linkage s Between Patriarchy and Female Offending Across Space. Feminist Criminology 3, 524. Sabol, W. J. & Couture, H. (June 2008). Prison Inmates at Midyear 2007. Bureau of Justice Statistics Bulletin. http://www.ojp.usdoj.gov/bjs/pub/pdf/pim07.pdf Singer, S. I. & Levine, M. (1988). Power Control Theory, Gender, and Delinquency: A Partial Replication with Additional Evidence on the Effects of Peers. Criminology 26, 627647. Stack, S., Gundlach, J., & Reeves, J. L. (1994). The Heavy Metal Subculture and Suicide. Suicide and Life Threatening Behavior 24, 1523. Stack, S. & Wasserman, I. (1992). The Effect of Religion on S uicide I deolog y: An Analysis of the Networks P erspective. Journal for the Scientific Study of Religion, 31, 457464.
54 Stack, S. & Wasserman, I. (1995). The Effect of Marriage, Family, and Religious Ties on African American Suicide Ideology. Journal of Marriage and the Family 57, 215222. Steffensmeier, D. & Haynie, D. (2000). G ender, Structural Disadvantage, and Urban Crime: Do Macrosocial Variables Also Explain Female Offending Rates? Criminology 38, 403438. Tjaden, P. & Thoennes, N. (2000). Extent, Nature, and Consequences of Intimate Partner Violence: Finding s from the Na tional Violence Against Women Survey. http://www.ncjrs.gov/pdffiles1/nij/181867.pdf Toro, P. A., Ur berg, K. A., & Heinze, H. J. (2004). Antisocial Behavior and Affiliation with Deviant Peers. Journal of Clinical Child & Adolescent Psychology 33, 336346. Uggen, C. (2000). Class, Gender, and Arrest: An Intergenerational Analysis of Workplace Power and Control. Criminology 38, 835862. Uniform Crime Report (2008). http://www.fbi.gov/ucr/cius2008/data/table_42.html Walby, S. (1990). Theorizing Patriarchy Oxford, UK: B. Blackwell. Wright E. O. (1978). Race, Class, and Income Inequality. American Journal of Sociology 83, 3681397. Wright E. O., Costello, C., Hachen, D., & Sprague, J. (1982). The American Class Structure. American Sociological Review 47, 709 126. Yllo, K. (1983). Sexual Equality and Violence Against Wives in American States. Journal of Comparative Famil y Studies 14, 6786.
55 Yllo, K. (1984). The Status of Women, Marital Equality, and Violence Against Wives. Journal of Family Issues 5, 307320. Yllo, K. & Straus, M. A. (1984). The Impact of Structural Inequality and Sexist Family Norms on Rates of Wife Beating. Journal of International and Comparative Social Welfare 1, 1629.
57 Appendix A. Factor Loadings for Latent Variables Factor Loading Eigenvalues Variance Cronbachs Alpha Maternal Control Scale I cant/can talk to her about anything. a She never/always trusts me. a I can never/always ask her for advice. a She never/always praises me. a I dont want/want to be the kind of person she is. a Mother knows where I am when I am not at home. b Mother knows who I am with when I am n ot at home. b .798 .734 .816 .708 .744 .537 .546 3.48 49.7 .83 Risk Perception c Skipping school. Stealing something. Hitting someone with the idea of hurting them. Using marijuana. .784 .764 .594 .786 2.17 54.2 .71 Risk Preference b Sometim es I will take a risk just for the fun of it. I test myself every now and then by doing something risky. I act on the spur of the moment without stopping to think. I often do whatever brings me pleasure here and now. Public Patriarchy Percent of females l iving below poverty level (<15K) Ratio of men to women aged 25 and older with a bachelors or higher degree Ratio of men to women in management and professional occupations .812 .833 .742 .767 .738 .560 .812 2.49 1.52 62.3 50.6 .80 a. 6 point Semantic Differential scale b. 1= strongly disagree= strongly agree. c. 1= not at all likely= very likely [to get caught by police]