Keywords

These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.

Criminal conduct is largely the purview of boys and men. Regardless of time, culture, country, or measurement strategy, scholars have repeatedly concluded that males, irrespective of their race or ethnicity, are responsible for the vast majority of crime, particularly crime of a violent and serious nature (Belknap, 2007; Blanchette & Brown, 2006; Muraskin, 2012). Although girls and women commit considerably less crime than boys and men, official statistics suggest that arrest and incarceration rates for females have increased substantially over the last two decades while the corresponding male rates have either remained constant or declined (Glaze, 2011; Snyder, 2011; Snyder & Sickmund, 2006). In 2010, women comprised 1.3 million of the 7.1 million adults in the USA under some form of correctional supervision (Glaze, 2011; Glaze & Bonczar, 2011; Guerino, Harrison, & Sabol, 2011; Minton, 2012). Importantly, non-White females are disproportionately incarcerated relative to their White female counterparts. For example, compared to the incarceration rate of White women (91 per 100,000), the incarceration rate of Black women and Hispanic women is respectively three times higher (260 per 100,000) and 1.5 times higher (133 per 100,000) (Glaze, 2011). Additionally, the American juvenile court system processed approximately 1.5 million youth in 2009, 28 % of whom were female. Although non-White female youth only accounted for 37 % of the 2009 female court statistics, they accounted for 61 % of the 2010 female youth residential placement population (Sickmund, Sladky, & Kang, 2012; Sickmund, Sladky, Kang, & Puzzanchera, 2011).

Despite the overrepresentation of non-White females in the criminal justice system, female-centered research and practice have concentrated primarily on girls and/or women falling into the majority demographic within the general population—that is, White females. Correctional psychologists have either ignored or paid little attention to the extent to which gender, race, and ethnicity may intersect (interact) and potentially result in unique causal pathways to the criminal justice system.Footnote 1 This is particularly problematic given that the existence of genuine differences may, in turn, necessitate the development and implementation of specialized correctional responses for non-White females.

Conversely, feminist criminologists have vigorously championed multiracial feminism or the intersectionality paradigm (Baca Zinn & Thornton Dill, 1996; Crenshaw, 1989/1993). This perspective posits that identities such as race, gender, and social class simultaneously intersect producing multiplicative rather than additive effects on an individual’s experience with the criminal justice system. The defining feature of this paradigm is the notion that marginalized individuals—namely, non-White, economically disadvantaged women—are most vulnerable to oppression. It is argued that this oppression, in turn, translates into systematic discrimination by the criminal justice system. For example, in accordance with the intersectionality paradigm, a low-income Black woman is said to experience the criminal justice system in profoundly different and inequitable ways in contrast to a middle-class White man or woman.

This chapter aims to synthesize what is currently known about the intersection of gender, race, ethnicity, and crime from a predominantly correctional psychological perspective with an emphasis on racial–ethnic minority women. That is, although the intersections of race, ethnicity, and gender impact racial–ethnic minority men (as can be seen by their disproportionate rates of arrests and incarceration), due to limited space as well as the fact that racial–ethnic minority females have garnered the greatest attention from criminologists working within the intersectionality paradigm, we focus primarily on concerns affecting girls and women. The chapter is divided into three sections: statistics, theory, and applications. In the statistics section, we briefly discuss what is currently known about how minority groups are represented in the criminal justice system. In the theory section, we first review the origins and evolution of the intersectionality paradigm. Next, we review prominent theories that have been proposed to explain why individuals of various races and ethnicities come in contact with the criminal justice system. Lastly, in the applications section, we describe mainstream correctional assessment and treatment practices and explore their applicability to racially–ethnically diverse people, including African Americans, Hispanics, and Aboriginal/indigenous people. In closing, we suggest directions for future research and practice.

Minority Groups in the Criminal Justice System: A Brief Statistical Overview

Studies conducted across jurisdictions internationally confirm the overrepresentation of ethnic minorities at all points in the criminal justice system (e.g., Bonta, LaPrairie, & Wallace-Capretta, 1997; Bonta, Rugge, & Dauvergne, 2003; Broadhurst, 1997; Hann & Harman, 1989; Lopez & Light, 2009; Sabol, West, & Cooper, 2009). The minority groups of primary focus in these investigations include indigenous populations (particularly in Canada, Australia, and New Zealand), as well as individuals of African and Hispanic descent (both in North America and the UK). For example, while African Americans account for 13 % of the general US population, they account for 28 % of all arrests. Moreover, 39 % of all violence-related arrests and 30 % of all property-related arrests are attributed to individuals of African-American descent (Rastogi, Johnson, & Hoeffel, 2011; Snyder, 2011).

Arrest Statistics

Given that the FBI’s Uniform Crime Reports (UCR) do not publish American arrest statistics by gender, race, and ethnicity simultaneously, it is difficult to discern to what extent minority female groups such as those of African or Hispanic descent account for all female arrests. However, some scholars have examined how specific arrest statistics vary as a function of gender and race using either unpublished UCR arrest data for select cities (e.g., Chilton & Datesman, 1987; Steffensmeier & Allan, 1988) or UCR supplementary homicide reports (e.g., Cooper & Smith, 2011). Additionally, there is some self-report research that has examined how crime varies jointly as a function of gender and race (e.g., Cernkovich & Giordano, 1979).

Cooper and Smith’s (2011) analysis of the 2008 Supplementary Homicide Reports (SHR) revealed that Black females were three to four times more likely to be arrested for homicide in comparison to White females. Interestingly, early scholars theorized (usually in the absence of sound data) that Black women and White men committed crime at similar rates (Adler, 1975; Pollak, 1950; Smith & Visher, 1980; Sutherland & Cressey, 1978). While Cooper and Smith’s analysis did provide some evidence in favor of this Black women/White men crime convergence hypothesis, the observed differences were not as pronounced as earlier scholars would have predicted. Specifically, an examination of the homicide arrest gap among 18–24-year-olds—the age group responsible for the vast majority of homicides—revealed the following. The homicide arrest gap was indeed smaller between Black women (10 per 100,000) and White men (20.4 per 100,000) in comparison to White women (2.2 per 100,000) and White men (20.4 per 100,000). However, the homicide rate for White men and Black women was still vastly discrepant. In fact, White men were twice as likely to be arrested for homicide (20.4 per 100,000) in comparison to Black women (10 per 100,000). Perhaps Cooper and Smith’s most noteworthy finding is that young Black males (aged 18–24) account for the vast majority of homicides in the USA (175 per 100,000 in 2008), followed by adolescent Black males (aged 14–17; 64.8 per 100,000 in 2008).

Chilton and Datesman (1987) analyzed changes in larceny/theft arrest rates using unpublished UCR data from 1960 to 1980 for five of the largest cities in the USA. The authors made two noteworthy conclusions. First, the gender gap in larceny arrest rates did start to close over the observed 20-year period. Second, even after controlling for demographic changes in age, the convergence was largely attributable to increased larceny arrests among non-White women (who were predominantly Black in this study). The convergence was so strong that by 1980, White men and Black women evidenced similar larceny arrest rates. Although these results again support the White men/Black women crime convergence hypothesis, more research exploring this hypothesis is required before conclusive statements can be made.

Similarly, Steffensmeier and Allan’s (1988) analysis of the Pennsylvania Uniform Crime Reports illustrated that Black women were in fact arrested at similar rates to White men for some (e.g., violent) but not all crimes (e.g., property). Moreover, a comparison between Black women and White women revealed that although Black women were disproportionately more likely than White women to be arrested for certain offenses such as assault, gambling, and family-related offenses, there were no differences for other types of crimes such as minor and major property offenses. Lastly and most importantly, additional regression analyses revealed that race could not predict the observed gender differences in arrest rates.

Interestingly, Young’s (1980) analysis of victim survey data collected between 1972 and 1975 from 26 of the largest US cities only provided partial support for the belief that Black women are more involved in crime than White women. Specifically, Young reported that White female offenders accounted for 72 % of reported violent victimizations involving multiple female perpetrators, whereas Black female offenders only accounted for 44 % of the reported violent victimizations involving multiple female perpetrators. Conversely, multiple perpetrator Black female offenders accounted for disproportionately more theft-related victimizations (56 %) than their multiple perpetrator White female counterparts (28 %). Interestingly, when racial differences were examined for reported victimizations involving solo female perpetrators, no racial differences emerged. Specifically, victimization reports that involved solo female offenders were just as likely to involve Black female offenders as White female offenders, irrespective of offense type.

In conclusion, the limited available research provides mixed support for the notion that Black females are more likely to engage in crime than White females, or just as likely as White males. However, the research is dated, reliant upon nonnational samples, and characterized by statistical techniques that would most likely not meet the methodological rigor of contemporary crime trend experts (e.g., Steffensmeier, Feldmeyer, Harris, & Ulmer, 2011). Thus, the more accurate conclusion is that we simply do not know how race, gender, ethnicity, and crime intersect in contemporary society to render firm conclusions. Clearly, more research is needed to examine the hypothesized intersecting effects of race, gender, and ethnicity on crime statistics—both official and unofficial. Moreover, not only would the field benefit from qualitative analyses to help contextualize some of the observed quantitative findings, but further research that deconstructs the “non-White” female along racial, ethnic, and cultural mosaics (e.g., Asian, Native American, Hispanic) is arguably needed. Lastly, the extent to which additional variables such as social class, offense severity, and offense history may be accounting for the apparent overrepresentation of non-White individuals in the criminal justice system requires investigation.

Incarceration Statistics

Early feminist scholars such as Rafter (1985) argued that the penal system systematically discriminated against Black women. Rafter conducted a narrative review and critical analysis of the extant, albeit scant literature about women in US state prisons between 1800 and 1935. Based on a qualitative analysis of quotes and archival prison records spanning the nineteenth and early twentieth century, Rafter concluded that the system was essentially two tiered. Black women were sent to prison largely because they were perceived as more masculine, more self-centered, volatile, and dangerous compared to White women. Thus, Black women were treated more like men. In contrast, White women were treated more chivalrously, were viewed to be in need of protection, and, consequently, were sent to reformatories as opposed to prisons. Rafter, like other feminist writers of her time, remained relatively silent on whether or not the same pattern emerged for Black versus White men.

Contemporary analyses confirm that the overrepresentation of Black Americans is more pronounced in the correctional system—particularly for Black males. While Black males only account for 13 % of the general population, they comprise approximately 40 % of the male custody population and 42 % of the male death row population (Guerino et al., 2011; Rastogi et al., 2011; Snell, 2011). Although not as disproportionately overrepresented as their Black male counterparts, Black females are also overrepresented in the custody population. Specifically, Guerino et al. (2011) reported that Black females accounted for 25 % of the female custody population. Also noteworthy, the proportion of female Hispanics in custody (18 %) is slightly higher than the proportion of Hispanic females in the general population (16 %) (Ennis, Rois-Vargas, & Albert, 2011; Guerino et al. 2011).

Perhaps more disturbing is the overrepresentation of minority youth in custody. Although non-White female youth only accounted for 37 % of the 2009 US female youth court statistics, they accounted for 61 % of the 2010 female youth residential placement population. Interestingly, the same findings emerged for non-White male youth. Specifically, similar to their non-White female counterparts, non-White males only accounted for 36 % of the 2009 male youth court statistics, but they accounted for 69 % of the 2012 male youth residential placement population (Sickmund et al., 2011, 2012). The slightly larger percentage of non-White male youth (69 %) versus non-White female youth (61 %) in residential custody is most likely attributable to differences in offense severity, with male youths more likely to commit more serious crimes and hence receive harsher punishments (e.g., residential placement over probation). Thus, despite the implicit assumption that non-White women are worse off than their non-White male counterparts—termed the double jeopardy effect (being female and non-White)—this cursory examination suggests that Black men and Black women are similarly overrepresented in the correctional system; this is particularly true of the youth residential system.

Similar patterns of overrepresentation pervade the Canadian Correctional System with federally incarcerated Aboriginal men representing 21 % of the federally incarcerated male population. Even more problematic is that 32 % of all federally incarcerated women are Aboriginal. Notably, only 3 % of the Canadian population is Aboriginal (Public Safety Canada, 2012). Thus, unlike men and women of African-American descent in the USA, it does appear that a double jeopardy effect may be in place for Aboriginal women, who are even more overrepresented in the federal Canadian correctional system than their Aboriginal male counterparts. The reason for the overrepresentation of Aboriginal women at the most severe form of correctional placement (i.e., the federal system) requires more thorough investigation.

While the statistics paint a clear picture of overrepresentation of ethnic minorities in the criminal justice system, some scholars suggest that the arrest rates for Black and White individuals are converging—arguably a sign of upward social mobility for individuals of African-American descent (LaFree, Baumer, & O’Brien, 2010; Tonry & Melewski, 2008). However, Steffensmeier et al. (2011) convincingly argue that the apparent convergence is merely an artifact of the Hispanic effect. In brief, the Hispanic effect refers to the following confluence of events. Individuals of Hispanic descent are not only the largest growing segment of the American population (43 % increase from 2000 to 2010 vs. 10 % for the general population) and consequently the arrest and prison populations, but they are also most likely to self-identify as White (56 %) rather than Black (2.6 %) in the US Census (Ennis et al., 2011). Additionally, Hispanic individuals are more likely to be arrested for crime than White individuals, but less likely than Black individuals. Based largely on these trends, Steffensmeier et al. demonstrated that the observed drop in the proportion of African-American arrests for crime, particularly violent crime, is actually a function of Hispanic individuals increasingly contributing to the proportion of White arrests. Stated differently, it is not that Black individuals are being arrested with less frequency; it simply appears that way because the composition of the comparison group of “White Americans” has shifted in recent years to comprise a higher percentage of individuals of Hispanic descent. It was not possible for Steffensmeier et al. to disaggregate the data further, and consequently, the extent to which the Hispanic effect applies to females is unknown.

In sum, racial and ethnic minorities are clearly overrepresented in the criminal justice system at all levels, particularly in the correctional system. However, the extent to which ethnic minority women are disadvantaged by the hypothesized double jeopardy effect requires further investigation; it is plausible that the effect may not be evident among all groups or may be particularly pronounced among other groups (i.e., Aboriginal women in Canada). At present, there is a paucity of sound research tapping this issue. The fact that not all crime agencies record ethnicity and race, or only started to do so relatively recently, makes it difficult to discern to what extent ethnicity factors such as the Hispanic effect discussed above may actually be masking the true state of affairs for minority groups in the system.

Theory: The Origins and Evolution of the Intersectionality Paradigm

American feminist groups organized along racial identities started to emerge in the 1970s. Examples included the Chicana group (a Mexican women’s group), the Asian Sisters, and a Native American women’s organization (Women of All Red Nations; WARN) (Campbell, 2002). In 1973, the National Black Feminist Organization (NBFO) was founded in New York City. One year later, a splinter chapter of the NBFO emerged in Boston, Massachusetts. Three members of the Boston NBFO chapter wrote the Combahee River Collective, commonly known as the first black feminist manifesto (Combahee River Collective, 1977/1995). Although the authors of the Combahee River Collective did not coin the term intersectionality, scholars generally credit this document as the origin of contemporary intersectionality paradigms (Baca Zinn & Thornton Dill, 1996; Cole, 1999; Thompson, 2002).

In sum, the manifesto documents the origins and evolution of the Combahee River Collective and, most importantly, the central tenets of the collective. The Combahee River Collective personified third wave feminism—a political movement that emerged in response to disenchantment with second wave feminism or the women’s liberation movement of the 1960s. In essence, critics decried the women’s liberation movement for being hegemonic—a movement for White, upper middle-class women who viewed sexism as the ultimate form of oppression at the expense of other forms of marginalization such as race, class, and sexual orientation (Burgess-Proctor, 2006; Thompson, 2002). Notably, Thompson (2002) has convincingly argued that intersectionality actually emerged in tandem with second wave feminism rather than in response to the hegemonic shortcomings of second wave feminism.

Initially, the collective functioned as a self-help group but eventually became a political agent of change through such actions as picketing hospitals that provided inadequate care for third world communities and establishing rape crisis centers. Eventually, the collective evolved into a body of scholars, forming a self-proclaimed study group that published their philosophical stances and began speaking at women’s conferences. Although the collective originally focused on combating racist and sexist political oppression, the group gradually targeted other institutions of oppression, namely, heterosexism and economic oppression under capitalism. The following self-described statement embodies not only the Combahee River Collective but arguably, modern day intersectionality:

The most general statement of our politics at the present time would be that we are actively committed to struggling against racial, sexual, heterosexual, and class oppression and see as our particular task the development of integrated analysis and practice based upon the fact that the major systems of oppression are interlocking (Combahee River Collective, 1977/1995, p. 232).

Kimberely Crenshaw (1989/1993), a legal scholar and critical race theorist, is credited with actually coining the term intersectionality in the context of analyzing how Black women plaintiffs experience discrimination in the courts. In essence, Crenshaw’s seminal discourse argues that Black women plaintiffs experience discrimination in one of four ways: (1) in similar ways to White women, (2) in similar ways to Black men, (3) in ways which discriminate on the basis of race and gender in an additive fashion, and lastly, (4) in a multiplicative simultaneous fashion—that is, “…they experience discrimination as Black women—not the sum of race and sex discrimination, but as Black women” (p. 385). Thus, in order to fully understand the experiences of Black women plaintiffs, one must not only consider the additive effects of gender and race but also the joint effects of gender and race, or in the language of social scientists who routinely use statistics—the interactive effects of gender and race.

It is also noteworthy that some feminist scholars (e.g., Burgess-Proctor, 2006; Thompson, 2002) prefer the term multiracial feminism to intersectionality. Originally coined by Baca Zinn and Thornton Dill (1996), multiracial feminism, like intersectionality, draws attention to all possible forms of oppression such as race, class, gender, sexuality, nationality, age, as well as other “defining social characteristics” that operate in interlocking and multiplicative ways to influence one’s social location—either one of power and privilege or marginalization and oppression. In a related vein, Collins (2001) refers to the matrix of domination. Similar to intersectionality and multiracial feminism, the matrix of domination refers to how various structural axes intersect, affording a given individual either a more or less privileged status; these axes include gender, race, ethnicity, sexual orientation, age, physical ability, and other locations of inequality. Anderson and Hill Collins (2007) underscore the importance of construing the various axes as structural forces that work in concert to produce structural systems of power and inequality.

While differences between the frameworks are challenging to discern, particularly when attempting to view the issue through a psychological lens, it does seem apparent that multiracial feminists afford race preeminent status in the causal pathway to oppression, as evidenced by the following statement: “race as a power system that interacts with other structured inequalities to shape genders” (Baca Zinn & Thornton Dill, 1996, p. 322). In contrast, feminists who either explicitly or implicitly prefer the term intersectionality afford gender preeminent status in the causal pathways to oppression. Perhaps scholars who prefer the term multiracial feminism do so as a mechanism of distancing themselves from the hegemonic feminism of the 1960s—an essentialist feminism that was inherently White, upper middle-class, and heterosexual, yet purported to represent all women. It is reasonable to suggest that those who embrace the term “intersectionality” are perhaps not as concerned with the hegemonic nature of second wave feminism.

Regardless, the centrality of intersectionality/multiracial feminism/the matrix of domination to sociological, feminist, and critical race perspectives cannot be understated. In sociological fields, the race/class/gender triad is affectionately known as a “mantra” (Fine & Burns, 2003). In fact, some have argued that intersectionality/multiracial feminism is the single most important contribution of feminist studies (McCall, 2005; Risman, 2004) and that it is indeed the future of feminist criminology (Burgess-Proctor, 2006). Importantly, both predating and postdating Crenshaw’s seminal work, sociological and/or feminist scholars have engaged in ongoing discussion around the multiplicative effects of intersecting identities that may potentially compound oppression in a criminal justice context (e.g., Anderson & Hill Collins, 2007; Chilton & Datesman, 1987; Collins, 2001; Hindelang, 1981; Lewis, 1981; Young, 1980).

In contrast, the discipline of psychology has historically been steeped in understanding individual differences with little regard for the intersecting effects of variables such as gender, race, and class (Cole, 1999). As a discipline, we have typically conceptualized these factors as nuisance variables that need to be controlled to study the preeminent issue (e.g., how personality or other individual-level factors impact productivity, happiness, relationship satisfaction, and other outcomes of interest). Some criticize or, perhaps more accurately, perceive the discipline of psychology as only being capable of developing additive and/or linear models of human behavior. While this “additive/linear” characterization might have been true of our earlier works, psychology has advanced considerably in recent years. Thanks to statistical advancements by quantitative methodologists, it is becoming increasingly easier and consequently more commonplace to accurately identify statistical interactions (“intersections”) between multiple variables. Like intersectionalist theorists, psychology is seeking out multiplicative or moderator effects among variables. Moreover, although psychology has a long tradition of examining gender differences (e.g., see Hyde, 2005 for a review), it is only very recently that we have started to assess how variables such as gender and race simultaneously interact and influence various psychosocial outcomes. In fact, Cole (1999) recently wrote an insightful review article in the American Psychologist entitled, “Intersectionality and Research in Psychology.” Thus, psychology in general is just starting to incorporate intersectionality.

We now turn our attention to discussing the theories of crime that have emerged to explain female crime committed specifically by girls and/or women of African or Hispanic descent. As will become evident, while sociologically orientated researchers have actually tried to develop specific theories for female minorities, psychologically oriented researchers generally prefer theories conceptualized as being applicable to all individuals, irrespective of race, ethnicity, or gender.

Theories of Crime

Historically, theories of delinquency and crime either explicitly or implicitly neglected girls and women, in large part due to the simple fact that boys and men account for the vast majority of crime, particularly crime of a serious nature. Over the last few decades, however, observed increases in female-perpetrated crime as measured by official crime statistics as well as the disparate increase in the female incarceration rate relative to males have catapulted all forms of female deviancy to the forefront of scholarly interest. Although female offenders may no longer constitute “correctional afterthoughts”—a term coined in the mid-1980s to reflect the lack of interest in developing female responsive interventions—non-White female offenders have garnered little, if any, attention in the correctional literature. Moreover, up until recently, feminist criminological theorizing focused almost exclusively on Black girls and women. However, girls and women of Hispanic origin have garnered increasing attention in the literature as have girls and women of Aboriginal or indigenous origin, particularly in Canada and New Zealand (see Dell, Lyons, Grantham, Kilty, & Chase, in press; Maher, 1997).

Apart from Lewis (1981) who reviewed extant theories to explain why Black women come in contact with the criminal justice system, virtually none of the feminist writings of the 1970s/1980s formulated concrete and testable theoretical perspectives based on sound empirical evidence. Similarly, contemporary correctional scholars have failed to develop detailed theoretical explanations for why non-White females come in contact with the criminal justice system. Nonetheless, three general positions characterize extant theoretical perspectives: (1) the gender/race similarity hypothesis—pathways to the criminal justice system are similar regardless of gender, race, or ethnicity, (2) the gender/race difference hypothesis—pathways to the criminal justice system are different for Black women and by extension all other race/gender permutations, and (3) the double jeopardy effect—in comparison to any other gender/race combination, Black girls and women are most disadvantaged in terms of systemic discrimination practiced by the criminal justice system (Chesney-Lind, 1996; Rafter, 1985; Simpson, 1989). Relatedly, Bloom, Owens, and Covington (2002a) speculate the existence of a triple jeopardy effect, contending that poor, minority women are among the most marginalized and disadvantaged groups within the criminal justice system. Notably, the notion of double and triple jeopardy effects is consistent with the general tenets of intersectionality and multiracial feminism paradigms. However, Anderson and Hill Collins (2007) clearly delineate double/triple/quadruple jeopardy effect models as additive; they are incapable of capturing the multiplicative structural effects of key forces (e.g., race/gender/class) that collectively work to shape systems of power and privilege. Thus, double/triple “etcetera” jeopardy effect models do not fall within the matrix of domination paradigm.

Several theories have been articulated to explain crime in general (see Andrews & Bonta, 2010 for a review) and, more recently, female crime (see Blanchette & Brown, 2006 for a review). However, very few scholars have thoroughly posited how the various intersectionality perspectives described above translate into specific theories of criminal conduct. Nonetheless, two perspectives that have emerged are the masculinity model and the racialized gender stereotype expectation model. Decidedly, both perspectives espouse the race/gender difference hypothesis. A third theoretical orientation—the economic deprivation model—emphasizes the multiplicative effects of gender and social economic status. Although this perspective does not explicitly discuss the role of race and hence does not fall within the race/gender difference framework, it is described nonetheless due to its potential for adherence to intersectionality frameworks. The next two models including the personal, interpersonal, and community-reinforcement (PIC-R) perspective and the developmental life-course perspective are clear examples of the gender/race similarity position. The final model to be described, the integrated structured life-course model, combines elements from both the gender/race similarity and gender/race difference perspectives.

Masculinity Model: Act Like a Man, Get Treated Like a Man

Adler (1975) argued that improved economic opportunity coupled with the tendency of women to take on masculine traits (e.g., aggression, competitiveness) caused the apparent increase in female crime, particularly violent crime, during the 1960s and 1970s. Adler’s liberation/emancipation theory has been soundly debunked for both theoretical and empirical reasons (see Blanchette & Brown, 2006). However, the hypothesis that the acquisition of masculine traits and behaviors somehow causes females to commit crime has been specifically applied to explain why Black women engage in crime. For example, Lewis (1981) hypothesized that Black children (particularly from working-class families) were more likely to be emancipated from gender roles in that it was considered normative behavior for Black females to exude both feminine and masculine characteristics depending upon the situation. According to Lewis, Black female children are socialized to be both aggressive and passive, to be conformists and non-conformists, and to be instrumental and expressive. In contrast, she argues that White children are comparatively raised along gender dichotomies—one can be aggressive or passive, conforming or nonconforming, instrumental or expressive, but never both. The notion that differential socialization practices between Black and White families exist and that they hold explanatory power for Black female criminal conduct has been popularized by some (Lewis, 1981) while heavily critiqued by others (Chilton & Datesman, 1987; Hemmons, 1980; Hershey, 1978). Although clearly more research is needed about how hypothesized differential socialization practices may vary across race and ethnicity, the answer to this question may be irrelevant. In contrast, understanding how perceived stereotypes (genuine or not) and related constructs such as personal beliefs and cognitive schemas (see Devine & Elliot, 1995; Fiske & Taylor, 1991) may lead the criminal justice system to differentially and unfairly treat minority girls and women warrants considerably more attention.

Racialized Gender Role Expectations Model

Some hold that the overrepresentation of ethnic minorities among offender populations is attributable to biased systemic practices at all junctures of criminal justice processing (Aboriginal Justice Inquiry of Manitoba, 1991; Harris, 1999; Rudin, 2006; Welch, 2007). For instance, there is typically a higher police concentration in communities comprised exclusively or predominantly of visible minorities (e.g., in Aboriginal communities), rendering the detection of criminal infractions of any kind more likely to occur (Aboriginal Justice Inquiry of Manitoba, 1991; Rudin, 2006). Similarly, for crimes of similar type and severity, visible minorities are routinely confined to police custody for longer periods and administered harsher legal sanctions compared to White offenders (Harris, 1999; Rudin, 2006; Welch, 2007). Gender was not examined within these studies.

In a related vein, racial profiling has been underscored as a factor that has fuelled the overrepresentation problem, whereby police officers selectively and proactively enforce laws on the basis of an individual’s race. This form of bias was apparent in a recent study focusing on police response to traffic violations in Cincinnati, Ohio (Ridgeway, Schell, Riley, Turner, & Dixon, 2006). A random sample of 325 traffic surveillance videos illustrated that compared to White drivers, Black drivers were more likely to be stopped for technical violations such as equipment malfunction (10 % vs. 19 %). Moreover, Black drivers experienced more intensive scrutiny during these stops, with police interactions lasting nearly 20 % longer than those involving White drivers. Finally, Black drivers were significantly more likely than White drivers to be subjected to vehicle searches (10 % vs. 3 %) and to be questioned proactively about drug or weapon possession (22 % vs. 7 %). Again, this study did not examine the interaction between gender and race.

Studies that have explored how minority girls and women may experience systemic biases from the criminal “justice” system have focused predominantly on the role of racialized gender role expectations and stereotypes. The chivalry hypothesis has been proposed to explain the underrepresentation of females in the criminal justice system relative to males; briefly stated, the proposition is that females are charged and convicted less often than males because the criminal justice system is more paternalistic and hence more lenient toward females (Moulds, 1980). However, some have argued that the chivalry effect only applies to upper class, White females who act in stereotypically “female” appropriate ways upon arrest—that is, crying and passive versus hostile and aggressive (Lewis, 1981; Rafter & Natalizia, 1981). Additional “White girl” stereotypes include being nonthreatening, treatable, maternal, and in need of protection (Bickle & Peterson, 1991; Franklin & Fearn, 2008; Gaarder, Rodriguez, and Zatz (2004). Similarly, theory and research suggest that Hispanic women are viewed as dependent and submissive, family oriented, domestic, and sexual (Cofer, 1993; Espin, 1984; Segura & Pierce, 1993). In contrast, some evidence suggests that Black girls are viewed as independent, aggressive, loud, pushy, rude, sexual, unfeminine, crime-prone, and deserving of violence (Collins, 2004; Miller, 2008; Moore & Hagedorn, 1996; Sinden, 1981).

Interestingly, Visher’s (1983) empirical study did suggest that race did in fact nullify the female chivalry effect in the context of whether or not police would make an arrest. Specifically, Visher reported that police chivalry existed toward White females but not Black females, even after controlling for legal variables. She hypothesized that Black females are treated more harshly than their White counterparts because they are less apt to display expected or traditional gender behaviors when they encounter a White, male police officer: “Although chivalry may be alive and well for White women, it appears to be dead (if it ever existed) for Blacks” (p. 614).

To date, the limited research involving Hispanic girls has been inconclusive; while some studies have found Hispanic girls to be disadvantaged (Miller, 1994), others have found that the criminal justice system treats them more leniently (Gaarder et al., 2004). Moreover, Simpson’s (1989) review of sentencing practices revealed that the overarching factor to influence sentencing practices after controlling for offense severity and criminal record was the concern for protecting nuclear families. In sum, Simpson concluded that lenient sentences were more common among offenders with families—both at the pretrial release and post-conviction stage. The finding was similar irrespective of race.

Moore and Padavic (2010) recently examined the hypothesis that court officials hold certain stereotypes about girls and women from different races and ethnicities that ultimately underlie sentencing disparities between similarly offending White, Black, and Hispanic females. The authors conducted a sophisticated quantitative analysis of all Black, Hispanic, and White girls (aged 10–18) processed by the Florida Department of Juvenile Justice (FLDJJ) in 2006, totaling over 18,000 girls. Three important findings emerged: (1) Black girls received more severe dispositions than White girls even after controlling for offense severity, prior record, and age; (2) Hispanic and White girls received similar dispositions after similar controls were in place; and, perhaps most importantly, (3) the relationship between race-ethnicity and disposition severity is moderated by legal variables—prior record and current offense severity. Specifically, the authors found that the criminal justice system afforded leniency to White girls only if they evidenced minor to average levels of offense severity and criminal history. In contrast, when White girls displayed above average levels of offense severity and criminal history, the criminal justice system meted out particularly severe dispositions. Unlike their White counterparts, the same pattern did not emerge for Black or Hispanic girls. Black and Hispanic girls who exceeded the “accepted threshold” of violence were not penalized any more harshly. Thus, the research supports the finding that stereotypes (or some related construct) may play a role in judicial processing. However, unexpectedly, the results also support the notion that stereotypes can work in favor of minority females. Simultaneously, stereotypes can work against majority females, namely, White females who arguably are situated above both Black and Hispanic females in the matrix of domination. This study is methodologically strong and an excellent first step to elucidating how stereotypes (and possibly related constructs) can work for or against girls and women of various ethnic and racial origins. The extent to which the results would be influenced by additional factors such as class as well as race-ethnicity of judicial officials requires examination.

Economic Deprivation Model

Scutt (1975) was one of the first researchers to propose an economic argument as a means of explaining the overrepresentation of Black women in the criminal justice system. In short, Scutt argued that because Black women are more likely than White women to be the family economic providers, they are under greater pressure to succeed and provide economically for their families; hence, Black women are more likely to commit crime if required. Similar perspectives were also shared by Adler (1975). Recent US Census Bureau 2010 (see Lofquist, Lugaila, O’Connell, & Feliz, 2012) statistics confirm Scutt’s original assertion. Black women account for the greatest proportion of single female-headed households (30.1 %), followed by Hispanic women (19.2 %), and lastly, non-Hispanic White women (9.2 %). Similarly, Black women account for the greatest proportion of single female-headed households with children (17.4 %), followed by Hispanic women (12.1 %), and lastly, non-Hispanic White women (4.7 %) (Lofquist et al., 2012). However, the link between “single, Black female-headed household” and “enhanced crime vulnerability” is not as clear (e.g., Velez, Krivo, & Peterson, 2003).

Perhaps one of the most well-developed contemporary theories created to explain female crime in general has particular merit for minority women—the integrated liberation and economic marginalization theory. Hunnicutt and Broidy’s (2004) integrated liberation and economic marginalization theory asserts that changing gender roles initially brought on by the women’s liberation movement has unintentionally increased the “economically marginal roles” of women by pushing them further into the economic fringes of society. Specifically, the women’s movement has perpetuated the belief that women have achieved greater financial advantage, a view that has caused society to overlook evidence to the contrary. Instead, the authors argue that divorce, a consequence of liberation, has actually increased the economic instability of women because there are now more single female-headed households with dependent children. The women’s movement having increased female expectations in regard to status and wealth, served to create a set of circumstances under which women are more likely to adopt illegitimate means to achieve wealth and power.

Hunnicutt and Broidy (2004) directly tested the theory using aggregate data pooled across ten countries. Based on a time series analysis that employed aggregate level independent variables (e.g., the number of divorces per 100,000 married persons) and an aggregate level dependent variable (crime rate per 100,000) they concluded “that liberation does indeed stimulate crime among women, to the extent that changing roles and expectations of gender equality further marginalize women” (p. 150). However, whether or not this finding may be explained by other individual-level factors remains unknown. Moreover, the authors did not specifically examine if the results varied as a function of race or ethnicity.

Personal, Interpersonal, and Community-Reinforcement Theory

The personal, interpersonal, community-reinforcement (PIC-R) theory (Andrews & Bonta, 2006, 2010) is a multidisciplinary perspective that integrates biological, sociological, cultural, familial, interpersonal, personal, and situational variables. Grounded heavily in social learning theory [i.e., crime is learned through the mechanisms of differential reinforcement, operant conditioning, and imitation (Akers & Jensen, 2003)], PIC-R posits that individuals commit crime when the rewards for doing so exceed the costs. Various factors influence the balance of rewards and costs ranging from highly proximal factors located in the immediate situation (e.g., opportunity) to more distally orientated factors (e.g., political/economic/cultural influences). The theory categorizes these factors along four dimensions: situational, personal, interpersonal, or community. Situational factors include opportunities (e.g., temptations), stressors (e.g., negative affect), facilitators (e.g., psychotic state), and disinhibitors (e.g., substance abuse), while personal factors include antisocial cognitions, history of antisocial behavior, antisocial personality, and biological factors. Interpersonal factors include variables such as antisocial associates and family functioning, while the community dimension encompasses factors such as the quality of one’s neighborhood. Although each category differs as a function of temporal proximity to the immediate situation, each one influences the probability that an individual will find him or herself in a situation conducive to committing crime. Additionally, each category also determines whether or not an individual will develop an internal dialogue consistent with definitions favorable towards criminal conduct (e.g., rewards exceed the costs) and, consequently, commit the criminal act.

PIC-R concurs with sociological perspectives that suggest that broad-based societal/structural factors are important; however, these factors are only deemed relevant to the extent that they control the distribution of rewards and costs within a social system. Further, the theory is primarily concerned with explaining individual differences in criminal conduct. With this in mind, Andrews and Bonta (2010) have identified which empirically determined risk factors account for the greatest individual variation in criminal conduct.

Each risk factor has been assigned to one of three predictive accuracy levels. The first and most powerful set includes (1) antisocial cognitions (attitudes, beliefs, values that support criminal conduct), (2) antisocial associates, (3) a history of antisocial behavior, and (4) antisocial personality (including indicators such as restless energy, adventuresomeness, impulsiveness, poor problem-solving skills, hostility, and callousness). Risk factors placed in the middle range include (5) substance abuse, (6) marital/family factors, (7) school/employment achievement, and (8) leisure/recreation. Collectively, these major predictors of criminal conduct are termed the Central Eight. In contrast, risk factors in the low range of predictive validity include lower-class origins, low verbal intelligence, and personal distress. Notably, with the exception of antisocial behavior, the Central Eight are also known as criminogenic need factors. Theory and empirical evidence indicate that appropriate correctional interventions can ameliorate criminogenic need factors, which, in turn, reduces criminal recidivism (Andrews & Bonta, 2010).

PIC-R is presented as a general theory that can account for individual differences in criminal conduct irrespective of gender, class, or ethnic origin. While gender is classified as a distal, personal variable that shapes both the person and the immediate situation, it is not central to the model. Although there is indirect evidence that supports the applicability of PIC-R to female offender samples the evidence is somewhat contradictory and mixed regarding African Americans and Hispanics (Andrews et al., 2012; Andrews & Bonta, 2010). Moreover, direct as well as indirect tests of the theory that simultaneously consider race and gender are largely absent.

Life-Course Perspectives

Developmental or life-course perspectives have emerged independently within sociology (e.g., Sampson & Laub, 1990, 1993) and psychology (Farrington, 2005; Loeber & LeBlanc, 1990; Moffitt, 1993; Patterson, 1992; Patterson & Yoerger, 1997). Life-course theorists assert that risk factors vary as a function of age or developmental stage. Specifically, during childhood, parental factors play a preeminent role in explaining criminal conduct whereas adolescent peer association and school attachment assume greater importance during adolescence. Finally, factors such as employment stability and marital attachment become increasingly relevant during adulthood. Life-course theorists are also recognized for their reliance on longitudinal research designs that examine how changes in developmentally salient risk factors translate into changes in criminal conduct. To date, only one of the developmental theorists (i.e., Moffitt, 1993) has explicitly addressed theoretical issues pertaining to female offending. Consequently, only Moffitt’s work is reviewed.

In 1993, Terrie Moffitt published a seminal article positing the existence of two distinct offender groups or developmental taxonomies: life-course-persistent and adolescent-limited. Each group has unique etiological pathways as well as differential antisocial trajectories that vary in terms of onset, severity, and desistence. According to Moffitt, antisocial behavior in life-course-persistent offenders emerges early in life. It results from both internal and external factors. Specifically, individuals considered high risk (i.e., inherited or acquired neuropsychological deficits resulting in mild cognitive impairment, difficult temperament, or hyperactivity) and raised in a high-risk social environment are likely to become life-course-persistent offenders. The high-risk social environment initially includes factors such as inadequate parenting and poverty but also incorporates additional developmentally relevant risk factors (e.g., peers and teachers) as the child begins to age. The theory argues that a series of negative, bidirectional interactions occurring between the high-risk child and high-risk environment eventually culminate in the development of a “disordered personality” characterized by persistent physical aggression and antisociality.

In contrast, Moffitt (1993) asserts that the onset of antisocial behavior in adolescent-limited individuals coincides with the onset of puberty. For this group, the primary causal factor accounting for the onset of antisocial behavior is the maturation gap, defined as a period “…when otherwise healthy youngsters experience dysphoria during the relatively roleless years between their biological maturation and their access to mature privileges and responsibilities” (Moffitt & Caspi, 2001, p. 356). The theory asserts that it is the norm rather than the exception for youths during this period to mimic the behaviors of life-course-persistent offenders as means of achieving respect from peers and autonomy from parents. Eventually, adolescent-limited individuals desist when they reach real maturity and are able to successfully pursue conventional ambitions. Unlike life-course-persistent individuals, they are able to successfully transition back to a prosocial lifestyle given that their behavioral repertoire already contains the necessary skill set to function in a prosocial world (e.g. educational attainment, positive interpersonal functioning). However, severe addiction or the existence of criminal records may impede the successful transition. Few individuals will meet the life-course-persistent criteria. However, those who do will account for a significant amount of crime, particularly serious crime, well into adulthood. In contrast, the theory asserts that crime committed by the adolescent-limited behavior is commonplace, minor, nonviolent, and relatively temporary, typically ending before early adulthood.

Moffitt (1994) hypothesizes that this developmental typology applies to both genders. She further asserts that observed gender differences in antisocial behavior exist largely because boys are more likely than girls to become life-course-persistent offenders. She suggests that this is the case because boys are more likely to be high-risk children (e.g., hyperactive, cognitive deficits, delayed speech, learning disabilities). The theory does not speculate in terms of whether or not environmental risk factors vary as a function of gender or race.

In regard to adolescent-limited behavior, Moffitt and Caspi (2001) suggest that boys and girls experience the same dysphoria associated with the maturation gap. Consequently, both genders are expected to mimic life-course-persistent offenders as a means of achieving independence from parents and respect from peers. While the theory posits that girls’ delinquency will be considerable, it predicts that female delinquency will be less frequent than male delinquency for two reasons. First, girls will have reduced access to antisocial male role models as a consequence of gender-segregated male antisocial groups. Second, girls will be less likely to perceive antisocial behavior as reinforcing due to the greater risk of personal risk or injury (e.g., pregnancy, injury from dating violence) and, as a consequence, will be less likely to adopt an antisocial pathway.

Interestingly, Belknap (2001) has categorized development perspectives, including Moffitt’s (1993) research, as “pro-feminist.” Moreover, with the exception of one study (Kratzer & Hodgins, 1999), the generalizability of Moffitt’s typology to females is mounting (Caspi, Lynam, Moffitt, & Silva, 1993; Moffitt & Caspi, 2001; Moffitt, Caspi, Rutter, & Silva, 2001). Unfortunately, developmental life-course theorists have paid little attention to race or ethnicity in general nor have they paid any attention to how race, ethnicity, and gender may intersect resulting in different life-course trajectories.

Structured Life-Course Model/Increased Exposure to Risk Model

Lynch (1997) describes the structured life-course perspective. This model nicely integrates traditional developmental life-course perspectives that emphasize individual risk factors with radical criminology that underscores structuring forces such as race, gender, and class. According to Lynch, one of the central tenets of radical criminology is that factors such as race, gender, and class not only influence how individuals behave, but also how the criminal justice system responds. Moreover, race, gender, and/or class can either individually or in combination improve or limit an individual’s access to economic and political power. This in turn shapes one’s available options and choices. Notably, in Andrews and Bonta’s (2010) PIC-R model, economic, political, and social forces also “shape” available rewards and costs—the posited drivers of criminal behavior. Beyond stating that “wealthy, white, males have access to the greatest number of choices in their life course, while poor, minority women would appear to have the fewest choices” (p. 7), Lynch does not provide an in-depth analysis of how the structured life-course model may inform theories of crime in general, or for ethnic minority females more specifically. Further, explicit or even implicit tests of the structured life-course perspective are sparse. However, Groves and Frank (1993) argue that individuals afforded less choice (e.g., assumedly poor, minority women) should be considered less accountable by the criminal justice system than those afforded more choice (e.g., assumedly wealthy, White men).

Closely related to the structured life-course perspective is the increased exposure to risk model. This model posits that marginalized groups may simply experience elevated levels of vulnerability across a range of criminogenic factors. It is this elevated level of vulnerability which in turn explains the overrepresentation of ethnic minorities in the criminal justice system (Fite, Wynn, & Pardini, 2009; Rugge, 2006). The increased exposure to risk model suggests that due to social inequalities faced by marginalized groups (e.g., lower levels of parental supervision, poverty, poor health care, fewer prosocial opportunities within the community), individual-level risk factors accumulate and increase one’s propensity to become enmeshed in a criminal lifestyle.

There is indeed evidence to support the contention that the principal criminogenic needs identified in aggregate samples and outlined in the Central Eight not only generalize to ethnic minorities, but are also more prevalent among ethnic minorities (Kempf-Leonard, 2007; Piquero, Moffitt, & Lawton, 2005). In Canada, for example, Aboriginal offenders consistently exhibit a greater magnitude of risk in the areas of substance abuse and family functioning compared to non-Aboriginal offenders. Specifically, Aboriginal offenders are more likely to have been abused or neglected, to have received custodial placements, and to have justice-involved family members (Hull, 2005; Rugge, 2006; Trevethan, Moore, & Rastin, 2002). In terms of education, Aboriginal offenders are significantly less likely than non-Aboriginal offenders to complete high school; specifically, upon admission to federal custody in Canada (i.e., reserved for offenders receiving a sentence of over 2 years), 26 % of Aboriginals have a level of education below Grade 8, compared to 18 % of non-Aboriginals (Trevethan et al., 2002). Not surprisingly then, Aboriginal offenders have greater ensuing difficulties in the area of employment, with 75 % of the latter being unemployed upon admission to federal custody compared to 66 % of non-Aboriginal offenders. Similar if not greater racial disparities in the prevalence of risk factors exist in samples of offenders under community supervision (Bonta et al., 1997).

Increased exposure to early risk factors extends across several ethnic minority groups. Based on a sample of 202 White and 279 Black youths, Fite and colleagues (2009) examined racial differences in the prevalence of multiple risk factors measured in childhood (Grade 2) to determine whether these could account for differences in prospective arrest rates at age 10–17. As predicted and consistent with previous research (e.g., Kempf-Leonard, 2007; Leiber, 2002; Piquero et al., 2005; Sampson, Morenoff, & Raudenbush, 2005), in contrast to their White counterparts, Black youths had significantly higher arrest rates across all offense categories including violent offending, theft, and drug-related crime (31.2 % vs. 56.6 %). Also as hypothesized, the 14 risk factors under consideration were pertinent to both racial groups in predicting subsequent arrest. That stated, race was significantly related to 10 of these 14 criminogenic variables; in other words, compared to White youths, Black offenders presented a significantly higher degree of risk across multiple domains. Factors on which risk level was particularly elevated for the African-American subsample were conduct problems (e.g., fighting, destruction of property), low academic achievement, family dysfunction (e.g., poor parent–child communication), association with antisocial peers, and community (e.g., disadvantaged neighborhood). However, noticeably absent is research examining the multiplicative effects of race and gender.

Correctional Applications

Although female-centered correctional research is growing exponentially, our knowledge remains infinitesimal relative to what we know about male offending and male-based effective correctional intervention strategies. In the same way that second wave feminism has been criticized for focusing on the oppression of White, middle-class women, correctional psychology feminism has been similarly afflicted. While we can unabashedly claim that female offenders are no longer “correctional afterthoughts,” the same cannot be said for female responsive programs that also attend to race, ethnicity, and culture. Although there are exceptions (e.g., Aboriginal strategy for women offenders in Canada), the norm has been to address questions of gender and race independently rather than simultaneously in the realm of female offender risk assessment and programming. Consequently, although considerable theorizing has emerged particularly from feminist criminologists (e.g., Chesney-Lind, 2006) and even from standard “gender-neutral” psychological or mainstream theories of crime and intervention (Andrews & Bonta, 2010), little empirical evidence actually exists regarding issues pertinent to correctional psychology. The following is a brief overview of how an offender is generally processed through the criminal justice system following a conviction, with particular attention accorded to issues of risk assessment and treatment. Whenever possible, we infuse our review with the available literature on gender and race.

Risk Assessment

Risk assessments are conducted at various junctures of criminal justice processing in an effort to determine an offender’s propensity to engage in a future criminal act (Andrews & Bonta, 2010; Bonta, 1996). Beyond a strict determination of risk to reoffend, information yielded from risk assessment protocols is integral to decisions regarding offender sentencing, security classification in custodial settings, requisite levels of supervision, the prioritization of treatment targets, and institutional release. Accordingly, the validity of risk assessment tools across the different populations to which they are being applied is paramount to the appropriate processing of the offender and to the preservation of public safety (Bonta, 1996).

Prior to the development of formally structured risk assessment protocols, professionals working in forensic settings relied exclusively on clinical judgment to render decisions regarding an offender’s recidivism risk (Hanson & Morton-Bourgon, 2009). This method of prediction is entirely subjective in that no parameters are imposed on the factors under consideration, or on the manner in which these factors are combined to produce an ultimate judgment of risk. Given its complete reliance on professional discretion, clinical judgment has often been criticized for its lack of transparency, replicability, and accountability (Andrews, Bonta, & Wormith, 2006; Bonta, 1996).

In response, the development and implementation of formal risk assessment tools have burgeoned over the last few decades (Andrews et al., 2006; Hanson & Morton-Bourgon, 2009). In stark contrast to clinical prediction, actuarial tools are highly structured protocols that feature a predefined set of items that are empirically linked to recidivism; moreover, such tools incorporate algorithms for combining items into a final estimate of risk (e.g., Quinsey, Harris, Rice, & Cormier, 2006). Alternatively, structured assessment protocols that do not strictly abide by all criteria inherent in the actuarial method have recently been distinguished and denoted as mechanical tools (Hanson & Morton-Bourgon, 2009). The construction of the latter is based on theoretical considerations and a comprehensive review of the literature; that is, item inclusion is not strictly empirically driven. Another point of distinction from actuarial tools is that mechanical methods do not typically ascribe a quantitative probability of recidivism to one’s total score.

Some researchers and practitioners advocate a compromise between actuarial/mechanical and clinical methods and, as such, endorse a “hybrid” approach to risk assessment termed structured professional judgment (e.g., Webster, Douglas, Eaves, & Hart, 1997). Tools grounded in this latter approach are structured in the sense that item inclusion is dictated by a review of the relevant literature. However, the ultimate judgment of risk is based on professional discretion—that is, the assessor subjectively combines item ratings and assigns the offender to a descriptive category of risk (e.g., low, moderate, high).

Which Method of Risk Assessment Is Preferable?

In gauging the relative merit of each aforementioned assessment approach, the measure of greatest interest is predictive validity; that is, the degree to which the protocol correctly predicts future criminal conduct. Spanning over 50 years, the vast body of research comparing actuarial to clinical prediction methods suggests that in complex decision tasks—particularly those conducted in forensic and medical settings—the predictive validity associated with actuarial prediction exceeds that of clinical judgment (e.g., Andrews et al., 2006; Bennell, Jones, & Taylor, 2011; Dawes, 1976; Grove & Meehl, 1996; Hanson & Morton-Bourgon, 2004, 2009; Meehl, 1954; Swets, Dawes, & Monahan, 2000). In addition to the lack of transparency associated with clinically based predictions, unguided human judgment is typically hindered by limitations inherent in one’s ability to effectively process and weigh multiple decision cues simultaneously. Notably, this limitation holds true regardless of the assessor’s level of expertise (Clarke, Nguyen, & Sweller, 2006; Dawes, 1976).

In the area of criminal risk assessment specifically, Hanson and Morton-Bourgon (2009) recently conducted a meta-analytic review comparing the relative levels of predictive accuracy achieved by various methods of assessment. Approaches under investigation included clinical judgment, structured professional judgment, mechanical, and actuarial. The authors considered 118 distinct samples of sexual offenders followed up over an average period of 70 months. In brief, unstructured clinical judgment was the least accurate method of risk assessment across all recidivism outcomes (e.g., general, violent, and sexual). Structured professional judgment tools offered a relative improvement in predictive accuracy, but the latter did not perform as well as mechanical and actuarial approaches. Both mechanical and actuarial methods achieved similar levels of accuracy in the prediction of sexual recidivism. However, consistent with previous literature in the area of risk assessment (e.g., Mossman, 1994), actuarial tools yielded the highest levels of predictive accuracy across all outcomes.

Theoretical Model Guiding Current Risk Assessment Practices

In sum, it is virtually uncontested that risk assessment methods with some degree of formalized structure achieve greater levels of predictive accuracy than unguided clinical judgment (e.g., Hanson & Morton-Bourgon, 2009). As such, the administration of structured risk assessment protocols has become normative in forensic contexts (Andrews et al., 2006). However, there is a paucity of research examining the extent to which currently adopted risk assessment tools genuinely transcend demographic lines. This gap in the literature is largely attributable to the paradigm governing the field of correctional psychology.

Namely, the principles underlying contemporary correctional assessment and intervention are captured through the operational derivative of PIC-R theory (Andrews & Bonta, 2010)—these are the risk, need, and responsivity (RNR) principles. In brief, the risk principle holds that the most intensive interventions should be reserved for the highest risk offenders. The need principle, in turn, posits that effective correctional treatment should prioritize dynamic risk factors that evidence a strong empirical relationship to recidivism (i.e., criminogenic needs). As outlined earlier in our brief description of PIC-R theory, the indicators most predictive of criminal conduct constitute the Central Eight (e.g., antisocial cognitions, antisocial associates, etc.) (Andrews & Bonta, 2010; Gendreau, Little, & Goggin, 1996). As such, it is reasonable that these factors would occupy a prominent place in the context of risk assessment so as to identify the level of need apparent within each area.

The third core principle of effective correctional intervention is the responsivity principle, under which demographic characteristics such as gender, ethnic background, and social class are encompassed. Specifically, such contextual variables are said to merit consideration in program delivery so as to optimize treatment efficacy; however, they are not explicitly factored into risk assessment protocols because they are not viewed as major predictors of criminal outcome (Andrews et al., 2006, 2012). Rather, variables such as gender and race are said to exert their influence more proximally through the Central Eight.

There is a vast body of literature to support the validity of the general RNR principles and derived tools across a variety of demographic groups (e.g., Andrews et al., 1990, 2012; Andrews & Bonta, 2010; Gendreau et al., 1996; Jung & Rawana, 1999; Laub & Sampson, 2003; Olver, Stockdale, & Wormith, 2009; Rettinger & Andrews, 2010; Snowden, Gray, & Taylor, 2010). Although adherence to RNR may indeed apply generally, most empirical evidence attesting to the universal relevance of individual risks/needs has traditionally rested on samples that were not disaggregated by gender or by race (Blanchette & Brown, 2006). Given this methodological shortcoming, the degree to which the Central Eight are germane to demographics beyond the Caucasian male offender remains a contentious issue. Perspectives opposing the dominant correctional theory, or at least facets thereof, hold that females and ethnic minorities have additional, potentially unique criminogenic needs that warrant consideration in the context of risk assessment so as to ensure that these individuals are processed in an equitable fashion (e.g., Blanchette & Brown; Blanchette & Taylor, 2007; Boer, Couture, Geddes, & Ritchie, 2003).

Given that the correctional literature has focused almost exclusively on issues of race and gender in isolation, the following subsections are organized in kind. Focusing on gender, we first provide a brief review of the empirical debate between those advocating gender-neutral versus gender-specific (or at the very least, gender-responsive) approaches to risk assessment. Second, we turn to the issue of race, exploring universal versus demographically specific perspectives. Finally, in keeping with intersectionality theory, we present the available literature that bridges issues of gender and race in the context of risk assessment.

Gender and Risk Assessment

Administered to males and females alike, the Level of Service/Case Management Inventory (LS/CMI; Andrews, Bonta, & Wormith, 2004)Footnote 2 and the Youth Level of Service/Case Management Inventory 2.0 (YLS/CMI 2.0; Hoge & Andrews, 2011)Footnote 3 stem from the RNR literature and are considered the gold standards in risk assessment. They serve a dual purpose of estimating risk for general recidivism and identifying viable treatment goals. Much evidence has been amassed to support the gender neutrality of the RNR principles and the applicability of these derived risk assessment tools to female offenders (Andrews et al., 2012; Dowden & Andrews, 1999b; Folsom & Atkinson, 2007; Rettinger & Andrews, 2010; Simourd & Andrews, 1994; Smith, Cullen, & Latessa, 2009).

For example, Smith and colleagues 2009 conducted a meta-analytic review of the LSI’s predictive accuracy for female offenders. Extracting results from 25 empirical studies, their sample consisted of 14,737 female offenders and 27 effect sizes (i.e., measure of association between total score and recidivism). Consistent with similar meta-analyses (e.g., Gendreau et al., 1996; Lovins, Lowenkamp, Latessa, & Smith, 2007; Olver et al., 2009), the measure yielded a respectable average effect size of .35 for females, which is statistically equivalent to the level of predictive validity obtained for males.

Also lending support to the gender-neutral perspective is a more recent investigation by Andrews and colleagues (2012). This study entailed the aggregation of 5 independent data sources, collectively achieving a sample of 354 females and 2,069 males. Supporting the position that RNR-grounded tools generalize to female offenders, the LS/CMI composite score was actually found to be more highly predictive of recidivism for females (AUC = .83) than for males (AUC = .74).Footnote 4 In addition, each of the eight domains featured on the tools—corresponding to the Central Eight—was found to be gender-neutral in predicting criminal outcome.

Despite ample evidence in support of the gender-neutral position, feminist scholars maintain that “gender-neutral” tools fail to capture criminogenic needs that may be particularly salient to girls and women—namely, factors related to trauma, mental health, self-esteem, and the quality of one’s relationships with family, intimate partners, and dependent children (e.g., Blanchette & Brown, 2006; Van Voorhis, Wright, Salisbury, & Bauman, 2010). Contrary to males who are taught to strive for individuation, relational-cultural theory and the feminist pathways literature collectively propose that healthy female development is inexorably tied to the quality of one’s relationships (e.g., Chesney-Lind, 2006; Daly, 1992; Jordan & Hartling, 2002; Miller, 1986). As such, early relational disruptions are said to catalyze a woman’s trajectory into crime and bear particular relevance on female offending patterns; these disruptions may begin with abuse or neglect in one’s family of origin and persist into adulthood in the context of unhealthy romantic attachments.

Although the feminist-grounded literature has largely been based on anecdotal accounts and qualitative research, Salisbury and Van Voorhis (2009) recently produced the first peer-reviewed, quantitative test of interrelationships among potentially unique initiating and sustaining factors in women’s criminal involvement. Applying path analysis to a sample of 313 adult females on probation in the state of Missouri, results suggest three distinct gendered trajectories into crime, each predicting recidivism (i.e., incarceration) within a 2-year follow-up period: (1) The childhood victimization model entails abuse sustained in childhood leading to mental health issues (namely, depression and anxiety), and subsequent attempts to achieve affective numbness through substance abuse; (2) the relational model suggests intimate relationship dysfunction as an initial catalyst, leading to an erosion of self-efficacy and further victimization, which in turn culminate in mental health issues and substance abuse; and finally, (3) the social and human capital model reflects challenges in the areas of education against a backdrop of familial and intimate relationship dysfunction. In turn, these deficits result in lowered self-efficacy coupled with employment and financial difficulty.

Salisbury and Van Voorhis (2009) offer a sound contribution in their efforts to supplement the qualitative pathways research with quantitative evidence. Indeed, their investigation does substantiate certain claims of feminist theorists regarding the etiology of female offending—specifically, that a conglomeration of factors tied to victimization, low self-efficacy, mental health issues, and poverty culminate in a criminal lifestyle among women (Bloom, Owen, & Covington, 2002b; Chesney-Lind & Shelden, 2003; Daly, 1992). Regrettably, the uniqueness of these pathways to women cannot be ascertained given the absence of a male comparison group; logically, one cannot conclude that women follow unique trajectories into crime without concurrently examining the extent to which these pathways generalize to male offenders. This limitation notwithstanding, Salisbury and Van Voorhis provide evidence for the criminogenic relevance to women of factors residing outside the Central Eight.

A question of greater practical relevance is the additional value of considering such gender-responsive items in the assessment of female offenders above and beyond the Central Eight. In a subsequent investigation, Van Voorhis and her colleagues (2010) explored this issue by examining the incremental predictive contribution of a collection of purportedly gender-responsive factors to gender-neutral models. In a large-scale investigation, they considered prison, probation, and prerelease samples of adult females across four American states, with sample sizes ranging from 134 to 313. The assessment of gender-neutral domains was based on the LSI-R, while the assessment of gender-responsive factors was achieved via two comprehensive supplements that entailed the administration of paper-and-pencil surveys, as well as an interview with each participant. Constructs were assessed through validated scales wherever possible (e.g., Rosenberg’s Self-Esteem Scale), and each domain featured a breadth of items intended to reflect the construct of interest as conceptualized in the feminist-based literature. The authors ultimately aimed to assess the incremental predictive validity of this gender-responsive supplement when used in conjunction with the currently adopted gender-neutral protocol.

Results of their analyses support the relationship of several gender-responsive scales to criminal outcome (Van Voorhis et al., 2010). Namely, the most highly predictive gender-responsive factors included current mental health needs, family support, parental stress, child abuse, and adult victimization. Most notably, the overall gender-responsive supplement (and subsets of these factors) did offer incremental predictive validity over the gender-neutral protocol. Despite the absence of a male comparison group, the seminal work of Van Voorhis and collaborators lends support to the position that modifications to current risk assessment protocols may be warranted for justice-involved females. Specifically, rather than hastily discount hypothesized gender-specific domains as responsivity factors, gender-responsive constructs assessed thoroughly and as intended can collectively enhance the predictive validity of gender-neutral tools with women offenders.

Race and Risk Assessment

Although not as fervent as the gender-neutral versus gender-specific debate, there is a similar divide on the issue of race. While RNR proponents contend that theories of criminal conduct and associated risk assessment strategies transcend racial lines (e.g., Andrews & Bonta, 2010; Bonta et al., 1997; Jung & Rawana, 1999; Schwalbe, 2009), others argue that at least a degree of demographic specificity is required in the construction and implementation of risk assessment protocols (e.g., Blanchette & Taylor, 2007; Blanchette, Verbrugge, & Wichmann, 2002; Boer et al., 2003; Hann & Harman, 1993).

Compared to White offender samples, an elevated level of risk is indisputably apparent among ethnic minorities (e.g., Fite et al., 2009; Moffitt, 1994). Albeit an important issue in its own right and pertinent in the context of devising interventions, the primary goal in the context of risk assessment is not to explain these disparities; rather, it is simply to determine the degree to which observed differences in recidivism rates between groups correspond to differences in magnitude of risk among a series of putative criminogenic factors. In other words, if the greater recidivism rates observed in ethnic minorities are attributable to elevated risk as gauged by currently adopted risk assessment tools, then one might argue that these measures are in fact appropriate for use across racial subgroups.

A collection of investigations have supported the position that risk assessment tools developed on normative, majority samples can successfully predict re-offending among ethnic minorities (e.g., Olver et al., 2009; Schwalbe, 2009; Snowden et al., 2010). For example, Bonta et al. (1997) sought to determine whether the Manitoba Risk-Needs Scale used in the context of probation in Manitoba, Canada, would accurately predict recidivism among Aboriginal offenders. Given a 3-year follow-up period, the total scores yielded from this tool predicted reconviction as accurately for Aboriginal offenders (n = 390; r = .23) as they did for non-Aboriginal offenders (n = 519; r = .32). Although Aboriginal offenders obtained significantly higher scores on the instrument compared to their non-Aboriginal counterparts, these elevated scores were congruent with the greater recidivism rates observed in the Aboriginal subsample (65.9 % vs. 47.8 %). Furthermore, the majority of the items featured on the instrument did indeed predict reconviction among the Aboriginal group, including substance abuse, criminal history, antisocial attitudes, and antisocial peers.

It bears noting that three domains featured on the Manitoba Risk-Needs Scale did not evidence predictive validity in the Aboriginal group: family/marital, academic/vocation, and mental ability. Although ranges were not provided for domain scores in Bonta and colleagues’ (1997) original publication, it is plausible that if members of the Aboriginal sample did consistently manifest a high level of risk on certain items, variability in the data may have been too restricted for such items to be predictively useful. With respect to the domain tapping mental ability, it is possible that high levels of risk observed across other important areas simply masked any predictive power that might otherwise have been associated with this domain. These exceptions notwithstanding, the overall findings of this investigation lend support to the position that risk assessment protocols developed predominantly on Caucasian samples can successfully be applied to Aboriginal minorities.

There has been a gradual accumulation of evidence supporting the pertinence of commonly used risk assessment tools to several ethnic minority subsamples. In a recent investigation, Snowden et al. (2010) considered a racially disaggregated sample of 874 White and 249 Black offenders (i.e., of African or Caribbean origin), all being discharged from a forensic psychiatric facility in the UK. Two instruments were featured in this investigation, both of which were specifically designed to predict violent recidivism: (1) the HCR-20 Risk Management Guide (Webster et al., 1997)—a structured professional judgement tool, and (2) the Violence Risk Appraisal Guide (VRAG; Quinsey et al., 2006)—an actuarial instrument.

Over a 2-year follow-up period, the capacity of each tool to predict reconviction for violent offending was comparable across ethnic groups (Snowden et al., 2010). Specifically, the VRAG produced moderate to high AUCs of .79 and .74 for the White and Black subsamples, respectively. In turn, the HCR-20 yielded moderate levels of accuracy across White and Black offenders, with respective AUCs of .72 and .66. Despite differences in magnitude, levels of predictive validity between ethnic groups on both instruments were not statistically significant. Notably, an independent investigation conducted by Fujii, Tokioka, Lichton, and Hishinuma (2005) confirmed the generalizability of the HCR-20 across samples of Asian Americans, Euro-Americans, and Native Hawaiians.

There is further evidence to support the efficacy of current risk assessment tools across ethnic minorities in juvenile populations (e.g., LeCroy, Krysik, & Palumbo, 1998; Olver et al., 2009; Schwalbe, 2009). Schwalbe (2009) examined the predictive validity of the Arizona Risk/Needs Assessment (ARNA; LeCroy et al., 1998), a protocol specifically developed for justice-involved youth in the state of Arizona. Over a 1-year follow-up period, the 10-item protocol showed no appreciable differences in predicting delinquent complaints (excluding technical violations) across Caucasians (AUC = .66), Latinos (AUC = .65), African Americans (AUC = .63), and Native Americans (AUC = .63). Similarly, Olver et al. (2009) recently published a meta-analytic review of the extant literature evaluating several variants of the YLS/CMI (Andrews et al., 2004)—as the reader will recall, this mechanical tool is the youth version of the LSI. Congruent with the RNR perspective, aggregate scores on the instrument significantly and comparably predicted general recidivism in both Aboriginal (r = .35) and non-Aboriginal juvenile samples (r = .32).

Despite the above evidence attesting to the generalizability of risk assessment protocols across marginalized populations, some scholars continue to express concern that traditional risk/needs assessment tools developed largely on Caucasian samples are deficient in cultural sensitivity. It is argued that introducing culturally informed modifications to current risk assessment tools through empirical testing can potentially enhance levels of predictive accuracy with minority groups (e.g., Blanchette & Taylor, 2007). Revisiting the investigation conducted by Bonta and colleagues (1997), although the majority of the domains featured on the Manitoba Risk-Needs Scale did indeed predict reconviction among Aboriginal offenders, recall that three risk areas did not: family/marital, academic/vocation, and mental ability. Although, as previously suggested, the failure of these items to predict recidivism among Aboriginal offenders may simply be an artifact of the data (i.e., restricted variance), an alternative explanation is that the manner in which these items are conceptualized and defined on the instrument genuinely does not provide an accurate assessment of risk in this minority group. Barring replication across samples, this particular question remains unanswered.

Some scholars maintain that a preferable course to introducing modifications to extant tools is to devise culturally specific risk assessment protocols from the ground up. Although there are few such tools in existence, one example was devised by a group of Canadian researchers for use with male and female Aboriginal offenders who are incarcerated for the commission of violent offenses (Boer et al., 2003). It is termed the Yokw’tol, which is a native word that translates to “the understanding of one is complete.” Unlike traditional risk assessments, the function of the Yokw’tol is not to predict recidivism. Rather, it is to facilitate one’s reintegration into the community, to foster personal responsibility, and to direct the individual towards the attainment of his or her full potential (Boer et al., 2003). As such, the protocol is more aptly termed a risk management guide.

The Yokw’tol is comprised of 20 items that were included based on consultation with Aboriginal Elders, Aboriginal correctional staff, and Aboriginal offenders themselves (Boer et al., 2003). Many of these items mirror those featured on traditional risk assessment tools (e.g., substance abuse, impulsivity). However, some of the content is intended to directly capture the unique cultural circumstances of indigenous peoples, both in terms of putative risk and resiliency factors (e.g., relationship to cultural heritage, respect for traditional customs). Primarily due to the purpose behind its derivation, the Yokw’tol has not yet been the subject of empirical validation. Therefore, it is not currently possible to compare levels of predictive validity between this tool and those yielded from traditional risk/needs assessments applied to Aboriginal populations.

Intersectionality and Risk Assessment

The core argument against adopting a universal approach to risk assessment—and in drawing conclusions on the matter from most published research—is that instruments developed and validated on aggregate samples potentially fail to capture risks/needs that are unique or salient to marginalized populations. Informed by intersectionality theory, the issue of multisource marginalization emerges, and more specifically, we are confronted with the question of how compound sources of oppression (i.e., being an ethnic minority female) might impact the validity of risk assessment tools among offender sub-populations.

Primarily due to sample size limitations, there are only a few empirical investigations conducted to date that examine the compound of effect of gender and race on the validity of risk assessment protocols. The first of these examined the validity of the Custody Rating Scale (CRS; Luciani, Motiuk, & Nafekh, 1996) with 68 Aboriginal and 266 non-Aboriginal female offenders (Blanchette et al., 2002). The CRS has been used by the Correctional Service of Canada since the early 1990s to guide security classification within federal institutions. The instrument is comprised of 12 items subsumed under two scales: (1) Institutional Adjustment (IA) (e.g., institutional misconducts, previous attempts to escape) and (2) Security Risk (SR) (e.g., severity of current offense, number of prior convictions). Confirming the elevated risk perspective (e.g., Moffitt, 1994), Aboriginal women scored higher on 11 of the 12 items on the CRS compared to non-Aboriginals and scored significantly higher on 6 of these items (i.e., alcohol/drug abuse, street stability, and several indices of criminal history).

However, scores yielded on this protocol did not necessarily cohere with criminal outcome gauged by institutional misconducts (Blanchette et al., 2002). Based on CRS total scores, the Aboriginal women in this sample were underrepresented in the minimum security category and overrepresented in medium and maximum security categories. Ideally, one should observe a linear increase in the proportion of institutional violations committed as risk category increases. This expected progression was evident among non-Aboriginal women, with the percentage of violations within each incremental risk category being 40.0, 52.4, and 80.0 %. However, this pattern was not discerned among their Aboriginal counterparts. Instead, among Aboriginal women, a higher proportion of institutional violations were evident in the medium risk (28.6 %) compared to the low-risk category (26.8 %).

Blanchette and colleagues (2002) additionally considered predictive validity estimates of the CRS based on 61 Aboriginal and 261 non-Aboriginal women offenders featured in their original sample. The IA total scale scores were moderately correlated with violent and nonviolent institutional incidents for both Aboriginal and non-Aboriginal women (average r = .43 and .20, respectively). With a lower degree of accuracy, the SR scale scores significantly predicted violent and nonviolent institutional misconducts for non-Aboriginal women (average r = .19). However, virtually no relationship existed between scores of Aboriginal women and the outcomes of interest (average r = .03).

Based on these findings, concerns have been voiced about the degree to which the CRS can appropriately guide security classification among female inmates—particularly those of Aboriginal descent. In response, Blanchette and Taylor (2007) developed the Security Reclassification Scale for Women (SRSW). This tool was empirically derived from an initial pool of 176 candidate variables. Only the 9 items most uniquely predictive of institutional misconduct in a development sample of 172 federally sentenced Canadian women were retained and included in the protocol (e.g., serious disciplinary convictions, prosocial family contact, etc.). Based on a validation sample of 103 non-Aboriginal and 45 Aboriginal women offenders, Blanchette and Taylor found that within a 3-month period, composite scores on the SRWS predicted institutional misconducts with an impressive level of accuracy across both female ethnic groups. For Aboriginal and non-Aboriginal women, respectively, AUCs were .72 and .75 for minor misconducts, and .74 and .68 for major misconducts. Interestingly, the magnitude of the AUCs was actually greater in the Aboriginal subsample, albeit these differences were not statistically significant.

While the research of Blanchette and her colleagues (2002, 2007) focused exclusively on Aboriginal versus non-Aboriginal Canadian women, a recent American study afforded a true disaggregation by gender and race. Holsinger, Lowenkamp, and Latessa (2006) assessed the predictive validity of the LSI in male and female subsamples of White (n = 263; 162 males, 101 females) and Native American offenders (n = 140; 100 males, 40 females). All subjects were under community supervision in the Midwestern USA. Total scores yielded from the LSI were moderately and significantly correlated with new arrests over a 17-month period for the combined sample (r = .18), as was the case for White males (r = .22) and White females (r = .26). That stated, the accuracy levels achieved by the LSI with Native Americans, particularly Native American females, were far less encouraging. The instrument did marginally predict rearrest in Native American males (r = .19). However, the relationship between LSI composite scores and rearrest in Native American females was in the direction opposite to that expected, albeit not statistically significant (r = −.13, p = .43).

Based on LSI total scores, Holsinger et al. (2006) further classified offenders into prescribed risk categories (i.e., low, low/moderate, moderate, high). An examination of recidivism rates by risk classification confirms the counterintuitive estimate of predictive validity observed among Native American females. Recall that an increasing linear trend should ideally emerge between recidivism rates and risk category. Although this pattern was evident across other racial/gender subgroups, it was not apparent among Native American women. For example, 50 % of indigenous women classified into the low-risk group reoffended, yet the recidivism rate associated with the moderate risk group was only 20 %. One caveat that bears noting is that these ethnic minority women constituted the smallest subsample in the study (n = 40), and as such, results should be interpreted with due caution. Although findings are tentative and require replication, this investigation does lend credence to concerns articulated by intersectionality theorists—namely, the questionable applicability of risk assessment tools developed on aggregate samples to minority subgroups.

Correctional Treatment

There has been considerable debate over the last several decades as to the effectiveness of treatment for offender populations. In the 1970s, it was claimed that “nothing works” in reference to the state of correctional programming of the era (Martinson, 1974). Although this conclusion was challenged on methodological grounds by some (e.g., Brody, 1976) and later recanted by Martinson (1979), the work was taken by many as proof that correctional treatment is ineffective, fitting with the political, right-winged climate of the time (Hollin, 1999). It became generally accepted that attempts at rehabilitation within a criminal justice setting were too individualistic to be useful (McGuire, 1995). However, throughout this period there were some who persisted in their belief that the treatment of offenders was viable, and continued to produce evidence of programs that did indeed work to reduce recidivism rates (e.g., Braukmann, Fixsen, Phillips, & Wolf, 1975; Palmer, 1975; Ross & Gendreau, 1980).

Using this preliminary evidence as a catalyst, the field of corrections has made considerable advances in offender rehabilitation over the past 30 years, largely based on the ardent efforts of several prolific Canadian researchers. Specifically, research by Don Andrews, Jim Bonta, and others in the 1980s and 1990s contributed substantially to our repository of knowledge around the psychology of criminal conduct by developing what has become known as the “What Works” model of effective correctional intervention. Broadly, this model states that programs that adhere to the principles of risk, need, and responsivity demonstrate the greatest reductions in recidivism (Andrews & Bonta, 2010; Andrews, Bonta, & Hoge, 1990; Gendreau, Goggin, French, & Smith, 2006; Smith, Gendreau, & Swartz, 2009). To reiterate, the intensity of an intervention should be proportional to an offender’s level of risk, treatment should target those factors deemed to be criminogenic (e.g., antisocial attitudes), and programming should be designed to address responsivity issues (Andrews & Bonta, 2010).

The responsivity principle can be further divided into two components. The general responsivity principle suggests that treatments abiding by a behavioral or multimodel approach are more successful than broad and less structured approaches (e.g., counseling; Lipsey, 1992). Programs that contain both a behavioral and cognitive component are most effective, with meta-analyses demonstrating reductions in recidivism rates of up to 30 % for treatment groups compared to control groups (Lipsey, Chapman, & Landenberger, 2001; Pearson, Lipton, Cleland, & Yee, 2002; Wilson, Bouffard, & MacKenzie, 2005). In turn, the specific responsivity principle suggests that treatment efficacy is maximized when interventions account for an individual’s level of motivation, preferred learning style, and abilities. Lastly, vital to effective correctional treatment is program integrity—that is, the quality with which the program is actually implemented. Although well-established treatment programs have had positive evaluations (e.g., Aggression Replacement Training; ART), they have been shown to fall short when not implemented by trained, competent, and motivated staff (Andrews & Dowden, 2005; Barnoski, 2004).

Numerous studies have supported the application of these principles in correctional settings; specifically, the extent to which offender programming adheres to the RNR principles bears a direct impact on its capacity to reduce recidivism rates (e.g., Andrews & Bonta, 2006, 2010; Andrews & Dowden, 2006; Bourgon & Armstrong, 2005; Gendreau & Andrews, 1990; Lipsey, 1995). Specifically, a number of meta-analyses have provided evidence for the effectiveness of RNR-grounded interventions in reducing prison misconducts (French & Gendreau, 2006) and rates of re-offending among violent offenders (Dowden & Andrews, 2000), sex offenders (Hanson, Bourgon, Helmus, & Hodgson, 2009), general offenders (Andrews & Dowden, 2006; Landenberger & Lipsey, 2005; Lipsey, 1992), and juvenile offenders (Dowden & Andrews, 1999a).

However, in line with some of the skepticism presented earlier with regard to the applicability of RNR-based risk assessments to special populations, some have questioned the applicability of programs designed for the White male offenders to groups that have been largely neglected in past research and practice (e.g., females, indigenous peoples) (Cameron & Telfer, 2004; Covington & Bloom, 2000; Hannah-Moffat & Maurutto, 2003). Given that the majority of treatment studies conducted to date have been based on samples comprised predominantly or entirely of Caucasian males, the effectiveness of these programs for White female offenders and ethnic minorities—both male and female—remains largely unknown. Further, despite some empirical support for the effectiveness of traditional correctional programs (i.e., developed for males) with female offenders (e.g., Dowden & Andrews, 1999b), it has been argued that because these programs are based upon our theoretical understanding of the middle-class male, they are unable to account for the social and economic constraints placed upon women (Hannah-Moffat & Maurutto, 2003) and arguably other minority groups (e.g., non-White offenders).

These criticisms around the nature of mainstream correctional programming have led to advancements in recent years. The following section will detail some of these developments and provide a review of correctional treatment options that have been developed or validated for female offenders in general, for ethnically diverse male offenders, and for ethnically diverse female offenders. Although scarce, treatment programs that have been developed for dually marginalized populations (e.g., female, aboriginal offenders) will subsequently be presented.

Gender and Correctional Treatment

As males of all ethnicities comprise the majority of the offender population, programs have generally been tailored to this group and then arbitrarily applied to female offenders (Covington & Bloom, 2000); in large part, this is a result of gender being conceptualized as a responsivity factor under the dominant correctional paradigm. Nonetheless, there is evidence that some traditional treatment programs work equally well for female offenders (e.g., Dowden & Andrews, 1999b).

Specifically, Dowden and Andrews (1999b) examined 45 effect sizes across 26 individual studies with samples comprised predominantly or entirely of females; recidivism rates were significantly lower when programs targeted criminogenic needs embedded within the Central Eight. Additionally, programming that adhered to the general responsivity principle demonstrated the largest reductions in recidivism. Although these results suggest that treatment designed for males can be equally effective for females, this study was unable to speak to whether or not a gender-responsive treatment paradigm could provide even better outcomes than those simply targeting the Central Eight.

In part, Dowden (2005) did address this limitation by extending his previous meta-analysis of treatment outcome studies to examine the effectiveness of gender-responsive programming. Overall, he found that counseling programs (i.e., general, family, individual, and group) generated mild reductions in recidivism ranging from 7 to 10 %. However, gender-specific programs that targeted family relationships (12 studies) demonstrated a 26 % reduction in recidivism, suggesting that gender-responsive programs may be more effective than those considered to be gender-neutral.

Dowden’s (2005) results are not entirely surprising as women who enter the criminal justice system typically have unique criminogenic and non-criminogenic needs that are not properly addressed with traditional (i.e., male-based) programming; these include histories of abuse and trauma, mental health issues, substance abuse, and relationship difficulties (Covington & Bloom, 2006; Van Voorhis et al., 2010). As gender-responsive interventions are designed to address these unique needs, they may ultimately be more successful than traditional treatment programs in reducing a woman’s level of risk.

Despite the conclusions generated from this study, the types of gender-responsive interventions included in Dowden’s (2005) meta-analysis were primarily limited to those that targeted family relationships, due in large part to the dearth of studies that have examined other gender-responsive targets (e.g., empowerment, self-efficacy, etc.). While a significant amount of knowledge has been generated and applied in a variety of women-centered services in fields such as mental health, substance abuse, and trauma, most programs within the criminal justice system have yet to implement this knowledge (Covington & Bloom, 2000, 2007). As such, empirical support for these types of interventions is sparse.

Gender-Responsive Treatment

By definition, gender-responsive programming entails abiding by a strengths-based approach that takes into account women’s pathways into crime while also addressing social, cultural, and psychological factors that have led to their criminal behavior (Bloom, Owen, & Covington, 2005). Although a number of correctional programs have been developed that claim to be gender-sensitive or gender-responsive, a chief criticism has been that most remain overly grounded in the RNR paradigm; by placing utmost emphasis on the risks and needs that women present as strictly defined within the parameters of the Central Eight, treatment providers are unable to take into account the context of offending, relationships that may have directly or indirectly led to the offending, socioeconomic status and constraints, and how these factors might interact to affect treatment outcomes (Bloom, 2003).

In contrast, some correctional programs can truly boast gender-responsivity. For example, one promising gender-responsive intervention developed in Ontario, Canada, is the Moving On program. This program takes a cognitive-behavioral approach to address risk-need factors that are ubiquitous among female offenders (e.g., histories of trauma/abuse, mental health issues, substance abuse). Abiding by a strengths-based framework, the program is designed to assist women in building healthy relationships, developing life skills and strategies for success (e.g., problem solving, emotional regulation, assertive skills), and assisting them with the challenges they face upon reintegration (Van Dieten & MacKenna, 2001).

A recent evaluation of this program compared the success of probationers who had participated in the program (n = 111) to a matched group of women probationers who had not completed the program (n = 111). Overall, the effect of treatment was apparent in both rates of rearrests and rates of convictions between program completers and the control group, with treatment effects ranging from 10.8 to 13.5 % (i.e., differences in rearrest/conviction rates between the treatment and control group). Importantly, these results suggest that treating gender-salient risk factors without additionally targeting gender-neutral risk factors (i.e., the Central Eight) can be effective in reducing recidivism. Furthermore, the authors conclude that cognitive-behavioral treatment (CBT) can be effective for women offenders. This is an important finding as CBT has been criticized by some feminist scholars as being too individualistic to be useful, namely, because it fails to take into account contextual influences in a woman’s life (e.g., poverty, racism, sexism, etc.) that may influence her pathways into crime (e.g., Hannah-Moffat, 2006). Despite these encouraging results, more research is necessary to substantiate whether treating gender-salient risk factors is sufficient on its own, or whether it would be more effective to combine the treatment of gender-neutral and gender-salient risk factors in correctional programs for women.

While the psychology of criminal conduct stems from cognitive theory and behaviorism, the psychology of women is closely tied to relational theories (Covington & Bloom, 2000; Miller, 1976). Specifically, as mentioned in the previous section on risk assessment, early relationship dysfunction can serve as a catalyst for future offending and is therefore important to consider in risk assessment; however, being cognizant of the importance of relationships for women is equally, if not more, important, when shaping treatment efforts. Specifically, some argue that the effectiveness of treatment for women would be enhanced if their structure was premised upon the relational aspects of women’s lives (Bloom, Owen, & Covington, 2002a).

According to relational-cultural theory (RCT), healthy relationships, characterized by empathy, empowerment, and mutuality, are necessary for the psychological development of both men and women. However, women are arguably more dependent upon the development of healthy relationships; whereas most psychological theories emphasize the importance of individuation (e.g., attachment theory), RCT conceptualizes connections with others as critical to the healthy development of women (Jordan, 1997, as cited in Liang et al., 2002). Therefore, disruptions or conflict in relationships will affect females to a greater degree than males, the latter more likely being interested in pursuing autonomy than forming meaningful connections. Furthermore, RCT suggests that women’s sense of self-worth improves when their actions arise out of connections with others. However, when women suffer disconnections in current or childhood relationships through abuse or neglect—as is often the case for female offenders—this can lead to psychological dysfunction and various forms of antisocial behavior (Bloom et al., 2002a). As such, feminist scholars have argued for the creation of programs grounded in gender-informed theories such as RCT (e.g., Bloom, Owen, & Covington, 2002b) that can readily address any gender-specific needs female offenders may present.

Ethnicity and Culturally Tailored Correctional Treatment

Although matters surrounding gender are still relatively nascent in the realm of correctional treatment, those stemming from the ethnic background of an offender are even further understudied. Given the vast racial disparities evident in correctional populations in both Canada and the USA, namely with the respective overrepresentation of Aboriginal and African-American offenders, the importance of developing culturally sensitive programs tailored to these populations is clear. However, most of the literature tends to minimize the potential importance of race in correctional treatment. Once again, this de-emphasis is likely a function of the fact that an individual’s cultural background is conceptualized as a responsivity issue under the RNR framework (Andrews & Bonta, 2010); that is, although treatment should be provided in a manner that is both understandable and culturally relevant to the offender, race and ethnicity are not prime considerations in program design.

Despite evidence for the applicability of the RNR principles across various demographic groups (see earlier section on risk assessment), the criticisms levied by feminist scholars are equally relevant here. Specifically, it is argued that the theoretical framework pervading corrections is unable to address issues of power, disadvantage, and marginalization facing not only women but male and female ethnic minority members as well. Furthermore, research on the specific responsivity principle has largely been neglected in the treatment literature; therefore, the extent to which mainstream programming is effective for various ethnic minorities has been accorded very little attention (Andrews & Bonta, 2010; Andrews, Bonta, & Hoge, 1990). Consequently, the question becomes whether interventions and programs should be modified (or created from the ground up) for offenders belonging to various ethnic minority groups to address these inequalities, or whether the programming designed for the majority population is equally effective across all demographic groups.

Although there have been a number of program evaluations to examine interactions between race and treatment, only one review of the literature was found that tested the effectiveness of general programs among different racial groups (Wilson, Lipsey, & Soydan, 2003). Specifically, Wilson and colleagues (2003) conducted a meta-analysis to examine the effectiveness of mainstream programs for minority juvenile offenders relative to Caucasian delinquents. They included 305 studies that had samples comprised of predominantly White or minority youth (greater than 60 % of the sample) and found that generally, mainstream delinquency programs were equally effective for minorities and Whites. However, the overall mean effect sizes for minority youth and White youth were relatively small (.11 and .17, respectively), and although both effects were positive and meaningful (i.e., indicative of a 10–16 % reduction in recidivism), the conclusions that could be drawn from the study are somewhat limited. As the authors note, the results are based on comparisons of different sets of studies (i.e., some including minority youth, and others majority youth). Thus, it is possible that differences in study characteristics may have impacted the results. Importantly, although this study found preliminary evidence that mainstream programs are equally effective for minority youth and White youth, they were unable to speak to whether or not culturally tailored programs would be more effective than general programs.

Although most program evaluations do not disaggregate results by race, a few studies have examined the interaction between race and treatment in relation to a number of outcomes. For example, Leiber and Mawhorr (1995) examined the extent to which the Second Chance program—a program that provides social skills training, preemployment training, and work placement opportunities to delinquent youth—is effective in reducing recidivism. Although the focus of the study was not on ethnic differences in the effectiveness of the program, the results were disaggregated by race. Overall, the authors found that the program had little effect on reducing recidivism. However, there was a reduction in the severity of offenses that were committed after program completion when the results were examined for the aggregate sample. However, when disaggregated by race, African-American males were found to be more likely to recidivate than White males. The authors argue that the program may not have been effective for African-American male offenders because it was “not sensitive to variations in the social and life situations of these populations” (p. 138).

In another study, a culturally tailored program for African-American youth, entitled the Community Corrections Partnership (CCP), was examined (King, Holmes, Henderson, & Latessa, 2001). This program was designed to expose youth to African-American culture and community while integrating five core treatment areas: (1) life skills, (2) behavior modification, (3) substance abuse, (4) cultural re-grounding (learning African heritage), and (5) sober leisure (i.e., learning how to have good clean fun). The program was run by African-American personnel, proud of their heritage and culture. An evaluation of this program revealed that African-American youth who participated had significantly lower recidivism rates relative to youth serving regular probation terms. However, the differences between these groups diminished, with both groups re-offending at the same rate after reaching adulthood. The authors postulate that while culturally specific treatment programs are likely beneficial, they should be provided in tandem with the core elements of effective intervention to realize their full potential (King et al., 2001).

In Canada, research has found that programs tailored to the Aboriginal population are successful in reducing levels of substance abuse (Weekes & Millson, 1994), improving rates of treatment completion (Ellerby & MacPherson, 2002; Trevethan, Moore, & Allegri, 2005), and reducing rates of recidivism (Sioui & Thibault, 2001). One example of a culturally tailored program currently in use by the Correctional Service of Canada (CSC) is the In Search of Your Warrior (ISOYW) program. This intervention was developed for federally incarcerated, male Aboriginal offenders with a history of violence. The program integrates Aboriginal customs and spirituality (e.g., using an elder to guide the offenders back to their spiritual roots) with traditional (i.e., Western) approaches to treatment that are more cognitive behavioral in nature (LaBoucane-Benson, 2002).

A preliminary evaluation of this culturally tailored program found that it had relatively high completion rates (87 % of 186 participants) (Trevethan et al., 2005). This finding was particularly important as other studies have found that Aboriginal offenders typically have lower completion rates relative to non-Aboriginal offenders (Ellerby & MacPherson, 2002). Additionally, change analyses demonstrated that several criminogenic needs, including substance abuse, associates, emotional regulation/cognitive skills, and antisocial attitudes changed in the desired direction after program completion. As well, offenders who completed the program were rated higher on reintegration potential, which suggests that their preparation for reentry into the community was also enhanced (Trevethan et al., 2005).

Overall, research on both African-American and Aboriginal offenders has provided some preliminary evidence to suggest that culturally tailored programming may be more effective than programming tailored to the mainstream prison population. However, further research is necessary to determine how culturally tailored treatment can contribute to existing treatment regimes.

Limitations to Developing Culturally Sensitive Treatment

Although there have been considerable advances in the last decade in the development of culturally tailored programming, particularly in Canada, a number of criticisms have been levied in regard to how culture is defined. For example, describing cultural programs as the “aboriginalization of prisons,” Martel, Brassard, and Jaccoud (2011) argue that the conceptualization of culture as proposed by correctional agencies is too narrow and ignores the multiple cultures and practices among Aboriginal peoples. Being culturally sensitive means “being sensitive to differences in ethnicity—including differences in language, customs, values, and beliefs—in order to create a sense of inclusivity” (Covington & Bloom, 2003, p. 12). However, if there are cultures within cultures, how programs will be tailored to accommodate these various subgroups of offenders remains problematic.

As well, there are a number of methodological issues to consider when studying race and ethnicity in the context of correctional treatment and evaluation. The melding of races and ethnicities is commonplace in North America; consequently, identifying someone as belonging to a particular race or ethnicity is not always a straightforward endeavor (Marks, 1995). Moreover, how researchers choose to determine the race or ethnicity of participants is sometimes vaguely described, and done so inconsistently. Allowing offenders to self-identify race and ethnicity may also complicate matters as demonstrated previously in the context of the Hispanic effect. Clearly, this is another area that warrants further research.

Intersectionality and Correctional Treatment

Further complicating matters of treatment is the interaction between race and gender within correctional settings. Feminist correctional scholars argue that one cannot limit focus to just gender, or just race (Bloom, 1996; Brubaker & Fox, 2010); rather the power imbalances that emerge as a result of the combined effects of race and gender require consideration. For example, it is important to consider to what extent the treatment needs of Black female offenders differ from those of White female offenders as well as White male offenders.

A review of the literature revealed few programs that consider the intersecting effect of race and gender simultaneously. However, two noteworthy programs are described here—one American and one Canadian. Both programs were developed specifically to address the needs of two groups overrepresented in these correctional populations (i.e., African-American women and Canadian Aboriginal women, respectively). In New York State, a residential treatment program entitled Greenhope Services for Women provides services to predominantly underprivileged African-American and Latino women on parole and to those referred by the courts as an alternative to incarceration (Greenhope Services for Women, n.d.) is designed to address the problems that led women to a life of drugs and/or crime by encouraging sobriety and treating factors that contribute to women’s addiction, namely, sexual abuse, domestic violence, and relationship issues. Although this program provides many anecdotal success stories and appears to tackle treatment issues linked to both race and gender, rigorous evaluations examining long-term outcomes are nonexistent.

In Canada, the Spirit of a Warrior program was designed to address both the gender-specific and culture-specific needs of Aboriginal women offenders (Bell & Flight, 2006). Adopting a cognitive-behavioral approach, the program uses culturally sensitive intervention strategies (e.g., cultural rituals) to address the attitudes and beliefs that led to women’s acts of violence. The program addresses a number of components including anger and violence awareness, family of origin awareness, self-awareness, individual and group skill development, cultural awareness, and cognitive learning (e.g., self-esteem, self-care, etc.). Preliminary results from an evaluation of the program found that by taking a nontraditional, holistic approach to healing, both facilitators and the women themselves believed the program provided supplemental value to existing programs in the institution. Furthermore, interviews with the program facilitators and the participants themselves indicated that both witnessed changes in cognitive processing, coping skills, and relationships with others including both staff and other women offenders. Lastly, changes in anger, self-control, and self-esteem were also evident through the administration of standardized tests pre- and post-program completion (Bell & Flight, 2006). Although these early results are promising, the number of participants who were involved in this evaluation was small (n = 29), and the methods to evaluate success were primarily self-report. There has yet to be a formal evaluation of the program to determine its efficacy in reducing rates of re-offending.

In sum, there is still a significant amount of work to be done to determine the value of correctional programs designed to account for both race and gender. Although beyond the scope of this chapter, the question of how tailored a program must be to effect positive change requires further exploration. The possibilities for tailoring programs are endless if one considers not only race and gender, but also offense type. Further research examining the extent to which general, gender-responsive, and/or culturally tailored programs are effective for various subtypes of offenders requires additional exploration.

Conclusions, Best Practices, and Directions for Future Research

In this chapter we have discussed crime statistics, theories of criminal conduct, and correctional assessment and intervention strategies from an intersectionality perspective. In brief, intersectionality (Crenshaw, 1989/1993), which is closely aligned with multiracial feminism (Baca Zinn & Thornton Dill, 1996) and the matrix of domination (Collins, 2001), is a theoretical paradigm that underscores the importance of understanding how social location markers—race, gender, class, sexual orientation, age, and physical disability—yield multiplicative effects that either marginalize (e.g., underprivileged women of African-American descent) or empower individuals (e.g., White, upper class men). Our review has made it abundantly clear that the discipline of sociology, particularly feminist criminology, has tirelessly championed intersectionality paradigms. In contrast, psychology—and by extension correctional psychology—has implicitly supported a similarities perspective; that is, human behavior is human behavior irrespective of gender, ethnicity, race, or class. However, this generalized perspective has begun to shift in recent years.

Noteworthy findings have emerged recently from scholarly camps as well as government organizations that would not readily identify with the intersectionality or multiracial feminist perspectives. Crime statistics clearly illustrate that minority groups, particularly youthful men of African descent, are grossly overrepresented in the American criminal justice system. Adolescent African Americans of both genders are particularly overrepresented in custody settings. In Canada, while both Aboriginal men and women are disproportionately represented in the federal correctional system, overrepresentation is particularly striking among Aboriginal women. Although such statistics provide critical information, gaps still remain. The unfortunate fact that the USA’s national crime reporting system (the FBI’s Uniform Crime Reports) does not readily publish arrest statistics that simultaneously consider race and gender are particularly problematic. Similarly, it would be equally desirable for the UCRs to record both race and ethnicity. In sum, incremental knowledge gains necessitate that all crime statistics variants (i.e., official vs. unofficial, arrest vs. incarceration, offender self-report vs. victimization reports) record gender and ethnicity in such a fashion that readily permits intersectionality-orientated analyses. Moreover, future research must seek to further deconstruct ethnicity beyond the Hispanic effect.

Six theories of crime were reviewed in this chapter. Three of these theories (i.e., masculinity model, racialized gender role expectations model, and the economic deprivation model) arguably were constructed within an intersectionality framework. Notably, these theories were not developed by psychologists. Each theory underscores the importance of considering either the joint effects of gender and race, or gender and social class. In contrast, the personal, interpersonal, community-reinforcement model and the developmental life-course perspectives focus on individual risk factors and either implicitly or explicitly highlight universality across gender, race, and social class. Interestingly, although the structured life-course model did explicitly integrate elements from the “person-focused” theories and “structural-focused” theories, structural forces—namely, race, gender, and class—were afforded preeminent status.

It is evident from the theory review that there is a paucity of tightly developed theories that translate into testable hypotheses. Further advancements require a shift from paradigm theorizing to empirical testing. Future theory-driven empirical testing should strive to explicitly examine whether or not unique theories developed from the ground up for ethnic minority women and men who engage in crime are preferable, or whether we can safely modify existing perspectives without falling victim to the “just add gender and race and stir” fallacy. In sum, scientific parsimony (Occam’s razor) and practical realities (i.e., how useful are an infinite number of theories?) must be balanced with the sacred tenets of each respective discipline. In the end, theoretical integration will be the hallmark of knowledge advancement, as will the acknowledgment that more often than not most theories are complimentary rather than disparate.

Considerable advancements have been made in the field of corrections. A number of reliable and valid correctional assessment tools are readily available to correctional decision makers. Countless studies have provided solid, evidence-based treatment guidelines. A once androgenic correctional field has become increasingly gender-responsive. Similarly, research is growing about the role of ethnicity and race. Although correctional scholars and policy makers have yet to reach consensus about “What Works” for offenders with multiple, intersecting identities (race, ethnicity, and gender), it is encouraging that they are now asking intersectionality-based questions.

In sum, it is clear that future research and practice must genuinely adopt an interdisciplinary perspective. This is simple in theory, but complex in practice. Disciplinary warfare and mutual misunderstanding (and sometimes mutual disdain) about our respective fields are obstacles not easily remedied. Nonetheless, integrative theoretical and methodological approaches are critical to advance theory and practice. For example, although quantitative models can readily analyze interactive effects involving two factors, models in excess of three or more terms are extremely hard to capture. Thus, qualitative approaches are required to fully understand how the experiences of a Black, impoverished, lesbian woman (for example) may be unique. Each discipline must recognize its own strengths and limitations as well as the strengths and limitations of alternative disciplines. Lastly, and perhaps most importantly, this entire chapter has focused on the intersection of sociological constructs—race, gender, and class. It is becoming increasingly apparent that no one discipline can fully explain the variance in human behavior. As such, future scholarship and practice in the study of crime must also attend to the intersecting disciplines of psychology, biology, and sociology.