A Quantitative Exploration of Gendered Pathways to Crime in a Sample of Male and Female Juvenile Offenders

Abstract

Feminist (gendered) pathways theorists maintain that female criminality is largely survival-based, and tied to a constellation of factors including early trauma, ensuing mental health issues, and poverty. Based on items drawn from the Youth Assessment and Screening Instrument, multidimensional scaling was performed to elucidate the respective thematic structure of background and offending characteristics of 663 female and 1,175 male juvenile offenders under community supervision in New York State. Although the gendered pathways theme emerged exclusively for females, a theme closely resembling the traditional antisocial pathway depicted in mainstream correctional literature was also evident among females. Theoretical integration is therefore recommended.

Keywords

feminist (gendered) pathways juvenile offenders multidimensional scaling case-based analysis

In 2009, the American juvenile court system processed approximately 1.5 million youth, 28% of whom were female (Sickmund, Sladky, & Kang, 2012). Noteworthy, over the last 30 years (1980-2009), the gender gap in arrest rates has narrowed considerably, with females seemingly “catching up” to their male counterparts in the perpetration of a number of crimes including burglary, aggravated assault, simple assault, larceny/theft, motor vehicle theft, and drug possession (Snyder, 2011). On closer examination, however, it becomes apparent that the changing arrest patterns accounting for the narrowed gender gap are not synchronous across crime categories. For example, the gender gap in aggravated assault has narrowed during the past 30 years, largely because the female arrest rate has more than doubled. Although the gender gap has also narrowed for burglary, it is not because females are committing more of these crimes, but rather because males are committing less. Additional nuanced differences of this nature also emerge within other crime categories; for example, arrest rates for drug use and simple assault have increased substantially for both genders since 1980, albeit the increase has been strikingly more pronounced for females.

Official arrest statistics for juvenile and adult populations would imply that relative to boys and men, girls and women are engaging in more crime than ever before. However, there is compelling evidence to suggest that changes in legislation (e.g., get-tough on drug policies) and police reporting practices (e.g., mandatory arrests for incidences of family violence) have disproportionately and negatively impacted females in comparison with their male counterparts. Specifically, these systemic changes have resulted in an apparent “surge” in certain types of female-perpetrated crime, and propagated the myth that girls and women are generally becoming more violent and predatory (Feld, 2009; Steffensmeier, Schwartz, Zhong, & Ackerman, 2005).

Despite potential misconceptions arising from official crime statistics, the very fact that females are entering the criminal justice system at an increasing rate has justifiably captured the attention of forensic scholars and policy makers. No longer branded as “correctional afterthoughts” (Ross & Fabiano, 1986), a growing body of literature is being devoted to elucidating the psychological and sociological landscape of female offending (e.g., Covington & Bloom, 2007; Dowden & Andrews, 1999; Van Voorhis, Wright, Salisbury, & Bauman, 2010).

The recent repository of research on the female offender depicts a vehement debate between gender-neutral and gender-responsive theorists. In brief, gender-neutral theorists, including developmental criminologists/psychologists (e.g., Loeber & Farrington, 2000; Moffitt, 1993) and mainstream correctional researchers (e.g., Andrews et al., 2012; Gendreau, Smith, & French, 2006), maintain that theories of criminal conduct transcend gender lines. Equally relevant to males and females, pathways to delinquency are said to be comprised largely of individual-level factors empirically linked to antisocial outcome and commonly termed the Central Eight: antisocial cognitions, antisocial peers, personality deficits (e.g., impulsivity), family/marital dysfunction, substance abuse, educational/vocational obstacles, and misguided use of leisure time (Andrews & Bonta, 2010). In contrast, feminist theorists contend that the etiology of female offending is unique, occurring against a backdrop of sociological factors contributing to women’s oppression (e.g., economic marginalization) and catalyzed primarily by relational trauma (e.g., Belknap, 2007; Daly, 1992). Fusing contributions of gender-neutral and gender-responsive perspectives, the present study has as its primary objective to determine whether a unique gendered pathway into crime is evident in a sample of female juvenile offenders.

At this juncture, it is important to underscore that the methodological prerequisites for elucidating true causal pathways to crime vary substantially across disciplines. Prominent developmental criminologists/psychologists (e.g., Loeber & Farrington, 2000) as well as mainstream correctional researchers (e.g., Andrews & Bonta, 2010) emphasize the importance of quantitative methodologies and longitudinal designs with multiple data waves to ascertain causal statements about crime trajectories. However, other than a few proposed typologies that essentially reflect differences in offending severity and/or persistence with offender “subtypes” implicitly captured along a continuum of risk (e.g., Chaiken & Chaiken, 1984; Moffitt, 1993; Onifade et al., 2008), no thematic subtypes have been delineated for general offender populations in an effort to highlight true qualitative differences between them.

In contrast, prominent feminist criminologists (e.g., Belknap, 2007) maintain that qualitative methodologies involving retrospective, in-depth, face-to-face interviews with justice-involved females are necessary and sufficient to understand the temporal sequencing of events that cumulatively propel girls and women toward a criminal justice pathway or trajectory. Although quantitative approaches have recently been adopted in an effort to supplement this body of literature (e.g., Salisbury & Van Voorhis, 2009), the goal remains to identify qualitative or thematic differences in female offenders by generating typologies representing distinct etiological pathways into crime.

Feminist (Gendered) Pathways Research

Over the last 5 years, there has been renewed interest in a body of literature collectively termed feminist pathways research (Belknap, 2007). Gaining prominence among feminist-oriented correctional researchers (Brennan, Breitenbach, Dieterich, Salisbury, & Van Voorhis, 2012; Reisig, Holtfreter, & Morash, 2006), this area emerged within feminist criminology in the late 1970s to explain why females become entrenched in the criminal justice system. In sum, theoretical advancements and empirical research falling within the feminist or gendered pathways paradigm seeks to explain how specific external events (e.g., trauma, economic marginalization) engender internal reactions (e.g., addiction and mental health issues) that collectively culminate in various externalizing behaviors such as prostitution, theft, assault, robbery, and drug use. Feminist scholars label these externalizing behaviors criminalized survival strategies (Belknap, 2007).

On the basis of in-depth qualitative interviews with incarcerated females, prostitutes, or runaways, feminist scholars maintain that the pathways to female offending are uniquely tied to certain sociological factors deemed to contribute to the oppression of girls and women—namely, economic marginalization, systemic poverty, lack of access to state capital/government funding, and lack of access to community resources. Importantly, feminist theorists fervently maintain that society’s patriarchal structure plays a pivotal role in the subjugation of girls and women and in the criminalization of their survival strategies (Chesney-Lind & Shelden, 2003). Last, individual-level factors that are relationally based, namely the experience of trauma and victimization in one’s family of origin, are considered particularly salient to elucidating female pathways to crime (Daly, 1992; Reisig et al., 2006).

Kathleen Daly’s (1992, 1994) research most strongly encapsulates the feminist pathways perspective. In her 1992 seminal study, Daly reviewed collateral information from presentence investigation reports (PSIs) for 40 women convicted of felonies. Qualitative analysis and the examination of descriptive statistics revealed five different pathways or typologies by which women come to the attention of the criminal justice system. She labeled each identified pathway as follows: street women (25%), harmed-and-harming women (38%), drug-connected women (15%), battered women (13%), and other (10%).

In brief, the street women scenario most closely approximates the theoretical prototype identified in the feminist literature (e.g., Chesney-Lind & Shelden, 2003). Following a history of childhood abuse and neglect, the young woman attempts to flee her abusive environment, turns to substance use in an effort to cope with early trauma, and ultimately becomes entrenched in a life of prostitution and petty crime in an effort to fulfill survival needs. The harmed-and-harming woman has also experienced an abusive and chaotic childhood, but engages in interpersonally aggressive crimes. She is also plagued with psychological difficulties and drug addictions. As the name implies, the battered woman suffers abuse within the context of an adult, intimate relationship with a male partner. Her attempts to fight back and protect herself result in criminal justice involvement. Last, the drug-connected woman evidences no childhood trauma or early onset criminality, but nonetheless becomes involves in the drug trade either through associations with a romantic partner or her own children.

Daly’s (1992) unclassifiable group (i.e., “other”) comprised 10% of the sample and did not appear to mesh with the feminist framework. This group consisted of women who were not particularly disadvantaged, tended to be motivated by greed rather than survival needs, and abused positions of trust to perpetuate crime (e.g., embezzlement). In contrast to the remaining groups, these women had no apparent history of victimization or substance abuse.

Building on Daly’s seminal work, Salisbury and Van Voorhis (2009) conducted the first quantitative test of the feminist pathways model using a longitudinal recidivism design. Applying path analysis to a sample of 313 adult females on probation, three distinct gendered trajectories into crime were identified, each predicting reincarceration within a 2-year follow-up period: (a) 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; (b) 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; finally, (c) the social and human capital model reflects challenges in the area 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.

In sum, Salisbury and Van Voorhis’s (2009) quantitative analysis supports Daly’s qualitative feminist pathways findings. However, Salisbury and Van Voorhis used a sample comprised primarily of lower risk offenders. The index offenses for this group were predominantly drug- and property-related, with only about 7% of women having committed a violent offense (e.g., assault). Moreover, among recidivists, nearly 75% were convicted as a result of violating a supervision order rather than for the commission of a new crime. Thus, the extent to which Salisbury and Van Voorhis’ findings will replicate on a moderate to high risk sample warrants investigation. In addition, the uniqueness of these pathways to women is debatable 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.

Notably, the qualitative and quantitative pathways research has either implicitly or explicitly ignored potential developmental differences that may be masked by failing to disaggregate girls and women or by focusing almost exclusively on women. Adolescence is characterized by rapid fluctuations in mood and behavior, and underdeveloped cognitive capacities (Cauffman & Steinberg, 2000). Generally less adept than adults in their ability to process information and manage impulsivity, adolescents are consequently more prone to engage in defiant, risky, and reckless behavior (Slobogin & Fondacaro, 2011). It is crucial to consider these important developmental differences when conducting feminist-oriented pathways research.

Study Rationale

In sum, much of the feminist-oriented pathways research has been dominated by qualitative studies. Not only are male comparison groups noticeably absent, but there is generally very little consideration for how other risk factors identified in the mainstream correctional literature may or may not contribute to female offending trajectories/pathways (e.g., impulsivity, criminal attitudes, criminal associates). Last, feminist pathways researchers have paid little attention to the extent to which feminist pathways models apply specifically to youth populations.

There is no question that the qualitative research produced by feminist pathways theorists through the thematic analysis of interview data has been invaluable to gaining an appreciation of the nuances and context surrounding the experiences of justice-involved females (Bottcher, 2001). However, it is important to supplement this literature with quantitative research. Fortunately, quantitatively-minded correctional feminists are beginning to take an interest in replicating the earlier work of qualitatively-driven criminological feminists using diverse clustering techniques (e.g., Brennan et al., 2012; Salisbury & Van Voorhis, 2009). However, further quantitative replication is required, particularly with juvenile offender samples that include male comparison groups.

Informed by qualitative research performed by feminist scholars (e.g., Daly, 1992; Reisig et al., 2006), and the recent quantitative work of Salisbury and Van Voorhis (2009), the present study reflects the first quantitative test of the pathways issue to include (a) a male comparison group and (b) additional indicators underemphasized by feminist scholars, yet extolled by mainstream correctional researchers (e.g., criminal attitudes). As depicted in the feminist pathways literature, a thematic structure prototypical of the “street woman” is hypothesized to emerge exclusively within our subsample of females—that is, a collection of items encompassing poverty, child neglect, mental health issues, substance abuse, attempts to run away from home, and status offenses. We expect this gendered pathway scenario to mirror the circumstances of a portion of the females in our study. However, it is also hypothesized that a second theme will emerge concurrently for females, depicting a more traditional “antisocial” pathway into crime as is reflected in the gender-neutral literature, and comprising such indicators as criminal attitudes, antisocial associates, impulsivity, and manifestations of violence. In contrast, this traditional antisocial pathway is expected to characterize the majority of male offenders in our sample. A further discussion of hypothesized indicator placement is available in the methodology.

Method

Sample

Entering into a memorandum of understanding with Orbis Partners Inc. and the New York State Division of Criminal Justice Services (DCJS), the first author was granted access to archival risk assessment data collected via the Youth Assessment and Screening Instrument (YASI; Orbis Partners, 2000). An initial sample of 2,369 youths (1,550 males and 819 females) across 54 counties in New York State was available. All youths had been processed by probation services under the direction of DCJS between October 2000 and January 2005. At intake, youths were identified either as Juvenile Delinquents (JDs) or Persons in Need of Supervision (PINS). According to New York State legislation (DCJS, 2012), a JD is defined as a person between the ages of 7 and 16 who commits an act that is also considered a criminal offense when committed by an adult. A PINS, in turn, refers to a person less than 18 years of age whose unlawful behavior is limited to status offenses (e.g., truancy, alcohol possession).

Based on the total sample, the mean age at intake was 14.59 years (SD = 1.66; M = 14.76, SD = 1.56 for females; M = 14.51, SD = 1.71 for males), whereas youths’ mean age at first contact with probation services was 12.91 years (SD = .65; M = 13.11, SD = 3.04 for females; M = 12.81, SD = 2.76 for males). Overall, the majority of participants were Caucasian (70.5%), with the remaining being primarily of African (17.4%) and Hispanic (8.7%) descents.

A significantly greater proportion of females than males were designated as PINS at intake (77.0% vs. 57.4%) compared with JDs (23.0% vs. 42.6%), χ² = 88.81, p < .001. With respect to criminal history, significantly more females had previous PINS complaints on record compared with males (38.0% vs. 32.9%; χ² = 6.06, p < .05). Compared with females, a significantly greater proportion of males had previous JD complaints (33.6% vs. 18.8%, χ² = 57.19, p < .001) and more specifically, previous felony-level complaints (13.3% vs. 3.9%, χ² = 52.21, p < .05).

For the purpose of examining the gendered pathways issue, the sample was restricted to youths aged 14 and above. The reduced sample comprised 1,838 juveniles on probation (663 females and 1,175 males), aged 14 to 19 at intake with probation services (M = 14.59, SD = 1.66). The rationale behind the selection of this older cohort was to maximize the likelihood that youth included in these analyses would have encountered the risk factors deemed salient to a female’s trajectory into crime according to feminist theory (e.g., diagnosed mental health issues, chronic substance abuse). The selection of age 14 as a minimum cutoff was based on the following rationale: First, approximately 90% of American girls have reached menarche by 13.75 years of age (Chumlea et al., 2003), which in conjunction with other risk factors, has been identified as a precursor to delinquency and other risk-taking behavior (e.g., engaging in sexual activity; Belsky, Steinberg, Houts, & Halpern-Felsher, 2010; Burt, McGue, DeMarte, Krueger, & Iacono, 2006); second, defining age 14 as the lower limit ensured the preservation of a sufficiently large sample.

Measures

YASI

In accordance with standard procedure, youths in the sample were subject to a thorough risk assessment protocol at intake with probation services. The risk assessment tool employed by DCJS to predict recidivism and guide case management efforts in the context of juvenile probation is the Youth Assessment and Screening Instrument (YASI) (Orbis Partners, 2000).¹ Although the YASI is principally grounded in the gender-neutral literature and its construction was based largely on male samples (Orbis Partners, 2007a), the tool does feature a number of gender-responsive items relevant to the present investigation (e.g., mental health issues, poverty). The YASI is scored by the youth’s assigned probation officer on the basis of a semi-structured interview, with input frequently offered by parents or an alternative legal guardian. Interview-based data are supplemented with a systematic review of collateral sources including police files, probation records, school records, and mental health reports (Orbis Partners, 2007a).

Designed primarily to inform case management and quantify incremental change in need and protective factors, the Full Assessment version of the YASI comprises 85 items (risks, needs, and protective factors) across 10 separate domains: legal history (11 items), family history (17 items), school (12 items), community and peers (9 items), alcohol and drugs (3 items), mental and physical health (7 items), attitudes (11 items), skills (7 items), employment (4 items), and use of free time (4 items). Whenever possible, individual items are scored on Likert-type scales, featuring up to six response options to maximize the sensitivity of the measure and enhance its capacity to quantify incremental change (Orbis Partners, 2007a). Within each domain, actuarially-based algorithms are used to generate static risk (SR), dynamic risk (DR), static protective (SP), and dynamic protective (DP) scores—higher scores are indicative of either greater risk or greater strength within a particular domain. Total scores on the Full Assessment are generated with respect to each domain (e.g., legal history, family history, etc.) relative to risk and protective components.

The YASI Pre-Screen features a subset of 28 SR and DR items from Full Assessment domains and is completed for all youth at intake with probation services. Pre-Screen risk scores associated with each relevant domain are then summed into a total score that can potentially range from 0 to 153, with higher scores reflective of greater overall risk. Serving a triage function, one’s aggregate score on the Pre-Screen constitutes the basis for classification into low, moderate, or high risk categories.

Based on the preliminary efforts of two independent research teams, the extant evidence supporting the YASI’s inter-rater reliability is somewhat mixed and seemingly contingent on rater characteristics—namely, researchers versus correctional staff. Brown, Geck, Harris, and Skilling (2012) reported encouraging results based on YASI interviews that were administered and scored by researchers, with intraclass coefficients (ICCs) ranging from .78 to .88 for risk domain totals and from .84 to .90 for protective domain totals. In contrast, Skeem, Hernandez, Kennealy, and Rich (2012) reported single ICCs ranging from .51 to .72 for YASI total scores based on field staff ratings obtained across four different sites in California; however, a 60% subset of these raters did achieve acceptable levels of agreement with expert raters (i.e., ICC ≥ .65). Skeem and colleagues did not identify specific characteristics discriminating “good” from “poor” raters, although their report did suggest that correctional officers performing most poorly were predominantly from one facility. Collectively, these results indicate that it is possible to achieve reliable ratings on the YASI under ideal conditions—presumably with sufficient training, motivation, and a block of protected time to integrate information sources and score the protocol. Regrettably, given the archival nature of the data, no indices of inter-rater reliability are available for the YASI with New York staff.

Internal consistency levels associated with risk and protective subcomponents of each domain featured on the YASI Full Assessment were determined for the present sample. Domains evidencing acceptable reliability with Cronbach’s alpha levels exceeding .70 include skills, attitudes, employment (protective component), and peers and community (risk component). The low alpha levels (α < .60) associated with remaining domains are potentially attributable to a combination of issues: (a) the low number of items per domain (i.e., several domains include four items or less; for e.g. alcohol and drugs), (b) the breadth of items subsumed under each construct (e.g., mental health), and (c) the potential unreliability of individual items with a low frequency of occurrence (e.g., escape from custody). Notably, measures of internal consistency for the YASI have also been gauged on the basis of 4,998 youths processed by juvenile probation services in the State of Illinois (Orbis Partners, 2007b). With respect to the Illinois sample, all but one YASI Full Assessment domain exhibited moderate to high levels of internal consistency (α = .67-.90 for risk domains and α = .64-.87 for protective domains).

Finally, moderate to high levels of predictive validity were yielded from a sample of 464 youths in Alberta, Canada, where YASI Pre-Screen and Full Assessment scores predicted convictions over an 18-month follow-up period with areas under the curve (AUCs) of .81 and .74, respectively (Orbis Partners, 2012). Regrettably, an insufficient number of females in the outcome sample precluded disaggregation by gender.

Procedure for Identifying “Pathways”

As underscored in our introductory remarks, we fully appreciate that in quantitative research, a longitudinal, multi-wave design is the optimal methodological approach for identifying a genuine pathway into crime. Being restricted to cross-sectional data, it may be more technically accurate to discuss typologies in lieu of pathways. This interpretational caveat notwithstanding, we opted to retain the term pathway to be congruent with feminist discourse. To reiterate, under the feminist paradigm, pathways are inferred retrospectively through events and circumstances identified in women’s narrative accounts (e.g., Chesney-Lind & Shelden, 2003). In our study, we also infer pathways retrospectively but through empirically informed thematic distinctions. Of greatest relevance is not the temporal order in which events transpired but the very fact that certain items co-occur with greater frequency and collectively form a theoretically-driven theme.

Variable selection

To test the pathways issue in our sample, 18 etiological and offense-related variables were extracted from the YASI Full Assessment. Items were selected on the basis of (a) thematic relevance to the feminist pathways perspective and/or (b) relevance to the assessment of criminal risk in the gender-neutral literature. Table 1 presents items categorized according to predicted thematic placement.

Table 1.

Item Placement According to Hypothesized Pathways.

Gendered pathways	Traditional antisocial
Previous PINS complaints (i.e., status offenses)	Previous JD complaints (i.e., criminal offenses)
Incorrigibility	Previous felony-level complaints
Kicked out of home	Defies parental authority
History of abuse	Antisocial peers
Family-level poverty	School suspensions
Runaway attempts	Manifestations of violence
Child neglect	Impulsivity
Substance abuse	Inadequate parental supervision
Mental health issues	Criminal attitudes

Note. PINS = Persons in Need of Supervision; JD = Juvenile Delinquent.

Despite vehement debate in the literature, it is not our contention that gender-responsive and gender-neutral perspectives are polarized, and we fully recognize that item overlap will exist. We hypothesized the a priori classification scheme presented in Table 1 not to imply mutual exclusivity but rather to ensure that we extracted items representative of each theoretical framework. While many of the items housed under the gendered pathways theme could arguably map onto the Central Eight, the specific items classified under this heading are considered integral to a female’s criminal trajectory according to feminist pathways theory and ostensibly reflect a unique context of offending.

For example, although substance abuse may be gender-neutral in its ability to predict criminal outcome (Andrews & Bonta, 2010; Dowden & Blanchette, 1998), there is evidence to suggest that the etiology and contextual factors surrounding substance abuse differ for males and females. In justice-involved women, substance abuse highly co-occurs with a sequence of more distal variables of which the “gendered pathway” is comprised; these principally include relational trauma, mental health issues, and poverty (Dowden & Blanchette, 1998; Salisbury & Van Voorhis, 2009). By extension, the argument is that substance abuse in females is generally grounded in a desire to self-medicate rather than to seek pleasurable effects, express defiance, or elicit reinforcement from an antisocial peer group. These apparent differences in motivation are confirmed through qualitative accounts of substance-abusing youths and adults in conflict with the law (Covington & Bloom, 2007; Greiner, Brown, & Skilling, 2012). Accordingly, pathways or thematic structures identified in this investigation are primarily dictated by item constellations reflective of context rather than individual items per se.

Of the 18 items extracted from the YASI, 11 were dichotomous indicators at the outset. For interpretative ease, the remaining indicators were also dichotomized to reflect their absence or substantial presence. For example, the YASI gauges parental supervision on a 5-point Likert-type scale ranging from supervision is always good to supervision is always inadequate. In this case, response options 4 (supervision is frequently inadequate) and 5 (supervision is always inadequate) were collapsed to signal significant need in this area.

Proximity scaling

For females and males, respectively, the underlying thematic structure of the 18 YASI indicators was explored through a multidimensional scaling (MDS) technique termed proximity scaling (PROXSCAL; Commandeur & Heiser, 1993). MDS was selected over alternative class-based data reduction methods (e.g., latent class analysis) for its high level of flexibility; namely, proximity scaling is suited to non-metric data, relatively free of distributional assumptions, allows structure to emerge from the data itself, and accommodates almost any measure of association (Canter & Heritage, 1990; Lingoes, 1979). Albeit novel to correctional psychology, MDS techniques have been ubiquitous in the field of investigative psychology and offender profiling for over two decades, guiding the construction of motivationally-grounded typologies in the study of homicide, arson, rape, and other sexually-based offenses (e.g., Canter, Bennell, Alison, & Reddy, 2003; Canter & Heritage, 1990; Canter & Wentink, 2004; Trojan & Salfati, 2008).

The PROXSCAL (proximity scaling) module of SPSS allows for the multivariate structure of the relationships among all variables to be represented in a geometric (visual) space. An initial correlation matrix is generated by considering the degree of association between all variable pairs. These correlations are then rank ordered, forming the basis of the spatial representation. The distance of the variables in the geometric space is inversely related to their degree of association; in other words, the closer the proximity of two given variables on the plot, the more highly they co-occur (Borg & Shye, 1995).

A second association matrix is derived from the distances between variables in the geometric space; the extent to which the rank orders between the two matrices are preserved dictates the degree of fit of the PROXSCAL solution. Note that PROXSCAL produces a series of sequential iterations in such a manner as to provide the minimal number of spatial dimensions required for a geometric representation of good fit (i.e., the smallest number of dimensions that will permit the rank order of correlations to be preserved to the greatest degree; Guttman & Greenbaum, 1998). The degree of fit between the two association matrices is given by the measure of normalized raw stress, which ranges from “0” (perfect fit) to “1” (complete lack of fit; Kruskal & Wish, 1978). Customarily, a stress measure less than .10 indicates a good degree of fit. A second measure of fit frequently considered in proximity scaling is Tucker’s measure of congruence, which ranges in value from −1.00 to 1.00; typically, .95 or greater is said to indicate a good degree of fit.

With dichotomous data that are archival, unverifiable, and reflect variation in human behavior (e.g., police data, risk assessment data from field implementations, etc.), use of a conservative measure of association is recommended (Canter, Hughes, & Kirby, 1998). Accordingly, the correlation coefficient selected for these analyses was the Lance and Williams measure (also known as the Bray–Curtis non-metric coefficient). This index reflects the degree of association between variable pairs, omitting joint non-occurrences; that is, the absence of any two indicators in an individual does not increase the degree of association between these variables.

Item placement on the proximity scaling plot is determined empirically according to frequency of co-occurrence between every variable pair. However, in accordance with a research approach known as facet theory (Canter, 1985), a priori theoretical considerations are the basis for interpreting the geometric space. The identification of structures inherent in the conglomeration of plotted variables is grounded in the principle of contiguity, which states that variables tied to a common construct or theme will be more highly correlated than those variables emerging from different constructs. Therefore, the former will be closer in proximity within the visual space. Conversely, variables appearing in different regions of the plot are considered dissimilar based on a low degree of co-occurrence and thus, ostensibly belong to different themes (Canter & Heritage, 1990). On this basis, it is possible to delineate partitions. If no interpretable themes are evident within regions of the plot, one could infer a random pattern of co-occurrences among indicators.

To partially circumvent the subjectivity inherent in partitioning the plot based solely on the face validity of included variables, Kuder-Richardson 20 (K-R 20) coefficients were calculated on various combinations of adjacent indicators. K-R 20 is an index of internal reliability, which tests the likelihood of variables in a particular region sharing a common theme (Anastasi, 1988). K-R 20 is the equivalent of Cronbach’s alpha, but the former is a measure specifically applied to dichotomous data. We attempted to partition variables so as to maximize K-R 20s while respecting theoretically-based predictions.

Results

PROXSCAL Solution for Females

The two-dimensional PROXSCAL solution generated for females is illustrated in Figure 1. The normalized raw stress index associated with the configuration is .07, achieved in 20 iterations. Tucker’s coefficient of congruence is .96. Both these values indicate a good degree of fit between the graphic representation of the items and the original association matrix.

Figure 1.

Proximity scaling plot for females.

As illustrated in Figure 1, the superior region of the plot—termed gendered pathways—depicts the abused, poverty-stricken young girl who engages primarily in minor forms of offending. However, a second structure representative of the traditional antisocial offender reflected in the gender-neutral literature emerged in conjunction (e.g., impulsivity, criminal attitudes, manifestations of violence, etc.). Two items were hypothesized to be subsumed under the gendered pathways theme, yet were located in the antisocial theme: parental neglect and substance abuse. A “misplacement” in this context is simply defined by an item’s lack of membership to its predicted conglomeration of items. As detailed in the methodology, these misplacements are likely attributable to the lack of theoretical exclusivity of said items. Substance abuse is pervasive among the sample of females (45.6%) and although integral to the pathways model, this indicator is also featured among the Central Eight predictors of criminal conduct according to gender-neutral theory. Parental neglect, in turn, might also be interpreted as a more extreme form of lack of supervision, and the close proximity of these two items on the plot supports this interpretation; notably, parental neglect and lack of supervision could also be included among the Central Eight under the family/marital umbrella.

Items located in respective regions of the PROXSCAL plot for females are listed in Table 2, along with their corresponding frequencies. K-R 20 coefficients associated with each theme were .47 and .56 for the pathways and antisocial themes, respectively. This low to moderate structural coherence was not entirely unexpected given the limited sample of items, the interpretational ambiguity and/or definitional breadth of certain items (e.g., diagnosed mental health issues), and the archival nature of the data. Studies in the field of investigative psychology have considered K-R 20s in the range of .50 to .60 to be acceptable given the potentially unreliable nature of police records (e.g., Canter et al., 2003).

Table 2.

Frequency of Items Featured in Proximity Scaling Analysis for Females According to Emerging Theme (n = 663).

Gendered pathways (7 items, K-R 20 = .47)	Traditional antisocial (11 items, K-R 20 = .56)
Previous PINS complaints (79.2%)	Defies parental authority (78.1%)
Incorrigibility (54.8%)	Antisocial peers (70.7%)
Kicked out of home (37.4%)	School suspensions (68.2%)
Abuse (28.2%)	Substance abuse (45.6%)^a
Poverty (individual-level) (25.6%)	Manifestations of violence (45.6%)
Runaway attempts (14.3%)	Previous JD complaints (32.0%)
Diagnosed mental health issues (8.4%)	Impulsivity (31.5%)
	Inadequate parental supervision (15.8%)
	Previous felony complaints (14.0%)
	Court finding of child neglect (13.7%)^a
	Antisocial attitudes (11.9%)

Note. PINS = Persons in Need of Supervision; JD = Juvenile Delinquent.

Items that did not fall within their hypothesized theme.

PROXSCAL Solution for Males

As illustrated in Figure 2, a two-dimensional PROXSCAL plot was also generated for the male subsample. Given 26 iterations, the achieved solution indicates a good degree of fit, with a normalized raw stress index of .07 and a congruence coefficient of .96. While the upper portion of the plot reflects the traditional antisocial offender (K-R 20 = .47), the inferior portion of the plot was termed mixed pathways (K-R 20 = .52). Notably, the analysis performed on the male subsample yielded regional clusters of items that were less thematically distinct than those emerging for females. Although elements of the gendered pathways theme are present among these items (e.g., runaway, abuse), other characteristics included in this collection render the label “gendered pathways” inappropriate (e.g., antisocial attitudes, impulsivity).

Figure 2.

Proximity scaling plot for males.

Item frequencies and apparent misplacements for the male sample are available in Table 3. As previously indicated, the mixed pathways portion of the plot contains the items impulsivity, school suspensions, and antisocial attitudes—indicators predicted to co-occur within the traditional antisocial offender. Conversely, items featured under the traditional antisocial theme include poverty and child neglect as potential misplacements. Overall, the portion of misplaced items from the analysis on males is 28% (five items), compared with only 11% (two items) in the analysis performed on females.

Table 3.

Frequency of Items Featured in Proximity Scaling Analysis for Males According to Emerging Theme (n = 1,175).

Mixed pathways (9 items, K-R 20 = .52)	Traditional antisocial (9 items, K-R 20 = .56)
School suspensions (74.2%)^a	Antisocial peers (71.6%)
Previous PINS complaints (64.0%)	Defies parental authority (71.2%)
Incorrigibility (48.4%)	Manifestations of violence (56.2%)
Impulsivity (36.9%)^a	Previous JD complaints (53.8%)
Kicked out of home (18.1%)	Substance abuse (42.7%)
Abuse (15.6%)	Previous felony complaints (28.8%)
Antisocial attitudes (11.4%)^a	Poverty (individual-level) (22.9%)^a
Diagnosed mental health issues (7.1%)	Inadequate parental supervision (13.3%)
Runaway attempts (4.2%)	Court finding of child neglect (12.0%)^a

Note. PINS = Persons in Need of Supervision; JD = Juvenile Delinquent.

Items that did not fall within their hypothesized theme.

Identifying Dominant Thematic Subgroups of Offenders

It is unreasonable to expect that offenders would exclusively exhibit traits associated with one particular thematic structure or the other. As such, dominant offender subtypes were identified among female and male samples based on a proportional method. Each youth was first assigned a percentage score based on the number of relevant items he or she exhibited from each theme. To maximize classification potential, a liberal criterion was used such that for a youth to be considered thematically dominant, the proportion of items he or she expressed from a given theme had to exceed the proportion of relevant items from the remaining theme (e.g., Salfati & Bateman, 2005). Those youths displaying an equal proportion of items across themes were termed unclassifiable.

Descriptive information on those subjects achieving thematic dominance is provided in Table 4. The distribution of females across thematic subgroups was approximately equal (47.8% in gendered pathways, 51.7% in traditional antisocial). Given that only two females did not observe thematic dominance, this unclassifiable category was omitted from further analysis. In contrast, the distribution of males across thematic groups was relatively disparate; more than half of males were classified into the antisocial category (59.1%) and nearly a quarter into the mixed pathways category (24.6%). Notably, 16.3% of males did not exhibit thematic dominance, but were rather a hybrid of these two themes. Also available in Table 4 are reconviction rates over a 2-year follow-up period and YASI Pre-Screen total scores for each female and male dominant subgroup; interestingly, within-gender comparisons of thematic groups were statistically equivalent based on results of a chi-square test and a one-way ANOVA, respectively.

Table 4.

Dominant Thematic Subgroup Distribution With Associated Reconviction Rates and YASI Pre-Screen Scores.

	Frequency	Reconviction	YASI Pre-Screen score
Offender subgroup	% (n)	% (n)	M (SD)
Females (n = 663)		16.7 (111)	39.73 (17.54)
Gendered pathways	47.8 (317)	16.1 (51)	39.95 (17.51)
Traditional antisocial	51.7 (343)	17.5 (60)	39.82 (17.33)
Unclassifiable	0.5 (2)
Males (n = 1,175)		31.4 (369)	38.42 (18.24)
Mixed pathways	24.6 (289)	33.6 (97)	35.75 (16.59)
Traditional antisocial	59.1 (694)	30.7 (213)	39.01 (18.88)
Unclassifiable	16.3 (192)	30.7 (59)	40.31 (17.94)

Note. YASI = Youth Assessment and Screening Instrument.

Discussion

The goal of the present research was principally to examine if a unique gendered pathway into crime emerges for female juvenile offenders. In contrast to the repository of feminist-grounded literature that infers etiological uniqueness based exclusively on female samples (e.g., Brennan et al., 2012; Daly, 1992, 1994; Salisbury & Van Voorhis, 2009), this study appropriately included a male comparison group. Furthermore, although largely ignored by feminist theorists, we sampled indicators deemed integral to the psychology and context of criminal conduct according to mainstream correctional literature (e.g., antisocial attitudes).

Informed by facet theory, results of MDS confirm the hypothesized gendered pathways theme for the female subsample, depicting the abused, poverty-stricken girl who suffers from mental health issues and engages primarily in minor forms of delinquency. Noteworthy, this gendered pathway previously identified in adult women has now been replicated in a sample of youth. Concurrently, however, a thematic structure representative of the traditional antisocial offender was also evident among females, largely comprising items subsumed under the Central Eight (e.g., criminal attitudes, impulsivity, etc.) and featuring more violent forms of offending. Of the total sample of girls included in the analysis, there was an approximately equal distribution across each theme (i.e., 47.8% gendered pathways and 51.7% traditional antisocial).

The liberal threshold for determining thematic dominance resulted in only two perfect thematic hybrids or unclassifiable cases. It is curious that the term “unclassifiable” has been adopted in the context of qualitative pathways research to denote a female offender adhering to the masculinized conception of offending etiology (Daly, 1992; Reisig et al., 2006)—in other words, in reference to offenders who display the traditional antisocial theme depicted in the present investigation. Notably, had this strict feminist template been imposed here, more than 50% of young girls in our sample would have been “unclassifiable.”

Based on the assumption that the majority of females are initiated into the criminal justice system though some variation of the pathways model (see Daly, 1992), feminist theorists lament that justice-involved girls and women are uniquely criminalized. Specifically, female offenders are considered victims of an oppressive chain of circumstances grounded in abuse and exploitation at the hands of males, ultimately culminating in the perpetration of survival-based crime (e.g., Belknap, 2007). However, the fact that nearly 50% of females in our sample predominantly exhibited features of the antisocial pathway challenges this central tenet, and suggests greater heterogeneity in female offending behavior and context than has been depicted in the feminist literature. We do not dispute the argument that many criminally entrenched females have endured challenging and oppressive circumstances; however, results of this investigation suggest that a portion of females will stray from the feminist-grounded prototype, share characteristics with their male counterparts, and are apt to be predatory and antisocial in nature.

With respect to the male sample, the collection of indicators subject to analysis demonstrated less thematic distinction than was the case for females. Specifically, nearly 30% of items were misplaced in reference to their hypothesized theme. While a traditional antisocial pathway did emerge as predicted, a “gendered pathway” could not be inferred, hence confirming our hypothesis that at least a portion of females follow a unique gendered pathway into crime. The second theme delineated on the proximity scaling plot for males was termed “mixed pathways,” because it encompasses a mix of hypothesized gendered and antisocial indicators (e.g., antisocial attitudes, diagnosed mental health issues).

Of the total sample of males included in the analysis, 59.1% predominantly displayed features of the traditional antisocial pathway, 24.6% fell into the mixed pathways category, and 16.3% were a perfect hybrid of each delineated theme. As such, the latter were subsumed under an additional category termed unclassifiable. It is particularly noteworthy that the mixed pathways theme typifying nearly one-quarter of male offenders features a selection of items underscored in the feminist literature (i.e., runaway attempts, early abuse, mental health issues). As such, while an argument can be posited for greater heterogeneity among female offenders than would be suggested by feminist scholars, male offenders may also be more heterogeneous in their etiology than portrayed in the mainstream correctional literature.

Results of the present case-based analysis strongly suggest a lack of mutual exclusivity between feminist and mainstream correctional perspectives, hence signaling a need for theoretical integration. At the thematic level, nearly half of female offenders did not adhere to a strict gendered pathway but rather, predominantly exhibited traits associated with an antisocial pathway. Further consideration of conventional gender-neutral indicators in feminist-grounded research is therefore advisable. In a similar vein, gender-neutral theories of crime might also be informed by considering the extent to which indicators emerging from the feminist literature (e.g., poverty, mental health needs) interact with traditional criminogenic needs.

Referring to Table 4, YASI Pre-Screen scores and reconviction rates within male and female subgroups were statistically equivalent. Notably then, thematic subgroups identified in this study are truly capturing qualitative differences rather than implicit variations in risk level. From a practical perspective, however, dominant membership to one theme or the other is likely to have an impact on optimal case management approaches. Contemporary (gender-neutral) models of correctional intervention are predicated on identifying and prioritizing a series of criminogenic needs (Andrews & Bonta, 2010). However, the emergence of distinct etiological themes or pathways based on constellations of co-occurring items provides quantitative support for appreciating risk indicators and other contextual variables holistically and interactively rather than in isolation. Echoing clinical recommendations of those developing and implementing services for female offender populations (e.g., Covington & Bloom, 2007), it is our contention that case management will be optimized by appreciating the manner and the context in which risk factors cluster within an individual. To illustrate, regardless of gender, strategies for targeting substance abuse in an offender whose behavior is fueled by antisocial peer influence should effectively be different from those strategies adopted with an offender who abuses drugs to self-medicate or cope with the emotional aftermath of a traumatic experience.

Limitations and Future Research

Pathways research is nascent and despite present evidence suggesting the existence of a gendered pathway into crime that is unique to females, there are a number of interpretational caveats. Albeit a commonly adopted statistical technique in the area of investigative psychology (e.g., Canter et al., 2003), proximity scaling applied for the purpose of delineating thematic structures or “pathways” of offending behavior is novel to the area of correctional psychology. Salisbury and Van Voorhis (2009) used path analysis to identify offending trajectories among women, Brennan et al. (2012) used quantitative taxonomic methods, while Daly (1992) and Reisig et al. (2006) adopted qualitative methodologies. Although results of the present MDS analysis uncovered a gendered pathway as observed in previous investigations, replication of MDS on an independent sample is recommended to establish the consistency and stability of the solution.

Recall that a liberal criterion for establishing thematic dominance was applied in this study to maximize classification potential or in other words, to reduce the number of unclassifiable (hybrid) cases; specifically, a youth had to display a greater proportion of items from one theme than the other. Researchers in the area of investigative psychology have been inconsistent in the stringency of the criterion set to allocate offenders to themes (e.g., proportional score double that of the other theme [Hakkanen, Lindlof, & Santtila, 2004]; proportional score 1.5 times that of the other theme [Trojan & Salfati, 2008]; proportional score of one theme exceeding that of the other theme [Salfati & Bateman, 2005]). Within investigative and correctional psychology, the ultimate goal would be to adopt a standard criterion for establishing thematic dominance that permits adequate discrimination while maximizing the number of classifiable cases.

Just as feminist scholars infer pathways retrospectively through narrative accounts (e.g., Chesney-Lind & Shelden, 2003), we inferred “pathways” retrospectively through empirically determined item clusters. As noted earlier, while we adopted the term “pathways” to be consistent with feminist terminology, we acknowledge that causality cannot be established given the cross-sectional sampling of items; longitudinal research is required to confirm the temporal placement of vulnerability factors along trajectories into crime for young females and males, respectively. Notwithstanding this caveat, of greatest relevance was not the temporal order of items but the high degree of concomitance between indicators (e.g., abuse and mental health issues in the subsample of females). Ultimately, highly correlated item subsets reflected hypothesized themes and confirmed the existence of a unique gendered pathway into crime for young female offenders.

With respect to the data itself, recall that there was a disproportionate gender distribution within PINS and JD categories (i.e., status offenders vs. criminal offenders), with significantly more females than males identified as PINS and significantly more males than females identified as JDs. While an examination of gender differences within PINS and JD groups would constitute an interesting study in itself, we opted to combine the groups for the purposes of this research. Youths in our database represent a stratified random sample of the juvenile offender population of New York State. Hence, these gender disparities are simply a reflection of a true difference within the justice population of interest.

Regrettably, we were restricted in the number of available and relevant items we could extract from the YASI Full Assessment, as well as by the relatively crude dichotomous measures of these variables. It is probable that a more comprehensive sample of items would have generated MDS solutions with higher reliability indices (Canter et al., 2003). Regarding the indicator ratings, it has been found that dichotomous decision-making in content analysis actually has the capacity to increase inter-rater reliability from 60% to 90% (Krippendorff, 1980), allowing the coder to focus on rendering one simple decision at a time. While dichotomous ratings do have the potential to increase reliability, quantifying the magnitude of expression of certain variables might ultimately be helpful in introducing important contextual elements to one’s criminal etiology. For example, a youth who has attempted to run away from home on one occasion to stay with friends or relatives likely faces different circumstances from a youth who perpetually runs away (or is kicked out of his or her home) and is subsequently forced to subsist without adequate shelter. In addition, more comprehensive information on offense history including severity, length, and context (e.g., victimology) may have permitted the identification of additional themes. For example, while we were able to infer a broad gendered pathways theme for females, the lack of contextual detail precluded the delineation of finer distinctions akin to those presented by Daly (1992).

Notably, being limited to the pool of available YASI items did not permit consideration of additional criminogenic factors that are ostensibly salient and/or specific to females (e.g., whether one’s criminal peer is a romantic partner or a simple acquaintance). In an attempt to adopt a gender-informed approach, Orbis Partners recently developed the YASI-Girls (YASI-G). Much of the tool parallels the YASI in its gender-neutral form, but further encompasses items deemed foundational to the criminal behavior of young females according to the feminist and gender-responsive literatures. Specifically, the measure includes additional items pertinent to the nature of one’s relationships, levels of emotional expression, self-efficacy, sexual vulnerability (e.g., prostitution), early parenthood, and features a broader array of mental health indicators. Although the YASI-G has not yet been implemented, it is presently being researched by the second author in the context of studying pathways to crime and desistance in a sample of Canadian justice-involved youth.

As indicated in the methodology, there are no available measures of inter-rater reliability on the YASI with New York State staff. As such, the extent to which probation officers and other frontline staff are actually observing the instructional specifications related to item scoring is presently unknown. Although we only extracted select items from the tool and some of these could arguably be scored with relative objectivity and corroborated by file review (e.g., court finding of child neglect, PINS/JD complaints), other indicators are inherently more subjective (e.g., antisocial attitudes). Encouragingly, quality assurance studies with the YASI are emerging (e.g., Skeem et al., 2012). Moreover, recent research on the field reliability of an alternative violence risk assessment tool tailored to juvenile justice populations suggests that with sufficient training and motivation, probation officers are highly capable of producing reliable ratings on risk assessment protocols (Vincent, Guy, Fusco, & Gershenson, 2012).

Conclusion

In sum, results of this investigation lend empirical support to the claims of feminist scholars that female offenders follow unique and distinct pathways into crime. However, these trajectories are not necessarily limited to those depicted in feminist pathways research (Daly, 1992), but additionally include pathways heavily comprised of factors articulated in the mainstream gender-neutral literature (e.g., Andrews & Bonta, 2010; Moffitt, 1993). It is our contention that these ostensibly polarized paradigms are not as distinct as their respective proponents would suggest, and that correctional theory is best advanced through theoretical and methodological integration.

Footnotes

Acknowledgements

We thank Orbis Partners and the New York State Division of Criminal Justice Services for access to data. We are also grateful to our anonymous reviewers for their valuable feedback.

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

Notes

Author Biographies

Natalie J. Jones, PhD, is the director of research at Orbis Partners in Ottawa, Ontario, Canada. In this role, she is responsible for managing justice-related research projects, including the validation of risk assessment tools, program evaluations, and the provision of research services to clients across the United States and Canada. Her primary research interests lie in strength-based and gender-responsive strategies for assessment and intervention with correctional populations.

Shelley L. Brown, PhD, is an associate professor within the Department of Psychology, Carleton University, Ottawa, Ontario, Canada. She uses mixed methods to improve the assessment and treatment of justice-involved girls and women.

Kayla A. Wanamaker is presently completing her MA in psychology at Carleton University in Ottawa, Ontario, Canada. Her research interests include the impact of childhood trauma on criminal behavior in youth populations, as well as gender differences among justice-involved youth.

Leigh E. Greiner is presently a PhD candidate at Carleton University in Ottawa, Ontario, Canada. Her research interests include gender and crime issues, and the effects of trauma, mental health, and substance use on the delinquent behavior of adolescent offenders.

References

Anastasi

(1988). Psychological testing (6th ed.). New York, NY: MacMillan.

Andrews

D. A.

Bonta

(2010). The psychology of criminal conduct (5th ed.). Cincinnati, OH: Anderson Publishing.

Andrews

D. A.

Guzzo

Raynor

Rowe

R. C.

Rettinger

L. J.

Brews

Wormith

J. S.

(2012). Are the major risk/need factors predictive of both female and male reoffending? A test with the eight domains of the level of service/case management inventory. International Journal of Offender Therapy and Comparative Criminology, 56, 113-133. doi:10.1177/0306624X10395716

Barnoski

(2003). Changes in Washington State’s jurisdiction of juvenile offenders: Examining the impact. Olympia: Washington State Institute of Public Policy.

Belknap

(2007). The invisible woman: Gender, crime and justice (3rd ed.). Belmont, CA: Thomson Wadsworth.

Belsky

Steinberg

Houts

R. M.

Halpern-Felsher

B. L.

(2010). The development of reproductive strategy in females: Early maternal harshness → earlier menarche → increased sexual risk taking. Developmental Psychology, 46, 120-128. doi:10.1037/a0015549

Borg

Shye

(1995). Facet theory: Form and content. Thousand Oaks, CA: Sage.

Bottcher

(2001). Social practices of gender: How gender relates to delinquency in the everyday lives of high risk youths. Criminology, 39, 893-932. doi:10.1111/j.1745-9125.2001.tb00944.x

Brennan

Breitenbach

Dieterich

Salisbury

E. J.

Van Voorhis

(2012). Women’s pathways to serious and habitual crime: A person-centered analysis incorporating gender responsive factors. Criminal Justice and Behavior, 39, 1481-1508. doi:10.1177/0093854804265179

10.

Brown

S. L.

Geck

Harris

Skilling

(2012). Reliability and validity of the Youth Assessment Screening Inventory (YASI) in a sample of Canadian youth. Manuscript in preparation.

11.

Burt

S. A.

McGue

DeMarte

J. A.

Krueger

R. F.

Iacono

W. G.

(2006). Timing of menarche and the origins of conduct disorder. Archives of General Psychiatry, 63, 890-896. doi:10.1001/archpsyc.63.8.890

12.

Canter

D. V.

(1985). Facet theory: Approaches to social research. New York, NY: Springer.

13.

Canter

D. V.

Bennell

Alison

L. J.

Reddy

(2003). Differentiating sex offences: A behaviorally based thematic classification of stranger rapes. Behavioral Sciences & the Law, 21, 157-174. doi:10.1002/bsl.526

14.

Canter

D. V.

Heritage

(1990). A multivariate model of sexual offence behavior: Developments in “offender profiling.” Journal of Forensic Psychiatry, 1, 185-212. doi:10.1080/09585189008408469

15.

Canter

D. V.

Hughes

Kirby

(1998). Paedophilia: Pathology, criminality, or both? The development of a multivariate model of offence behaviour in child sexual abuse. Journal of Forensic Psychiatry, 9, 532-555. doi:10.1080/09585189808405372

16.

Canter

D. V.

Wentink

(2004). An empirical test of Holmes and Holmes’ serial murder typology. Criminal Justice and Behavior, 31, 489-515. doi: 10.1177/0093854804265179

17.

Cauffman

Steinberg

(2000). (Im)maturity of judgment in adolescence: Why adolescents may be less culpable than adults. Behavioral Sciences & the Law, 18, 741-760. doi:10.1002/bsl.416

18.

Chaiken

M. R.

Chaiken

J. M.

(1984). Offender types and public policy. Crime & Delinquency, 30, 195-226. doi:10.1177/0011128784030002003

19.

Chesney-Lind

Shelden

R. G.

(2003). Girls, delinquency, and juvenile justice (3rd ed.). Belmont, CA: Wadsworth Publishing Company.

20.

Chumlea

W. C.

Schubert

C. M.

Roche

A. F.

Kulin

H. E.

Lee

P. A.

Himes

J. H.

Sun

S. S.

(2003). Age at menarche and racial comparisons in US girls. Pediatrics, 111, 110-113. doi:10.1542/peds.111.1.110

21.

Commandeur

J. J. F.

Heiser

W. J.

(1993). Mathematical derivations in the proximity scaling (PROXSCAL) of symmetric data matrices (Report No. RR-93–03). Leiden, Netherlands: Leiden University.

22.

Covington

S. S.

Bloom

B. E.

(2007). Gender responsive treatment and services in correctional settings. Women & Therapy, 29, 9-33. doi:10.1300/J015v29n03_02

23.

Daly

(1992). Women’s pathways to felony court: Feminist theories of lawbreaking and problems of representation. Review of Law and Women’s Studies, 2(11), 11-52.

24.

Daly

(1994). Gender, crime, and punishment. New Haven, CT: Yale University Press.

25.

Dowden

Andrews

D. A.

(1999). What works for female offenders: A meta-analytic review. Crime & Delinquency, 45, 438-452. doi:10.1177/0011128799045004002

26.

Dowden

Blanchette

(1998). An investigation into the characteristics of substance-abusing women offenders: Risk, need, and post-release outcome (Research Report R-81). Ottawa, Ontario, Canada: Research Branch, Correctional Service Canada.

27.

Feld

B. C.

(2009). Violent girls or relabeled status offenders? An alternative interpretation of the data. Crime & Delinquency, 55, 241-265. doi:10.1177/0011128708330629

28.

Gendreau

P. C.

Smith

French

(2006).The theory of effective correctional intervention: Empirical status and future directions. In Cullen

Wright

Coleman

(Eds.), Taking stock: The status of criminology theory (pp. 419-446). Piscataway, NJ: Transaction Press.

29.

Greiner

Brown

S. L.

Skilling

(2012, March). What the raw score doesn’t tell us: Capturing context in risk assessment. Paper presented at the 2012 American Psychology and Law Society Conference. San Juan, Puerto Rico.

30.

Guttman

Greenbaum

C. W.

(1998). Facet theory: Its development and current status. European Psychologist, 3, 13-36. doi:10.1027//1016-9040.3.1.13

31.

Hakkanen

Lindlof

Santtila

(2004). Crime scene actions and offender characteristics in a sample of Finnish stranger rapes. Journal of Investigative Psychology and Offender Profiling, 1, 17-32. doi:10.1002/jip.001

32.

Krippendorff

(1980). Content analysis: An introduction to its methodology. Beverly Hills, CA: Sage.

33.

Kruskal

J. B.

Wish

(1978). Multidimensional scaling. Beverly Hills, CA: Sage.

34.

Lingoes

J. C.

(1979). Geometric representation of relational data. Ann Arbor, MI: Mathesis Press.

35.

Loeber

Farrington

D. P.

(2000). Young children who commit crime: Epidemiology, developmental origins, risk factors, early interventions, and policy implications. Development and Psychopathology, 12, 737-762. doi:10.1017/S0954579400004107

36.

Moffitt

T. E.

(1993). “Life-course persistent” and “adolescent limited” antisocial behavior: A developmental taxonomy. Psychological Review, 100, 674-701. doi:10.1037/0033-295X.100.4.674

37.

New York State Division of Criminal Justice Services. (2012). Frequently used terms and phrases (glossary). Retrieved from http://www.criminaljustice.ny.gov/opca/glossary.htm

38.

Onifade

Davidson

Livsey

Turke

Horton

Malinowski

. . . Wimberly

(2008). Risk assessment: Identifying patterns of risk in young offenders with the Youth Level of Service/Case Management Inventory. Journal of Criminal Justice, 36, 165-173. doi:10.1016/j.jcrimjus.2008.02.006

39.

Orbis Partners. (2000). General Youth Assessment Screening Inventory (YASI). Ottawa, Ontario, Canada: Author.

40.

Orbis Partners. (2007a). Long-term validation of the Youth Assessment and Screening Instrument (YASI) in New York State juvenile probation. Retrieved from http://dpca.state.ny.us/pdfs/nyltyasifullreport20feb08.pdf

41.

Orbis Partners. (2007b). YASI: Brief report. Unpublished manuscript.

42.

Orbis Partners. (2012). Validation of SPIn and YASI for use by Alberta Solicitor General and Public Security. Ottawa, Ontario, Canada: Author.

43.

Reisig

M. D.

Holtfreter

Morash

(2006). Assessing recidivism risk across female pathways to crime. Justice Quarterly, 23, 384-405. doi:10.1177/0093854809349438

44.

Ross

R. R.

Fabiano

E. A.

(1986). Female offenders: Correctional afterthoughts. Jefferson, NC: McFarland.

45.

Salfati

C. G.

Bateman

A. L.

(2005). Serial homicide: An investigation of behavioral consistency. Journal of Investigative Psychology and Offender Profiling, 2, 121-144. doi:10.1002/jip.27

46.

Salisbury

E. J.

Van Voorhis

(2009). Gendered pathways: A quantitative investigation of women probationers’ paths to incarceration. Criminal Justice and Behavior, 36, 541-566. doi:10.1177/0093854809334076

47.

Sickmund

Sladky

Kang

(2012). Easy access to juvenile court statistics: 1985-2009 (National Centre for Juvenile Justice). Retrieved from http://www.ojjdp.gov/ojstatbb/ezajcs/

48.

Skeem

Hernandez

Kennealy

Rich

(2012). CA-YASI reliability: How adequately do staff in California’s Division of Juvenile Justice rate youths’ risk of recidivism? Irvine: University of California.

49.

Slobogin

Fondacaro

(2011). Juveniles at risk: A plea for preventive justice. New York, NY: Oxford University Press.

50.

Snyder

H. N.

(2011). Patterns and trends: Arrests in the United States, 1980-2009. U.S. Department of Justice, Office of Justice Programs, Bureau of Justice Statistics. Retrieved from United http://bjs.ojp.usdoj.gov/content/pub/pdf/aus8009.pdf

51.

Steffensmeier

Schwartz

Zhong

Ackerman

(2005). An assessment of recent trends in girls’ violence using diverse longitudinal sources: Is the gender gap closing? Criminology, 43, 355-406. doi:10.1111/j.0011-1348.2005.00011.x

52.

Trojan

Salfati

(2008). Methodological considerations of determining dominance in multidimensional analyses of crime scene behaviours and offender characteristics. Journal of Investigative Psychology and Offender Profiling, 5, 125-145. doi:10.1002/jip.88

53.

Van Voorhis

Wright

E. M.

Salisbury

Bauman

(2010). Women’s risk factors and their contributions to existing risk/needs assessment: The current status of a gender-responsive supplement. Criminal Justice and Behavior, 37, 261-288. doi:10.1177/0093854809357442

54.

Vincent

G. M.

Guy

L. S.

Fusco

S. L.

Gershenson

B. G.

(2012). Field reliability of the SAVRY with juvenile probation officers: Implications for training. Law and Human Behavior, 36, 225-236. doi:10.1037/h0093974