Starting With Girls and Their Resilience in Mind: Reconsidering Risk/Needs Assessments for System-Involved Girls

• Author: Linsey A. Belisle,Emily J. Salisbury
• DOI: 10.1177/0093854820983859
• Published date: 01 May 2021
• Date: 01 May 2021
• Subject Matter: Articles

CRIMINAL JUSTICE AND BEHAVIOR, 2021, Vol. 48, No. 5, May 2021, 596 –616.

DOI: https://doi.org/10.1177/0093854820983859

Article reuse guidelines: sagepub.com/journals-permissions

© 2021 International Association for Correctional and Forensic Psychology

596

STARTING WITH GIRLS AND THEIR

RESILIENCE IN MIND

Reconsidering Risk/Needs Assessments for

System-Involved Girls

LINSEY A. BELISLE

University of Nevada, Las Vegas

EMILY J. SALISBURY

University of Utah

Despite the numerous differences between boys and girls in the juvenile justice system, there are currently no gender-

responsive risk/needs assessments for system-impacted girls, thereby limiting agencies to the use of gender-neutral risk/needs

assessments. This article examines commonly used gender-neutral risk/needs assessments and illustrates how these instru-

ments are not truly “neutral.” We argue that predictive validity is not enough to demonstrate effectiveness; these tools can

harm and possibly discriminate against girls by placing them in similarly labeled risk categories (i.e., high, medium, low) as

boys, despite engaging in less delinquency. This practice of force-fitting girls to assessments primarily developed for boys

results in over- and misclassification of girls’ risk and fails to capture their gendered needs and resilience. We see this as an

opportunity to reconsider risk assessments altogether for girls and propose a resilience/needs assessment may be better suited

to identify girls’ needs and predict future behavior.

Keywords: risk assessment; gender; juvenile justice; delinquency; gender differences

INTRODUCTION

Risk and needs assessments guide many decisions within correctional and juvenile jus-

tice agencies, including risk level, custody placement, and access to treatment (Bonta &

Andrews, 2016). The literature has broadly categorized these assessments as either: (a)

gender-neutral or (b) gender-responsive. Gender-neutral risk/needs assessments are devel-

oped primarily based on justice-involved boys/men and then applied to girls/women with

AUTHORS’ NOTE: Special thanks to the anonymous reviewers and the editorial team at Criminal Justice and

Behavior, especially Dr. Jaime Henderson, Managing Editor. We would also like to thank the staff and girls at

Oak Creek Youth Correctional Facility for their incredible insight and willingness to share their views on this

topic. Correspondence concerning this article should be addressed to Emily J. Salisbury, Utah Criminal Justice

Center, College of Social Work, University of Utah, 395 South 1500 East #111, Salt Lake City, UT 84112;

e-mail: emily.salisbury@utah.edu.

983859CJBXXX10.1177/0093854820983859Criminal Justice and BehaviorBelisle, Salisbury / SHORT TITLE

article-commentary2021

Belisle, Salisbury / GIRLS’ RISK/NEEDS ASSESSMENT 597

little consideration of gendered differences (Van Voorhis et al., 2010). Gender-responsive

risk/needs assessments are created specifically for justice-involved girls/women and incor-

porate both gender-neutral and gender-responsive scales (Salisbury et al., 2016). Gender-

responsive assessments originated from the conceptualization of gender-responsive

correctional strategies (Bloom, Owen, & Covington, 2003). Despite the considerable

amount of research demonstrating that justice-involved boys and girls1 have different path-

ways to delinquency, risk factors, needs, and strengths (Belknap & Holsinger, 2006; Bloom,

Owen, Rosenbaum, & Deschenes, 2003; Zahn, 2007; Zahn et al., 2008), no gender-respon-

sive juvenile risk/needs assessment exists for system-impacted girls.

Although some studies found gender-neutral instruments to have acceptable predictive

accuracy for both boys and girls in the juvenile justice system (see Pusch & Holtfreter,

2018; Schwalbe, 2008), numerous studies have shown that gender-neutral risk/needs assess-

ments do not work as well for girls as they do for boys and tend to (a) overclassify girls’ risk

levels and (b) fail to capture girls’ unique gendered needs (see Baird et al., 2013; Emeka &

Sorensen, 2009; Onifade et al., 2009). Overclassification of risk occurs when risk/needs

assessments overestimate a group’s likelihood of committing a future offense (see Hardyman

& Van Voorhis, 2004; Skeem et al., 2016). This malpractice harms girls by placing them in

more restrictive custody and supervision levels, limiting their eligibility to participate in

certain programming (Bloom & Covington, 2001; Salisbury et al., 2016; Skeem et al.,

2016), and can contribute to staff assuming the level of danger girls (or women) present as

more inflated than it actually is (Skeem et al., 2016). In addition, instruments built primarily

for boys fail to capture girls’ unique gendered needs, reducing the likelihood that girls

receive the treatment required to divert them from further involvement in the system (Bloom

& Covington, 2001; Bloom, Owen, & Covington, 2003; Bloom, Owen, Rosenbaum, &

Deschenes, 2003) This is not unlike the assessment research with justice-involved women

(e.g., inaccurate placement, overclassification, failure to identify gendered needs; Corbett-

Davis & Goel, 2018; Hamilton, 2019; Hardyman & Van Voorhis, 2004; Skeem et al., 2016).

Although some juvenile gender-neutral tools have demonstrated acceptable predictive

validity for girls (see Pusch & Holtfreter, 2018; Schwalbe, 2008), we argue that focusing

solely on predictive validity is not enough to deem an instrument effective for girls. As we

illustrate below, this is partly because gender-neutral risk instruments do not function nearly

as well for girls as they do for boys. The myriad disadvantages to girls and women persist

because of the over-reliance on a single psychometric property to demonstrate an instrument’s

effectiveness across gender: predictive validity. While prediction is a central goal among

research and criminal justice decision-making (Gottfredson, 1987), it does not mean much if

the content validity of assessments is in question. Content validity measures the extent to

which all aspects of a construct are captured (Rust & Golombok, 1989). Importantly, con-

structs can vary greatly across cultures and gender, and investigating these distinctions are

common sources of inquiry in the fields of psychological and educational measurement (e.g.,

Gilligan, 1982; Prince, 2008). Thus, criminologists too have a responsibility to step back and

address the significant content validity problem in these gender-neutral assessments: Are we

truly measuring all facets of all constructs for all groups of people among each risk factor, and

what content might be missing altogether for certain groups, such as girls (and girls of color)?2

Moreover, criminal justice agencies are ethically and legally required to provide equal

treatment to all (see American Bar Association, 2018; United & Nations, 1985, 2010). Over

the decades, agencies have translated “equal treatment” to mean the same treatment (where

the male-based approach is the normative approach); but identical treatment is not equal