Evaluating the Clinical Utility of the MAYSI-2 Among African American Male Juvenile Offenders

Abstract

This study examined the clinical utility of the Massachusetts Youth Screening Instrument–Second Version (MAYSI-2) among African American (AA) incarcerated youth and used White incarcerated youth as a comparison group. Data were analyzed for 314 incarcerated youth (193 AA offenders and 121 White offenders) of ages 13–17 years who were adjudicated delinquent from a southeastern United States medium security residential facility. Seven logistic regression and receiver operating characteristic curve (ROC) models were built to determine whether the MAYSI-2 subscales accurately identify committed AA male incarcerated youth who have a mental illness diagnosis on file. Analyses also examined how well the MAYSI-2 subscales identify specific mental illnesses among AA-committed male incarcerated youth. Results demonstrated that no MAYSI-2 subscales accurately identified and categorized AA-committed male incarcerated youth who have mental disorders, and only two subscales (Alcohol/Drug Use, Depressed/Anxious) identified and categorized White committed male incarcerated youth who have a mental disorder. Additional results and implications for research and practice are provided.

Keywords

juvenile corrections juvenile justice MAYSI-2 African American mental health

The United Nations (U.N.) has long asserted a commitment to the rehabilitation of incarcerated youth as evidenced by the U.N. Rules for the Protection of Juveniles Deprived of their Liberty (United Nations General Assembly, 1990). Similarly, rehabilitation is the underlying premise and promise of the U.S. juvenile justice (JJ) system (Gagnon et al., 2022). Broadly, the U.S. approaches incarcerated youth rehabilitation with the notion that they are qualitatively different than adults, a view that has been confirmed by recent neuroscience research (Arain et al., 2013). However, the goal of rehabilitation is often complicated by the complex needs of incarcerated youth, which often includes mental disorders. While data does not exist for youth in commitment facilities (i.e., those adjudicated delinquent), prevalence data from detained incarcerated youth indicates that 70% of them have at least one diagnosed mental health disorder in comparison to only 20% of adolescents in the general population (Seiter, 2017). In addition, Seiter (2017) noted that their comorbidity rates are ten times higher for incarcerated youth (e.g., 30% have 4 or more diagnoses). Conduct disorder and substance use disorder are the most common comorbid disorders, followed by mood, disruptive, and trauma disorders (Seiter, 2017). Over 60% of incarcerated youth with a mental health disorder also have a co-occurring substance use disorder and almost 30% have mental health disorders that are serious enough to require immediate and significant treatment (National Center for Mental Health and Juvenile Justice [NCMHJJ]). Moreover, incarcerated youth with mental illnesses who are incarcerated may have their symptoms actually worsen due to facility overcrowding, lack of available treatment/services, and separation from support systems (i.e. familial supports) while incarcerated (National Institute of Justice, 2016).

Identifying and addressing the mental health needs of incarcerated youth are critical for rehabilitation, given that there may be links between adequate mental health services recidivism rates (Zeola et al., 2017). A lack of appropriate services also disproportionally affects minority incarcerated youth. African American (AA) incarcerated youth account for 41% of the incarcerated youth in detention and commitment facilities and AA males constitute 42% of the residential placement population (Hockenberry, 2018). These incarcerated youth have the highest rates of mental health illness (Seiter, 2017) and are less likely to receive mental health treatment while incarcerated (Baglivio et al., 2017).

Mental Health Screening

Despite decades of reform, the JJ system has yet to determine a systematic method to address the mental health needs of incarcerated youth (NCMHJJ, 2013). Some authors posit that this issue is an acute crisis that must be addressed (Grisso et al., 2005). Researchers (see Seiter, 2017) argue that the first step to providing effective mental health services to incarcerated youth who enter JJ facilities is to formalize screening and assessment procedures and use validated tools that can quickly and correctly identify problems that need evaluation. Screening tools have the potential to be effective in JJ settings because of their ability to identify the needs of incarcerated youth in an expeditious manner. Mental health screening is a brief process (10–15 minutes) that is often completed by non-clinical staff within the first 48 hours of the incarcerated youth’ arrival (NCMHJJ, 2013). The Children’s Defense Fund (CDF, 2012) advocates for the use of screening tools in JJ facilities and conveys five main reasons why they are effective. They argue: (a) screening provides facilities with a method of briefly evaluating each incarcerated youth, identifying high-risk incarcerated youth, and responding to their immediate mental health needs; (b) it helps identify serious mental health issues in incarcerated youth in a timely manner before their condition deteriorates; (c) it guides the allocation of time and money toward the care of incarcerated youth with greatest needs; (d) it increases efficiency of decision-making related to assessments post-screening; and (e) it assists facilities in providing mental health care appropriate to the needs of each incarcerated youth.

Mental health screening is important for all incarcerated youth and especially for minority incarcerated youth, given that youth of color are overrepresented in the JJ system. However, researchers (Mallett & Stoddard-Dare, 2010; Sickmund et al., 2019) reported that minority incarcerated youth are less likely than their White counterparts to be identified as needing mental health services. This raises the question of whether current screeners are accurately identifying mental health needs for minority incarcerated youth in the JJ system. Best practice for mental health assessment requires that mental health screeners and assessments be equivalent in utility for incarcerated youth across all races (Groth-Marnat, 2009). If mental health screeners have less utility in identifying mental health needs among minorities, then it places already vulnerable populations at a further disadvantage because they would be less likely to, (a) be identified, (b) have comprehensive mental health assessments completed, and (c) receive treatment referrals and other specific programming supports (Louden et al., 2017).

Massachusetts Juvenile Offenders Screening Instrument–Second Version (MAYSI-2)

The Massachusetts Youth Screening Instrument–Second Version (MAYSI-2; Grisso & Barnum, 2000) is a brief self-report mental health screening instrument that was designed to be administered upon intake into JJ facilities or a transitional placement point in the JJ system (Grisso & Barnum, 2006). Its purpose is to identify youth ages 12–17 within JJ settings who are experiencing thoughts, feelings, or behaviors that may be indicative of mental disorders and/or who are in acute emotional crisis that requires immediate attention (Grisso et al., 2005). The MAYSI-2 was designed specifically for JJ intake personnel to meet the need for a standardized, reliable, and valid screening instrument that can be used with every juvenile entering juvenile facilities. The MAYSI-2 is the most widely used mental health screener in JJ settings nationwide (CDF, 2012) and includes statewide use in all intake probation, detention and/or corrections facilities in 44 states (Grisso et al., 2012).

While the MAYSI-2 is a popular screening tool in JJ settings, studies have demonstrated differences in scores across races for detained youth. MAYSI-2 scores tended to be higher for White (non-Hispanic) detained youth than for detained youth of minority backgrounds (Cauffman & MacIntosh, 2006; Wasserman et al., 2005). For example, Vincent et al. (2008) found that White (Non-Hispanic) detained youth were more likely than Latino or AA detained youth to self-report alcohol and drug use and suicide ideation. While research results have been mixed, Cauffman and MacIntosh (2006) found that differences in symptom reporting is a result of differential item functioning. Overall, these data demonstrate that there are differences in symptom reporting by race.

Yet, none of the studies described above evaluated the degree to which the MAYSI-2 accurately categorized mental disorders by race for incarcerated youth in diversion programs or those who have been adjudicated delinquent and are in commitment facilities. Specifically, “diversion” is a set of procedures and/or programs that address youth delinquent behavior in a system other than the formal juvenile court (e.g., community-based mental health services; Juvenile Law Center, 2023). Additionally, “adjudication of delinquency” refers to, “a juvenile court judge’s determination as to whether or not a youth committed a delinquent offense. A juvenile adjudication is like an adult criminal conviction but generally does not subject the youth to the same direct and collateral consequences” (Juvenile Law Center, 2023). Louden et al. (2017) recognized this crucial literature gap and addressed it for Latino youth in diversion programs and represents the only study that specifically evaluated the clinical utility of the MAYSI-2 by race. From this research, Louden et al. found that the MAYSI-2 had similar clinical utility in identifying mood and anxiety disorders for Latino and White youth in diversion programs. However, the ROC analyses demonstrated that the MAYSI-2 was less sensitive to behavioral and substance use disorders among Latinos than it was for Whites.

Statement of the Problem

Cauffman and MacIntosh (2006) stated that the disproportionate number of incarcerated youth from minority backgrounds, especially AA-incarcerated youth, poses a significant concern with regard to the racial sensitivity of mental health screeners. Seiter (2017) argues that the lack of mental health services for justice-involved AA-incarcerated youth may indicate bias on the screeners. This is based on the fact that the screeners identify mental illness among Whites in JJ settings at a higher rate than AAs, despite the fact that AA-incarcerated youth in JJ samples have greater mental health diagnoses in their records. It is vital that the mental health screeners used in JJ facilities, have strong clinical utility for AA-incarcerated youth, since they are the least likely to receive services (Prud’homme, 2018). This study is of particular importance because of how frequent the MAYSI-2 is used in JJ facilities (Swank & Gagnon, 2017), the overrepresentation of AA males, and the lack of research on the clinical utility of its use on adjudicated youth in residential facilities.

Purpose of the Study

The purpose of this exploratory quantitative study is to assess the clinical utility of the most widely used mental health screener in JJ facilities, specifically for a population of AA and White adjudicated male youth male incarcerated in a secure residential facility. The two main objectives of the data analyses are to: (a) determine whether the MAYSI-2 subscales accurately identify AA-committed male incarcerated youth with mental illness on file and (b) examine how well the MAYSI-2 subscales identify mental illness among AA-committed male incarcerated youth. To determine how well the MAYSI-2 subscales identify AA-committed male incarcerated youth with mental illnesses, White youth were the comparison group. To address these objectives, logistic regressions, and the receiver operating characteristic curve (ROC) analyses were utilized. The following research questions and hypotheses guided the study.

Research Question 1: Do the MAYSI-2 subscales (Alcohol/Drug Use, Angry-Irritable, Depressed/Anxious, Somatic Complaints, Traumatic Experiences) accurately identify (i.e., follow up for potential disorder/no follow-up necessary) AA-committed male incarcerated youth who have mental disorders? Due to the previous research, albeit limited, it is hypothesized that each of the subscales does not accurately identify AA-committed incarcerated youth with the associated mental disorder.

Research Question 2: How well do the MAYSI-2 subscales (Alcohol/Drug Use, Angry-Irritable, Depressed/Anxious, Somatic Complaints, and Traumatic Experiences) identify A-committed male incarcerated youth’ diagnosed mental disorders? Based on available research, it is hypothesized that each subscale, with the exception of Traumatic Experiences, moderately identifies AA-committed male incarcerated youth with mental disorders.

Methods

Research Design

In the current study, a cross-sectional approach was utilized (Hulley, 2007; Smith et al., 2011). Cross-sectional designs allow for the evaluation of correlations between independent and dependent variables to be able to draw conclusions on the outcome of interest (Levin, 2006). The current study seeks to evaluate the association of mental health diagnoses, race, and MAYSI-2 subscale scores.

Sample

Following institutional review board approval, de-identified archival data on youth diagnosis of mental disorders and MAYSI-2 data collected at the facility were used in this study. The data were collected between August 2009 and August 2013 and have not been utilized in any previous study. Participants were adolescent males ages 12–17 who were adjudicated delinquent and serving their sentence in a rural private medium security juvenile correctional residential facility in the southeastern United States. Incarcerated youth in this facility are from a wide swath of the east coast.

The initial sample included 439 male participants, with 51.7% (n = 227) AA, 37.4% (n = 16) White, 9.4% (n = 41) Latino, 1.1% (n = 5) Mixed Race (i.e., incarcerated youth who identify as both AA and White), and .5% (n = 2) Asian/Pacific Islander. Due to the small number of incarcerated youth identifying as Latino, Mixed Race, and Asian/Pacific Islander, these incarcerated youth were not included in the final sample and analysis. Also, 38 incarcerated youth in the initial sample were excluded because they were not within the age range necessary (i.e., 12–17) for valid MAYSI-2 scores. As such, the final sample was 314 incarcerated youth. Of these, 2.3% were 13 years old (n = 10), 15% were 14 years old (n = 66), 26% were 15 years old (n = 114), 26% were 16 years old (n = 114) and 30.8% were 17 years old (n = 135).

Data Collection

The current study uses two sources of data. Upon entry into the facility, each student was provided a code number to ensure data was de-identified and to maintain confidentiality. The MAYSI-2 was administered to the incarcerated youth during their intake into the facility. Study data were collected by two groups of trained graduate students. Both groups of graduate students received training for the standardized administration of the screener. One group administered the MAYSI-2 via reading the items to each incarcerated youth, and the other graduate students recorded their responses. The other graduate students were also directed to review the medical records for each student’s most recent diagnosis or diagnoses, if any existed, as well as information on race. Information was recorded for each student using their code number. Examiners were blind to the diagnoses of each incarcerated youth.

Measures

The MAYSI-2 (Grisso & Barnum, 2000) is a 52-item brief self-report mental health screening instrument designed to be administered upon intake into JJ facilities or a transitional placement point in the JJ system. The MAYSI-2 is intended to identify incarcerated youth ages 12–17 who are experiencing thoughts, feelings, or behaviors that may be indicative of mental disorders and/or who are in acute emotional crisis that requires immediate attention (Grisso et al., 2005). The norms for the MAYSI-2 were established in 2001, and previous psychometric studies of the MAYSI-2 demonstrated that it has good internal consistency (Grisso et al., 2001), test-retest reliability (Cauffman, 2004; Grisso et al., 2001) and concurrent validity for specific subscales (Alcohol/Drug Abuse, Suicidal Ideation; Grisso et al., 2001) with the CBCL-youth Self Report (YSR) and the Millon Adolescent Clinical Inventory (MACI).

The MAYSI-2 does not provide a total score. Rather, the MAYSI–2 yields scores on seven subscales: Alcohol/Drug Use, Angry–Irritable, Depressed–Anxious, Somatic Complaints, Suicide Ideation, Traumatic Experiences, and Thought Disturbance (Grisso et al., 2001). Each of the subscales, except the Traumatic Experiences subscale, produces both a numerical score and a categorical score that identifies incarcerated youth by the severity of their symptoms. The Traumatic Experiences subscale is not a measure of clinical symptoms. Rather, it indicates whether an incarcerated youth has been exposed to traumatic events. Scores for the other subscales categorize incarcerated youth into one of three categories: low risk, caution, and warning (Grisso & Barnum, 2006; Louden et al., 2017).

Cut-off scores for the MAYSI-2 were developed based on its comparability with clinically significant scores on the MACI and Achenbach’s Incarcerated YSR (Grisso & Barnum, 2006). The cut-off scores ranges vary for each MAYSI-2 subscale score. Specifically, the Alcohol/Drug Use subscale scores range from 0–8. Scores of 0–3 are in the low risk range, scores of 4–6 are in the warning range and scores of 7–8 are in the caution range. The Angry/Irritable subscale scores range from 0–9. Scores of 0–4 are in the low-risk range, scores of 5–7 are in the warning range and scores of 8–9 are in the caution range. The Depressed/Anxious subscale scores range from 0 to 7. Scores of 0–2 are in the low-risk range, scores of 3–5 are in the warning range and scores of 6–7 are in the caution range. The Somatic Complaints subscale scores range from 0 to 6. Scores of 0–2 are in the low-risk range, scores of 3–5 are in the warning range and a score of 6 is in the caution range. Finally, the Traumatic Experiences subscale scores range from 0 to 4. A score of 0 is in the low-risk range and any score 1 or above is in the elevated-risk range. In terms of interpretation, Caution cut-off scores indicate that there are some problems in that specific domain. The caution category was designed to identify adolescents who would be likely to obtain clinically significant scores on a more in depth assessment of psychopathology (Archer et al., 2010; Grisso & Barnum, 2006; Louden et al., 2017). The warning cut-off score identifies incarcerated youth with the highest, most severe needs in that domain. These incarcerated youth scored in the top 10% of incarcerated youth in terms of mental health problems when compared to other incarcerated youth in the JJ system (Archer et al., 2010; Grisso & Barnum, 2006; Louden et al., 2017).

Factor Structure

Based on prior research of the MAYSI-2’s factor structure (Grisso & Barnum, 2006; Grisso et al., 2001), those results were used to determine the current seven subscale structure. Congruence coefficients ranged from .80 (TE and TD subscales) to .94 (ADU subscale). As noted by Tavakol and Dennick (2011), a good alpha coefficient is at least .70. When examining the internal consistency of items within subscales for detained youth, the alpha coefficients on subscales ranged from .61 to .86, in both samples except the Traumatic Experiences subscale (.51). The average alpha coefficient was .75. Further, the average corrected item total correlations were calculated and ranged from .37 to .63 (Grisso & Barnum, 2000, as cited in Archer et al., 2010). This finding suggests that there is an adequate association between items and their cluster subscales. Additionally, in this psychometric study, intercorrelations of the MAYSI–2 subscales ranged from .24 to .61 (M = .39 for boys) in the Massachusetts sample of detained youth (Grisso & Barnum, 2000, as cited in Archer et al., 2010).

Archer et al. (2004) replicated the initial psychometric study and performed an exploratory factor analysis with several factors (i.e., seven MAYSI-2 subscales). Archer et al. (2004) computed factor congruence coefficients for best-matched factor pairs. This specific analysis used the matrix of factor-pattern coefficients, and to complete it, the researchers used the factor loading table published by the MAYSI-2 developers. The results demonstrated strikingly similar factor structure to the original MAYSI-2 psychometric study, with strong item loadings on all subscales except Thought Disturbance, which was weaker in item loadings and alpha coefficients (e.g., .65 on Traumatic Experiences, .87 on Alcohol/Drug Use and Somatic Complaints). Archer et al. (2004) stated that the findings from their study are encouraging to generalize the MAYSI-2 reliability and validity findings to AA participants, who made up the majority of the sample.

Validity and Reliability

Grisso and Barnum (2006) tested concurrent validity by comparing MAYSI-2 results to similar constructs as measured by the MACI (Millon & Davis, 1993) and Child Behavior Checklist-Youth Self Report (CBCL-YSR; Achenbach, 1991; Grisso and Barnum, 2006). All subscales were compared to similar typed categories on the MACI and CBCL and were in the acceptable range. However, the concurrent validity was not examined with either the MACI or the YSR for the Traumatic Experiences MAYSI-2 subscale. Further, test-retest reliability coefficients were acceptable (Grisso & Barnum, 1998, cited in Archer et al., 2010; Bisbee, 2010; Cauffman, 2004).

Data Analysis

The two main objectives of this study were to: (a) determine whether the MAYSI-2 subscales accurately identify AA-committed male incarcerated youth with mental illness, and (b) examine how well the MAYSI-2 subscales identifies mental illness among AA-committed male incarcerated youth. To determine how well the MAYSI-2 subscales identify AA-committed male incarcerated youth with mental illnesses, White youth were the comparison group. To address these objectives, logistic regression and the ROC analysis were utilized.

Homotypic Mappings

The homotypic mappings of MAYSI-2 subscales onto clusters of psychiatric disorders has been completed by several researchers (e.g., Leenarts et al., 2016; Louden et al., 2017; Shulman et al., 2018; Wasserman et al., 2004) and considered an effective method for determining the MAYSI-2’s diagnostic utility in identifying psychiatric disorders. In the current study, the determination of homotypic mappings onto clusters of psychiatric disorders was based on subscale questions and their relation to the diagnostic criteria for the respective disorders (Wasserman et al., 2004). Further, based on previous research (Louden et al., 2017; Shulman et al., 2018), some disorders (e.g., Attention Deficit Hyperactivity Disorder (ADHD) are better identified by another MAYSI-2 subscale than its homotypic mapped disorder. Therefore, additional analyses were completed to assess the MAYSI-2’s ability to identify those disorders.

The MAYSI-2 subscales and disorders that were utilized in this analysis were mapped based off of the mappings by previous researchers (see Leenarts et al., 2016; Louden et al., 2017; Shulman et al., 2018; Wasserman et al., 2004) and are noted in Table 2. It is of note that they did not classify ADHD as a disruptive behavior disorder. However, the present study classified ADHD as a disruptive behavior disorder based on the premise that disruptive behavior disorders can be viewed as a continuum from least severe (i.e., ADHD) to the most severe (Conduct Disorder [CD]; Austerman, 2015; A. Ghosh et al., 2017; S. Ghosh & Sinha, 2012).

Logistic Regression

A logistic regression is a specific regression analysis used to evaluate the relationship between predictor variables and a binary outcome (Stoltzfus, 2011). Peng and So (2002) argue that using a logistic regression has advantages over other regression analyses. First, a logistic regression overcomes the limitations of OLS regression in handling dichotomous outcomes. Second, a logistic regression is less restrictive than discriminant analyses for categorical outcomes because the data is not assumed to have equal variances and covariances or multivariate normality (Peng & So, 2002). As such, in this study, a binary, or simple logistic regression was used to determine whether the MAYSI-2 subscales accurately identify AA-committed male incarcerated youth with homotypic mental illnesses.

Based on the categorization for each subscale, analyses were conducted for AAs, as well as Whites to provide information related to the probability of an AA-incarcerated youth having a mental health disorder compared to Whites while controlling for the MAYSI-2 subscales.

Seven separate logistic regression models were built to determine if the MAYSI-2 subscales accurately identify and categorize AA-committed male incarcerated youth who have mental disorders as defined by the Diagnostic and Statistical Manual of Mental Disorders (5th ed.; DSM-5; American Psychiatric Association [APA], 2013). The logistic regressions evaluated, (a) Alcohol/Drug Use, (b) Angry-Irritable (two models: one for mood/affective disorders, and one for disruptive behavior disorders), (c) Depressed/Anxious (two models: one for mood/affective disorders and one for anxiety disorders), (d) Somatic Complaints, and (e) Traumatic Experiences. The additional two subscales Thought Disturbances and Suicide Ideation were omitted due to the Thought Disturbances and the Suicide Ideation scale not having Diagnostic and Statistical Manual of Mental Disorders (4th ed.; DSM-IV; APA, 1994) homotypic disorder(s) to match with. The aim was to determine if there is an interaction between AA males, and the MAYSI-2 subscales. Including the interaction terms will allow for additional understanding of the relationship between the variables in the model.

ROC Curve Analysis

In the present study, the ROC curve analysis was used to examine how well the MAYSI-2 subscales identify and categorize mental illness among AA-incarcerated youth. ROC analyses were calculated for AA males and White males to compare the MAYSI-2’s diagnostic abilities. To do so, ROC analyses were run once with all possible scores. The ROC curve is a graphical display that evaluates the forecasting precision of a regression model (Raju & Schumacker, 2015). It is applied to binary targets and describes the discriminative capacity of the predictive model. In other words, it evaluates the identification ability of an independent (predictor) variable that is continuously measured and a dependent (outcome) variable that is dichotomously measured (Provost & Fawcett, 1997). In addition, the area under the ROC curve (AUC) was also calculated. The goal of the AUC is to identify the cut-off score that maximizes both sensitivity and specificity. Gallop et al. (2003) has established criteria for interpreting the values in the AUC. Those criteria are: (a) 90%–99% (.90–.99) = excellent; (b) 80%–89% (.80–.89) = good, or highly useful; (c) 70%–79% (.70–.79) = fair, or moderately useful; and (d) 60%–69% (.60–.69) = poor, not useful. These criteria were also used in the current study to determine how well the MAYSI-2 identifies homotypic mental illness among AA and White committed male incarcerated youth. The AUC was calculated for AA males and White males, which permitted comparisons of the AUC values.

Results

Descriptive Statistics

After removing those who did not meet the age and race requirements, there were a total of 314 participants, which included 193 AA and 121 White incarcerated youth. Based on the file review, most participants in the sample were 15–17 years old. Of the 314 participants, 305 (97%) had at least one diagnosis. The most prevalent disorders in the sample were substance use disorder, and disruptive behavior disorder. Specifically, 85% (n = 103) of White and 89% (n = 172) of AA-incarcerated youth had a previously diagnosed substance use disorder on file (see Table 1). Additionally, 96% (n = 116) of White and 98% (n = 189) of AA-incarcerated youth had a previously diagnosed disruptive behavior disorder (i.e., ADHD, ODD, or CD; see Table 2). Fewer AA than White incarcerated youth had a previously diagnosed mood/affective disorder (42.0%, n = 81; 58.7, n = 71, respectively), anxiety disorder (15.7%, n = 19; 7.8, n = 15, respectively), or post-traumatic stress disorder (PTSD) (14.9%, n = 18; 5.7, n = 11, respectively).

Table 1.

Characteristics of Participants in the MAYSI-2 Study.

Variable/ethnicity	Number (%)
Race/Ethnicity
African American	193 (61.4)
Caucasian	121(38.6)
Substance use disorder diagnosis demographics
Caucasian
No substance use disorder diagnosis on file	18 (14.9)
Substance use disorder diagnosis on file	103 (85.1)
African American
No substance use disorder diagnosis on file	21 (10.9)
Substance use disorder diagnosis on file	172 (89.1)
Disruptive behavior disorder diagnosis
Caucasian
No disruptive behavior disorder diagnosis on file	5 (4.0)
Disruptive behavior disorder diagnosis on file	116 (96.0)
African American
No disruptive behavior disorder diagnosis on file	4 (2.0)
Disruptive behavior disorder diagnosis on file	189 (98.0)
Mood/Affective disorder diagnosis
Caucasian
No mood/affective diagnosis on file	50 (41.3)
Mood/affective diagnosis on file	71 (58.7)
African American
No mood/affective diagnosis on file	112 (58.0)
Mood/affective diagnosis on file	81 (42.0)
Anxiety disorder diagnosis
Caucasian
No anxiety diagnosis on file	102 (84.3)
Anxiety diagnosis on file	19 (15.7)
African American
No anxiety diagnosis on file	178 (92.2)
Anxiety diagnosis on file	15 (7.8)
Post-traumatic stress disorder (PTSD) diagnosis
Caucasian
No PTSD diagnosis on file	103 (85.1)
PTSD diagnosis on file	18 (14.9)
African American
No PTSD diagnosis on file	182 (94.3)
PTSD diagnosis on file	11 (5.7)

Note. MAYSI-2 = Massachusetts Youth Screening Instrument–Second Version (Grisso & Barnum, 2000).

Table 2.

Homotypic Mappings of Disorders.

MAYSI-2 subscale	DSM-IV-TR disorder category
Alcohol/Drug use	Substance-related disorders Substance abuse disorder, substance dependence disorder
Angry-Irritable	Disruptive behavior disorders Attention-deficit/hyperactivity disorder, oppositional defiant disorder, and conduct disorder
Angry-Irritable	Mood/Affective disorders Major depressive disorder, dysthymia, bipolar disorder, and other depressive or bipolar disorders
Depressed/Anxious	Mood/Affective disorders Major depressive disorder, dysthymia, bipolar disorder, and other depressive or bipolar disorders
Depressed/Anxious	Anxiety disorders Generalized anxiety disorder, obsessive compulsive disorder, and other anxiety disorders
Somatic complaints	Anxiety disorders Generalized anxiety disorder, obsessive compulsive disorder, and other anxiety disorders
Traumatic experiences	Post-traumatic stress disorder

Note. MAYSI-2 = Massachusetts Youth Screening Instrument–Second Version; (Grisso & Barnum, 2000); DSM-IV-TR = Diagnostic and Statistical Manual of Mental Disorders (4th ed., text rev.; American Psychiatric Association, 2000).

Additional descriptive analyses were run for White and AA-incarcerated youth to gain information on the range of scores and the number of incarcerated youth who scored in the low-risk, caution, or warning ranges for each MAYSI-2 subscale. The ranges used for each subscale were the pre-determined ranges established by the MAYSI-2 developers and slightly varied for each subscale (Grisso et al., 2001). Several salient results were noted for each subscale (see Table 3). Concerning the Alcohol/Drug Use subscale scores, most White incarcerated youth, scored in the low-risk range (62%, n = 75), followed by the warning range (32.2%, n = 39). For AA-incarcerated youth, most (76.2%, n = 147) scored in the low-risk range. In terms of the Angry/Irritable subscale scores, White incarcerated youth most frequently scored in the low-risk range (69.4%, n = 84), and AA-incarcerated youth typically scored in the low-risk range (65.9%, n = 127). For the Depressed/Anxious subscale scores, White and AA-incarcerated youth, commonly scored in the low-risk range (67.8%, n = 82; 65.8%, n = 127, respectively). Similarly, for the Somatic Complaints subscale scores, White and AA-incarcerated youth most frequently scored in the low-risk range (65.3%, n = 7; 65.4%, n = 126, respectively). Lastly, for the Traumatic Experiences subscale scores range, White and AA-incarcerated youth most commonly scored in the clinically elevated range (65.3%, n = 79; 36.8%, n = 72, respectively), followed by the low-risk range (34.7%, n = 42; 36.8%, n = 72, respectively).

Table 3.

MAYSI-2 Juvenile Offenders Descriptive Data

Variable/ethnicity	Subscale score	Number (%)
MAYSI-2 Alcohol/Drug Use Subscale
Caucasian
Low risk	0	0 (0)
	1	11 (9.1)
	2	15 (12.4)
	3	11 (9.1)
Warning	4	11 (9.1)
	5	13 (10.7)
	6	15 (12.4)
Caution	7	6 (5.0)
	8	1 (0.8)
African American
Low risk	0	87 (45.1)
	1	23 (11.9)
	2	21 (10.9)
	3	16 (8.3)
Warning	4	14 (7.3)
	5	14 (7.3)
	6	7 (3.6)
Caution	7	10 (5.2)
	8	1 (0.5)
MAYSI-2 Angry/Irritable Subscale
Caucasian
Low risk	0	26 (21.5)
	1	13 (10.7)
	2	16 (13.2)
	3	14 (11.6)
	4	15 (12.4)
Warning	5	12 (9.9)
	6	9 (7.4)
	7	8 (6.6)
Caution	8	5 (4.1)
	9	3 (2.5)
African American
Low risk	0	43 (22.3)
	1	20 (10.4)
	2	20 (10.4)
	3	23 (11.9)
	4	21 (10.9)
Warning	5	13 (6.7)
	6	19 (9.8)
	7	10 (5.2)
Caution	8	15 (7.8)
	9	9 (4.7)
Depressed/Anxious Subscale
Caucasian
Low risk	0	38 (31.4)
	1	26 (21.5)
	2	18 (14.9)
Warning	3	22 (18.2)
	4	6 (5.0)
	5	6 (5.0)
Caution	6	3 (2.5)
	7	2 (1.7)
African American
Low risk	0	63 (32.6)
	1	37 (19.2)
	2	27 (14.0)
Warning	3	28 (14.5)
	4	17 (8.8)
	5	8 (4.1)
Caution	6	10 (5.2)
	7	3 (1.6)
MAYSI-2 Somatic Complaints Subscale
Caucasian
Low risk	0	34 (28.1)
	1	33 (27.3)
	2	12 (9.9)
Warning	3	12 (9.9)
	4	15 (12.4)
	5	10 (8.3)
Caution	6	5 (4.1)
African American
Low risk	0	69 (35.8)
	1	32 (16.6)
	2	25 (13.0)
Warning	3	23 (11.9)
	4	26 (13.5)
	5	12 (6.2)
Caution	6	6 (3.1)
MAYSI-2 Traumatic Experiences Subscale
Caucasian
Low risk	0	42 (34.7)
Elevated risk	1	34 (28.1)
	2	24 (19.8)
	3	14 (11.6)
	4	7 (5.8)
African American
Low risk	0	72 (36.8)
Elevated risk	1	45 (23.3)
	2	40 (20.7)
	3	24 (12.4)
	4	13 (6.7)

Note. MAYSI-2 = Massachusetts Youth Screening Instrument–Second Version (Grisso & Barnum, 2000).

Logistic Regression Findings

The first research question asked whether the MAYSI-2 accurately identified committed male incarcerated youth who had specific diagnoses noted in their medical records. In order to determine whether the MAYSI-2 subscales accurately identified AA-committed male incarcerated youth with mental illnesses, White committed male incarcerated youth was the comparison group, and logistic regressions were completed (see Table 4). As a result of conducting repeated logistic regression analyses it was important to also control for type 1 error. Type 1 error was controlled by reducing the significance level from 0.05 to 0.01.

Table 4.

Logistic Regression Models

Ethnicity	Predictor	B	SE	OR	[Lower CI (95%), Upper CI (95%)]	p
Alcohol/Drug Use (AD) MAYSI-2 Subscale Predicting Diagnosis for Substance Abuse Disorders
Caucasian^a	AD caution warning	2.58	1.05	13.2	[1.69, 102.87]	.014
African American^b	AD caution warning	1.96	1.04	7.09	[0.924, 54.33]	.060
Angry/Irritable (AI) MAYSI-2 Subscale Predicting Diagnosis for Disruptive Behavior Disorders
Caucasian^c	AI caution warning	0.445	0.278	1.56	[0.905, 2.69]	.110
African American^d	AI caution warning	0.024	0.181	1.02	[0.719, 1.46]	.895
Angry/Irritable (AI) MAYSI-2 Subscale Predicting Diagnosis for Mood/Affective Disorders
Caucasian^e	AI caution warning	0.21	0.40	1.23	[0.56, 2.72]	.606
African American^f	AI caution warning	0.50	0.31	1.65	[0.90, 3.00]	.104
Depressed/Anxious (DA) MAYSI-2 Subscale Predicting Diagnosis for Mood/Affective Disorders
Caucasian^g	DA caution warning	0.36	0.40	1.40	[0.63, 3.07]	.404
African American^h	DA caution warning	0.31	0.30	1.36	[0.74, 2.47]	.311
Depressed/Anxious (DA) MAYSI-2 Subscale Predicting Diagnosis for Anxiety Disorders
Caucasianⁱ	DA caution warning	1.29	0.52	3.63	[1.33, 9.97]	.012
African American^j	−.042	0.570	0.96	0.314	[2.93, 0.941]	−.042
Somatic Complaints (SC) MAYSI-2 Subscale Predicting Diagnosis for Anxiety Disorders
Caucasian^k	SC caution warning	0.110	0.519	1.12	[0.404, 3.09]	.832
African American^l	SC caution warning	−0.410	0.605	0.664	[0.203, 2.17]	.498
Traumatic Experiences (TE) MAYSI-2 Subscale Predicting Diagnosis of Post-Traumatic Stress Disorder
Caucasian^m	TE caution warning	1.11	0.664	3.05	[0.829, 11.20]	.094
African Americanⁿ	TE caution warning	0.464	0.694	1.59	[0.408, 6.20]	.504

Note. OR = odds ratio; CI = confidence interval; MAYSI-2 = Massachusetts Youth Screening Instrument–Second Version (Grisso & Barnum, 2000).

Caucasian: Nagelkerke R² = .164. ^bAfrican American: Nagelkerke R² = .64. ^cCaucasian: Nagelkerke R² = .104. ^dAfrican American: Nagelkerke R² = .00. ^eCaucasian: Nagelkerke R² = .003. ^fAfrican American: Nagelkerke R² = .018. ^gNagelkerke R² = .008. ^hAfrican American: Nagelkerke R² = .007. ⁱCaucasian: Nagelkerke R² = .8. ^jAfrican American: Nagelkerke R² = .00. ^kCaucasian: Nagelkerke R² = .001. ^lAfrican American: Nagelkerke R² = .006. ^mCaucasian: Nagelkerke R² = .048. ⁿAfrican American: Nagelkerke R² = .007.

For all the regression analyses, assumptions were met. As noted, seven separate logistic regression models were built to determine if the MAYSI-2 subscales accurately identify and categorize AA and White committed male incarcerated youth who have mental disorders and/or a substance abuse disorder. Of the seven regression models completed, zero were statistically significant for AA-incarcerated youth. For White incarcerated youth, two of the seven models were statistically significant (Alcohol/Drug Use regression, Depressed/Anxious regression).

Data from the two statistically significant regressions are described. The regression model for the Alcohol/Drug Use subscale was calculated to determine if White incarcerated youth and AA-incarcerated youth who have a substance abuse disorder would have scores in the caution or warning range on the Alcohol/Drug Use MAYSI-2 subscale. For White incarcerated youth the logistic model was statistically significant (odds ratio [OR] = 13.2, p = .014), indicating that the Alcohol/Drug Use MAYSI-2 subscale accurately predicts group membership in the caution/warning range for these youth. However, for AA-incarcerated youth, the logistic model was not statistically significant (OR = 7.09, p = .06), indicating that the Alcohol/Drug Use MAYSI-2 subscale does not accurately predict group membership in the caution/warning range for these incarcerated youth.

The regression model for the Depressed/Anxious MAYSI-2 subscale was calculated to determine if White youth and AA-incarcerated youth who have an anxiety disorder would have scores in the caution or warning range on the Depressed/Anxious MAYSI-2 subscale. For White incarcerated youth, the logistic model was statistically significant (odds ratio [OR] = 3.63, p = .012), indicating that the Depressed/Anxious MAYSI-2 subscale does accurately predict group membership in the caution/warning range for these incarcerated youth. However, for AA-incarcerated youth, the logistic model was not statistically significant (OR = .96, p = .942), indicating that the Depressed/Anxious MAYSI-2 does not accurately predict group membership in the caution/warning range for these incarcerated youth.

ROC and AUC Findings

The second research question asked about how well the MAYSI-2 identified AA-committed male incarcerated youth who had specific diagnoses noted in their medical records. To determine how well the MAYSI-2 did this, White committed male incarcerated youth were used as the comparison group and ROC analyses were calculated for each of the five MAYSI-2 subscales utilized in this study (see Table 5). They were used to assess the sensitivity and specificity of each subscale. Each subscale demonstrated higher sensitivity for White incarcerated youth than AA-incarcerated youth, except for the Somatic Complaints and the Traumatic Experiences subscales.

Table 5.

Receiver Operator Characteristic (ROC) and Area Under the Curve (AUC) Findings.

Analysis/ethnicity	Sens	Spec	PPV	NPV	AUC (95% CI)	AUC p	AUC SE	Optimal cutoff	Current cutoff
MAYSI-2 Alcohol/Drug Use Subscale ROC and AUC estimates and optimal elevated risk cutoff scores
Caucasian	.78	.88	.98	.42	.85 [.76, .93]	.000	.042	4	4
African American	.58	.76	.95	.18	.69 [.59, .80]	.003	.051	4	4
MAYSI-2 Angry/Irritable Subscale for ROC and AUC estimates and optimal elevated risk cutoff scores for disruptive behavior disorders
Caucasian	.80	.60	.98	.11	.74 [.54, .93]	.068	.100	5	5
African American	.77	.25	1.00	.02	.50 [.25, .75]	.971	.126	3	5
MAYSI-2 Angry/Irritable Subscale for ROC and AUC estimates and optimal elevated risk cutoff scores for mood/affective disorders
Caucasian	.81	.26	.61	.50	.52 [.42, .63]	.632	.054	4	5
African American	.80	.24	.52	.39	.54 [.46, .62]	.280	.042	3	5
MAYSI-2 Depressed/Anxious Subscale for ROC and AUC estimates and optimal elevated risk cutoff scores for mood/affective disorders
Caucasian	.69	.40	.64	.44	.47 [.18, .76]	.850	.148	4	5
African American	.67	.25	.45	.63	.56 [.37, .74]	.677	.096	3	5
MAYSI-2 Depressed/Anxious Subscale for ROC and AUC estimates and optimal elevated risk cutoff scores for anxiety disorders
Caucasian	.84	.35	.28	.90	.67 [.54, .81]	.014	.069	5	5
African American	.60	.32	.14	.93	.48 [.31, .64]	.812	.083	3	5
MAYSI-2 Somatic Complaints Subscale for ROC and AUC estimates and optimal elevated risk cut-off scores for anxiety disorders
Caucasian	.73	.29	.91	.29	.52 [.38, .67]	.695	.074	3	3
African American	.80	.38	.91	.14	.56 [.43, .69]	.392	.065	3	3
MAYSI-2 Traumatic Experiences Subscale for ROC and AUC estimates for post-traumatic stress disorder
Caucasian	.83	.38	.90	.24	.69 [.55, .82]	.010	.069	—	—
African American	.72	.38	.92	.15	.60 [.42, .79]	.236	.095	—	—

Note. Sens = sensitivity rate; Spec = specificity rate; PPV = positive predictive value; NPV = negative predictive value; CI = confidence interval; MAYSI-2 = Massachusetts Youth Screening Instrument–Second Version (Grisso & Barnum, 2000).

Concerning the Alcohol/Drug Use subscale, ROC analyses were performed to assess the sensitivity and specificity of the MAYSI-2 Alcohol/Drug Use subscale for its ability to detect incarcerated youth with previously diagnosed substance abuse disorders. The Alcohol/Drug Use subscale was more sensitive to substance abuse disorders for White incarcerated youth than for AA-incarcerated youth (0.78 vs. 0.58, respectively). This indicates that this subscale correctly identifies White incarcerated youth with a substance use disorder at a higher rate than AA-incarcerated youth. There was also higher specificity for White incarcerated youth, than for AA-incarcerated youth (0.88 vs. 0.76, respectively). This indicates that this subscale accurately detects White incarcerated youth without a substance use disorder and who did not score in the elevated risk range at a higher rate than AA-incarcerated youth.

Further, the AUC for White incarcerated youth (.85, SE = .042, p = .000) revealed that the MAYSI-2 Alcohol/Drug Use subscale is effective for detecting White incarcerated youth with previously diagnosed substance abuse disorders. However, concerning the MAYSI-2 Alcohol/Drug Use subscale, the AUC for AA-incarcerated youth did not accurately detect AA-incarcerated youth with previously diagnosed substance use disorders. Additionally, the optimal elevated risk cut-off scores for both White incarcerated youth and AA-incarcerated youth was four, which is consistent with the pre-established cut-off scores by the MAYSI-2 developers.

Discussion

Mental health screening is a critical first step to ensuring incarcerated youth at-risk are provided a needed mental health evaluation and those that are in crisis are provided immediate services (Swank & Gagnon, 2017). The guarantees of rehabilitation via the provision of appropriate mental health services to incarcerated youth hinges on accurate screening results. While limited research exists concerning differences in MAYSI-2 scores across youth race for detained youth (see Cauffman & MacIntosh, 2006; Wasserman et al., 2004), this is the first study to address the clinical utility, and use as a diagnostic tool with incarcerated youth that have been adjudicated delinquent in a juvenile detention facility and that focuses on AA and White incarcerated youth.

Overall, 97% of the sample had at least one diagnosis and the most frequent disorders for both White and AA-incarcerated youth was substance use disorders (89.1%, n = 172; 85.1%, n = 103, respectively) and disruptive behavior disorders (98%, n = 189; 96%, n = 116, respectively). Regarding the MAYSI-2 subscales, the results for substance use disorders (Alcohol/Drug Use) and disruptive behavior disorders (Angry-Irritable) varied greatly by race. Specifically, for the Alcohol/Drug use analyses for White incarcerated youth, the logistic model was statistically significant (odds ratio [OR] = 13.2, p = .014), indicating that the Alcohol/Drug Use MAYSI-2 subscale accurately predicts group membership in the caution/warning range for these incarcerated youth. However, for AA-incarcerated youth, the logistic model was not statistically significant (OR = 7.09, p = .06) for the Alcohol/Drug Use MAYSI-2 subscale and as such, it does not accurately predict group membership in the caution/warning range for these youth. Additionally, the Angry-Irritable logistic model was not statistically significant for both White and AA-incarcerated youth (odds ratio [OR] = 1.56, p = .110; OR = 1.02, p = .895, respectively), indicating that the subscale does not accurately predict group membership in the caution/warning range for these incarcerated youth.

Based on the results of the current study, it is evident that the MAYSI-2 does not accurately identify AA-committed male incarcerated youth who have mental disorders and does so accurately for White youth on just two subscales (Alcohol/Drug Use, Depressed/Anxious). The immediate need for an appropriate mental health screening tool for incarcerated youth is a pressing issue, given that the MAYSI-2 is the most widely used. It is important to consider possible explanations for why the MAYSI-2 is an ineffective measure and to identify other or additional approaches that may provide greater assurance of accurately identifying youth at-risk and in immediate need for a mental health evaluation and services. For example, as the MAYSI-2 is a self-report measure, concerns exist that incarcerated youth may, “underreport, deny, or minimize” (Semel, 2017, p. 222) their behaviors and symptoms. As such, the use of multiple informant reports is a potential remedy (De Los Reyes et al., 2012). Although not focused on the MAYSI-2, “self-reports are less predictive of caseness than are parent reports; while the combination of parent and self-reports generally does best” (Kuhn et al., 2017, p. 389). Granted, relying on both incarcerated youth and parent/guardian informants may result in logistical problems and make the process somewhat less rapid and economical (Kuhn et al., 2017). Further, there are also complexities with the interpretation of information from multiple informants that would need to be addressed. However, if these challenges are met, such an approach has the potential to provide more accurate screening data, particularly for AA-incarcerated youth.

Because the current study has limitations in terms of the sample, it is premature to advocate for the complete discontinuation of the MAYSI-2. However, the results may suggest that practitioners should supplement the MAYSI-2 with other measures of incarcerated youth. For example, researchers (Williams et al., 2019; Yurasek et al., 2021) suggest that the CRAFFT (Knight et al., 1999) may also be a useful screening tool concerning incarcerated youth substance use. Additionally, the Behavior Assessment System for Children, 3rd edition (BASC-3; Reynolds et al., 2015), should be considered. Although, the BASC has not been evaluated solely for AA populations, it has a large, stratified norming sample, can yield gender and clinical norms, and has validity indices to address overly positive/negative ratings and malingering. Further, it has a computer scoring system that includes possible DSM-5 diagnoses and treatment recommendations.

If use of the MAYSI-2 is to continue in facilities, it is also suggested that facilities adjust the cut-off scores used to identify incarcerated youth. This finding is imperative for several categories, specifically for AA-incarcerated youth. Based on the results, the optimal elevated risk cut-off score for the Angry/Irritable, Depressed/Anxious subscales is three. Similarly, for the Depressed/Anxious subscale, the optimal cut-off score is three. These optimal cut-off scores are all lower than the pre-established cut-off score of five by the MAYSI-2 developers. Additionally, there are subscales in which the screener did not accurately identify White incarcerated youth. For White incarcerated youth, the optimal elevated risk cut-off score for Angry/Irritable subscale and the Depressed/Anxious subscale is four, which are lower than the pre-established cut-off scores. This indicates to maximize the number of incarcerated youth correctly identified as having a disorder, and the cut-off score should be adjusted accordingly. Should facilities adjust the cut-off score, it may accurately identify more incarcerated youth who may be suffering from these disorders. This would provide the treatment team with more accurate information, leading to the earlier implementation of critical interventions, including individual/group therapy.

Limitations

There are a number of limitations to the current study that must be noted. First, this study included data on incarcerated youth who were screened by the MAYSI-2 and as a result, the analyses that were conducted were based on the possibility of true and false positives based on MAYSI-2 scores alone. Additionally, given the frequency of reading difficulties among incarcerated youth (Houchins et al., 2018), trained graduate students read each MAYSI-2 question to the incarcerated youth during screening, rather than each incarcerated youth reading and responding to the questions on their own. Third, the data was collected between August 2009 and August 2013, and DSM-IV-TR homotypic mappings of diagnoses were used to evaluate the MAYSI-2’s ability to identify AA and White incarcerated youth with mental disorders. As the DSM-5 is the current edition, an additional limitation is not having data for the incarcerated youth from DSM-5 diagnoses. However, it is also worth noting that there was little to no change in diagnostic criteria between DSM editions for the diagnoses included in this study. While it is also important for this study to be replicated with more recent data on DSM-5 diagnoses, the fact that this is the only study to focus on committed incarcerated youth, rather than those who are detained, provides an important contribution to the literature. Fourth, the use of a cross-sectional design is not allowed.

Other limitations relate to the data. For example, the current study relied on nonconcurrent data collection, including diagnosis data from medical records. Inclusion of updated evaluations and identification of evaluation procedures would strengthen future studies. It is also important to note that the study is from one facility with a relatively low sample size. As such, future studies are necessary to determine if the results hold true with larger samples and in other facilities. Additionally, due to the small sample size and the fact the facility of interest housed only males, future research should work to include Latino and female offenders.

Conclusions

Despite the limitations of the study, it is clear that the MAYSI-2 is not an optimal tool for screening AA male incarcerated youth with mental illnesses. Though the study’s focus was on evaluating the MAYSI-2’s utility for AA populations, the findings for White incarcerated youth also demonstrate that it is not an optimal tool for that population either. This is critical, particularly if the MAYSI-2 is the sole screening tool used for determining incarcerated youth’ require a comprehensive mental health evaluation and/or urgent mental health services.

However, the MAYSI-2 continues to be a popular tool to use in juvenile facilities and as such, future research efforts should continue to evaluate this screener for its utility among AA offenders. Future studies may build upon the current study by gathering a larger sample size in different regions to determine if the findings hold true. Further, expanding the analyses to include the suicidality and thought disturbance subscales would be beneficial, as this can provide additional information related to its utility for schizoaffective disorders and suicide risk. Finally, evaluating the MAYSI-2’s ability to identify AA-incarcerated youth with mental illnesses with the newest version of the DSM would be helpful in drawing additional conclusions about its utility for the offenders that represent the largest group in juvenile facilities but are currently the least likely to receive mental health treatment (Prud’homme, 2018).

Footnotes

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.

Ethical Approval

All procedures performed in studies involving human participants were in accordance with the ethical standards of the institutional and/or national research committee and with the 1964 Helsinki Declaration and its later amendments or comparable ethical standards.

Funding

This study was conducted as part of Project LIBERATE, funded by the Institute of Education Sciences (IES), Award #R324A080006. The opinions expressed here are the authors’ and do not necessarily represent those of IES or the U.S. Department of Education.

Informed Consent/Assent

Informed assent was received from all youth and informed consent from the guardian of each youth.

ORCID iDs

Brittany LaBelle

Joseph Calvin Gagnon

References

Achenbach

T. M.

(1991). Manual for the Child Behavior Checklist/4-18 and 1991 profile. University of Vermont.

American Psychiatric Association. (1994). Diagnostic and statistical manual of mental disorders (4th ed.).

American Psychiatric Association. (2000). Diagnostic and statistical manual of mental disorders (4th ed., text rev.).

American Psychiatric Association. (2013). Diagnostic and statistical manual of mental disorders (5th ed.). American Psychiatric Publishing.

Arain

Haque

Johal

Mathur

Nel

Rais

Sandhu

Sharma

(2013). Maturation of the adolescent brain. Neuropsychiatric Disease and Treatment, 9, 449–461. http://doi.org/10.2147/NDT.S39776

Archer

R. P.

Sionds-Bisbee

Spiegel

Handel

Elkins

(2010). Validity of the Massachusetts Youth Screening Instrument-2 (MAYSI-2) in juvenile justice settings. Journal of Personality Assessment, 92, 337–348. https://doi.org/10.1080/00223891.2010.482009

Archer

R. P.

Stredny

R. V.

Mason

J. A.

Arnau

R. C.

(2004). An examination and replication of the psychometric properties of the Massachusetts Youth Screening Instrument-(MAYSI-2) among adolescents in detention settings. Assessment, 11(4), 290–302. https://psycnet.apa.org/doi/10.1177/1073191104269863

Austerman

(2015). ADHD and behavioral disorders: Assessment, management, and an update from DSM-5. Cleveland Clinic Journal of Medicine, 82, S2–S7. http://doi.org/10.3949/ccjm.82.s1.01

Baglivio

M. T.

Wolff

K. T.

Piquero

A. R.

Greenwald

M. A.

Epps

(2017). Racial/ethnic disproportionality in psychiatric diagnoses and treatment in a sample of serious juvenile offenders. Journal of Youth and Adolescence, 46(7), 1424–1451. https://doi.org/10.1007/s10964-016-0573-4

10.

Bisbee

(2010). Massachusetts Youth Screening Instrument long-term outcomes and scale stability. Dissertation Abstracts International: Section B: The Sciences and Engineering, 71(1), 650.

11.

Cauffman

(2004). A statewide screening of mental health symptoms among juvenile offenders in detention. Journal of the American Academy of Child & Adolescent Psychiatry, 43(4), 430–439. https://doi.org/10.1097/00004583-200404000-00009

12.

Cauffman

MacIntosh

(2006). A Rasch differential item functioning analysis of the Massachusetts Youth Screening Instrument: Identifying race and gender differential item functioning among juvenile offenders. Educational and Psychological Measurement, 66(3), 502–521. https://doi.org/10.1177/0013164405282460

13.

Children’s Defense Fund. (2012). Mental health screening in juvenile detention facilities [Issue brief]. https://www.cdfohio.org/wp-content/uploads/sites/6/2018/07/mental-health-juvenile-issue-brief-OCT2012.pdf

14.

De Los Reyes

Aldao

Thomas

S. A.

Daruwala

Swan

A. J.

Van Wei

Goepel

Lechner

W. V.

(2012). Adolescent self-reports of social anxiety: Can they disagree with objective psychophysiological measure and still be valid? Journal of Psychopathology and Behavioral Assessment, 34, 308–322. https://doi.org/10.1007/s10862-012-9289-2

15.

Gagnon

J. C.

Ross Benedick

Mason-William

(2022). Mental health interventions for youth who are incarcerated: A systematic review of literature. American Journal of Orthopsychiatry, 92, 389–390. https://doi.org/10.1037/ort0000587

16.

Gallop

R. J.

Crits-Christoph

Muenz

L. R.

X. M.

(2003). Determination and interpretation of the optimal operating point for ROC curves derived through generalized linear models. Understanding Statistics, 2, 219–242. https://psycnet.apa.org/doi/10.1207/S15328031US0204_01

17.

Ghosh

Ray

Basu

(2017). Oppositional defiant disorder: Current insight. Psychology Research and Behavior Management, 10, 353–367. https://doi.org/10.2147/PRBM.S120582

18.

Ghosh

Sinha

(2012). ADHD, ODD, and CD: Do they belong to a common psychopathological spectrum? A case series. Case Reports in Psychiatry, 2012, 520689. https://doi.org/10.1155/2012/520689

19.

Grisso

Barnum

(1998). Massachusetts Youth Screening Instrument: Preliminary manual. University of Massachusetts Medical School.

20.

Grisso

Barnum

(2000). Massachusetts Youth Screening Instrument, Second Version (MAYSI-2): User’s manual and technical report. University of Massachusetts Medical School.

21.

Grisso

Barnum

(2006). Massachusetts Youth Screening Instrument, Version 2: MAYSI-2: User’s manual and technical report. Professional Resource Press.

22.

Grisso

Barnum

Fletcher

Cauffman

Peuschold

(2001). Massachusetts youth Screening Instrument for mental health needs of juvenile justice youths. Journal of the American Academy of Child & Adolescent Psychiatry, 40, 541–548. https://doi.org/10.1097/00004583-200105000-00013

23.

Grisso

Fusco

Paiva-Salisbury

Perrauot

Williams

Barnum

(2012). The Massachusetts Youth Screening Instrument-Version 2 (MAYSI-2): Comprehensive research review. University of Massachusetts Medical School. https://www.nysap.us/MAYSI-2%20Review.pdf

24.

Grisso

Vincent

Seagrave

(Eds.). (2005). Mental health screening and assessment in juvenile justice. Guilford Press.

25.

Groth-Marnat

(2009). Handbook of psychological assessment. John Wiley.

26.

Hockenberry

(2018). Juveniles in residential placement, 2015. Office of Juvenile Justice and Delinquency Prevention, Office of Justice Programs, U.S. Department of Justice. https://ojjdp.ojp.gov/sites/g/files/xyckuh176/files/pubs/250951.pdf

27.

Houchins

D. E.

Gagnon

J. C.

Lane

H. B.

Lambert

R. G.

McCray

E. D.

(2018). The efficacy of a literacy intervention for incarcerated adolescents. Residential Treatment for Children & Youth, 35, 60–91. https://doi.org/10.1080/0886571X.2018.1448739

28.

Hulley

S. B.

(Ed.). (2007). Designing clinical research. Lippincott Williams & Wilkins.

29.

Juvenile Law Center. (2023). Glossary: A. https://jlc.org/glossary/glossary

30.

Knight

J. R.

Shrier

L. A.

Bravender

T. D.

Farrell

Vanderbilt

Shaffer

H. J.

(1999). A new brief screen for adolescent substance abuse. Archives of Pediatrics & Adolescent Medicine, 153, 591–596. https://doi.org/10.1001/archpedi.153.6.591

31.

Kuhn

Aebi

Jakobsen

Banaschewski

Poustka

Grimmer

Goodman

Steinhausen

H.-C.

(2017). Effective mental health screening in adolescents: Should we collect data from youth parents, or both? Child Psychiatry & Human Development, 48, 385–392. https://doi.org/10.1007/s10578-016-0665-0

32.

Leenarts

L. E. W.

Dölitzsch

Schmeck

Fegert

J. M.

Grisso

Schmid

(2016). Relationship between Massachusetts Youth Screening Instrument-second version and psychiatric disorders in youths in welfare and juvenile justice institutions in Switzerland. BMC Psychiatry, 16(1), Article 340. https://doi.org/10.1186/s12888-016-1032-1

33.

Levin

K. A.

(2006). Study design III: Cross-sectional studies. Evidence-Based Dentistry, 7(1), 24–25. https://doi.org/10.1038/sj.ebd.6400375

34.

Louden

Kang

Ricks

E. P.

Marquez

(2017). Using the MAYSI-2 to identify mental disorder among Latino juvenile offenders. Psychological Assessment, 29(6), 727–739. https://psycnet.apa.org/doi/10.1037/pas0000399

35.

Mallett

C. A.

Stoddard-Dare

(2010). Predicting secure detention placement for African-American juvenile offenders: Addressing the disproportionate minority confinement problem. Journal of Ethnicity in Criminal Justice, 8(2), 91–103. https://doi.org/10.1080/15377931003761011

36.

Millon

Davis

R. D.

(1993). The millon adolescent personality inventory and the millon adolescent clinical inventory. Journal of Counseling & Development, 71(5), 570–574. https://doi.org/10.1002/j.1556-6676.1993.tb02244.x

37.

National Center for Mental Health and Juvenile Justice. (2013). Mental health screening in juvenile justice settings.

38.

National Institute of Justice Annual Report 2016. National Institute of Justice. (n.d.). https://nij.ojp.gov/library/publications/national-institute-justice-annual-report-2016

39.

Peng

C. Y. J.

T. S. H.

(2002). Logistic regression analysis and reporting: A primer. Understanding Statistics: Statistical Issues in Psychology, Education, and the Social Sciences, 1(1), 31–70. https://doi.org/10.1207/S15328031US0101_04

40.

Provost

F. J.

Fawcett

(1997). Analysis and visualization of classifier performance: Comparison under imprecise class and cost distributions. Knowledge Discovery and Data Mining, 97(1), 43–48. https://www.researchgate.net/publication/2637219_Analysis_and_Visualization_of_Classifier_Performance_Comparison_Under_Imprecise_Class_and_Cost_Distributions/citation/download

41.

Prud’homme

(2018). Policy trends in the south on youth mental health in juvenile justice. Journal of Law and Criminal Justice, 6(1), 113–119. https://doi.org/10.15640/jlcj.v6n1a10

42.

Raju

Schumacker

(2015). Exploring student characteristics of retention that lead to graduation in higher education using data mining models. Journal of College Student Retention: Research, Theory & Practice, 16(4), 563–591. https://dx-doi-org.web.bisu.edu.cn/10.2190/CS.16.4.e

43.

Reynolds

C. R.

Kamphaus

R. W.

Vannest

K. J.

(2015). BASC3: Behavior Assessment System for Children. PscyhCorp.

44.

Seiter

(2017). Mental health and juvenile justice: A review of prevalence, promising practices, and areas for improvement. National Technical Assistance Center for the Education of Neglected or Delinquent Children and Youth. https://neglected-delinquent.ed.gov/sites/default/files/NDTAC-MentalHealth-JJ-Brief-508.pdf

45.

Semel

R. A.

(2017). Utility of the ASEBA Youth Self-Report (YSR) in juvenile delinquency assessments. EC Psychology and Psychiatry, 1(6), 217–225. https://www.researchgate.net/publication/2637219_Analysis_and_Visualization_of_Classifier_Performance_Comparison_Under_Imprecise_Class_and_Cost_Distributions/citation/download

46.

Shulman

E. P.

Bechtold

Kelly

E. L.

Cauffman

(2018). Mental health screening in juvenile justice settings: Evaluating the utility of the Massachusetts Youth Screening Instrument, Version 2. Criminal Justice Policy Review, 29(8), 849–872. https://doi.org/10.1177/0887403416650250

47.

Sickmund

Sladky

Kang

(2019). Easy access to juvenile court statistics: 1985-2017. https://www.ojjdp.gov/ojstatbb/ezajcs/

48.

Smith

A. K.

Ayanian

J. Z.

Covinsky

K. E.

Landon

B. E.

McCarthy

E. P.

Wee

C. C.

Steinman

M. A.

(2011). Conducting high-value secondary dataset analysis: An introductory guide and resources. Journal of General Internal Medicine, 26(8), 920–929. https://doi.org/10.1007/s11606-010-1621-5

49.

Stoltzfus

J. C.

(2011). Logistic regression: A brief primer. Academic Emergency Medicine, 18(10), 1099–1104. https://doi.org/10.1111/j.1553-2712.2011.01185.x

50.

Swank

Gagnon

J. C.

(2017). A national survey of mental health screening and assessment practices in juvenile correctional facilities. Child & Youth Care Forum, 46, 379–393. https://doi.org/10.1007/s10566-016-9379-5

51.

Tavakol

Dennick

(2011). Making sense of Cronbach’s alpha. International Journal of Medical Education, 2, 53–55. https://doi.org/10.5116/ijme.4dfb.8dfd

52.

United Nations General Assembly. (1990). United Nations rules for the protection of juveniles deprived of their liberty (General Assembly Resolution 45/13). https://www.ohchr.org/en/instruments-mechanisms/instruments/united-nations-rules-protection-juveniles-deprived-their-liberty#:~:text=Juveniles%20deprived%20of%20their%20liberty%20shall%20not%20for%20any%20reason,14

53.

Vincent

G. M.

Grisso

Terry

Banks

(2008). Sex and race differences in mental health symptoms in juvenile justice: The MAYSI-2 national meta-analysis. Journal of the American Academy of Child & Adolescent Psychiatry, 47(3), 282–290. https://doi.org/10.1097/CHI.0b013e318160d516

54.

Wasserman

G. A.

McReynolds

L. S.

S. J.

Katz

L. M.

Cauffman

Haxton

Lucas

C. P.

(2004). Screening for emergent risk and service needs among incarcerated youth: Comparing MAYSI-2 and Voice DISC-IV. Journal of the American Academy of Child & Adolescent Psychiatry, 43(5), 629–639. https://doi.org/10.1097/00004583-200405000-00017

55.

Wasserman

G. A.

Vilhauer

McReynolds

Shoai

John

(2005). Mental health screening in the juvenile justice system: A comparison between the Voice-DISC-IV and the MAYSI-2. Journal for Juvenile Justice and Detention Services, 19(1), 7–17.

56.

Williams

M. M.

Rogers

Hartigan

S. E.

(2019). Dimensions of psychopathology and behavioral problems: A CFA study of the MAYSI-2 scales of legally involved juveniles. Translational Issues in Psychological Science, 5, 143–153. https://dx-doi-org.web.bisu.edu.cn/10.1037/tps0000189

57.

Yurasek

A. M.

Kemp

Otero

Tolou-Shams

(2021). Substance use screening and rates of treatment referral among justice-involved youth. Addictive Behaviors, 122, 107036. https://doi.org/10.1016/j.addbeh.2021.107036

58.

Zeola

M. P.

Guina

Nahhas

R. W.

(2017). Mental health referrals reduce recidivism in first-time juvenile offenders, but how do we determine who is referred? Psychiatric Quarterly, 88(1), 167–183. https://doi.org/10.1007/s11126-016-9445-z