Gina Vincent, Ph.D., Thomas Grisso, Ph.D., Anna Terry, B.A., & Steve Banks, Ph.D.

1. Law and Psychiatry Program MAYSIs Across the Nation: What do 70,000 Delinquents Tell Us About Youths’ Mental HealthSymptoms? Gina Vincent, Ph.D., Thomas Grisso, Ph.D., Anna Terry, B.A., & Steve Banks, Ph.D. University of Massachusetts Medical School Supported by William T. Grant Foundation

2. Prevalence of MH Disorder in JJ (Teplin, 2002; Wasserman, 2002) • Based on current studies conducted at a few JJ facilities, it seems • 65% of JJ adolescents meet DSM criteria for at least one disorder (vs. 20% in general population) • Rates of disorders vary by • Gender (higher for girls) • Race (highest for Whites and lowest for Blacks)

3. History of the MAYSI Project 1994 1996 1998 2000-8 2002-5 2003-5 2006-8 Identified the need, developed the prototype Field testing, norms, initial validation Preparation for release Released to JJ agencies, developed technical support office, National Youth Screening Assistance Project Developed national norms and MAYSIWARE Evaluation of impact of MH screening on MH services in JJ programs Technical Assistance for MacArthur Foundation’s “Models for Change” Initiative

4. Used Statewide in Probation, Detention or Corrections In 38 States NH WA ME MT VT ND MN OR MA NY WI ID SD WY MI RI PA NE IA CT OH NJ DC NV IN UT IL CA CO WV VA DE KS MO KY MD NC TN OK AZ NM SC AR GA AL MS TX LA AK FL HI

5. Purpose/Research Questions • Develop national norms for the MAYSI-2 • Are there differences between gender, age, and racial groups in the reporting of psychological disturbance (as measured by the Massachusetts Youth Screening Instrument-2; MAYSI-2) among youth in the Juvenile Justice (JJ) System? • Are these differences consistent (reliable, generalizable) across JJ systems nationwide?

6. NH WA ME MT VT ND MN OR MA NY ID WI SD WY MI RI PA IA CT NE OH NJ NV DC IN UT IL CA CO WV DE VA KS MO KY MD NC TN OK AZ NM AR SC GA AL MS TX LA AK FL HI MAYSI-2 Use & Norm Study Donors = Statewide MAYSI-2 User= Norm study donor

7. Method • We gathered archival MAYSI-2 records and demographic information from 283 JJ facilities located in 19 states • Started With 155,835 youths • Removed cases that were: 1) duplicates, 2) outside of the age range, or 3) from any community or treatment facility

8. Final Sample (N = 70,423)

9. Final Sample (N = 70,423) • Gender: Boys - 78% Girls - 22% • Age (years): 12-14 - 29% 15-17 - 71% • Ethnicity/Race: Black- 33% Hispanic - 24% White - 39% Asian – 1% AK Native/Amer Indian - >1% • Legal Status:Pre-adj - 77% Post-adj - 23% • Time of MAYSI-2 Administration (hrs after intake): First few hours - 56% =/> 48 hrs - 12% Within 24 hrs - 32%

10. MAYSI-2 Data • Gathered Item Responses and Scale Scores ALCOHOL / DRUG USE 8 items ANGRY-IRRITABLE 9 items DEPRESSED-ANXIOUS 9 items SOMATIC COMPLAINTS 6 items SUICIDE IDEATION 5 items THOUGHT DISTURBANCE (boys) 5 items TRAUMATIC EXPERIENCES 5 items • Caution Cutoffs: Most generalizable. Based on “clinically significant” scores from corresponding MACI and YSR scales of youth in the general population. • Warning Cutoffs: Least generalizable. Compares juveniles to other juvenile offenders to identify the top 10% of the distribution of Massachusetts scores.

11. Research Question • Are the national norms for each scale comparable to the original Massachusetts norms?

12. Massachusetts vs National Sample% Above Caution Cut-Off

13. Massachusetts vs National Sample% Above Caution Cut-Off

14. Massachusetts vs National Sample% above Warning Cut-Off

15. Massachusetts vs National Sample% above Warning Cut-Off

16. Findings Massachusetts > National % youth above cutoffs % youth above cutoffs • Alcohol/Drug Scale • Angry/Irritable Scale Officially, decreased warning cutoff on Alc/Drug Scale from 7 to 6

17. What Methods of Analysis Could Be Used to Answer The Following Questions? • What are the differences between gender, age, and race groups in the reporting of clinically significant levels of symptoms? • Where differences exist, were these differences consistent (homogenous) across all sites and across all possible interactions?

18. Meta-analytic Procedures • Unit of Analysis – Site (JJ Facility) • Separate Analyses for Each MAYSI-2 Clinical Scale -TRAUMATIC EXPERIENCES not included • Examined Odds of Scoring > Caution Cut-offs • Past research was done to determine clinically significant levels of disturbance. • Variables for Testing Interactions/Controls • Gender • Age Group (12-14, 15-17) • Race (White, Black, Hispanic) • Legal Status (pre vs. post-adjudication) • Time of MAYSI-2 Administration

19. Meta-analytic Procedures (cont.) Cochran-Mantel-Haenszel (CMH) Calculated Odds Ratios ORs weighted for reliability General Linear Modeling Produce Avg OR across all sites for each effect (Gender, etc) I2 - Test Homogeneity of ORs If Yes Consistent Effect If No Explain w/interactions?

20. Overall Gender Differences (Main Effects) % Above “Clinical” Cut Off

21. Overall Gender Differences (Main Effects) % Above “Clinical” Cut Off

22. Are Gender Differences Consistent Across Sites & Demographics? • Mostly – girls at most all sites, on average, are more likely than boys to report clinically significant levels of symptoms on most MAYSI-2 scales • Large Effects: • Suicide Ideation OR = 2.4 • Medium Effects: • Angry-Irritable OR = 1.8 • Depressed-Anxious OR = 1.95 – 2.14 • Somatic Complaints OR = 1.8

23. When do Gender Differences Vary Across Youths? • Alcohol-Drug Scale – gender interacts w/age consistently across sites • 12-14 year olds – girls > boys (OR=1.6) • 15-17 year olds – no gender difference

24. Overall Age Differences (Main Effects)% Above “Clinical” Cut-Off

25. Overall Age Differences (Main Effects)% Above “Clinical” Cut-Off

26. Are Age Differences Consistent Across Sites & Demographics • Mostly – at most all sites, there were no differences or only small differences between younger and older youths in reporting clinically significant levels of symptoms on most MAYSI-2 scales. • Small Effects • Angry-Irritable OR = 1.3 • Thought Disturbance OR = 1.2 Younger youths > older youths

27. When Do Age Differences VaryAcross Youths? • Alcohol-Drug Scale Older youths > Younger youths OR = 1.7 • No appreciable age differences among girls • No appreciable differences among youths in custody post-adjudication

28. Overall Race Differences (Main Effects) % Above “Clinical” Cut-Off

29. Overall Race Differences (Main Effects) % Above “Clinical” Cut-Off

30. Overall Race Differences (Main Effects) % Above “Clinical” Cut-Off

31. Overall Race Differences (Main Effects) % Above “Clinical” Cut-Off

32. Overall Race Differences (Main Effects) % Above “Clinical” Cut-Off

33. Overall Race Differences (Main Effects) % Above “Clinical” Cut-Off

34. Are Racial Differences Consistent Across Sites & Demographics? • No. Whites, on average, were more likely to report clinically significant levels of symptoms than Blacks or Hispanics, but • Results generally were not consistent across sites, and • The Meta-analysis was unable to determine the source of the variability in most cases

35. When Do Racial Differences VaryAcross Youths? • Alcohol Drug Scale • Whites > Blacks OR = 2.3 Large ES • Hispanics > Blacks OR = 1.7 Medium ES • Variability in the magnitude of the odds ratios between Whites & Blacks: • Age (larger for older youths) • Gate (larger for pre-adjudicated youths) • State

36. Site-Level Variables (% of cases) • Gate:Probation - 36% Pretrial Detention - 42% Corrections (post-adjudicated) - 22% • Density:Urban - 78% Rural - 22% • Region:West - 17% Midwest - 44% South - 17% Northeast - 22% Test Administration • Staff:Non-professional - 70% Professional - 30% • Setting: Individual - 87% Group - 13% • Method: Voice CD - 28% Self-administer - 64% Staff Administer - 8%

37. Differences by Gate% Above “Clinical” Cut-Off

38. Differences by Region% Above “Clinical” Cut-Off

39. Differences by Region% Above “Clinical” Cut-Off

40. Advantages to the Meta-analytic Approach for Norming Tests • Provides the strength and consistency of “prevalence rate” differences across settings and youth characteristics (gender, race, etc.) • Provides confidence in the generalizability of these findings across the population of interest

41. Conclusions • % of youths scoring above cut-off not substantially different between Massachusetts and National norms • Girls in JJ much more likely than boys to report clinically significant symptoms – generalizes across JJ sites

42. Conclusions 3. Wide variability in racial differences: • White youths most likely to report problem levels of alcohol or drug use, but the disparity varies across the US • White youths, on average, are consistently more likely than black youths to report suicide ideation

43. Future Applications • Translating the results to improve juvenile justice programs’ understanding of mental health needs of girls and ethnic minority youths • Published a revision of MAYSI-2 manual for mental health screening in juvenile justice • Studying whether race differences are “true” differences or measurement bias… Study in progress: Item Response Theory and PsychologicalDisturbance in Young Offenders (NIMH) (PI: Gina Vincent)