Psychological Services 2009, Vol. 6, No. 2, 139 –153

© 2009 American Psychological Association 1541-1559/09/$12.00 DOI: 10.1037/a0015307

Evaluation of a Voluntary Military-Style Residential Treatment Program for Youths With Conduct Problems: 6- and 36-Month Outcomes Robert Weis and Erin E. Toolis Denison University The authors investigated the socioemotional and behavioral outcomes of adolescents referred to voluntary military-style residential treatment. Adolescents (N ⫽ 232) with conduct problems were classified into three groups: adolescents who completed treatment, adolescents who withdrew from treatment, and wait list controls. Six months after treatment, boys and girls who completed military-style residential treatment showed fewer externalizing problems, greater adaptive skills, and better behavioral outcomes (e.g., high school completion, employment, lower recidivism) than comparison youths. Results were not maintained at 36-month follow-up. Adolescents who enlisted in the military after treatment showed better outcomes at 36-month follow-up than youths who returned home after treatment. Results indicate that the benefits of treatment might be tangible but short-lived for adolescents who return to their communities. Keywords: residential treatment, treatment outcomes, adolescents, conduct problems, military

these programs would “shock” youthful offenders into adopting more prosocial behaviors (Benda, 2005a). Subsequent research, however, has indicated that traditional military-style residential treatment programs are associated with only short-term changes in adolescents’ attitudes and may not lead to long-term reductions in disruptive behavior (Steiner & Giacomazzi, 2007). For example, adolescents who attend military-style residential treatment programs often show significant improvement in their perceptions of themselves, their attitudes toward school, and their self-esteem relative to youths in residential treatment centers or detention facilities (MacKenzie, Gover, Armstrong, & Mitchell, 2001; Wood, May, & Grasmick, 2005). However, youths who attend these programs do not show decreased likelihood of reoffending compared with youths assigned to traditional residential facilities (Weis & Toolis, 2007; Wilson, MacKenzie, & Mitchell, 2005). More recently, military-style residential treatment programs for adolescents have deemphasized the confrontational nature of the paramilitary environment and increased the variety and amount of therapeutic programming available to participants (Austin et al., 2002; MacKenzie, 2006). Contemporary programs for

Military-style residential treatment programs for adolescents with disruptive behavior problems have gained the attention of mental health professionals, corrections specialists, public policy experts, and the general public in recent years (MacKenzie, 2006). These “boot camp” programs combine a highly structured residential environment, paramilitary drill and discipline, and full-time education to reduce recidivism and improve adolescents’ social, emotional, and behavioral functioning (Armstrong, 2004; MacKenzie & Rosay, 2004). When adolescent military-style residential treatment programs began appearing in the juvenile corrections system during the mid-1990s, they were viewed as mid-level sanctions for firsttime offenders. Corrections officials hoped that

Robert Weis and Erin E. Toolis, Department of Psychology, Denison University. Erin E. Toolis is now at the Sanctuary Art Center, Seattle, and is affiliated with Lutheran Social Services. This research was supported by a grant from the Carl Arvid Anderson Family Fund for Science Research. The authors thank Col. (R) M. G. MacLaren and Maj. (R) Mike Brown for their help with data collection. Correspondence concerning this article should be addressed to Robert Weis, Department of Psychology, Denison University, 100 South Road, Granville, OH 43023. E-mail: [email protected] 139

140

WEIS AND TOOLIS

juveniles often include individual and group psychotherapy, social skills and anger management training, substance abuse counseling, job skills development, and other psychoeducational interventions designed to improve adolescents’ functioning after release (Wilson, Bouffard, & MacKenzie, 2005). Indeed, military-style residential treatment programs with the highest degrees of therapeutic programming tend to have the greatest effectiveness in reducing recidivism and improving long-term outcomes (Weis & Toolis, 2007). Voluntary, military-style residential treatment programs, in particular, have shown promise as a means of reducing adolescents’ disruptive behavior. Most juvenile, militarystyle residential treatment programs are compulsory; adolescents are either assigned to programs after adjudication and sentencing or are allowed to choose between military-style treatment or a longer period of traditional incarceration (MacKenzie, 2006). It is, perhaps, not surprising that mandatory militarystyle residential treatment programs are not associated with improvements in functioning or reduced recidivism. Court-ordered psychosocial treatment often yields poor outcomes (Shearer, 2003), especially for individuals with histories of substance use problems and antisocial behavior (McCullough, Engel, & Wright, 2006). In contrast, voluntary military-style residential treatment programs do not accept referrals from juvenile courts. Instead, adolescents and parents must consent to participate in treatment and adolescents are free to withdraw from treatment at any time. The voluntary nature of treatment is believed to ensure that participants are willing to acknowledge their behavior problems, set personally relevant treatment goals, and work with staff members to change their behavior (Lee, Uken, & Sebold, 2007; Velicer & Prochaska, 2008). Some data indicate that voluntary militarystyle residential treatment for youths with conduct problems can improve adolescents’ socioemotional and behavioral functioning. Weis, Whitemarsh, and Wilson (2005) compared the outcomes of adolescents who completed a voluntary military-style residential treatment program, adolescents who prematurely withdrew from treatment, and adolescents assigned to a wait list control group. Youths were assessed before treatment and 6 months after return to the

community. Results showed that program graduates displayed a significant reduction in externalizing behavior problems not shown by youths who withdrew from treatment and adolescents on the wait list. Graduates and those who withdrew also showed significant increases in adaptive skills (e.g., leadership, social skills) compared with controls. Program completion was also associated with improvements in adolescents’ overt behavior at 6-month follow-up. Graduates from the treatment program were more likely to earn a high school degree, more likely to hold a full-time job, less likely to have an alcohol or drug use problem, and less likely to be arrested for a nonmisdemeanor than adolescents in the control group. The evaluation provided initial evidence for the effectiveness of voluntary military-style residential treatment. However, the study was limited by its relatively short follow-up period. Previous research investigating the effectiveness of mandatory military-style residential treatment indicates that most programs are not able to maintain treatment gains 1 year after termination (Weis & Toolis, 2007). We do not know whether the gains associated with participation in voluntary military-style residential treatment programs can be maintained several years after treatment. It is possible that youths acquire certain skills during the course of voluntary treatment that allow them to avoid antisocial behaviors and engage in prosocial activities after they return home. Indeed, programs that emphasize educational, vocational, and psychosocial skills development and programs that deliberately prepare youths for transition to the community tend to have lower recidivism than programs that do not emphasize these aspects of rehabilitation (Benda, 2005b; Wilson, Gallagher, & MacKenzie, 2000). Vocational training and cognitive– behavioral interventions are particularly associated with reduced likelihood of reoffending (Mitchell, Wilson, & MacKenzie, 2007; Wilson, Bouffard, & MacKenzie, 2005). Alternatively, it is possible that the treatment gains observed by Weis, Whitemarsh, and Wilson (2005) at 6-month follow-up may not be maintained over time. The short-term reductions in disruptive behavior shown by program graduates may reflect the fact that these adolescents distanced themselves from many of the familial stressors and problematic peer

EVALUATION OF MILITARY-STYLE TREATMENT

interactions that may have contributed to their disruptive behavior problems in the first place (McMahon & Kotler, 2006). Indeed, adolescent behavior problems are associated with hostile and coercive parent– child interactions (Dodge, 2006; Patterson, Reid, & Dishion, 1992), parental psychopathology and substance abuse (Crnic & Low, 2002; Lovejoy, Graczyk, O’Hare, & Neuman, 2000), lack of parental monitoring (Patterson & Yoerger, 2002), and association with deviant peers (Dishion, Spracklen, Andrews, & Patterson, 1996). The longer adolescents remained in the residential program, the less time they may have been exposed to these potential risk factors. The purpose of this study was to examine the long-term outcomes of youths who participated in the original evaluation of a voluntary military-style residential treatment program (Weis, Whitemarsh, & Wilson, 2005). As reported in the original study, youths were separated into three groups: those who completed treatment, those who prematurely withdrew from the program, and those assigned to a wait list control group because of limited space in the treatment program. Participants were assessed before treatment, 6 months after treatment, and again 36 months after treatment. If treatment was effective, we expected the socioemotional gains observed at 6-month follow-up to be present at 36-month follow-up. We also expected youths who completed treatment to show superior behavioral outcomes at 36-month follow-up. Specifically, we hoped that program graduates would be more likely to have earned a high school degree and to be employed full time, less likely to have problems with alcohol and other drugs, and less likely to have reoffended than youths in the comparison groups. We were especially interested in the outcomes of adolescents who selected a specific career path after completing treatment: the military. Weis, Wilson, and Whitemarsh (2005) found that adolescents who returned to their families and peer groups after treatment showed similar outcomes at 6-month follow-up as youths who relocated to new communities. However, we wondered whether adolescents who enlisted in the armed forces shortly after completing treatment might show better outcomes at 36-month follow-up compared with their counterparts who did not enlist. We reasoned that military service might protect youths

141

from the potentially deleterious effects of family conflict and problematic peer relationships by providing them with physical distance from their former communities, stable employment and job training, and social support. Finally, we wanted to investigate whether gender played a significant role in the socioeomotional or behavioral outcomes of youths at 36-month follow-up. Several critics have argued that military-style residential treatment may be less effective for girls than for boys. For example, girls may not respond to the confrontational nature of the military milieu and military-style programs may even exacerbate girls’ socioemotional problems (Lutze, 2006; Lutze & Bell, 2005; Marcus-Mendoza, Klein-Saffran, & Lutze, 1998). The previous evaluation indicated that boys and girls were equally likely to complete treatment. Furthermore, the 6-month outcomes for youths who completed treatment did not vary as a function of gender (Weis, Whitemarsh, & Wilson, 2005). Consequently, we expected no gender differences in outcomes at 36-month follow-up as well. A lack of gender differences would speak to the effectiveness of voluntary military-style residential treatment for all youths, even girls, who have received far less attention in the treatment literature.

Method Participants Adolescents (154 boys, 78 girls) who participated in the original evaluation study of military-style residential treatment (Weis, Whitemarsh, & Wilson, 2005) provided complete, valid data for this study. At the beginning of the research study, 12% were 18 years old, 67% were 17 years old, and 21% were 16 years old. Ethnicities included White (71%), African American (15%), Latino (11%), and Asian American or Native American (3%). Psychologists, social workers, guidance counselors, and corrections officials from across Wisconsin referred adolescents. Reasons for referral included chronic truancy or school dropout (91%), substance abuse (72%), nonviolent antisocial behavior leading to arrest (e.g., stealing, vandalism, running away; 51%), and physical aggression leading to arrest (e.g., assault, robbery; 31%). Indeed, 81% had been arrested prior to referral and 16% had been removed

142

WEIS AND TOOLIS

from their parents’ homes because of behavioral or emotional problems. Sixty percent participated in outpatient psychotherapy; mean age for psychological referral was 12.13 years (SD ⫽ 4.11). Sixteen percent also participated in inpatient psychiatric treatment; mean age for first hospitalization was 13.21 years (SD ⫽ 2.39). Adolescents displayed significant school problems; 53% had failed a grade and 44% had stopped attending classes at the time of referral. Most (66%) were receiving special educational services while in school. Family backgrounds of adolescents were mixed. Eighty-three percent were living in maternal-headed, single-parent households or foster homes. Family histories were significant for externalizing problems: alcohol abuse (53%), other drug abuse (32%), aggression leading to arrest (18%), other legal problems leading to arrest (i.e., theft, robbery; 17%), drug trafficking (8%), domestic violence or physical child abuse (9%), and child sexual abuse (5%). Family histories were also significant for internalizing problems: depression (44%), anxiety (19%), suicide attempts (13%), and bipolar disorder (10%).

Measures The Structured Developmental History questionnaire (Reynolds & Kamphaus, 1998) was administered to parents. This questionnaire assesses reason for referral, family background and structure, the adolescent’s medical and psychosocial history, and demographic information. The Behavior Assessment System for Children (BASC; Reynolds & Kamphaus, 1998) was completed by parents to assess adolescent socioemotional functioning before treatment and at 6-month follow-up. The BASC was standardized for parents of youths between 12 and 18 years of age. It yields composite T scores on four dimensions used in this study: Externalizing Problems (e.g., aggression, conduct problems), Internalizing Problems (e.g., anxiety, depression), Behavioral Symptoms Index (e.g., composite of externalizing and internalizing symptoms), and Adaptive Skills (e.g., leadership, social skills). Scores on behavior problem composites greater than 60 indicate significant elevations. Scores on the Adaptive Skills composite less than 40 indicate significant deficits in

adaptive behavior. The BASC possesses adequate psychometric properties; validity has been demonstrated by associations with other measures of adolescent behavior and changes associated with intervention. The Behavior Assessment System for Children—Second Edition (BASC–2; Reynolds & Kamphaus, 2004) was administered to parents to assess adolescent socioemotional functioning at 36-month follow-up. The BASC–2 includes the same item content as the original BASC and yields the same four composite scores. However, the BASC–2 is based on updated and extended normative data so that it can be administered to the parents of youths between 18 and 21 years of age (the age of the youths in our study at 36-month follow-up). Standardized T scores generated by the BASC–2 can be compared with scores generated by the original BASC by adding or subtracting correction values presented in the manual (Reynolds & Kamphaus, 2004). Validity is supported by factor analysis, theoretically meaningful correlations with other child behavior rating scales, and differences in BASC–2 scores as a function of diagnosis. Adolescent behavioral outcomes were assessed at 6- and 36-month follow-ups using a semistructured clinical interview. Parents answered questions regarding their adolescent’s (a) educational status, (b) employment status, (c) alcohol use, (d) other substance use, and (e) arrest history. The Adolescent Drinking Index (ADI; Harrell & Wirtz, 1990) and the Adolescent Drug Involvement Scale (ADIS; Moberg, 1991) were administered to identify problematic alcohol and other drug use, respectively. Both are norm-referenced measures that permit the identification of potentially problematic substance use. The ADI shows adequate internal consistency and reliability; evidence of convergent and discriminant validity is presented in the manual. The ADIS also shows adequate psychometric properties; however, validity information comes primarily from middle-class samples.

Procedure Description of the treatment program. The National Guard Challenge Program is a military-style residential treatment program for delinquent adolescents. The program was

EVALUATION OF MILITARY-STYLE TREATMENT

authorized by Congress in 1993 and renewed in 2003 by the Department of Defense Authorization Bill. The Challenge Program is managed conjointly by the National Guard Bureau under the auspices of the Assistant Secretary of Defense. Currently, 28 states participate in the program. Participants are 16- to 18-year-old U.S. citizens or legal residents who are at risk for school dropout and who are not gainfully employed. Adolescents typically have histories of behavior problems, including truancy, substance abuse, and antisocial behavior; however, participants must not be convicted of a felony and must be drug-free on the day of admission to the program. The program is voluntary; parent and adolescent consent is required for participation. The combined federal and state cost for each participant is approximately $15,000. Adolescents and their families participate at no charge. The Challenge Academy (CA), located at Ft. McCoy, Wisconsin, began administering the National Guard Challenge Program to adolescents in 1998. CA enrolls approximately 100 adolescents twice yearly. Like all treatment sites that administer the National Guard Challenge Program, CA consists of 22 weeks of residential treatment. During the first 2 weeks, adolescents adjust to the physical and mental demands of the military environment. This phase emphasizes discipline, code of conduct, and physical fitness. During the subsequent 20 weeks, adolescents participate in military-style and academic activities designed to improve educational, behavioral, and socioemotional functioning. The core components of this phase include (a) educational training; (b) job skill development; (c) physical fitness; (d) leadership skills; (e) health, sex, and nutrition education; (f) life coping skills; (g) citizenship; and (h) community service. Participants attend school throughout the residential phase. Instructional staff includes licensed teachers and counselors, medical personnel, and National Guard members. After treatment, adolescents return to their families or live independently and attempt to implement a life plan developed in the program. Recruitment, assignment, and retention. Approximately 2 to 4 weeks before treatment, 289 adolescents (199 boys, 90 girls) and their parents attended a 1-day orientation at CA. During the orientation, dyads were asked to participate in the study. Adolescents and parents were

143

told that the purpose of the research study was to investigate the effectiveness of military-style residential treatment (not to evaluate individual youths) and that the research study was approved by a university institutional review board independent of CA. Dyads were told that admission to CA was not contingent on participation in the evaluation study; indeed, CA staff would not know which adolescents and parents agreed to participate in the evaluation. Furthermore, participation would not affect the quality of services adolescents received during treatment. Finally, dyads were also informed that their responses would be used for research purposes only; CA staff would not have access to data unless they indicated danger to the adolescent or others. The parents of 282 adolescents (193 boys, 89 girls) consented to participate. Assent was obtained verbally from adolescents. Families were not paid for their participation. Of the 282 adolescents who agreed to participate in the study, 210 subsequently participated in the intervention, whereas 72 were assigned to a wait list control group because of limitations in class size. Financial and logistical considerations prohibited true random assignment to intervention and control groups. Instead, assignment was based on (a) the date youths attended orientation (not the date youths were referred to the program), and (b) gender. Within each gender, youths were admitted into the program until all treatment slots for male and female adolescents were filled. Of the 210 adolescents who began treatment, 58 (27.6%) withdrew before graduation, an attrition rate similar to that seen in other juvenile military-style residential treatment programs (Peters, Thomas, Zamberlan, & Caliber Associates, 1997). In summary, 152 adolescents graduated, 58 withdrew, and 72 served as wait list controls. None of the youths who served as wait list controls in this study later participated in the treatment program. Approximately 6 months after residential treatment, parents were contacted by telephone and asked to participate in follow-up testing. Parents answered questions about adolescents’ functioning over the previous 6 months, including (a) educational status/attainment, (b) employment status, (c) substance use (i.e., ADI, ADIS), and (d) arrests. Parents also completed the BASC regarding their adolescents’ socioemotional functioning over the previous 6 months. A total of 252 parents of adolescents provided complete, valid outcome

144

WEIS AND TOOLIS

data at 6-month follow-up (89.4% of the original sample). Approximately 36 months after treatment, parents were contacted again to participate in a final round of follow-up testing. Data were obtained from parents, rather than from adolescents, for theoretical and practical reasons. First, previous research has indicated that parents are often more reliable informants of youths’ disruptive behavior problems than youths themselves (Hart & Lahey, 1999; Sattler, 2002). Second, we wanted to collect data at 6- and 36-month follow-up periods from the same informants (i.e., parents) to reduce the possibility that informant biases might lead to differences in youths’ outcomes at the two follow-up periods (Kazdin, 2003). Third, we were unable to contact many of the youths themselves at 36-month follow-up, especially those serving in the armed forces and stationed overseas. Data collection followed the same procedures as those described at 6-month follow-up except that the BASC–2 was administered instead of the BASC. A total of 232 parents (82.3% of the original sample) provided complete, valid outcome data at 36-month followup. A transparent reporting of evaluations with nonrandomized design (TREND; Cisler, Barnes, Farnsworth, & Sifers, 2007; Des Jarlais, Lyles, Crepaz, & TREND Group, 2004) flowchart is presented as Figure 1. Data regarding these 232 adolescents were used in all analyses.

Results Socioemotional Outcomes Overall program effectiveness. We conducted a mixed between–within multivariate analysis of covariance (MANCOVA) to evaluate changes in adolescents’ socioemotional wellbeing as a function of treatment and across time. The independent variable was adolescents’ completion status with three levels: (a) completed treatment, (b) withdrew from treatment, and (c) wait list controls. Dependent variables were the four BASC/BASC–2 composite standard scores at pretreatment, 6-month follow-up, and 36-month follow-up: Externalizing Problems, Internalizing Problems, Behavioral Symptoms, and Adaptive Skills. Age, ethnicity,1 and gender served as covariates. Data screening indicated that assumptions of normality and linearity were met. There

were no missing data. A significant Box’s M test indicated possible violation of the assumption of homogeneity of variance– covariance matrices. Consequently, multivariate significance was evaluated using Pillai’s trace instead of Wilks’s ⌳, following the recommendations of previous authors (Olson, 1976; Tabachnick & Fidell, 2007). The Greenhouse–Geisser correction was used to evaluate univariate significance to adjust degrees of freedom and to control for family-wise error (Tabachnick & Fidell, 2007). In evaluating results within each group of tests, we used Bonferonni correction to further control for inflated error. Results indicated a significant multivariate effect for time, Pillai’s trace ⫽ .149, F(8, 219) ⫽ 4.80, p ⬍ .001, ␩2 ⫽ .149, and a significant multivariate effect for completion status, Pillai’s trace ⫽ .139, F(8, 448) ⫽ 4.04, p ⬍ .001, ␩2 ⫽ .070. Results also showed a significant Completion Status ⫻ Time interaction, Pillai’s trace ⫽ .476, F(16, 440) ⫽ 8.59, p ⬍ .001, ␩2 ⫽ .238. Follow-up ANCOVAs revealed significant interactions for Externalizing Problems, F(4, 452) ⫽ 8.81, p ⬍ .001, ␩2 ⫽ .072; Internalizing Problems, F(4, 452) ⫽ 15.05, p ⬍ .001, ␩2 ⫽ .118; Behavioral Symptoms, F(4, 452) ⫽ 3.82, p ⫽ .008, ␩2 ⫽ .033; and Adaptive Skills, F(4, 452) ⫽ 9.86, p ⬍ .001, ␩2 ⫽ .080. Table 1 shows results of follow-up tests, examining composite scores as a function of completion status and time. Prior to treatment, there 1 We conducted three analyses of covariance (ANCOVAs) to evaluate adolescents’ socioemotional functioning at 6- and 36-month follow-ups as a function of completion status and ethnicity. For each ANCOVA, dependent variables were BASC/BASC–2 Externalizing Problems, Internalizing Problems, Behavioral Symptoms, and Adaptive Skills composite scores. Independent variables were completion status (i.e., completed treatment, did not complete treatment) and ethnicity (i.e., White, non-White). BASC composite scores at baseline served as covariates. We were especially interested in significant Completion Status ⫻ Ethnicity interactions, that is, differential treatment response as a function of ethnicity. At 6-month follow-up, results did not show significant interactions for Externalizing, F(1, 227) ⫽ 0.057, p ⫽ .812, Internalizing, F(1, 227) ⫽ 0.858, p ⫽ .355, Behavioral Symptoms, F(1, 227) ⫽ 0.792, p ⫽ .374, and Adaptive Skills, F(1, 227) ⫽ 2.39, p ⫽ .123, composites. Similarly, at 36-month follow-up, results did not show significant interactions for Externalizing, F(1, 227) ⫽ 0.193, p ⫽ .661, Internalizing, F(1, 227) ⫽ 0.985, p ⫽ .322, Behavioral Symptoms, F(1, 227) ⫽ 0.167, p ⫽ .683, and Adaptive Skills, F(1, 227) ⫽ 0.786, p ⫽ .376, composites.

EVALUATION OF MILITARY-STYLE TREATMENT

were no significant differences in adolescents’ behavior by completion status. Six months after the intervention, adolescents who completed treatment displayed significantly lower Externalizing, Internalizing, and Behavioral Symptoms scores and significantly higher Adaptive Skills scores than controls. Adolescents who withdrew from treatment showed significantly fewer Behavioral Symptoms and significantly greater Adaptive Skills than controls. At 36month follow-up, there were no significant differences between adolescents who completed treatment, withdrew from treatment, and controls on any of the four dependent variables. Effects of military service and gender. We conducted additional analyses to examine whether adolescents’ gender or their decision to enlist in the military was associated with their socioemotional outcomes at 36-month followup. Adolescents were classified into four groups on the basis of their completion status and their subsequent military service: (a) completed CA and enlisted in military, (b) completed CA and did not enlist in military, (c) withdrew from CA, and (d) controls.2 In the first analysis, we conducted a MANCOVA examining differences in BASC–2 composite scores as a function of military service and gender. The independent variables were military service with four levels and gender with two levels. The dependent variables were adolescents’ four BASC–2 composite scores at 36-month follow-up. Age and ethnicity served as covariates. Results (see Table 2) indicated a significant multivariate effect for military service, Pillai’s trace ⫽ .208, F(12, 663) ⫽ 4.12, p ⬍ .001, ␩2 ⫽ .069. Follow-up ANCOVAs revealed significant main effects for Externalizing Problems, F(3, 222) ⫽ 6.12, p ⫽ .001, ␩2 ⫽ .076, and Adaptive Skills, F(3, 222) ⫽ 5.86, p ⫽ .001, ␩2 ⫽ .073, but not for Internalizing Problems, F(3, 222) ⫽ 1.75, p ⫽ .157, or Behavioral Symptoms, F(3, 222) ⫽ 0.625, p ⫽ .599. Adolescents who enlisted in the military after completing CA showed significantly fewer externalizing problems and significantly greater adaptive skills than adolescents in the other three groups at 36-month follow-up. Results also showed a significant multivariate effect for gender, Pillai’s trace ⫽ .276, F(4, 219) ⫽ 10.87, p ⬍ .001, ␩2 ⫽ .176. Follow-up

145

ANCOVAs revealed a significant main effect for Internalizing Problems, F(1, 222) ⫽ 35.61, p ⬍ .001, ␩2 ⫽ .154, but not Externalizing Problems, F(1, 222) ⫽ 2.03, p ⫽ .156, Behavioral Symptoms, F(1, 222) ⫽ 3.32, p ⫽ .070, or Adaptive Skills, F(1, 222) ⫽ 3.98, p ⫽ .047. Overall, girls showed significantly more internalizing symptoms than boys. Finally, results showed a significant Military Service ⫻ Gender interaction, Pillai’s trace ⫽ .189, F(12, 663) ⫽ 3.72, p ⬍ .001, ␩2 ⫽ .063. Follow-up ANCOVAs indicated significant interactions for Internalizing Problems, F(3, 222) ⫽ 7.28, p ⬍ .001, ␩2 ⫽ .090, and Behavioral Symptoms, F(3, 222) ⫽ 8.63, p ⬍ .001, ␩2 ⫽ .104, but not for Externalizing Problems, F(3, 222) ⫽ 0.336, p ⫽ .800, or Adaptive Skills, F(3, 222) ⫽ 3.30, p ⫽ .021. Military service was associated with significantly fewer internalizing problems and behavioral symptoms for boys but not for girls.

Behavioral Outcomes Overall program effectiveness. We conducted a series of contingency analyses to determine whether there was a relationship between adolescents’ completion status and their behavioral outcomes at 6- and 36-month follow-ups. Each analysis tested the null hypothesis that adolescents’ completion status was independent of their likelihood of (a) earning a high school degree, (b) being employed or attending school full time, (c) experiencing problems with alcohol, (d) experiencing problems with other drugs, or (e) being arrested for a nonmisdemeanor.3 2

No adolescents who withdrew from treatment or assigned to the wait list control group enlisted in the military. 3 We conducted a series of two-way contingency analyses to examine possible associations between adolescents’ ethnicity (i.e., White, non-White) and their behavioral outcomes at 6and 36-month follow-ups. At 6-month follow-up, ethnicity was independent of high school degree attainment, ␹2(1, N ⫽ 252) ⫽ 1.70, p ⫽ .193, employment, ␹2(1, N ⫽ 252) ⫽ 0.035, p ⫽ .852, alcohol use problems, ␹2(1, N ⫽ 252) ⫽ 0.090, p ⫽ .765, other drug use problems, ␹2(1, N ⫽ 252) ⫽ 1.25, p ⫽ .264, and arrest, ␹2(1, N ⫽ 252) ⫽ 0.434, p ⫽ .510. Similarly, at 36-month follow-up, ethnicity was independent of high school degree attainment, ␹2(1, N ⫽ 232) ⫽ 1.70, p ⫽ .193, employment, ␹2(1, N ⫽ 232) ⫽ 0.111, p ⫽ .739, alcohol use problems, ␹2(1, N ⫽ 232) ⫽ 0.793, p ⫽ .373, other drug use problems, ␹2(1, N ⫽ 232) ⫽ 0.207, p ⫽ .649, and arrest, ␹2(1, N ⫽ 232) ⫽ 1.74, p ⫽ .190.

146

WEIS AND TOOLIS

Asked to participate in study (N = 289)

Consented (n = 282) Refused to participate (n = 7)

Nonrandom assignment Allocated to intervention group (n = 210)

Completed treatment (n = 152)

Withdrew (n = 58)

105 boys 47 girls

42 boys 16 girls

Allocated to control group (n = 72)

6 month follow-up Completed treatment (n = 135)

Withdrew (n = 50)

97 boys 38 girls

37 boys 13 girls

Controls (n = 67) 37 boys 30 girls

36 month follow-up Completed treatment (n = 124)

Withdrew (n = 43)

88 boys 36 girls

31 boys 12 girls

Controls (n = 65) 35 boys 30 girls

Figure 1. TREND (transparent reporting of evaluations with a randomized design) flowchart showing recruitment, assignment, and retention of adolescents in the study.

Results (see Table 3) of the first contingency analysis showed a significant association between completion status and attainment of a high school degree at 6-month follow-up, ␹2(2, N ⫽ 252) ⫽ 89.17, p ⬍ .001, ␸ ⫽ .62, and 36-month follow-up, ␹2(2, N ⫽ 232) ⫽ 89.65, p ⬍ .001, ␸ ⫽ .63. Adolescents who completed treatment were more likely to earn a degree than controls at 6-month followup, ␹2(1, N ⫽ 202) ⫽ 71.31, p ⬍ .001, ␸ ⫽ .61,

and 36-month follow-up, ␹ 2 (1, N ⫽ 189) ⫽ 68.38, p ⬍ .001, ␸ ⫽ .60. Furthermore, adolescents who completed treatment were more likely to earn a degree than adolescents who withdrew from treatment at 6-month follow-up, ␹2(1, N ⫽ 185) ⫽ 66.73, p ⬍ .001, ␸ ⫽ .63, and 36-month follow-up, ␹ 2 (1, N ⫽ 167) ⫽ 70.66, p ⬍ .001, ␸ ⫽ .65. We discovered a significant association between completion status and adolescents’ likeli-

Note. N ⫽ 232. COM ⫽ completed treatment; WITH ⫽ withdrew from treatment; CON ⫽ control. Numbers reflect T scores with M ⫽ 50 and SD ⫽ 10. Means in the same row that do not share subscripts differ significantly at p ⬍ .001.

CON WITH

70.05a (15.70) 63.93a (12.86) 64.65a (10.57) 32.40a (7.02) 65.16a (11.94) 63.46a (12.54) 63.80a (13.26) 35.56a (9.09)

COM CON

73.46a (11.71) 61.32a (16.19) 65.23a (17.65) 38.80a (11.90) 70.40a (15.56) 56.42a (13.00) 58.98b (13.19) 45.21b (10.17)

WITH COM CON

77.60a (15.12) 65.12a (8.33) 68.42a (10.78) 33.57a (8.67)

WITH COM

76.90a (15.01) 65.38a (11.72) 67.92a (13.01) 33.37a (8.38)

Outcome

Externalizing problems Internalizing problems Behavioral symptoms Adaptive skills

75.88a (14.48) 64.47a (10.41) 67.28a (10.83) 32.53a (8.11)

61.10b (12.87) 49.35b (10.70) 57.19b (12.17) 47.02b (8.63)

36-month follow-up 6-month follow-up Pretreatment

Table 1 Means (Standard Deviations) for Adolescent Socioemotional Outcomes as a Function of Completion Status and Time

71.03a (13.26) 64.18a (10.98) 64.62a (12.57) 33.97a (8.12)

EVALUATION OF MILITARY-STYLE TREATMENT

147

hood of being employed or attending school full time at 6-month follow-up, ␹2 (2, N ⫽ 252) ⫽ 12.35, p ⫽ .002, ␸ ⫽ .24. Adolescents who completed treatment were significantly more likely to return to school or to be employed than controls, ␹2(1, N ⫽ 202) ⫽ 10.97, p ⫽ .001, ␸ ⫽ .24, and adolescents who withdrew from treatment, ␹2(1, N ⫽ 185) ⫽ 6.81, p ⫽ .009, ␸ ⫽ .21. However, there was no significant association between completion status and employment or school outcomes at 36-month follow-up, ␹2(2, N ⫽ 232) ⫽ 2.83, p ⫽ .243. There was a significant association between completion status and adolescents’ likelihood of having an alcohol problem at 6-month follow-up, ␹2(2, N ⫽ 252) ⫽ 9.30, p ⫽ .001, ␸ ⫽ .21. Adolescents who completed treatment were less likely to experience alcohol problems than controls, ␹2(1, N ⫽ 202) ⫽ 9.19, p ⫽ .002, ␸ ⫽ .22, and adolescents who withdrew from treatment, ␹2(1, N ⫽ 185) ⫽ 4.42, p ⫽ .036, ␸ ⫽ .16. However, there was no significant relationship between completion status and likelihood of alcohol problems at 36-month follow-up, ␹2(2, N ⫽ 232) ⫽ 5.98, p ⫽ .061. There was also a significant association between completion status and adolescents’ likelihood of having problems with other drugs at 6-month follow-up, ␹2(2, N ⫽ 252) ⫽ 9.29, p ⫽ .009, ␸ ⫽ .20. Adolescents who completed treatment were less likely to experience drug problems than controls, ␹2(1, N ⫽ 202) ⫽ 9.19, p ⫽ .002, ␸ ⫽ .22, but they were equally likely to have drug problems as adolescents who withdrew from treatment, ␹2(1, N ⫽ 185) ⫽ 1.12, p ⫽ .290. There was not a significant association between completion status and drug problems at 36-month follow-up, ␹ 2 (2, N ⫽ 232) ⫽ 3.84, p ⫽ .147. Finally, there was a significant association between completion status and adolescents’ likelihood of being arrested for a nonmisdemeanor at 6-month follow-up, ␹2(2, N ⫽ 252) ⫽ 24.09, p ⬍ .001, ␸ ⫽ .32. Adolescents who completed treatment were less likely to be arrested than controls, ␹2(1, N ⫽ 202) ⫽ 18.89, p ⬍ .001, ␸ ⫽ .32, and adolescents who withdrew, ␹2(1, N ⫽ 185) ⫽ 16.30, p ⬍ .001, ␸ ⫽ .31. However, there was not a significant relationship between completion status and arrest for nonmisdemeanor at 36-month follow-up, ␹2(2, N ⫽ 232) ⫽ 3.77, p ⫽ .152.

148

WEIS AND TOOLIS

Table 2 Means (Standard Deviations) for Adolescent Socioemotional Outcomes as a Function of Military Service and Gender at 36-Month Follow-up Military service Outcome Externalizing problems Boys Girls Internalizing problems Boys Girls Behavioral symptoms Boys Girls Adaptive skills Boys Girls

Completed/ enlisted

Completed/ civilian

Withdrew

Control

55.48a (11.68) 61.88b (10.60)

67.09b (9.49) 69.57b (13.54)

69.65b (15.22) 71.08b (15.70)

70.31b (13.19) 71.87b (13.51)

55.96b (6.93)ⴱ 73.00b (5.63)

59.22b (8.32)ⴱ 73.61b (13.47)

62.16b (12.08) 68.50b (14.23)

60.57b (9.84)ⴱ 68.40b (10.88)

53.48a (11.25)ⴱ 74.00a (9.61)

64.28b (12.30) 68.25a (13.16)

66.35b (10.20) 60.25b (10.65)

66.83b (13.16) 62.03b (11.54)

39.61a (7.31) 40.50a (12.22)

33.82b (8.92) 34.89b (8.76)

34.10b (7.11) 28.00b (7.02)

34.77b (6.73) 30.70bc (8.49)

Note. N ⫽ 232. Completed/enlisted ⫽ Completed CA and enlisted in military; Completed/civilian ⫽ Completed CA and did not enlist in military; Withdrew ⫽ Withdrew from CA and did not enlist in military; Control ⫽ Served in control group and did not enlist in military. Means in the same row with different subscripts differ significantly at p ⬍ .008. ⴱ Significant gender difference at p ⬍ .0125.

Effects of military service and gender. We conducted a series of contingency analyses to determine whether adolescents’ behavioral outcomes were associated with their military service or gender. In these analyses, adolescents who completed CA were separated into two

groups on the basis of whether they enlisted in the military after completing treatment. Outcomes were assessed at 36-month follow-up. Results (see Table 4) showed that adolescents who completed CA and enlisted in the military were more likely to be employed or attend

Table 3 Percentage of Adolescent Behavioral Outcomes as a Function of Completion Status and Time

Table 4 Percentage of Adolescent Behavioral Outcomes as a Function of Military Service and Gender at 36Month Follow-up

Outcome High school degree Completed treatment Withdrew from treatment Controls Employed or in school Completed treatment Withdrew from treatment Controls Alcohol problems Completed treatment Withdrew from treatment Controls Other drug problems Completed treatment Withdrew from treatment Controls Arrest for nonmisdemeanor Completed treatment Withdrew from treatment Controls

6 months

36 months

91.9 30.2 33.8

91.9 30.2 35.4

88.7 72.1 69.2

76.6 65.1 78.5

3.2 11.6 15.4

21.0 32.6 36.9

3.2 7.0 15.4

12.1 16.3 23.1

14.5 44.2 43.1

41.9 55.8 53.8

Gender Outcome High school degree Enlisted Did not enlist Employed or in school Enlisted Did not enlist Alcohol problems Enlisted Did not enlist Other drug problems Enlisted Did not enlist Arrest for nonmisdemeanor Enlisted Did not enlist

Boys

Girls

Total

100.0 90.8

100.0 85.7

100.0 89.2

100.0 67.7

100.0 71.4

100.0 68.8

13.0 26.2

0.0 21.4

9.7 24.7

4.3 15.4

0.0 14.3

3.2 15.1

17.4 46.2

12.5 60.7

16.1 50.5

Note. Cell ns are as follows: Boys enlisted ⫽ 23, boys not enlisted ⫽ 65, girls enlisted ⫽ 8, girls not enlisted ⫽ 28.

EVALUATION OF MILITARY-STYLE TREATMENT

school, ␹2(1, N ⫽ 124) ⫽ 12.62, p ⬍ .001, ␸ ⫽ .32, and less likely to be arrested, ␹2(1, N ⫽ 124) ⫽ 11.31, p ⫽ .001, ␸ ⫽ .30, than adolescents who completed CA but did not serve in the military. The association between military service and greater likelihood of high school graduation, ␹2(1, N ⫽ 124) ⫽ 3.63, p ⫽ .057, ␸ ⫽ .17, reduced likelihood of alcohol problems, ␹2(1, N ⫽ 124) ⫽ 3.18, p ⫽ .075, ␸ ⫽ .16, and reduced likelihood of drug problems, ␹2(1, N ⫽ 124) ⫽ 3.06, p ⫽ .080, ␸ ⫽ .16, only approached significance. Adolescents’ behavioral outcomes at 36month follow-up were not significantly associated with gender: attainment of a high school degree, ␹2(1, N ⫽ 124) ⫽ 0.635, p ⫽ .426; employment, ␹2(1, N ⫽ 124) ⫽ 0.038, p ⫽ .845; alcohol problems, ␹ 2 (1, N ⫽ 124) ⫽ 0.566, p ⫽ .452; other drug problems, ␹2(1, N ⫽ 124) ⫽ 0.046, p ⫽ .830; and legal problems, ␹2(1, N ⫽ 124) ⫽ 1.36, p ⫽ .244.

Discussion To our knowledge, our study provides the first quantitative data investigating the longterm outcomes of adolescents referred to voluntary military-style residential treatment. Youths with disruptive behavior problems showed significant reductions in externalizing and internalizing problems and improvement in adaptive skills 6 months after treatment. Treatment was also associated with more desirable behavioral outcomes at 6-month follow-up. Youths who completed treatment were more likely to earn a high school degree and to be employed than controls. Furthermore, youths who completed treatment were approximately three times less likely to be arrested for a nonmisdemeanor and approximately five times less likely to have an alcohol or other drug problem than adolescents in the control group. Unfortunately, most of the socioemotional and behavior gains displayed by graduates at 6-month follow-up disappeared 30 months later. At 36-month follow-up, we observed no differences between graduates, adolescents who withdrew, and controls on any of the socioemotional outcome measures. Although program graduates’ socioemotional functioning was within the normal range at 6-month follow-up, their Externalizing Problems score was approximately 1.5 SDs above the mean and their Adaptive Skills

149

score was approximately 1.5 SDs below the mean at 36-month follow-up. On average, program graduates reverted to the same levels of behavior problems that prompted their referral to treatment 3 years earlier. The data concerning adolescents’ long-term behavioral outcomes yielded mixed results. On one hand, program graduates were more likely than youths in the comparison groups to earn a high school degree by 36-month follow-up. Indeed, almost all youths in the comparison groups who had not completed high school by 6-month follow-up continued to lack a high school degree at 36-month follow-up. This finding highlights one benefit of treatment: It is effective in helping youths with disruptive behavior problems acquire a high school degree. Given the financial, health, social, and emotional benefits associated with earning a high school degree, this tangible benefit of treatment should not be overlooked (Caputo, 2005; Suh-Ruu, 2008). On the other hand, we saw no other differences in the behavioral outcomes of program graduates and comparison youths at 36-month follow-up. Although adolescents who completed treatment showed higher rates of employment and lower rates of alcohol use problems, drug use problems, and rearrest than comparison youths, these differences were not statistically significant. These results are somewhat disappointing, given that behavioral outcomes, such as employment and recidivism, are of primary interest to corrections officials and policy analysts (MacKenzie, 2006). Furthermore, these behavioral outcomes have the greatest impact on the general public who must deal with the financial and social costs of continued substance abuse and antisocial behavior in their communities (Benda, 2005a). Results of our evaluation are consistent with several previous evaluations of mandatory military-style residential treatment programs in the juvenile justice system (MacKenzie, 2006; Weis & Toolis, 2007). Although previous evaluations have been limited by methodological shortcomings such as high attrition and lack of control groups, available data indicate that military-style residential treatment is associated with only short-term changes in adolescents’ attitudes and self-esteem. When changes in socioemotional functioning and behavior are observed, they tend to dissipate several years

150

WEIS AND TOOLIS

after treatment despite the voluntary nature of treatment. Our results also indicated that the long-term effects of military-style residential treatment may differ for boys and girls. Boys who completed treatment showed similar socioemotional functioning at 36-month follow-up as boys who withdrew from treatment and controls. However, girls who completed treatment showed significantly more overall symptoms than girls in the two comparison groups. The negative outcomes shown by girls who completed treatment are consistent with previous research indicating potential iatrogenic effects of group therapy for disruptive youths (Dishion & Stormshak, 2007). It is possible that the girls in our intervention program selectively reinforced each other’s antisocial behavior during the course of treatment, a process labeled deviancy training (Dishion & Dodge, 2005). However, the symptom increase shown by girls in our study was largely due to elevations in internalizing symptoms, rather than disruptive behavior problems; furthermore, symptom exacerbation was shown only by girls, not boys. Consequently, it is possible that other mechanisms, besides deviancy training, may explain girls’ increase in symptom presentation— mechanisms unique to girls. For example, military-style treatment may stigmatize girls in ways differently than it does boys and may contribute to girls’ feelings of low self-esteem and self-worth (Weiss et al., 2005). Similarly, girls who are separated from family and friends during the course of residential treatment may be more susceptible to loneliness and depression than boys (Lutze, 2006). Another possibility is that girls, during the course of residential treatment, may model and selectively reinforce depressogenic cognitions that exacerbate their social and emotional problems (Stevens & Prinstein, 2005). Clearly, possible iatrogenic effects of residential treatment for disruptive girls deserve greater attention in the empirical literature. Although the results of our evaluation are disappointing, they should not be disheartening. The data seem to show that a short-term intervention program, no matter how carefully structured or intense, cannot compensate for the great number of psychosocial risk factors encountered by delinquent youth. For example, researchers have identified several risk factors

that place children on developmental trajectories toward disruptive behavior. These risks include genetic and biological predispositions, difficult temperament, hostile and coercive parent– child interactions, academic problems and school failure, peer rejection, and association with deviant peers (McMahon & Kotler, 2006). Given that adolescent conduct problems emerge across childhood and adolescence, it may not be realistic to expect a 22-week residential treatment program to dramatically alter the developmental pathways on which genetic, psychosocial, and cultural risk factors have placed these youth. Furthermore, the core components of the National Guard Challenge Program may not address many of these psychosocial risks. The intervention might have yielded more lasting results had treatment focused on teaching youths new skills to improve academic and work functioning, avoid problematic substance use and associations with deviant peers, and increase their capacity for emotion regulation (Benda, 2005b; Mitchell et al., 2007; Wilson, MacKenzie, & Mitchell, 2005). Our evaluation revealed one possible moderator for the relationship between military-style residential treatment and adolescents’ socioemotional and behavior outcomes: military service. Adolescents who enlisted after completing treatment displayed fewer externalizing problems and greater adaptive skills at 36-month follow-up than adolescents who completed treatment and returned to their communities. Adolescents who completed treatment and enlisted in the military also showed superior behavioral outcomes than adolescents who completed treatment and did not enlist. Although adolescents’ outcomes were largely independent of gender, the benefits of military service were more pervasive for boys than for girls. Boys who enlisted in the armed forces after treatment showed reductions in externalizing and internalizing symptoms, whereas girls continued to show significantly high internalizing symptoms after treatment completion, irrespective of whether they enlisted or returned to their homes. Military service might protect adolescents from returning to previous patterns of antisocial behavior in at least three ways. First, adolescents who enlist in the armed forces experience the same degree of structure and behavioral monitoring that they had during residential

EVALUATION OF MILITARY-STYLE TREATMENT

treatment. Adolescents who enlist in the military may be less able to engage in delinquent behaviors than youths in civilian life given these environmental constraints. Indeed, careful monitoring of adolescent behavior is strongly associated with the prevention of disruptive disorders (Patterson & Yoerger, 2002). Second, adolescents who enlist in the military are surrounded by peers with common, prosocial goals. In contrast, youths in civilian life may associate with deviant peers who reinforce substance use and antisocial behaviors (Piquero, Gover, MacDonald, & Piquero, 2005). Third, adolescents who enlist in the armed forces distance themselves from many of the psychosocial stressors that can contribute to the development of socioemotional and behavioral problems. In contrast, youths who return to their former communities must again cope with family, interpersonal, and environmental stressors that place them at risk for continued relapse. Our data do not allow us to conclude with certainty that military service protects youth from relapse. It is possible that military service alone is sufficient to produce long-term socioemotional and behavioral gains without the need for residential treatment. For example, recent research has indicated that at-risk youths who enlist in the military experience financial and social benefits not experienced by their civilian counterparts (Teachman & Tedrow, 2007). At the very least, it is likely that one of the benefits of military-style residential treatment is that it prompts some adolescents to select a career in the armed forces that they otherwise would not have considered. Future research might be directed at comparing the outcomes of youths who select military careers after treatment with at-risk youths who enlist in the military without participating in a treatment program. It is also possible that the observed benefit associated with enlistment is an artifact of the screening process recruitment officers use when deciding whether to accept an individual for military service. A second, related avenue for future study would be to examine the potential iatrogenic effects of military service for at-risk youths. In some instances, military service may place young adults at increased risk for psychosocial stress and socioemotional problems. Indeed, the prevalence of mood disorders, substance use disorders, and posttraumatic stress disorder is

151

high among individuals currently serving in the armed forces, especially those individuals who have experienced prolonged tours of duty or combat-related traumas (Lapierre, Schwegler, & LaBauve, 2007; Milliken, Auchterlonie, & Hoge, 2007). Although both baseline and follow-up data for our study were collected during the United States military conflicts in Iraq and Afghanistan, only five (16%) youths who enlisted after treatment actually served in the Middle East, and it is unknown whether they engaged in combat. It is possible that military service may not protect youths from developing behavioral and socioemotional problems if service is associated with prolonged separation from loved ones and combat-related psychosocial stress. A final direction for future research is to investigate the impact of aftercare services on the long-term outcomes of adolescents referred to military-style residential treatment programs. In recent years, many programs have begun to help adolescents transition from residential treatment to life in the community by providing mentors, apprenticeships in local businesses, and job placement. Anecdotal evidence indicates that aftercare services may be critical components of treatment and may help prevent relapse (MacKenzie, 2006). Indeed, the benefits of military service seen in this study might be attributed to the fact that youths who enlisted after program completion received aftercare services provided by the military, whereas youths who did not enlist were not given such support. Research is sorely needed to investigate the long-term effects of aftercare services for youths referred to military-style residential treatment to determine whether these services are critical to helping adolescents maintain treatment gains over time.

References Armstrong, G. S. (2004). Boot camps as a correctional option. In D. L. MacKenzie & G. S. Armstrong (Eds.), Correctional boot camps (pp. 7–15). Thousand Oaks, CA: Sage. Austin, J., Camp-Blair, D., Camp, A., Castellano, T., Adams-Fuller, T., Jones, M., et al. (2002). Multisite evaluation of boot camp programs: Final report. Washington, DC: U.S. Department of Justice. Benda, B. B. (2005a). Boot camps revisited: Issues, problems, prospects. Journal of Offender Rehabilitation, 40, 1–25.

152

WEIS AND TOOLIS

Benda, B. B. (2005b). Self-control, gender, and age: A survival analysis of recidivism among boot camp graduates in a 5-year follow-up. Journal of Offender Rehabilitation, 40, 87–113. Caputo, R. K. (2005). The GED as a predictor of mid-life health and economic well-being. Journal of Poverty, 9, 73–97. Cisler, J. M., Barnes, A. C., Fransworth, D., & Sifers, S. K. (2007). Reporting practices of dropouts in psychological research using a wait-list control: Current state and suggestions for improvement. International Journal of Methods in Psychiatric Research, 16, 34 – 42. Crnic, K., & Low, C. (2002). Everyday stresses and parenting. In M. H. Bornstein (Ed.), Handbook of parenting (pp. 243–268). Mahwah, NJ: Erlbaum. Des Jarlais, D. C., Lyles, C., Crepaz, N., & the TREND Group. (2004). Improving the reporting quality of nonrandomized evaluations of behavioral and public health evaluations: The TREND statement. American Journal of Public Health, 94, 361–366. Dishion, T. J., & Dodge, K. A. (2005). Peer contagion in interventions for children and adolescents: Moving towards an understanding of the ecology and dynamics of change. Journal of Abnormal Child Psychology, 33, 395– 400. Dishion, T. J., Spracklen, K. M., Andrews, D. W., & Patterson, G. R. (1996). Deviancy training in male adolescent friendships. Behavior Therapy, 27, 373–390. Dishion, T. J., & Stormshak, E. A. (2007). Intervening in children’s lives: An ecological, familycentered approach to mental health care. Washington, DC: American Psychological Association. Dodge, K. A. (2006). Translational science in action: Hostile attributional style and the development of aggressive behavior problems. Development and Psychopathology, 18, 791– 814. Harrell, T. H., & Wirtz, P. W. (1990). Adolescent drinking index. Odessa, FL: Psychological Assessment Resources. Hart, E. L., & Lahey, B. B. (1999). General child behavior rating scales. In D. Shaffer, C. P. Lucas, & J. E. Richters (Eds.), Diagnostic assessment in child and adolescent psychopathology (pp. 65– 87). New York: Guilford Press. Kazdin, A. E. (2003). Research design in clinical psychology (4th ed.). Boston: Allyn & Bacon. Lapierre, C. B., Schwegler, A. F., & LaBauve, B. J. (2007). Posttraumatic stress and depression symptoms in soldiers returning from combat operations in Iraq and Afghanistan. Journal of Traumatic Stress, 20, 933–943. Lee, M. Y., Uken, A., & Sebold, J. (2007). Role of self-determined goals in predicting recidivism in domestic violence offenders. Research on Social Work Practice, 17, 30 – 41.

Lovejoy, M. C., Graczyk, P. A., O’Hare, E., & Neuman, G. (2000). Maternal depression and parenting behavior: A meta-analytic review. Clinical Psychology Review, 20, 561–592. Lutze, F. E. (2006). Boot camp prisons and corrections policy: Moving from militarism to an ethic of care. Criminology and Public Policy, 5, 389 – 400. Lutze, F. E., & Bell, C. A. (2005). Boot camp prisons and masculine organizations: Rethinking recidivism and program design. Journal of Offender Rehabilitation, 40, 133–152. MacKenzie, D. L. (2006). What works in corrections: Reducing the criminal activities of offenders and delinquents. New York: Cambridge University Press. MacKenzie, D. L., Gover, A. R., Armstrong, G. S., & Mitchell, O. (2001). A national study comparing the environments of boot camps with traditional facilities for juvenile offenders. Washington, DC: U.S. Department of Justice. MacKenzie, D. L., & Rosay, A. B. (2004). Correctional boot camps for juveniles. In D. L. MacKenzie & G. S. Armstrong (Eds.), Correctional boot camps (pp. 26 – 45). Thousand Oaks, CA: Sage. Marcus-Mendoza, S. T., Klein-Saffran, J., & Lutze, F. (1998). A feminist examination of boot camp prison programs for women. Women and Therapy, 21, 173–185. McCullough, J. J., Engel, I., & Wright, S. (2006). Families and juveniles. In Law and mental health professionals (pp. 152–258). Washington, DC: American Psychological Association. McMahon, R. J., & Kotler, J. S. (2006). Conduct problems. In D. A. Wolfe & E. J. Wash (Eds.), Behavioral and emotional disorders in adolescents (pp. 153–225). New York: Guilford Press. Milliken, C. S., Auchterlonie, J. L., & Hoge, C. W. (2007). Longitudinal assessment of mental health problems among active and reserve component soldiers returning from the Iraq war. Journal of the American Medical Association, 298, 2141–2148. Mitchell, O., Wilson, D. B., & MacKenzie, D. L. (2007). Does incarceration-based drug treatment reduce recidivism? A meta-analytic synthesis of the research. Journal of Experimental Criminology, 3, 353–375. Moberg, D. P. (1991). The Adolescent Drug Involvement Scale. Journal of Adolescent Chemical Dependency, 2, 75– 88. Olson, C. L. (1976). On choosing a test statistic in multivariate analysis of variance. Psychological Bulletin, 83, 579 –586. Patterson, G. R., Reid, J. B., & Dishion, T. J. (1992). Antisocial boys. Eugene, OR: Castalia. Patterson, G. R., & Yoerger, K. (2002). A developmental model for early- and late-onset delinquency. In J. B. Reid, G. R. Patterson, & J. Snyder (Eds.), Antisocial behavior in children

EVALUATION OF MILITARY-STYLE TREATMENT

and adolescents: A developmental analysis and model for intervention (pp. 147–172). Washington, DC: American Psychological Association. Peters, M., Thomas, D., Zamberlan, C., & Caliber Associates. (1997). Boot camps for juvenile offenders: Program summary. Washington, DC: Office of Juvenile Justice and Delinquency Prevention. Piquero, N. L., Gover, A. R., MacDonald, J. M., & Piquero, A. R. (2005). The influence of delinquent peers on delinquency: Does gender matter? Youth and Society, 36, 251–275. Reynolds, C. R., & Kamphaus, R. W. (1998). Behavioral Assessment System for Children manual. Circle Pines, MN: AGS. Reynolds, C. R., & Kamphaus, R. W. (2004). Behavioral Assessment System for Children—2nd Edition manual. Circle Pines, MN: AGS. Sattler, J. M. (2002). Assessment of children: Behavioral and clinical applications (4th ed.). San Diego: Author. Shearer, R. A. (2003). Court ordered counseling. Annals of the American Psychotherapy Association, 6, 8 –11. Steiner, B., & Giacomazzi, A. L. (2007). Juvenile waiver, boot camp, and recidivism in a northwestern state. The Prison Journal, 87, 227–240. Stevens, E. A., & Prinstein, M. J. (2005). Peer contagion of depressogenic attributional styles among adolescents: A longitudinal study. Journal of Abnormal Child Psychology, 33, 25–37. Suh-Ruu, O. (2008). Do GED recipients differ from graduates and school dropouts? Urban Education, 43, 83–117. Tabachnick, B. G., & Fidell, L. S. (2007). Using multivariate statistics. Boston: Pearson. Teachman, J., & Tedrow, L. (2007). Joining up: Did military service in the early all volunteer era affect subsequent civilian outcomes? Social Science Research, 36, 1447–1474. Velicer, W. F., & Prochaska, O. (2008). Stage and non-stage theories of behavior and behavior

153

change: A comment on Schwarzer. Applied Psychology: An International Review, 57, 75– 83. Weis, R., & Toolis, E. E. (2007). Military-style residential treatment for disruptive adolescents: A critical review and look to the future. In T. C. Rhodes (Ed.), Focus on adolescent behavior research (pp. 75–118). Hauppauge, NY: NOVA Science Publishers. Weis, R., Whitemarsh, S. M., & Wilson, N. L. (2005). Military-style residential treatment for disruptive adolescents: Effective for some girls, all girls, when, and why? Psychological Services, 2, 105–122. Weis, R., Wilson, N. L., & Whitemarsh, S. M. (2005). Evaluation of a voluntary, military-style residential treatment program for adolescents with academic and conduct problems. Journal of Clinical Child and Adolescent Psychology, 34, 692– 705. Weiss, B., Caron, A., Ball, S., Tapp, J., Johnson, M., & Weisz, J. R. (2005). Iatrogenic effects of group treatment for antisocial youth. Journal of Consulting and Clinical Psychology, 73, 1036 –1044. Wilson, D. B., Bouffard, L. A., & MacKenzie, D. L. (2005). A quantitative review of structured, grouporiented, cognitive– behavioral programs for offenders. Criminal Justice and Behavior, 32, 172– 204. Wilson, D. B., Gallagher, C. A., & MacKenzie, D. L. (2000). A meta-analysis of corrections-based education, vocation, and work programs for adult offenders. Journal of Research in Crime and Delinquency, 37, 347–368. Wilson, D. B., MacKenzie, D. L., & Mitchell, F. N. (2005). Effects of correctional boot camps on offending. New York: Campbell Collaboration. Wood, P. B., May, D. C., & Grasmick, H. G. (2005). Gender differences in the perceived severity of boot camp. Journal of Offender Rehabilitation, 40, 153–175. Received June 4, 2008 Revision received October 7, 2008 Accepted December 12, 2008 䡲