The effectiveness of EQUIP on sociomoral development and recidivism reduction: A meta-analytic study

The effectiveness of EQUIP on sociomoral development and recidivism reduction: A meta-analytic study
of 9
All materials on our website are shared by users. If you have any questions about copyright issues, please report us to resolve them. We are always happy to assist you.
Related Documents
  This article appeared in a journal published by Elsevier. The attachedcopy is furnished to the author for internal non-commercial researchand education use, including for instruction at the authors institutionand sharing with colleagues.Other uses, including reproduction and distribution, or selling orlicensing copies, or posting to personal, institutional or third partywebsites are prohibited.In most cases authors are permitted to post their version of thearticle (e.g. in Word or Tex form) to their personal website orinstitutional repository. Authors requiring further informationregarding Elsevier’s archiving and manuscript policies areencouraged to visit:  Author's personal copy The effectiveness of EQUIP on sociomoral development and recidivismreduction: A meta-analytic study M.A. van Stam a, ⁎ ,1 , W.A. van der Schuur a,1 , S. Tserkezis a,1 , E.S. van Vugt a , J.J. Asscher a , J.C. Gibbs b , G.J.J.M. Stams a a University of Amsterdam, The Netherlands b Ohio State University, USA a b s t r a c ta r t i c l e i n f o  Article history: Received 13 September 2013Received in revised form 3 January 2014Accepted 3 January 2014Available online 15 January 2014 Keywords: EQUIPEffectivenessMeta-analysisSociomoral developmentRecidivism Juvenile offenders Two multilevel meta-analyses, consisting of 10 studies and 33 effect sizes ( N   = 796 subjects) and 6 studies and22effectsizes( N  =1179subjects)wereconductedtoexaminetheeffectsoftheEQUIPinterventiononthelevelof sociomoral development and recidivism, respectively; moderating effects of participant, program, and studycharacteristics were also examined. A signi fi cant overall effect on sociomoral development was found ( d  =.27). The effect of EQUIP on sociomoral development was moderated by the sample size: studies with a largersamplesizehadsmallereffectsizes.Forrecidivism,anon-signi fi cantoveralleffectsize( d =.13)wasattributabletostrong moderator effects bygender (asigni fi canteffectfor girls of   d  = .55), yearof publication (older studiesyielded smaller effect sizes), ethnicity (smaller effects in non-Caucasians), and especially region, indicating thatstudiesconductedintheUSA(showinghightreatmentintegrity)weremoreeffective( d =.32)thanstudiescon-ductedoutsidetheUSA(showinglowornegligibletreatmentintegrity, d = − .31).Theseresultssuggesttheim-portance of considering potential effects of study, participant characteristics, and program integrity whendelivering EQUIP.© 2014 Elsevier Ltd. All rights reserved. 1. Introduction Cognitivedevelopmentaltheory(Blasi,1999;Colby&Kohlberg,1987;Nucci, 2002) assumes an important role for moral cognition in the moti-vation of (im)moral acts and thus in delinquent behavior. According tothis view (see Gibbs, 2013), antisocial behaviors are caused in part by adelay in moral development, as characterized by immature or super fi cialmoral judgment, and a strong egocentric bias (DiBiase, Gibbs, Potter, &Blount, 2012; Gibbs, Arnold, Ahlborn, & Cheesman, 1994). Indeed, themoral judgment of delinquents has been found to be less mature thanthat of non-delinquents (Blasi, 1980; Gregg, Gibbs, & Basinger, 1994;Nelson, Smith, & Dodd, 1990; Stams et al., 2006). In line with this cogni-tive – behavioral approach, a number of intervention programs havebeen developed, aiming to improve sociomoral development in order toreduce antisocial and delinquent behavior among youth. One of theseintervention programs is EQUIPping Youth to Help One Another trainingprogram (EQUIP), a group-based cognitive – behavioral program that wasdeveloped by Gibbs, Potter, and Goldstein (1995) and Gibbs, Potter, and DiBiase (2013).Bycombiningapeerhelping(ormutualhelp)andskills-streaming(orcognitivebehavioral)approach,EQUIPaimstoteachjuveniledelinquentstothinkandactresponsibly.BasedonaPositivePeerCulture(PPC)model(Vorrath & Brendtro, 1985), the peer helping component of the EQUIPprogramaimstotransformtheself-destructiveandantisocialpeercultureintoacultureinwhichindividualsfeelresponsibleforeachotherandhelponeanother(Gibbsetal.,1995).Apeerhelpingapproachaloneisineffec-tive in preventing recidivism (Gottfredson, 1987). Although juveniledelinquents participating in PPC programs desire to help fellow groupmembers, they often lack the helping skills and moral maturity to do soin a meaningful way. Therefore, EQUIP, along with the PPC program,adaptsandexpandsupon “ skills-streaming ” materialfoundinAggressionReplacement Training (ART, Glick & Goldstein, 1987).Beyond skills training, the EQUIP cognitive behavioral approachemphasizes social perspective taking and cognitive restructuring. Theseemphases are evident in EQUIP's curriculum addressing three typicalyouthful offender problematic tendencies or limitations (in terms of developmentaldelaysorde fi ciencies):immaturemoraljudgment,cogni-tive distortions, and a lack of social skills.The  fi rst limitation, a delay in sociomoral development, is de fi ned as “ thepersistence ” beyondearlychildhoodofanimmaturemoraljudgmentand a pronounced  “ me-centeredness ”  or  “ egocentric bias ”  (Gibbs et al.,1995, p. 43). Indeed, research has shown that the moral judgment of delinquents is less mature than the moral judgment of non-delinquents(e.g., Stams et al., 2006). However, Gibbs (1991, p. 98) suggested that a developmental delay in moral judgment without the presence of cogni-tive distortions does not lead to antisocial behavior per se. Empiricalresearch has shown that cognitive distortions, which constitute the sec-ond limitation, mediate the relation between sociomoral development Children and Youth Services Review 38 (2014) 44 – 51 ⁎  Corresponding author. 1 These authors contributed equally to this paper.0190-7409/$  –  see front matter © 2014 Elsevier Ltd. All rights reserved. Contents lists available at ScienceDirect Children and Youth Services Review  journal homepage:  Author's personal copy and antisocial behavior (Barriga, Morrison, Liau, & Gibbs, 2001; Van derVeldenetal.,2010).Cognitivedistortionsarecharacterizedby “ inaccurateor rationalizing attitudes, thoughts, or beliefs concerning own or other'sbehavior ” (Gibbsetal., 1995,p.108).Lastly, thethirdlimitationconcernslack of social skills. Antisocial youth often show  “ imbalanced and uncon-structivebehaviorindif  fi cultinterpersonalsituations ” (Gibbsetal.,1995,p.105).EQUIPisthoughttoimprovesocialskillsthroughmodeling,role-playing, providing feedback to the role-player, and practicing social skills(Gibbs et al., 1995).Rehabilitation or prevention versions of the EQUIP program are cur-rently widely used in North America, Europe, and Australia, althoughlow levels of program integrity (i.e., inadequate program implementa-tion) at a number of facilities have been noted (Helmond, Overbeek, &Brugman, 2014). For example, in The Netherlands EQUIP constituted anation-wide basic methodology for all juvenile correctional facilities(Dienst Justitiële Inrichtingen, 2010). In addition, a preventive versionfor the srcinal juvenile offenders (Gibbs et al., 1995) has been adaptedfor students from the general population (DiBiase et al., 2012). Severalstudies have analyzed the effectiveness of EQUIP for different outcomes(e.g., social skills, cognitive distortions, moral judgment, and recidivism).To date there is no consensus about the effects of the EQUIP program.To provide a clear overview of the effectiveness studies of EQUIP, wewill perform two meta-analyses (one examining the effects of EQUIP onsociomoral development, and one examining the effects of EQUIP onrecidivism), both including moderators that describe participant, pro-gram and study characteristics that might in fl uence the effectiveness of EQUIP or EQUIP-based programs. 1.1. Present study Thepresentmeta-analyticstudyaimedtoinvestigatetheeffectivenessofEQUIPandEQUIP-basedinterventionsinbothcorrectionalfacilitiesandschool settings. Study outcome variables were sociomoral development(cognitive distortions, social skills, and moral judgment) and recidivism.Included in our investigation was an assessment of whether certainfactors act as moderators in the linkage between EQUIP or EQUIP-basedinterventions and program effectiveness. Eleven moderators were avail-able for inclusion in this meta-analytic study. 1.2. Moderators1.2.1. Participant characteristics Thesocio-demographiccharacteristicsofthestudyparticipantsmighthave an important in fl uence on the outcomes of the study. We thereforeincluded  fi ve moderators that describe the characteristics of the partici-pants.First,severalstudiesindicatedthatitisnecessarytoconsiderdiffer-ences between male and female delinquents. For example, juveniledelinquency has been shown to be more prevalent in males than infemales (Mullis, Cornille, Mullis, & Huber, 2004; Snyder & Sickmund,1994), and females have generally committed fewer violent crimes thanmales(Acoca,1999).However,inrecentdecadesfemalearrestsinviolentcrime increased while male arrests decreased (Garcia & Lane, 2013). Inaddition, it is known that female and male juveniles often react differen-tially in intervention programs: what works for boys does not automati-cally work for girls (Hipwell & Loeber, 2006). We included gender asthe fi rst moderator (M1: Gender).Other moderator variables pertained to other participant characteris-tics and to program or facility related factors. A prevention version of EQUIP has been delivered to non-delinquents, that is, students from thegeneral population. Accordingly, we also examined whether the type of target group (students or juvenile delinquents) moderated the impactof the intervention on sociomoral development (M2: Type of respon-dents). The third variable was age, given indications that age could mod-eratetheimpactoftheintervention(M3:Averageageofparticipants).Forexample, Van der Put, Dekovi ć , Stams, Hoeve, and Van der Laan (2012)found that the impact of criminogenicriskfactorsdecreased signi fi cantlywhenageincreased.Thissuggeststhatinterventionsfocusingonreducingthe impact of criminogenic risk factors could be more effective for youn-ger adolescentscompared to older adolescents and young adults. Fourth,theaveragetimeofparticipantsspentinprisonisincludedasamoderat-ing factor (M4: Average time in prison). The EQUIP program, aiming todecrease delinquent behavior by changing the mental life of juveniles, isnot delivered to the participants in an identical manner. Based on thebackground and availability of the participants, the duration, intensityand speci fi c goal setting of the program might differ. As a consequence,participants who spend a relatively long time in prison may pro fi t moreor perhaps less from the intervention than those who were imprisonedonly for a limited period of time. A  fi fth possible moderating variablewas ethnicity. To take into account the role of cultural and ethnic back-ground characteristics in individuals' sociopsychological adjustment(Cox, Lobel & McLeod, 1991; Duarte & Summers, 2013), we examinedwhether the percentage of non-Caucasian participants in the samplemoderated the linkage between intervention and program effectiveness(M5: Percentage of non-Caucasians). 1.2.2. Study/program characteristics In addition to the characteristics of the participants, study design andprogramcharacteristicsmightin fl uencetheresultsoftheEQUIPeffective-nessstudies.Accordingly,thefollowingsixstudy-andprogramcharacter-istics were included. The  fi rst study characteristic, and the sixthmoderator, was whether the study used a quasi-experimental or a ran-domizedcontrolleddesign(M6:Studydesign).Next,theEQUIPinterven-tion is often differently implemented from one country or region toanother (Helmond, Overbeek & Brugman, 2012). Therefore, the regionof collected data (USA versus Europe) was included as a moderator(M7:Regionofcollecteddata).Studiescanbedividedinthosethatexam-inedacompleteversionoftheEQUIPprogram,andstudiesthatexamineda program ostensibly based on EQUIP principles, or implemented only apart of the program. Because the effectiveness of the program might berelated to whether the program was adopted completely or partially,we included this factor as a moderator (M8: Program completeness). Tounravel the individual effect of EQUIP on the underlying constructs of sociomoral development (social skills, cognitive distortions, and moral judgment) and examine whether the selection of an outcome indicatorconcerningrecidivism(actualrecidivismversusriskforrecidivism)in fl u-ences the effect of EQUIP on respectively sociomoral development orrecidivism, we included outcome type as moderator (M9: Type of out-come).Furthermore,toexaminewhetherthenumberofincludedpartic-ipantshasaneffectontheeffectivenessofEQUIP,weincludedsamplesizeas a moderator (M10: Sample size). Lastly, we tested whether year of publication of the studies moderated the relation between interventionand program effectiveness (M11: Year of publication). 2. Method  2.1. Sample of studies Twocriteriawereformulatedfortheselectionofstudies.First,studieshad to focus on the effect of EQUIP or interventions based on EQUIP(interventionswithcomponentsofEQUIPorPPC)onsociomoraldevelop-mentand/orcriminaloffenserecidivism.Inthecurrentstudy,sociomoraldevelopment consisted of cognitive distortions, social skills, and moral judgment. Second, manuscripts had to present the results of a bivariateanalysis of the association between EQUIP and aspects of sociomoraldevelopment and/or recidivism or to provide enough details to calculatean effect size.The studies were collected between November 2012 and January2013 according to the following procedure. First, electronic databasessuch as ERIC, PsycINFO, PsycLIT, Sociological Abstracts, and DissertationAbstracts,wereused.Wesearchedforarticles,books,chapters,paperpre-sentations, dissertations and reviews. Our purpose was to  fi nd as manystudies as possible. We therefore used a variety of terms related to 45 M.A. van Stam et al. / Children and Youth Services Review 38 (2014) 44 – 51  Author's personal copy EQUIP, sociomoral development and recidivism. The search term EQUIPwascross-referencedwithtermsascognitivedistortion,recidivism,effec-tiveness, and moral judgment.Thereafter, manual searches were conducted. The reference lists of reviews and other articles were checked in order to fi nd relevant studiesthat were not found in the electronic databases.We found 12 controlled studies on the basis of the information in theabstract. Finally, for both meta-analyses 11 independent studies wereselected.TenofthesestudiesexaminedtheeffectsofEQUIPonsociomoraldevelopment (social skills( n  = 7), moral judgment ( n =6), and/or cog-nitive distortions ( n  = 7)) and six examined its effect on risk for recidi-vism ( n  = 3) or recidivism ( n  = 5). One study examined the effects of EQUIP on only recidivism. The ten studies examining effects onsociomoral development (total  N   = 796 subjects) reported 33 effects,and the six studies examining recidivism (total  N   = 1179 subjects)reported 22 effects. The studies that were included in the meta-analysesare displayed in Table 1.  2.2. File drawer problem Publication bias refers to the tendency of journals to accept papersthat mainly report signi fi cant associations, which may have implicationsfor the conclusions derived from the meta-analysis (Rosenthal, 1991;VanIJzendoorn,1998).Thistypeofbiasislabeledasthe fi ledrawerprob-lem (Rosenthal, 1979).Methodsexisttoaddresspotentialeffectsofpublicationbias. Howev-er,accordingtoRothstein(2008),eachmethodhasitsshortcomings.Thebest solution to publication bias is to try to prevent effects of publicationbias by obtaining all unpublished material as best as possible (e.g.,Mullen, 1989; Rosenthal, 1991). Therefore, unpublished dissertationswere considered too. We found two dissertations, but only one met theselection criteria and was included in our study. Additionally, we appliedone of the conventional methods to examine publication bias, which isadvisedbyRothstein(2008).Namely,afail-safenumberisprovided N   fs  ¼ N  0 =  Z  2 t     N  0  Z  20 −  Z  2 t    , whereby  N  0  is the number of effect sizes,  Z  t   is thecriticalvalueof   Z  (  p = 0.1),and  Z  0  isthemean  Z  obtainedforthe N  0 stud-ies, and this fail safe number should exceed the critical value ((5 ∗ N  0 )+ 10), as suggested by Rosenthal (1979, 1991).  2.3. Coding of the study outcomes and characteristics First, we retrieved the study results (test statistic and value). Second,the following characteristics of participants were coded: gender (male,female, and mixed), background of respondents (delinquents vs. stu-dents), average time in prison, average age of participants, percentagenon-Caucasians. We then coded program characteristics: program com-pleteness(EQUIPvs.EQUIPBased)andwhetherornotprogramintegritywas reported. Finally, study/program characteristics were coded: theregion where the data was collected (USA vs. Europe), sample size, yearof publication, study design (Randomized Controlled Trial vs. quasi-experimental), type of outcome assessed (cognitive distortions, socialskills, moral judgment). Because three out of six studies reporting onrecidivism outcomes did not use of  fi cial records of recidivism but riskforrecidivism,thetypeofoutcome(recidivismversusriskforrecidivism)was coded as a categorical moderator.  2.4. Analysis Effect sizes were calculated for all studies. Cohen's  d  was computedto represent the difference between the experimental and controlgroup, using formulas from Lipsey and Wilson (2001) and Mullen (1989). For most studies Cohen's  d  was computed based on meansand standard deviations. If necessary, the  F  , chi-square test statistics,or probabilities were used. When studies did not provide statisticalinformation on non-signi fi cant differences that are needed to calculatethe effect size, an effect size of zero was assigned based on a one-tailed  p  of .50 (  Z   = 0.00).WeusedtheprogramMLwiNforconductingmultilevelanalysesandused an adapted set up to make our models suitable for meta-analysis(Hox, 2002). To deal with dependency of effect sizes within studies,we used a multilevel random effects model for the calculation of com-bined effect sizes and moderator-analyses (Hox, 2002; Van denNoortgate, & Onghena, 2003). A multilevel random effects model ac-counts for the hierarchical structure of the data, in which the effectsizes or study results (the lowest level) are nested within studies (thehighest level). Iterative maximum likelihood procedures were appliedto estimate unknown parameters. First the empty model or interceptonly model (without moderators) was estimated. We conducted a testfor heterogeneity of effect sizes to detect signi fi cant variation in effectsizes across primary studies (Rosenthal, 1991). Thereafter discrete andcontinuous moderators, divided in methodological moderators andsample descriptors, were included to test their effects on the heteroge-neity of effect sizes (did the addition of this moderator decrease thelevel of heterogeneity found between the effect sizes signi fi cantly)and improvement of model  fi t (did the model  fi t the data signi fi cantlybetter after the addition of the moderator). 3. Results  3.1. Sociomoral development  The overall mean effect size for the association between EQUIP andsociomoral development was signi fi cant ( d  = 0.266,  Z   = 3.023,  p b  .01), indicating that participation in EQUIP was related to sociomoraldevelopment (see Table 2). In addition, the homogeneity test revealedthat there was a trend towards heterogeneity (  Z   = 1.515,  p  b  .10).Given the small number of studies included in the analysis, this trendstrongly suggests that study outcomes are in fl uenced by study  Table 1 Overview of the studies that were included in this meta-analytic study.Sociomoral development RecidivismStudy Background Participants Total  N   Social skills Moral judgment Cognitive distortions Risk for recidivism RecidivismNas, Brugman, and Koops (2005) Delinquents 56  ● ● ● Liau et al. (2004) Delinquents 316  ● ● ● ● Claypoole, Moody, and Peace (2000) Delinquents 46  ● ● Brugman and Bink (2011) Delinquents 77  ● ● Helmond, Overbeek, and Brugman (2012) Delinquents 115  ● ● ● ● Devlin and Gibbs (2010) Delinquents 538  ● ● Leeman, Gibbs, and Fuller (1993) Delinquents 54  ● ● Van der Meulen, Granizo, and del Barrio (2010) Students 50  ● DiBiase (2010) Students 45  ● ● Van der Velden, Brugman, Boom, and Koops (2010) Students 622  ● ● ● Stuurman (2009) Students 152  ● ● ● 46  M.A. van Stam et al. / Children and Youth Services Review 38 (2014) 44 – 51  Author's personal copy characteristics (see moderator analysis results in the next sections). It isnoteworthy that the fail-safe number was 6025 based on 33 aggregatedeffect sizes. Therefore a  fi le-drawer problem was not considered likely,because the fail-safe number exceeded the critical value of 175 (5 ∗ 33+ 10), as suggested by Rosenthal (1979, 1991).  3.1.1. Participant characteristics Participant characteristics or incarceration circumstances did not sig-ni fi cantlymoderatetheimpactoftheinterventionsstudiedonsociomoraldevelopment. Speci fi cally, no moderating effects for sociomoral develop-ment outcomes were found for gender (M1), and type of respondent(M2) (see Table 2). Furthermore, there was no moderating effect of ageof the respondents (M3), time spent in prison (M4), and ethnicity (M5)on the effectiveness of EQUIP in improving sociomoral development(Table 3).  3.1.2. Study/program characteristics Of the discrete study/program moderators region of collected data(M7), program completeness (M8), and type of outcome (M9) did notsigni fi cantly moderate the effect of EQUIP on sociomoral development(see Table 2). However, a trend was found towards signi fi cance forstudy design (M6:  Z   = − 1.369,  p  b  .10), indicating that studies with arandomized controlled design yielded smaller effect sizes ( d  = .05)than studies with a quasi-experimental design ( d  = .35). Furthermore,a trend was found for year of publication (M11:  Z   = − 1.316,  p  b  .10;Table 3), indicating that more recent studies showed smaller effectsizes. Moreover, sample size (Table 3) was signi fi cantly associated withthe effect size (M10:  Z   = − 1.75,  p  b  .05), indicating that studies with alarger sample size had smaller effect sizes.  3.2. Recidivism TheoverallmeaneffectsizefortheimpactofEQUIPonrecidivismwasnot signi fi cant ( d  = .128,  Z   = .790,  p  = .215) (see Table 4). However,there was a trend towards heterogeneity (  Z   = 1.533,  p  b  .10). Based onthis possible inconsistency in outcomes among the included studies, weexamined moderators in order to explain heterogeneity of effect sizes.Nopossible fi le-drawerproblemwasfoundbasedonthefail-safenumber  Table 2 Results for the overall mean effect size and discrete moderators for sociomoral development (bivariate models).Moderator variables # Studies # ES  Β 1  (SD)  Z  B1  (  p ) Mean  d Z  d  (  p ) Heterogeneity (  p )  ∆ fi t (  p )Overall 10 33 0.266 3.023 ⁎  (.001) 1.515 + (.065)  – Participant characteristicsM1: Gender 1.515 + (.065) 0.373 (.830)Male (RC) 4 9  –  0.308 2.702 ⁎  (.003)Female 2 4  − 0.066 (0.167)  − 0.395 (.346) 0.242 1.424 + (.077)Mixed sample 6 20  − 0.063 (0.110)  − 0.573 (.283) 0.245 2.552 ⁎  (.005)M2: Type of respondents 33 1.619 + (.053) 1.148 (.284)(Youth) delinquents(RC) 5 16  –  0.178 1.380 + (.084)Students 5 17  − 0.205 (0.166)  − 1.235 (.108) 0.383 2.969 ⁎  (.002)Study/program characteristicsM6: Study design 1.231 (.109) 1.900 (.168)Quasi-experimental (RC) 4 11  –  0.347 2.649 ⁎  (.004)Randomized control 2 9  − 0.297 (0.217)  − 1.369 + (.086) 0.050 0.289 (.386)M7: Region of collected data 1.556 + (.060) 1.456 (.228)USA (RC) 4 13  –  0.423 0.367 (.357)Europe 6 20  − 0.238 (0.191)  − 1.246 (.106) 0.185 1.609 + (.054)M8: Program completeness 1.564 + (.060) 0.059 (.808)EQUIP (RC) 8 21  –  0.303 ⁎  2.806 ⁎  (.003)Based on equip 2 10  − 0.119 (0.227)  − 0.524 (.300) 0.184 0.920(.179)M9: Type of outcome 1.529 + (.063) 1.338 (.512)Social skills (RC) 7 11  –  0.216 2.160 ⁎  (.015)Moral judgment 6 11 0.081 (0.074) 1.095 (.137) 0.297 3.000 ⁎  (.001)Cognitive distortions 7 11 0.073 (0.088) 0.830 (.203) 0.289 2.726 ⁎  (.003)RC = referencecategory;# Studies = numberof independent studies; # ES = number ofeffect sizes; Β 1  = regressioncoef  fi cient ofthe corresponding category,with its standarderrorbetweenbrackets,showingthedifferenceineffectsizewiththereferencecategory;  Z  B1 isthez-scoredemonstratingwhetherthisdifferenceissigni fi cant;mean d =meaneffectsize(d);  Z  d  = signi fi cance of the effect size; heterogeneity = within class heterogeneity (  Z  ); Δ fi t = difference with model without moderators ( χ  2 ). +  p  b  0.1 (tested one-sided). ⁎  p  b  0.05 (tested one-sided).  Table 3 Results for continuous moderators for sociomoral development (bivariate models).Moderator variables # Studies # ES  Β 0  (SD)  Β 1  (SD)  Z   (  p ) Heterogeneity (  p )  ∆ fi t (  p )Participant characteristicsM3: Average age of participants 9 32 0.250 (0.095)  − 0.022 (0.017)  − 1.294 (.449) 1.514 (.065) 1.782 (.182)M4: Average time in prison 5 15 0.342 (0.211) 0.010 (0.058) 0.172 (.432) 1.256 (.105) 0.489 (.484)M5: Percentage of non-Caucasian 5 19 0.396 (0.177)  − 0.007 (0.008)  − .087 (.465) 1.278 (.101)  − 0.591 (.442)Study/program characteristicsM10: Sample size 10 33 0.268 (0.078)  − 0.001 (0.000) a − 1.75 ⁎  (.040) 1.375 (.085) 3.464 + (.062)M11: Year of publication 10 33 0.278 (0.089)  − 0.025 (0.019)  − 1.316 + (.094) 1.500 (.067) 1.718 (.190)# Studies = numberof independentstudies; #ES= number ofeffectsizes; Β 0  = interceptwith its standarderror betweenbrackets; Β 1 = regression coef  fi cient withits standarderrorbetween brackets;  Z   = signi fi cance of the regression coef  fi cient; heterogeneity = within class heterogeneity (  Z  ); Δ fi t = difference with model without moderators ( χ  2 ). +  p  b  0.1 ⁎  p  b  0.05 (tested one-sided). a − 0.0007 (0.0004).47 M.A. van Stam et al. / Children and Youth Services Review 38 (2014) 44 – 51
Similar documents
View more...
Related Search
We Need Your Support
Thank you for visiting our website and your interest in our free products and services. We are nonprofit website to share and download documents. To the running of this website, we need your help to support us.

Thanks to everyone for your continued support.

No, Thanks

We need your sign to support Project to invent "SMART AND CONTROLLABLE REFLECTIVE BALLOONS" to cover the Sun and Save Our Earth.

More details...

Sign Now!

We are very appreciated for your Prompt Action!