Methods: A sequential cohort design was used to examine the responses of third-grade students who received the MC program in two elementary schools. In 2000-1, students entering the third grade comprised the first comparison cohort (n=177). They received a routine health curriculum. In 2001-2 (n=173) and 2002-3 (n=198), third-grade students received MC plus the routine curriculum. These two cohorts constituted the intervention condition. Last, MC was withdrawn and instruction returned to the routine health curriculum in 2003-4. After a 1-year lag, additional comparison data were collected from the 2004-5 third-grade cohort (n=140).
For each cohort, teachers conducted fall and spring child behavioral assessments. Prior research with similar scales found teacher assessments to reliably report child behavior within the school context (Huesmann et al., 1994). Measures included overt aggression (α=.79), relational aggression (α=.80), cognitive concentration (α=.97), social competence (α=.91), and peer relations (α=.81). No differences between cohorts on baseline covariates or outcome variables were observed.
Results: A person-centered model was estimated by cohort over time. LCA evaluation using fit statistics (e.g., BLRT, LMR, BIC, and entropy) revealed a four-class latent profile solution at each data wave for each cohort. In addition to fit statistics, substantive information was used in confirming the four-class model. Specifically, the percentage of students fitting a high-risk class (6-9%) was substantiated against prior studies (see, e.g., 9%; Fast Track, CPPRG, 2002; 8%, OJJDP, 2001; 5-10%, Seattle Social Development Project, Ayers et al., 1999; 5-10%, Sugai & Horner, 2008). Extending the LCA, a LTA model was fit.
Based on LTA, significantly more MC students appear to have remained in low-risk profiles compared to students in comparison cohorts. Additionally, more students exposed to MC fit progressive transitions and fewer fit digressive transitions. That is, more high-risk students progressed to lower-risk status in the MC versus comparison cohorts. Conversely, more students in the comparison cohorts digressed from lower to higher risk profiles. Mobility tables and figures will be presented to graphically portray these patterns.
Implications: The findings suggest that social-cognitive skills training reduces risk for high-risk children while buffering children in low-risk profiles. In practice, students fitting high-risk profiles comprise only 5-10% of school populations; but they are responsible for over 60% of all disciplinary referrals (Kaufman et al., 2010; Sugai et al., 2002). Prevention programs that alter the status of high-risk children have both individual and environmental effects. They directly affect developmental trajectories by strengthening skills, and they indirectly affect learning opportunities by reducing classroom disruptions. The PCM analysis suggests MC may hold dual promise for high-risk children.