TPC Journal V7, Issue 2 - FULL ISSUE

178 The Professional Counselor | Volume 7, Issue 2 high school students. Both found evidence of adequate construct validity, criterion-related validity and internal consistency reliability (McCoach & Siegle, 2002; Suldo et al., 2008). Data Analysis SAAS-R scores, attendance, discipline and credit accrual pre- and post-intervention data, and control data were entered into Statistical Package for the Social Sciences (SPSS Version 21) for analysis. Next, we screened for missing data. Then we conducted preliminary analyses to examine statistical assumptions (e.g., normality, outliers, linearity, homogeneity of regression, multicollinearity and singularity, and homogeneity of variance-covariance matrices). A repeated measures multivariate analysis of variance was performed to determine if there was a significant difference in participants’ school attitudes, credit accrual, discipline and attendance scores pre- and post- intervention intervals and control intervals (Pallant, 2016). Four dependent variables were used: SAAS-R (assessment), percentage of courses passed (credit and grade accrual), discipline referrals (incidents), and attendance. There were two forms of independent variables: treatment and control, and Time 1 and Time 2. Treatment and control were the between-subjects independent variables and Time 1 and Time 2 were the within-subjects independent variables. This study had four dependent variables (e.g., assessment, grades, incidents, attendance) and one grouping variable with two levels (time and control). The dataset should include more cases than dependent variables, which we satisfied (Pallant, 2016). The power analysis helped to decrease the probability of a Type II error (Balkin & Sheperis, 2011; Cohen, 1992; Faul, Erdfelder, Lang, & Buchner, 2007). For these reasons, a post hoc power analysis was conducted for the means of this study and established sufficient power for the overall model (.98). Results There was no significant main effect due to treatment (time by treatment/control): Wilks’ Lambda = .890, F (4, 47) = 1.451, p = .232. However, the multivariate test did reveal a significant main effect for time: Wilks’ Lambda = .654, F (4,47) = 6.219, p < .001 (see Table 1.1). Because of the significant main effect for time, each dependent variable was investigated further by reviewing the univariate results. Examination of the simple effects indicated a significant difference between pre- and post-values for grades: F (1,50) = 13.178, p < .001. Both treatment and control grades decreased between pre- and post- grades. The simple effects indicated a significant difference in pre- and post-values for discipline: F (1,50) = 6.206, p < .05. Both treatment and control had a decrease in discipline referrals between pre- and post-values. All univariate effects are reported in Table 1.2. Overall multivariate results revealed that time was significant and time by treatment and control was not significant. The test of between- subjects effects results show that there was a significant effect of treatment on SAAS-R: F (1,50) = 5.159, p < .027. All between-subjects univariate effects are reported in Table 1.3. The effect of treatment on SAAS-R revealed a significant result, which indicated that participants who received the intervention scored higher on the SAAS-R at the end of the school year. The participants in the treatment group had higher attitudes toward school than the participants who did not receive the intervention. Table 1.1 Multivariate Effects Wilks’ Lambda F (4,47) p Time .654 6.219 .001 Time by Treatment/Control .890 1.451 .232