Volume 1 - Issue 1 - DBU Journal for K-12 Educational Leadership - Page 50

48 Research Question 1 (RQ1) Does the number of disci- pline referrals differ based on the degree of TBSI implemen - tation? H1₀: µ campuses with no TBSI = µ campuses with TBSI common area implementation = µ campuses with TBSI school-wide im - plementation H1: At least one of the population means is different from the others. Setting, Target Population and Sample The setting for the study was a suburban school district in the North Texas region with a total student population in grades kindergarten through 12 of approximately 33,254. The pop- ulation for the current study was kindergarten through fourth grade students enrolled in elementary campuses in a suburban school district in the North Texas region. A total of 22 elementary campuses serve kindergarten through fourth grade students in the selected district. The sample for the current study was derived from nine purposely selected el - ementary campuses representing approximately 41% of the elementary campuses in the selected district. The nine campuses selected consists of 5,578 students. Of the total, 223 students accounted for 463 total discipline referrals. The 223 students were reviewed, and two outli - ers (students) were removed due to the drastic difference in total discipline referrals received resulting in a total sample size of 221 students and 425 discipline referrals. Of the 425 total discipline referrals, 282 were coded as aggressive and 143 were coded as non-aggressive. Based on data support - ing an increase in student aggression towards adults and information discovered in the data collected, the researcher conducted additional statistical analysis by further dividing the aggressive discipline referrals into two sub-categories: aggressive actions directed toward other students and ag - gressive actions directed toward adults. Two hundred four - teen acts of aggression toward students were identified, and 68 acts of aggression toward adults were identified. Instrumentation, Measures and Findings To answer the research question, a one-way MANOVA was utilized with the computed mean scores of each level of the three constructs; aggressive discipline referrals, non-aggres- sive discipline referrals, and total discipline referrals as the dependent variable. The independent variables—full, com- mon, or none—were levels of TBSI implementation. Addi - tional analysis was conducted based on added dependent variables of total aggressive referrals, student to student aggressive referrals (StS), or student to teacher (StT) ag- gressive referrals. The data were examined to determine if differences were present between the three groups for the three constructs. Table 1 presents the means and standard deviations for the three dependent variables represented in the research question and the two additional dependent variables analyzed. (See Table 1 on page 51.) Statistical Package for the Social Sciences (SPSS) out- puts four rows, which each shows the statistics of a separate multivariate test method: Pillai’sTrace, Wilks’ Lambda, Ho- telling’s Trace, and Roy’s Largest Root. For the purposes of the current study, the Wilks’ Lambda was used, as it is the most widely used and accepted (AERD Statistics, 2013; Crichton, 2000). Table 2 reveals no statistically significant difference among the level TBSI implementation groups on a linear combination of the three dependent variables, F (6, 432) = 1.498, p > .05; Wilk’s Λ = 0.96, partial η2 = .020. (See Table 2on page 52.) Because the MANOVA did not reveal a statistically sig- nificant result, further follow-up tests were performed. A Tests of Between-Subjects Effects was conducted to deter - mine how each dependent variable differed. A statistically significant difference was present in non-aggressive disci - pline referrals based on the level of TBSI implementation, F (2,218) = 3.748; p ≤ .025; partial η2 = .033. Bonferroni correc - tion was applied to alpha ( p ≤ .025) to account for multiple analysis of variance. Because statistical significance was detected in the Test of Between-Subjects Effects for non-aggressive referrals, it was necessary to complete a Tukey’s post-hoc pairwise comparison to determine how the groups differ (AERD Sta - tistics, 2013; Fields, 2013). Results of Tukey’s post hoc test Kevin S. Hood, Ed.D.