Journal of K-12 Educational Research 35 Research Question 2 (RQ2) Is there a difference in STAAR math performance for emergent bilingual students in Grades 3, 4, and 5 between campuses led by principals with and without bilingual teacher certification? Research Question 3 (RQ3) Is there a difference in STAAR reading performance for Hispanic students in Grades 3, 4, and 5 between campuses led by principals with and without bilingual teacher certification? Resarch Question 4 (RQ4) Is there a difference in STAAR reading performance for emergent bilingual students in Grades 3, 4, and 5 between campuses led by principals with and withou bilingual teacher certification? Data Analysis Procedures To analyze the data, independent samples t-tests were conducted using SPSS Version 28 to compare the STAAR reading and math performance of Hispanic and EB students across campuses led by principals with and without bilingual certification. To handle these multiple comparisons, a total of 12 independent samples t-tests were performed: three grade levels (third, fourth, fifth) x two subject areas (reading and math) x two student groups (Hispanic and EB). Conducting multiple t-tests increases the risk of committing a Type I error, which may incorrectly lead to rejecting a true null hypothesis. To mitigate this risk, a Bonferroni correction was applied to adjust the significance level. The Bonferroni correction divides the standard alpha level (.05) by the number of tests conducted. In this case, with 12 t-tests, the adjusted alpha level was set at .004 (.05/12) for each individual test (Bonferroni, 1936). This adjustment ensured that the familywise error rate was controlled, reducing the likelihood of false positives across the multiple comparisons. Prior to conducting the independent samples t-tests, several key assumptions were evaluated to ensure the validity of the analyses (Field, 2018). The assumption of independence was met through the study design, as each campus’s data represent distinct independent student populations (Creswell & Creswell, 2018). Normality was assessed using the Shapiro-Wilk test and examination of Q-Q plots for each grade level and subject area combination (Tabachnick & Fidell, 2019). While some deviation from normality can be tolerated given the robustness of t-tests with sufficient sample sizes, any substantial violations were noted as limitations (Cohen, 1988). Homogeneity of variance were tested using Levene’s test, with the appropriate t-test version selected based on results (Field, 2018). Additionally, outliers were identified through box plots and standardized scores, with any extreme values (z > ±3.29) investigated for potential data entry errors or legitimate extreme scores (Tabachnick & Fidell, 2019). Cases identified as outliers were reviewed individually to determine appropriate handling methods that maintain data integrity while not unduly influencing results (Field, 2018). Additionally, since the current study focused only on English STAAR results, any assessments administered in Spanish were excluded from the analysis to maintain consistency. The decision to analyze only English results was based on ensuring comparability across campuses and avoiding potential confounding variables related to language proficiency differences. By conducting separate analyses for each grade level and applying appropriate statistical corrections, the current study aimed to provide accurate and reliable insights into how bilingual-certified principals impact student achievement in Texas public schools. Summary of Findings and Interpretation of Results No statistically significant differences were found in STAAR reading or math scores for Hispanic or EB students in Grades 3 through 5 based on whether the campus principal held bilingual certification. In math, Hispanic students across all three grade levels exhibited non-significant differences (p > .05), with effect sizes under 0.05, indicating negligible impact. Similarly, EB students demonstrated no significant differences in math performance across grades (p > .05), with effect sizes remaining below 0.01. For reading, Hispanic students’ mean scores showed minimal differences, and all associated p-values exceeded the .05 threshold. EB students likewise showed no significant differences in
RkJQdWJsaXNoZXIy ODc4ODgx