completing this method, missingness was reduced to an average of 16% across the six two-week intervals. To analyze the diferences in change over the course of the six two-week intervals, we selected Repeated Measures Analysis of Variance (ANOVA). This was deemed appropriate given the small sample size, multiple occasions, and the homogeneity of variance established by Mauchly’s test of Sphericity ( χ2(14) = 4.25, p = .99). SPSS 28 requires that longitudinal data be complete – otherwise, participants are excluded from Repeated Measures ANOVA. A total of 16% of responses were missing, which was sufcient to justify conducting Multiple Imputation (MI) to impute the missing responses. All presented longitudinal FastBridge fndings are a result of the pooled results of the fve MI-generated data sets. Description of HLM and any covariates used in the model For the Star and FAST assessments, students were tested at the start and end of the study period. There was sufcient sample size in each grade-test combination for grades 1, 2, and 3-5 to be included in these analyses. Three-level hierarchical linear regression models (HLMs) help account for any diferences (e.g., neighborhood efects) that could be measured by the fact that students are “clustered” or “nested” in schools within the district and that schools were assigned to be either treatment or comparison schools. HLMs with time (level 1) nested within students (level 2) nested with schools (level 3) were employed to examine growth in literacy scores. All models contained a series of covariates including gender (“female”; 0 = male, 1 = female), ELL status (“ELL_Status”; 1 = ELL, 0 = non-ELL), Free and reduced price lunch (“Free_Reduced_Lunch”; 1 = FRPL, 0 = non-FRPL), SPED status (“SPED”; 1 = SPED, 0 = non-SPED), Minority (“Minority_Ethnicity”; 1 = Minority, 0 = White), an indicator of time (“Time”; 1 = Middle of Year (MOY), 2 = End of Year (EOY)), an indicator of whether the student was in the treatment or comparison group (“intervention”; 0 = comparison, 1 = Treatment), and an interaction between time and group calculated as the product of Time*group (“Tigr”). We explored main efects of treatment versus comparison groups by considering the signifcance of the interaction between time and group (“Tigr”). A signifcant interaction term would suggest that the slope (i.e., growth) in literacy scores is diferent for the treatment versus comparison groups. All analyses were conducted separately by grade using the statistical software package R 3.6.2. In the current sample, diferent assessments were administered for diferent grade levels at the Middle of Year (MOY) and End of Year (EOY). Table 9 summarizes the assessments administered for each grade.
LXD Research -RISE and RISE UP Winter 2022-Spring 2023 Report
24
Made with FlippingBook - Online catalogs