Chapter 3. Primary School Skill Development: from First to Sixth Grade 3.5 Results 3.5.1 First-sixth-grade relationship First, we analyze sixth-grade skill scores using first-grade skill scores as predictors. Table 3.3 presents four models: (1) model A includes only initial skill levels, (2) model B adds parental education, (3) model C includes school fixed effects, and (4) model D is the full model with student characteristics.13 Additionally, model E replaces school fixed effects with the school performance measure. Table 3.3 shows a strong positive relationship between first- and sixthgrade skill scores. In model A, a one standard deviation higher skill level of first-grade students is associated with a 0.51 SD, 0.41 SD, and 0.57 SD higher sixth-grade reading, spelling, and math skill level, respectively. With additional controls in models B-D, the estimates remain qualitatively robust, with slight adjustments. In model D, after accounting for parental education, schooling, and other factors, the associations are 0.46 SD, 0.42 SD, and 0.54 SD for reading, spelling, and math, respectively. Last, we examine the relationship between sixthgrade performance and average school performance. When controlling for school heterogeneity with fixed effects, the R² of the model increases, indicating that schools explain some variation in sixth-grade skills. However, including school fixed effects has little impact on the first-grade skill coefficients, suggesting that unobserved school heterogeneity only slightly mediates the relationship between first- and sixthgrade skills. 13The regression estimates of these student characteristics can be found in Table B.4. 60
RkJQdWJsaXNoZXIy MjY0ODMw