Chapter 3. Primary School Skill Development: from First to Sixth Grade erogeneity (Panel I), while the latter model controls for the average school performance (Panel II). Table 3.4 shows OLS and IV estimates. The table demonstrates that the OLS estimates for the second-grade skill score serve as lower bounds, as the IV estimates are consistently higher across all skills. The difference between both estimates is substantial and ranges from 0.19 SD to 0.28 SD. When we compare the OLS and IV estimates for parental education, two notable observations emerge. First, the estimates on the learning advancement of students of low-educated parents compared to students of middle-educated parents do not significantly differ. Second, the OLS estimates overestimate the learning advancement of students of high-educated parents compared to that of middle-educated parents. While this overestimation appears negligible for spelling, it is evident in reading and math. Last, the OLS and IV estimates for the average school performance measure remain roughly consistent. Beyond the relationship between initial skill levels in primary education and those in sixth grade, we analyze how this relationship evolves throughout primary education. Figure B.2 in the appendix illustrates that the OLS estimates are consistently lower than the IV estimates. It seems that this difference between OLS and IV estimates diminishes during primary education. Furthermore, the figure demonstrates that skill levels in later grades are better predictors for sixth-grade performance than first-grade performance, which is due to their temporal proximity. Differences in IV estimates among skills seem rather subtle. 64
RkJQdWJsaXNoZXIy MjY0ODMw