In addition to contributing to the understanding of the development of achievement gaps in schooling, this chapter advances the literature on measurement error in models using lagged dependent variables as predictors. The IV approach we use and its findings are consistent with the existing literature (e.g., Bradbury et al., 2015; Holm et al., 2025; Jerrim & Vignoles, 2013; Passaretta & Skopek, 2018b; Passaretta et al., 2022; Skopek & Passaretta, 2018; Van Huizen, 2018). Failing to account for measurement error could lead to opposite findings in some cases. Using OLS, we would have concluded that female students make smaller learning advancements than their male peers in math, whereas the IV approach shows that this is not the case. These findings provide important information to researchers, policy makers, practitioners, schools, and teachers. Students do not start with a clean slate in primary education. Therefore, early skill development must remain high on the agenda. With respect to existing achievement gaps in education, our findings highlight that closing these achievement gaps, if desired, implies considering a student’s initial skill level and a student’s learning advancement during primary education. was not statistically significant in his study. Also here, this could be explained by the fact that we include additional covariates in our analyses. 75
RkJQdWJsaXNoZXIy MjY0ODMw