Exploring family life circumstances and their relationship to a child’s school achievement – an econometric analysis in large data contexts
This thesis is motivated by the need to find parental characteristics other than demographic indicators which exhibit substantial links to a child’s success in school. Therefore, the associations between various indicators of family background and a child’s school achievement are empirically explored. A latent variable model is assumed to describe the underlying relationships. A data set, large in terms of selectable variables, gives rise to the need to use statistical methods that reduce its dimensionality. Since there many possible methods, simulations comparing their predictive ability are conducted. A regularized version of Factor Regression turns out to be most suitable. Using data from the German Socio-Economic Panel, the method extracts several factors with relevant associations to the outcome. While factors related to demographic indicators play a major role, other factors related to participation in cultural life or the interest and participation in politics or in further education substantially contribute to explaining variance in the outcome. The results suggest that demographic indicators alone might not be sufficient to describe the relevant facets of family background.
Use and reproduction:
No CC License (german copyright law applies)