Measuring the Effect of Educational Attainment on Fertility Intentions: A Machine Learning Approach

Maria Rita Testa, Wittgenstein Centre (IIASA, VID/ÖAW, WU)
Fabian Stephany, Wittgenstein Centre (IIASA, VID/OEAW, WU)

Recent studies have shown a positive correlation between educational attainment and fertility intentions in Europe. This empirical finding contradicts the positive links commonly observed between educational attainment and actual fertility and the fact that staying longer in education prevents women to have large families. In this paper, we investigates the meaning of this positive correlation using a machine learning approach still largely underdeveloped in demographic research. We aim at uncovering the responsiveness of a positive education-fertility intentions link to the categorizations of the educational attainment measure in scholarly papers. We briefly introduce machine learning algorithms used to build decision trees and then apply such algorithms to delineate the key feature which distinguish the categories of educational attainment using Generations and Gender Surveys data for 10 European countries pooled together. From the analysis emerges that the most discriminatory cut-off level of the educational attainment variable is at the lower secondary level.

Presented in Session 1160: Fertility