CASMIN Versus ISCED: Validating Standard Education Variables for German Microdata

Abstract

In German surveys, educational attainment is typically assessed through two questionnaire items: the highest general school-leaving qualification and the highest vocational qualification or higher education degree. These items are often combined into one variable for analysis, but there is no standard method for doing so. This hinders comparisons across publications and the aggregation of research data. This article presents the results of a large-scale validation of candidates for a German standard education variable based on the official International Standard Classification of Education (ISCED) and the sociological Comparative Analysis of Social Mobility in Industrial Nations education scheme (CASMIN), both commonly used in Germany. Many survey datasets offer a derived ISCED variable that reflects the main ISCED levels; fewer offer CASMIN. The validation uses ALLBUS 2018 data and a data-driven selection of 157 validation variables. Candidate standard education variables that retain a higher relative partial explanatory power across validation variables in linear multiple regression models compared to the most detailed education variable are considered more valid. The results show that CASMIN-based candidates generally outperform ISCED-based candidates of the same number of categories. An ISCED-based variable performs better only when measured in a highly detailed fashion that accounts for the stratification of the German educational system. The commonly used aggregation into three broad education levels loses about half the explanatory power of the detailed variable. Research data centers are called upon to provide a CASMIN variable to data users, and researchers to use CASMIN rather than ISCED whenever possible.