Historical trend in educational homophily: U-shaped or not U-shaped? Or, how to set a criterion to choose a criterion?

Abstract: Measuring changes in overall inequality between different educational groups is often performed by quantifying variations in educational marital homophily across consecutive generations. However, this task becomes challenging when the education level of marriageable individuals is generation-specific. To address this challenge, various indicators have been proposed in the assortative mating literature. In this paper, we review a set of criteria that indicators must satisfy to be considered as suitable measures of homophily and inequality. Our analytical criteria include robustness to the number of educational categories distinguished and the negative association between intergenerational mobility and homophily. Additionally, we also impose an empirical criterion on the identified qualitative historical trend in homophily between 1960 and 2010 in the US at the national and sub-national levels. Our analysis reveals that while a specific cardinal indicator meets all three criteria, many commonly applied indices do not. We propose the application of this well-performing indicator to quantify the trend in overall inequality in any country, including European countries, with available population data on couples' education level.