“No End to Hypergamy when Considering the Full Married Population,” our new article in Population and Development Review, upends existing understandings of how marriage responds to educational context. Whereas recent literature has documented a global pattern of declining prevalence of educational hypergamy (marriages in which wives are less educated than their husbands), we show that, in many contexts, the documented declines are only observed among couples with unequal levels of education and disappear once all marriages are included in the analysis.
Over the past five decades women’s educational opportunities have expanded around the world. In many countries, girls went from trailing boys in educational attainment to catching up with and even surpassing their male peers. In a 2012 article in Population and Development Review, Esteve et al. set out to examine how these gendered changes in educational opportunities transformed marriage patterns, identifying a “nearly worldwide and substantial decline” in educational hypergamy. These findings led the authors to predict that as women approach and overtake men in their level of education, hypergamy will become less prevalent, eventually leading to an “end of hypergamy.” In the years since this publication, a similar methodological approach has been used to confirm declining prevalence of hypergamy across various geographic contexts.
Based on our familiarity with this “end of hypergamy” body of literature, we expected to document near-universal declines in hypergamy in our own work on Latin America and sub-Saharan Africa. Much to our surprise, we did not (Lopus and Frye 2020, Urbina 2021). When providing feedback on drafts of this work, readers asked us to explain why our results departed from the well-documented worldwide trend of declining hypergamy. That is when we realized that the prior analyses had measured hypergamy only among couples with unequal levels of educational attainment, and not as a proportion of all marriages.
In our new publication, we argue that this methodological choice has led scholars to overstate the ubiquity of hypergamy decline around the world and to overlook concurrent changes in the prevalence of homogamy (equal educational pairings). If the prevalence of homogamous couples had remained stable over time, excluding them from analyses would not have much effect on our understanding of hypergamy trends. But, because homogamy has decreased substantially in many countries, excluding homogamous couples can give a false impression of declining hypergamy when its overall prevalence is actually constant, or even increasing. Given these implications, we argue for the importance of considering all three categories of coupling (marrying up, marrying down, and marrying across) in combination when analyzing trends in marital patterns.
We make our case by using census data to empirically examine assortative mating across birth cohorts in Latin America. Our findings show that when homogamy is excluded from the analysis, our results align with prior work pointing to an “end of hypergamy.” But, when all marriages are included in the analysis, the outcome is the opposite: hypergamy has increased – not decreased – over time in most of the countries we study, and it remains relatively stable across an axis of female educational advantage.
The persistence of hypergamy amid a reversal of the educational gender gap is surprising, representing a case where a change in composition does not lead to the expected change in behavior. Substantively, this finding is a reminder of how entrenched gender inequalities are within the private sphere, even as a rising number of girls go to school. More broadly, our results also highlight the importance of carefully selecting and defining denominators, as a seemingly straightforward methodological decision can lead to vastly different understandings of the topic being studied.
About the Authors
Daniela R. Urbina, University of Southern California
Margaret Frye, University of Michigan
Sara Lopus, California Polytechnic State University, San Luis Obispo