Thursday, February 04, 2010

Educational assortative mating

The choices that men and women make in who to marry have implications that go well beyond the social fabric of a society. Income mobility, for instance, is an important consideration. If highly educated women always marry highly educated men, and the least educated intermarry, then little income mobility may take place.

The 2001 Armenian population census provides an opportunity to gauge the extent of educational assortative mating in Armenia. Do Armenian women marry men of similar education, do they marry down, or do they marry up? These questions can be explored by comparing the educational level of husbands and wives.

The top panel of the table below summarizes the choices made by women as gleaned from a 10 percent sub-sample of the 2001 census data (Armstat). About 60 percent married men with similar educational attainment (sum of main diagonal), 20 percent married down and 20 percent married up. About 8.2 percent of women had education below the fourth grade (primary or less). Close to 18 percent had completed tertiary education (university), but only 15 percent married university graduates.

The middle panel reports similar statistics but restricted to women born in the 1920s. About 62 percent of women married men with similar education, 23 percent married up, and 15 percent married down. About one third have an education below the fourth grade level, and around 10 percent are university graduates; 15 percent of male partners are university graduates.

The bottom panel replicates the analysis for those born in the 1960s. Here 60 percent of women married men of similar education, 19 percent married up, and 21 percent married down; more marry men with lower education when contrasted with the older generation. About one percent of the women have an education at the third grade level or below in contrast to one third of those born in the 1920s. This highlights the spectacular educational expansion from the formative years of Soviet Armenia.

This is part of a paper co-authored with Shushan which contains further description of the data and additional analyses that address the choices made in the 1930s, 1940s, and 1950s, as well as estimation of the probability of the various choices made (ordered probit). One finding of interest is that women who marry younger men are more likely to have married down. Drop a note if interested.


Hans Gutbrod said...

very interesting. So is this published already? It would be interesting to compare different cohorts here, too.

A first glance at some of our data (incomplete) indicated that people with incomplete tertiary education in the 1990s sometimes were higher achievers, as quality of provision went down.

Great topic, good luck in exploring it further.


David said...

Thanks Hans.
Has not been submitted for publication yet. Initially started with household survey and then moved to the census data when it became available; the census data is so large that one is able to address the various cohorts.
We did look for Georgia census data (and other countries in the region) to provide some comparisons but without success (not on IPUMS). It will be good to see what you get for Georgia. On a positive note, the household survey provides similar results for the population, but obviously cannot address the various cohorts given the sample size.
As for achievements by education in Georgia, I would like to hear more. Any chance of posting it on the CRRC blog?