Posted on April 13, 2015

Race, Assortative Mating and Inequality

Edward Rodrigue and Richard V. Reeves, Brookings, April 9, 2015

There is a growing trend in the United States towards assortative mating–a clunky phrase that refers to people’s tendency to choose spouses with similar educational attainment. Rising numbers of college-educated women play a key role in this change. It is much easier for college graduates to find and marry each other when there are more equal numbers of each gender within an educational bracket.


One implication of assortative mating is greater household income inequality, since education is a strong–and strengthening–predictor of earnings. Households with two college graduates multiply that earnings power by two and are doing much better than households with less-educated couples. Jeremy Greenwood of the University of Pennsylvania and colleagues estimate that assortative mating pushes up the Gini coefficient (a measure of income inequality) from 0.34 to 0.43. {snip}


Race is a factor in patterns of assortative mating. Black women face more difficult “marriage markets” than white women, given current rates of intermarriage according to work from University of Maryland sociologist Philip N. Cohen. Black women have the lowest rates of “marrying out” across race lines, in part because of racist attitudes to inter-marriage. Just 49 percent of college-educated black women marry a well-educated man (i.e., with at least some post-secondary education), compared to 84 percent of college-educated white women, according to an analysis of PSID data by Yale sociologist Vida Maralani.

In this Long Memo, we examine race gaps in marriage patterns in terms of educational sorting, using 5-year estimates from the 2008-2012 waves of the American Community Survey. We focus in particular on college graduates.


Black-white gaps in marriage rates reflect different levels of education by race, but there is an important gender gap too. Young white women–aged between 25 and 35–are the most likely to have at least a BA (37%), followed by white men (29%), black women (23%) and black men (16%), according to our analysis of the ACS. We focus on the 25 to 35 year-old age cohort because these are the years during which most women, particularly college graduates, enter into their first marriage.

The chance for a college graduate to marry another college graduate is likely to be greater if there are more marriages across race lines, since this will expand the pool of potential mates. This is especially true for those from minority racial groups. The good news is that there has been an increase in inter-racial marriage in recent decades, as Bill Frey’s new book “Diversity Explosion,” documents. {snip}

There is a gender gap here too: Frey reports that three-quarters of black-white marriages involve a black man, rather than a black woman.

Marriage rates are lower among black women compared to white women, even among those with a college education. The proportion of black college graduates aged 25 to 35 who have never married is 60 percent, compared to 38 percent for white college-educated women.

By definition, the black female college graduates who do not marry are not assortatively mating, since they are not mating–defined as marrying–at all. This helps to explain why white women with college degrees are more than twice as likely as their black counterparts (29% v 13%) to be married to someone of equal or greater educational status.

Race gaps in assortative mating

If we narrow our focus to those college graduates who do marry, the race gap remains clear: compared to whites, black college graduates are much more likely to have “married down,” in terms of education. White, married college graduates are slightly more likely (11% v 8%) to have a better-educated husband (i.e., with post-graduate qualfications). But the real race gap appears between those with equally-educated and less-educated husbands. Married, black college graduates are much more likely to have a husband with a lower level of education, compared to whites of a similar background (58% v 48%):