Sunday, March 16, 2014

Assortative mating declining?

++Addition++Henry Harpending reminds us that there weren't many changes in educational affinities from 1940 through 2000. Presumably, steady state continues to be the story into the early 21st century.


A recent Pew Research report contained the following graph:

At first blush it might appear as though, contra Charles Murray, assortative mating is actually declining, as more bosses marry their secretaries and more high-powered women pair up with charismatic stay-at-home dads. While the pair rates among dual grads has increased, so has the percentage of people who go to college. Among those in their twenties today, about half will end up in either the "college" (bachelor's or more) or "some college" categories. The percentages of those who attain a high school diploma or less has dropped in the last 50 years as well, but not nearly at the magnitude suggested by the graph.

We need clarifying data on the frequencies of educationally unalike marriages to similarly compare these with the provided figures over time. Irritatingly, Pew doesn't provide as much and delving into primary census data is too daunting a requirement on this amateur's time. As is, the only thing to glean with certainty is that educational increases have occurred. Nothing novel there.

In 1960, 79.6% of married couples involved two people with the same levels of broadly defined educational attainment (high school or less, some college, college graduate). By 2012, this fraction had declined to 59.4%. Again, at face value it appears as though assortative mating is more a thing of the past than of the future. However, If we break a population into three groups and then pair members randomly, we'd expect the lowest frequency of same-group pairings if the groups were split 33.3%-33.3%-33.3% (intragroup pairings an expected 33% of the time). Conversely, we'd expect the highest frequency of same-group pairings if the groups were split 100%-0%-0% (100% intragroup pairings). The past was closer to the latter; the present closer to the former. Consequently, it's difficult to disentangle the genuine changes in (dis)assortative mating and when they're occurring from what is simply an artifice of the measurement approach.

Parenthetically, in checking to see if anyone else had asked Pew about the missing data, I noticed the report's comment thread. I happened to do so not long after listening to the Derb discuss his hope, a la 1984, in the commentariat. The title of the report from which the graph comes is "Record share of wives are more educated than their husbands". The comments include snark along the lines of women needing to be careful what they wish for because unhappiness and underperforming husbands go together; the subjects women major in are not as academically rigorous as the fields men study; after a generation of tilting the scales in favor of women, women now achieve higher average levels of educational attainment than men do; and the like. Not a single comment echoes what is contained in the canon of the Cathedral. Refreshing.


Steve Sailer said...

The three marriages of Kelly Johnson (1910-1990), the aeronautical engineer who founded Lockheed's Skunk Works, are interesting. His first wife died of cancer while he was in his 50s. Just before she died, she instructed him to marry his secretary. Then his second wife got terminally ill too, so she arranged for him to marry her best friend.

Johnson was a famous leader of men, but all this romantic stuff he delegated to his wife.

Anonymous said...

Just before she died, she instructed him to marry his secretary.


Maybe he was just a little disorganized and she thought the secretary would help him stay organized! Ditto for the 3rd wife!

Joe Schmoe said...

My guess is that there are two things going on here.

First, we are under-estimating the prevalence of assortative mating in the past. It is a bad idea to look at educational achievement as a proxy for social class / intelligence / socioeconomic status, because prior to @1960 or so, many women from upper class backgrounds did not attend college.

So while it is true that in the "Mad Men" era, executives often married their secretaries, it is also true that the secretaries tended to come from the same social class as the executives. For example, who did George H.W. Bush marry? Barbara Pierce Bush, the daughter of another wealthy WASP family (her dad was the CEO of a publishing company) who grew up in Rye, NY and attended exclusive private schools. Bush Sr. did not marry Rosie the Riviter, he married a girl from his own social class. I’m pretty sure that the same is true of most other men of Bush Sr.’s generation.

The social mobility of that era occurred when Rosie the Riveter married Michael McCafferty, a guy who grew up on the same block as Rosie in a working-class section of Queens, and he went to school on the GI Bill and became a middle manager for Westinghouse and moved up into the upper middle class.

When rich guys married the secretary back then, the secretary was someone from a similar family background.

Now, "family background" and "intelligence" might not be as strongly correlated as "college degree" and "intelligence," but it is just common sense to know that smart people seldom marry stupid spouses. It happens, but you just don't see it that often.

So I think that we tend to under-estimate the amount of assortative mating that went on in the past. We tend to think of the 1950's as an era when a good-looking secretary could marry a rich man, but in truth that seldom happened. Then, just as now, people tended to marry others like themselves.

Today, I think that we OVER-estimate the prevalence of assortative mating in the working classes. My impression is that for lower-middle class girls, it is relatively easy to get a "bachelor's" degree in medical office administration from Fourth Rate State College. For some reason, these days it's a lot harder for lower-middle-class guys to make it through college. So if you assume that people from similar backgrounds tend to marry one another, lower-middle women girls with BA's often have to marry cops, construction workers, enlisted soldiers, etc. It isn't really possible for all of those women to marry college graduates, there simply aren't enough of them to go around.

Anonymous said...

forgive me, but does it not mean that "high school or less" just are *not* getting married at all?

Jokah Macpherson said...

Anon 4:43 is right - this is how I interpret the data as well. To their credit, Pew at least mentioned this in the write-up, although I agree that some data on inter-educational marriages would be useful as well.

Audacious Epigone said...


When you hit a relatively slow time, you should put together a post relating your most interesting anecdotes to your readership, even if they are seemingly disparate and variegated. Unless you're saving the bulk of them for a book...


I'm not qualified to offer much in affirmation or refutation, but much thanks for the thought provoking.


The graph is only for those who are actually married, not the total population.

Which, when you think about it, should probably count as a bit more circumstantial evidence in favor of the possibility that assortative mating may actually have declined over the last fifty years or so. As a decreasing share of the adult population is currently married at a given time, wouldn't we expect more alignment among the minority of those who actually go through with marriage than we would've in the past when it was more commonplace, all other things being equal (which, of course, they're not)?

Audacious Epigone said...


Or even just a rehashing of all the ones you've dispensed over the years in various posts. Your instinct is probably to avoid doing something like that since you'll feel it to be overly self-indulgent as a consequence of the natural humility in your personality that bleeds through to what you write, but I'm sure I'm not alone in knowing I'd thoroughly enjoy it.

staffanspersonalityblog said...

It's odd that the category "both some college has half the percentage of the others. Is that people who never finish anything? They do some college but never graduate, have on and off relationships but never marry?

Audacious Epigone said...


It's defined here as having taken some college courses but not enough to have earned a bachelor's degree. So associate's degrees and people who probably originally intend to make it through at least four years but end up falling short.

Henry Harpending said...

This kind of data is hard to interpret because one ought to “remove” the effects of different and changing subpopulation sizes.

There is a discussion on our blog at

Doing this by “adjusting” the data to what it would be if every group were the same size, it turned out that there was almost no change at all between 1940 and 2000 in what one might call “affinity” for education levels.

Audacious Epigone said...

Professor Harpending,

Thanks. I can't usefully rake the table with the missing data points, though, can I? Might as well see if steady state has remained the norm over the last decade, too, if I were able to.

MM said...

De-lurking: Kudos for attempting to figure out what % of marriages are between college and non-college graduates. I've searched for that kind of detail myself, but no dice. Given that married college graduates have a divorce rate somewhere below 20% I believe, I was curious whether that changed if one spouse was say, only a HS graduate. Probably does, but by how much?

BTW, thanks a million for being one of the first online to use the GSS for analysis of social trends. I'd never heard of them before visiting your site, and have gone over there often just to run some correlations whenever the news media reports on marriage, divorce, etc.

Audacious Epigone said...


Thanks for the confirmation of what the acronym stands for.

In seriousness, it just takes a few minutes to get the hang of the GSS and it's openly accessible to the public. It's a shame that one has to come to search the blogosphere to find people who actually make use of such a comprehensive resource to try and make sense of the society and culture they live in, but alas.

MM said...

Sorry, can't seem to get Wordpress to display my Gravatar.

Anyway, whenever some journalist or politician makes some statement about what "most Americans believe", I pay the GSS a visit.

Helps keep my own confirmation biases in check. For example, I found that a majority of respondents who self-identified as upper class or have over $500,000 in wealth (net worth, I suppose) consider themselves politically moderate. And almost 30% consider themselves liberal or very liberal.

Not so surprising: Controlling for whether respondents smoke or not, just that one variable, makes a dramatic difference in number of lifetime sexual partners reported.