Who Ends Up With Whom

People pair up with people like themselves. This is called assortative mating, it appears in every society anybody has measured, and the interesting question is never whether it happens. It is which trait is doing the sorting.

For most of history the answer was mostly geography and class, which were the same answer. You married within the parish, within the trade, within the range your family could reach. The sorting was done by physical proximity and by who was introduced to whom, and it was extremely effective without anybody consciously filtering for anything.

In rich countries now, the strongest sorting variable is education.

Graduates marry graduates. People without degrees marry people without degrees. The correlation has strengthened over the period in which higher education expanded, which is not what you might naively expect: more graduates means more opportunity to marry across the line, and the sorting tightened instead.

Here is why that matters outside the couple.

Household income is the sum of two people's earnings. If those two earnings are uncorrelated, households sit in a relatively narrow band around the average of two draws. If the earnings are strongly correlated, households spread out, because the high earner is paired with another high earner and the low earner with another low earner.

So two graduates in one household and two non-graduates in another produce a gap between those households that is wider than the gap between any two of the individuals. Educational sorting takes individual wage inequality and amplifies it at the household level, which is the level at which people actually live, buy houses and raise children.

This is one of very few places in this subject where private romantic choices aggregate into a measurable macroeconomic effect. Nobody involved intended it. Nobody could notice it from inside their own relationship, where the decision looks like falling for a person you met in a seminar.

Which brings us to a question the previous lessons make unavoidable, and which does not have a settled answer.

Does online dating increase or reduce sorting?

The case for reduction is straightforward. The pool is enormous and geographically wide. You can now meet people you would never have been introduced to, outside your workplace, your university, your social circle and your neighbourhood. The old sorting mechanism was proximity, and proximity has been largely dissolved.

The case for increase is just as straightforward. Sorting by a trait requires being able to see the trait, and a profile shows you several of them at once, in a filterable form. Age, education, income band, height, religion, politics, whether somebody wants children. A parish dance displayed almost none of that explicitly; an interface displays it as a field you can set a range on. Filtering is easier than noticing, and the platforms make it a single tap.

The honest position is that these two effects both exist, they pull in opposite directions, and which one dominates is contested in the literature. Anybody telling you confidently that apps have broadened who ends up with whom, or that they have narrowed it, is going further than the evidence does.

What is not contested is the direction of the consequence if sorting does tighten. It shows up in household inequality, a generation later, in the form of children who start from very different places, which is where this subject stops being about feelings.