What we measure when we talk about homogamy

In 1964, Alain Girard published Le Choix du conjoint, based on a survey of married people by the Institut national d'études démographiques, France's national demographic research institute. In it he documents marked social homogamy, that is, the tendency to form unions within one's own social milieu.

Michel Bozon and François Héran took up the question on the basis of a new survey by the institute, and in 1987 and 1988 published in the journal Population two articles brought together under the title La découverte du conjoint. In them they describe the places where couples meet and how these are distributed across social milieus.

Measuring homogamy requires a classification: occupation, qualifications, social background, age, religion. The result depends on how the categories are drawn. The broader the categories, the more alike couples appear; the finer they are, the rarer resemblance seems.

A point of comparison is then needed. Noting that many couples bring together two people from the same milieu says nothing in itself: if a milieu is very widespread, its members will often meet through sheer force of numbers. Homogamy is therefore measured as a deviation from what random matching would have produced.

Sociologists thus distinguish observed homogamy, the raw share of similar couples, from relative homogamy, which neutralises structural effects. The two can move in opposite directions, for example when the composition of the population changes without preferences changing.

The lengthening of education provides a textbook case. When the proportion of graduates rises among men and women alike, the share of couples in which both are graduates mechanically increases. Only a relative measure can tell whether preferences have shifted too.

The same figure admits two readings. It may reflect a preference for people like oneself, or simply the fact that one mostly frequents places where one meets people like oneself: school, work, the leisure activities specific to each milieu. The work of Bozon and Héran gives a large place to this second mechanism.

The two explanations are not mutually exclusive, and survey data can hardly decide between them. What is observed is couples that have formed, not the encounters that led nowhere or those that never took place, which leaves preference beyond the reach of direct observation.

Dating services have opened up a new source, since they keep a record of interactions that come to nothing. Marie Bergström drew, among other material, on data from these sites in Les Nouvelles Lois de l'amour, published in 2019.

These data have their own limits: they describe a particular setting and people who have chosen it. Above all, they are a reminder that homogamy is an outcome, produced by a chain of steps, and not a disposition that each person carries within them.