Measuring sociability in minutes

Time-use surveys ask people to fill in a diary for a randomly drawn day, describing their activities in short intervals, most often of ten minutes. The respondent does not estimate a usual amount of time; they record what they do as the day goes on.

This format reduces the memory and social desirability biases that weigh on frequency questions. Insee, the French national statistics institute, has conducted such surveys since the mid-1960s, the United States has had an annual survey since 2003, and a harmonised European framework allows certain comparisons between countries.

Sociability first appears as an activity: conversations, visits, entertaining and calls appear in the classifications under a dedicated heading. Measured in this way, it takes up a modest share of the day, because most moments spent with others are coded according to their main activity.

The second route goes through the question about who was present, asked for each interval in most of these surveys. A meal, a walk or a sports session shared with friends then becomes identifiable, even if the main activity is nothing specifically social.

The diary records moments, not relationships. It tells how much time a person spent with friends, not with how many friends or with which ones. The category of friends is moreover often grouped with other people from outside the household, which limits precision.

Written exchanges scattered through the day are poorly captured by a division into intervals: brief and often simultaneous with another activity, they are rarely noted, or relegated to a secondary activity where the diary provides for one.

A randomly drawn day describes a population well and a person badly. An evening with friends once a week appears in the diary of about one respondent in seven, and not in those of the others, who nevertheless have the same habit. Averages are reliable, individual distributions misleading.

Some surveys ask for two diaries, one on a weekday and one on a weekend day, to mitigate this problem without eliminating it. Since sociability is concentrated at weekends, the choice of days bears directly on the estimates.

Diaries and frequency questionnaires do not measure the same thing and do not always give the same picture. The former capture time actually spent, the latter a representation of habits, which the respondent reconstructs at the moment of answering.

The gaps between the two sources therefore do not point to one right source and one wrong one. They indicate that sociability has several dimensions, duration, frequency, number of people, that no single instrument covers, and that long series based on one or the other tell partly different stories.