Filed under measurement

What a time-use diary records, and what falls outside it

Diary surveys describe when activity happens rather than how much of it there is. That design choice explains both their value and their blind spots.

Most labour statistics count hours. Time-use surveys do something narrower and more unusual: they ask a respondent to reconstruct a single day in sequence, from one point to the next, and record what was happening at each moment. The result is a record of timing rather than a total, which makes this family of surveys the main source of evidence on when work occurs rather than how much of it there is.

How the American Time Use Survey works

The American Time Use Survey has been conducted by the Bureau of Labor Statistics since 2003. Its sample is drawn from households completing their final months in the Current Population Survey, which gives it a probability basis and links each diary to a fuller record of employment and household composition. One person per household is interviewed about one designated day. The interviewer walks through the previous twenty-four hours in order, recording each activity, its duration, where it happened and who else was present.

The designated-day design is deliberate. Assigning days across the week and the year, rather than letting respondents choose, prevents the sample from filling up with memorable or convenient days. It is also what allows population estimates of activity by hour of the day.

The single-day limit

One diary describes one day. Aggregated across thousands of respondents, that produces sound population averages. It produces very little about variability within a person. A worker whose hours differ sharply from week to week and a worker with a fixed schedule can contribute identical diaries if their designated days happen to look alike. Questions about instability — how much a schedule moves, how often it changes — sit largely outside what a single-day instrument can answer, and are usually addressed with separate panel or survey modules.

Sample size compounds this. Uncommon schedule patterns occupy small cells, and estimates for narrow subgroups carry wide intervals. Published tables often suppress them for that reason.

Simultaneous and ambiguous activity

Diaries ask for a primary activity. Work that happens in short bursts alongside something else — a message answered during a domestic task, an hour of availability that is never called on — is recorded inconsistently or not at all. Secondary-activity fields exist in some instruments but are known to be under-reported, a limitation documented in the methodological literature on time-use measurement.

Classification is a further source of drift. Paid work done at home outside normal hours may be recorded as work, as a household activity, or as leisure, depending on how a respondent describes it and how a coder interprets the description.

Comparing across countries

Cross-national comparison relies on harmonisation. The Multinational Time Use Study, developed over many years under Jonathan Gershuny and colleagues, maps national surveys onto a common set of categories so that diaries collected under different designs can be compared. Harmonisation necessarily loses detail, and categories that exist in one national instrument may be folded into a broader class in the harmonised file. Comparative estimates should be read with that compression in mind.

What the design is good for

None of this makes diary data weak. It makes it specific. For describing the distribution of activity across the hours of a day and across the days of a week, at population level, it remains the strongest instrument available, and it is the reason evening, overnight and weekend work can be described at all. For describing how one person's schedule moves over time, a different instrument is required, and treating diary estimates as though they answered that question is a common misreading.