>
> Well a time diary is a form of self-reporting, but that is semantics.
> More to the point I doubt time diaries are filled out as the work is
> done. I filled out an arbitron diary once, and I did not track my
> listening as it happened, but reconstructed from memory they day
> before mailing it. Anecdotal, but I wonder if diaries really differ
> significantly from other forms of self-reporting.
The paper addresses that:
> No matter how extensive a set of codes is, each survey will have a
> different way of coding and aggregating what might seem like the
> same activity to an observer. Time diaries have the virtue of
> forcing respondents to provide a time allocation that adds to 24
> hours in a day. Also, unlike retrospective data about last week's
> or even last year's time spent working, while the time-diary
> information is necessarily based on recall, the recall period is
> only one day. The shorter recall period and the implicit time-
> budget constraint suggest that information on market work from time
> diaries is likely to be more reliable than the recall data on time
> use from standard household surveys; and, of course, time diaries
> provide information on non-market activities that is unavailable
> from labor-force surveys.
[WS:] The problem with diaries is not recollection - on that diaries score better than interviewer administered surveys that cover longer time spans - but willful misrepresentation. Respondents have more time and opportunity to "doctor" their responses to provide "desirable" answers. This was a known problem of diaries of TV watching - people often reported what they wanted to see rather than what they actually watched. Consequently, diaries were replaced with electronic devices recording the time, channel, and whether someone was in the room.
The same problem pertains to time use - people will probably skip questionable or illicit activities (like jerking off, watching porn, smoking dope, or even taking a nap) and report more socially desirable ones. And by doing diaries they have more opportunities to do than undetected i.e. provide logical responses that add to 24 hours.
My point is not to discount time use surveys, but to show that there are no easy fix for socially desirable response bias in self-reported surveys - regardless of the format or the protocol.
PS - We are currently working with the ILO to develop an instrument for measuring some non-market activities using labor force surveys, which are not only more economical but easier to create time series.
Wojtek