Time Use Estimates for Local Areas
The American Time Use Survey (ATUS) sample size is sufficient to generate estimates for each of the four Census regions, but the sample sizes are generally inadequate to support direct estimates for states and local areas at standard levels of statistical confidence. (The ATUS sample includes 13,038 respondents for 2005, 12,943 for 2006, and 12,246 for 2007, the three survey years used in developing the local estimates.)
For many business and public policy applications, it is useful to have estimates of time use for the populations of local areas. To meet this need, the Time Use Institute has devised a method for reweighting the ATUS sample observations to match the demographic profile of local areas.
The reweighting involves using basic records from the 2000 Census to develop a profile of the local population across 7 characteristics. The local adult (age 15+) population is allocated among 2,016 different categories, based on each individual’s combination of age (7 categories), sex (2 categories), employment status (3), family income (3), race (4), marital status (2), and parent status (2). Time use estimates for local areas are generated by applying the resulting weight variable for the local area to the unweighted responses of ATUS participants of the same demographic profile.
The seven demographic characteristics used to generate the local weight are among those most influencial in determining how people use time. However, there may be additional determinants of time use that are unique to a local ares and are attributable to the local economy, climate, or some other geographically based feature.
Statistical tests by the Time Use Institute indicate that over 90 percent of the variation in time use across adults that can be explained by either personal characteristics or location is attributable to personal characteristics. Therefore, differences from place to place in summary measures of time use are attributable mostly to differences in population composition. Thus, our estimates of time use for local areas are unlikely to be systematically biased.
Although the time use estimates for local areas are unbiased, it should be noted that, because of the reweighting of the national ATUS responses, it is not possible to generate standard errors and confidence intervals for the local area estimates. More detailed information on the local area time use estimation is available upon request from Jack Goodman, Director of the Time Use Institute, at email@example.com
More Information on the American Time Use Survey is available at www.bls.gov/tus