Hi! Are there well-known comparability issues across US time use surveys in the 1990s. I’m finding that trend estimates display a clear break in the 1990s surveys and I wonder if there are additional adjustments I should be making to make those surveys comparable with the rest of the US time use series. Thank you!
The American Time Use Survey (ATUS) began in 2003; samples available before then in the AHTUS were collected for other studies with different survey designs. For this reason, there are some comparability issues between the historical surveys that are outlined in the AHTUS codebook. You might find useful the overview of study aims, target populations, and sample restrictions for all surveys (1965 - 2011) in the table on page 5. Starting on page 15, a more detailed description of features unique to each survey is given. After reading through it, a few of the comparability issues you will need to account for in your analysis include the following:
- Age restrictions (example: the 1965-66 survey restricts to persons age 19-65)
- Industry restrictions (example: the 1965-66 survey restricts to households with a person employed outside of Agriculture)
- Relationship restrictions (example: the 1991-2001 sample only covers parents whereas the 1998-99 survey is a national sample)
- Seasonality issues (ex. While the 1965-66 sample only covers 6 months of the year, the 1975-76 survey covers the whole year if all four waves are included)
- Caution should be used if combining the 1992-94 and 1994-95 surveys in which the person-level identifier 8011 was used in each survey; denoting both diarists completing a diary in October of 1992 and October of 1994. If the surveys are combined with a single variable, you will need to distinguish between the two diarists.
Very helpful, thank you! I’ll restrict the sample to make it comparable across surveys and hope this helps fixing the trend. If there are any other important differences across surveys (another one that I had considered was about the ratio of weekday/weekend diaries) let me know.