What Are You Doing Now? Activity-Level Responses and Recall Failures in the American Time Use Survey

Abstract

Questions about people's pasts are common in many surveys, but memories are error prone. The current research focuses on recall failures in the American Time Use Survey (ATUS). The ATUS most commonly encourages respondents to report all of their activities of the previous day in a forward chronological fashion, from the beginning to the end of the day. Even with a short reference period, the ATUS is prone to recall errors. We explore these errors, taking into account the response process, respondent, and interviewer as possible contributors to a recall failure. Importantly, we posit that the chronological recall of events leads to earlier activities affecting recall of the current activity. Events are more easily recalled when they are more distinct (less frequent) or additional contextual information about the event is available. While research has focused on these characteristics of the target event, the previous event recalled may also provide distinctiveness and context. Results suggest that periods following a more frequent activity are likely to be followed by a failure, although this is modulated by the duration of the event. The presence of others and the location of the event also have significant effects. The elapsed time since the event is also important, with a higher chance of recall failure for more distant activities. Although results highlight the importance of the response level in understanding outcomes, respondent characteristics still matter, as those with apparently lower cognitive ability are more likely to have a failure. Interviewers also contribute to the variance of recall failures, with interview experience not having an apparent effect, while interviewers who make other types of errors, surprisingly, show lower likelihoods of recall failure. The results shed light on the relationship between memory and survey errors and have implications for future survey design.