Measuring household spending and payment habits: The role of typical and specific time frames in survey questions

We designed and fielded an experimental module in the American Life Panel (ALP) where we ask individuals to report the number of their purchases and the amount paid by debit cards, cash, credit cards, and personal checks. The design of the experiment features several stages of randomization. First, three different groups of sample participants are randomly assigned to an entry month (July, August, or September, 2011) and are to be interviewed four times during a year, once every quarter. Second, for each method of payment a sequence of questions elicits spending behavior during a day, week, month, and year. At the time of the first interview, this sequence is randomly assigned to refer to specific time spans or to typical time spans. In all subsequent interviews, a specific sequence becomes a typical sequence and vice versa. In this paper, we analyze the data from the first wave of the survey. We show that the type - specific or typical - and length of recall periods greatly influence household reporting behavior.