In 2006 an initiative was taken to re-design the Survey of Income and Program Participation (SIPP), a major longitudinal survey. This was done in order to reduce costs and respondent burden as well as improve data collection. As part of this re-design effort, the reference period was extended from four months to one year and an Event History Calendar (EHC) was introduced. A major concern in making this change was that the longer recall  period would result in fewer reports of spells of program participation and employment, especially during the early months of the recall period. Several studies are ongoing to evaluate the effectiveness of the re-design; this paper seeks to study the effect of the recall period on reported job duration and program participation. Careful evaluation of how well the data in surveys are collected is important for ensuring the accuracy of these empirical studies, and recall is an important component of collecting quality data.

As a result of testing the new survey instrument, a natural experiment has been created where households were randomly assigned to the current production version of the survey or the re-designed survey. Using two years of field test data and the current SIPP, we will exploit the experimental nature of this design to look at the effect of the recall period on the reported job duration and duration of program participation (in particular Food Stamp (SNAP) or WIC participation). This involves carefully mapping the data to a common standard. We can also compare different uses of the EHC for the two years of field testing. We will also examine differences by sex and marital status. We use a proportional hazards model to estimate whether the recall period influences the reported duration of jobs and program participation, and we find that there is little evidence of differences in recall over the two periods.