Reducing underreports of behaviors in retrospective surveys: the effects of three different strategies

Panel surveys routinely ask respondents retrospectively about events or facts that occurred since the last interview. Earlier research has documented that events occurring in between two interviews remain underreported. This paper studies the effect of three question formats for estimating annual use of a Family Physician (FP), i.e., 1-month retrospective questions, a 12-month retrospective question without a recall aid, and a 12-month retrospective question with a timeline recall aid. Our results show that the three question formats lead to very different estimates for annual FP use which were all lower than a population estimate based on register data. The timeline with landmarks did not significantly improve retrospective recall and reporting of Family Physician use for the one-year period.