"MY WIFE KNOWS BEST" A COMPARISON OF EVENT DATING ACCURACY BETWEEN THE WIFE, THE HUSBAND, THE COUPLE, AND THE BELGIUM POPULATION REGISTER

The problem of memory bias in retrospective data has been studiedfor short-term retrospective recall and for frequently occurringevents, but little has been written on how questionnaire designcan improve memory of personal events over a very long retrospectivetime period. This study offers a new approach to the problemby using a sample survey conducted in Belgium that allows thecomparison of recall accuracy not only between the husband andthe wife but also the couple together through the use of NationalPopulation Register data. The sample data support findings fromcertain psychological experiments that respondents' errors inevent dating are generally in years, and not in months, andfurthermore that this pattern is not general and is relatedto the event itself. Subsequently, the year of the birth ofa child, which has an annual significance, is generally recalledaccurately; the year of a change of residence, however, is oftenerroneous. Results of the log-linear models show that the oddsof recalling an incorrect year of a move are significantly decreasedif an event with annual significance occurred around the sametime period. Design of retrospective questionnaires should takeinto account the fact that time units are not recalled equallyfor differing autobiographical events, and that dating accuracycould be improved through use of parallel event questioning.Correlations between dating errors and omissions, and the possibilityof a "chain of errors" in responses, are also examined.