Effects of varying recall periods on reported food intakes

A 32-item food frequency questionnaire was administered with different time periods over which intake frequencies were to be recalled. In replicate Studies 1 and 2, 154 students were randomly allocated to six recall periods, ranging from the "the past 3 days" through to "the past year" and "usual intake". Perceived "oftenness" of intake was estimated on seven-point ratings. In Study 3, 78 students were randomly allocated to five short recall periods, e.g. last weekend, the past 5 days. In addition to rating "oftenness", they estimated their frequencies of intake numerically. In the first two studies, recall period was related to "oftenness" ratings in three ways. Responses for some foods (mainly snacks) were unaffected by recall period; others (mainly staples) were monotonically related to the requested duration of the recall; and, for a third set of foods, periods greater than one month (including "usual intake") yielded equivalent ratings but these were greater frequencies than those associated with shorter recall periods. In Study 3, marked differences were seen between frequency ratings over "average" durations and those over recent periods of similar duration. Correlations between numerical frequency estimates and "oftenness" ratings were high. The implications of the results for uses of food frequency methodology in nutritional epidemiology are discussed.