Sources of Variability in the Reproducibility of Food Frequency Questionnaires

Abstract

The reproducibility of food frequency questionnaires varies widely. Since reports of past intake are known to be biased toward the present and the forces of supply and demand affect what people eat at a given point in time, the questionnaire may capture an atypical snapshot of consumption rather than the intended view of usual consumption.

The consumption of regularly consumed foods is the same throughout the year. The consumption of these foods is likely to be highly reproducible at another point in time. The consumption of seasonally consumed foods, however, fluctuates throughout the year, and may have peaks in winter or summer or particular holidays. There may be no common denominator among these foods necessary for the purpose of assessing reproducibility. Therefore, questionnaires that contain a combination of regularly and seasonally consumed foods, will be likely to have problems with reproducibility, the variance depending upon the number of seasonally consumed foods in the questionnaire.

This explanation for variability in reproducing food frequency questionnaires raises a question about the importance of assessing reproducibility as a way of evaluating the worth of questionnaires. Perhaps an improved method of collecting data for seasonal foods is what is really needed to improve the quality of data collected.