The problem of memory in nutritional epidemiology research

A technical report reviews theories of memory, focusing primarily on applications to reports of food consumption and eating patterns in the distant past as evidenced in epidemiological studies using various retrospective methods to estimate consumption that occurred months, years, or decades earlier. Specific attention is given to 4 types of memory error: (1) failing to report foods actually eaten; (2) reporting foods that were never eaten; (3) inaccurate estimates of amounts and kinds of foods eaten; and (4) other factors (e.g., current diet) that influence recall of diet in the distant past. The implications of these findings for dietary recall studies are discussed.(wz).