Validity of the 24-hr. recall. Analysis of data obtained from elderly subjects.

Tests of the validity of the 24-hr. dietary recall were done by comparing actual with recalled intakes for eight nutrients and the MAR (mean adequacy ratio) for a sample of seventy-six subjects age sixty years or older. Validity was tested by using paired-t tests and regression analysis. In the paired-t test, no significant difference was found between the mean recalled and the mean actual intake of nutrients, with the exception of calories. Using regression analysis, results indicated that for three of the eight nutrients considered (calories, protein, and vitamin A), small intakes tend to be over-reported and large intakes under-reported (p less than .05). Thus, for these three nutrients, the recall seems to be statistically conservative for group comparisons; it would seldom, if ever, indicate a difference in intake where no difference exists. But, it could yield a false negative, i.e., an indication of no significant difference, when, in fact, a difference does exist. Clearly, more research is needed, both to replicate this study and to develop techniques with greater internal validity for comparing the dietary intakes of groups.