Validity of a food frequency questionnaire for the determination of individual food intake

Abstract

The validity of a food frequency questionnaire for dietary assessment was tested with 31 college students living and dining in a dormitory. A food frequency questionnaire was mailed to participants requesting them to estimate their usual intake of specific food items. Actual food consumption was determined at each meal with self-report forms. These forms required participants to indicate foods chosen at that meal. To test the validity of the frequency questionnaire, regression equations were calculated for each individual using the frequency questionnaire data as the independent variable (X) and actual intake data as the dependent variable (Y). Eighty-four percent of the r values were greater than 0.50 and 55% were greater than 0.71. All were significant at the 0.002 level or below. As an alternative test of validity, the percentage of over- and underestimation of each food group was also assessed. Foods which are often major components of a meal (eg, flesh foods) were estimated with greater accuracy than those foods which could be considered accessory (eg, nuts or seeds). While it appears that a large percentage of individuals could accurately estimate their intake with this technique, some individuals were unsuccessful in doing so. Further research is needed to identify individual characteristics which may predict success with this technique.

Validation of a short food frequency questionnaire to assess consumption of cereal foods, fruit and vegetables in Chinese Singaporeans

“…More detailed FFQs have generally been found to over-estimate mean intakes for both nutrients and foods (Jain et al, 1980;Mullen et al, 1984;Pietinen et al, 1988;Salvini et al, 1989;Hankin et al, 1991). Krebs-Smith et al (1995) found that estimates of total fruit and vegetables were positively associated with the number of food items included in food frequency questionnaires.…”

“…Most food frequency questionnaires (FFQs) have been developed and validated as measures of nutrient intake (Jain et al, 1980;Willet et al, 1985;Block et al, 1986;Willett et al, 1987;Pietinen et al, 1988;Hankin et al, 1991;Block et al, 1992;Eck et al, 1996). Several studies (Jain et al, 1980;Mullen et al, 1984;Pietinen et al, 1988;Salvini et al, 1989;Hankin et al, 1991;Serdula et al, 1993) have evaluated FFQs at the foodbased level, but with the exception of one (Serdula et al, 1993), the FFQs were relatively long (55±278 items). Serdula et al (1993) evaluated a six-item fruit and vegetable questionnaire for telephone administration in the US Behavioural Risk Factor Surveillance System, and generally found the mean number of daily fruit and vegetable servings to be similar to those estimated by more detailed methods.…”

“…The authors concluded that the short FFQ is likely to be useful for surveillance purposes. However, Serdula et al (1993) and others investigating validity at the food group level (Jain et al, 1980;Mullen et al, 1984;Pietinen et al, 1988;Hankin et al, 1991) did not evaluate the FFQs in terms of their ability to identify low and high consumers, nor did these studies adequately assess agreement between methods at an individual level. Correlation coef®cients in the range of 0.23±0.91 have been interpreted as indicating moderate agreement.…”

History of food acceptance research in the US Army

“…The comparison of dietary recalls, records, and food frequency lists is one of the thorniest issues in the field of food habits measurement. Herbert Meiselman ran this study for Natick, working with Howard Schutz and Nora Krantzler of UC Davis and Nathan Macoby of Stanford University (Krantzler et al, 1982;Mullen, Krantzler, Grivetti, Schutz, & Meiselman, 1984).…”

Using cross-check questions to address the problem of mis-reporting of specific food groups on Food Frequency Questionnaires

“…Studies that have included an attempt to validate the FFQ suggest that fruits and vegetables are more likely to be over-reported than other food groups, (Mullen et al, 1984;Feskanich et al, 1993) although such results may not be generalizable because of the variation in design of the FFQs used and the different populations studied. However, if such a reporting bias exists it could dilute any association between low fruit and vegetable consumption and high incidence of cancer.…”