Improving Food Frequency Questionnaires: A Qualitative Approach Using Cognitive Interviewing
Section snippets
General Methods
In each of two iterative phases, conducted 2 months apart, 24 different participants between 50 and 70 years of age were recruited by a community newspaper advertisement and word of mouth to participate in a 2-hour interview. Each participant was paid $50. We attempted to recruit equal numbers of men and women, but made no attempt to stratify by or collect data on other demographic characteristics. For phase 1, 11 men (mean age=61.3 years; range=51 to 73 years) and 13 women (mean age=63.6
Phase 1: Design
Questionnaires developed for phase 1 contained an abbreviated 43-item food frequency questionnaire based on the HHHQ (1), (33). Eight variants of the questionnaire were designed to test different approaches, which we describe later. Each respondent was asked to complete three of these questionnaires in the 2-hour period.
Portion Size
In all cases, people tended to skip portion size questions after completing frequency. Of the three approaches to portion size (Fig. 1), the one providing ranges of portion size within the columns for small, medium, and large appeared easiest for respondents to understand and complete. The meaning of the word, “helpings,” to describe portion size was unclear to respondents.
In the standard HHHQ format, the reference medium portion size is placed to the left of the “S,” “M,” and “L” columns (for
Phase 2: Design
The intent of phase 2 was to develop a questionnaire incorporating the most successful approaches from phase 1 and to continue experimenting with format and wording. We developed two questionnaires, each with a complete list of foods (100 items based primarily on the HHHQ questionnaire) intended to capture intake of most nutrients and commonly consumed foods in the United States from 1976 to 1980. Each respondent was asked to complete one questionnaire in the cognitive interview.
Findings from
Portion Size
Portion sizes in ranges appeared to be preferable to approaches that used a reference medium portion size. Although wordier, this method allowed respondents to select a portion size disassociated with the terminology “small,” “medium,” or “large.” The reference medium portion size approach with the portion size in the “M” column was also satisfactory; respondents were able to choose a portion size and did not confuse the reference with “S” as in the standard HHHQ approach. As in phase 1, many
Discussion
Our findings suggest that, among the many tasks required to develop a food frequency instrument, focusing on the respondent's perceptions of the meaning of the questions is fundamental to good questionnaire design. Listening to and then probing respondents as they verbalize their thought processes while completing a food frequency-type questionnaire is a sobering experience. We were forced to reconsider fundamental issues in questionnaire design to obtain the best data from the most people.
Applications
Researchers who design new dietary instruments should conduct cognitive interviewing to pinpoint problems encountered by respondents. Our findings suggest approaches that we believe may facilitate the collection of more useful dietary data from older men and women. We did not, however, examine validity of the portion size or frequency estimates. Thus, we plan to conduct a validation study of the final version of this cognitively designed food frequency questionnaire using multiple days of food
References (35)
- et al.
Reproducibility and validity of food intake measurements from a semiquantitative food frequency questionnaire
J Am DietAssoc.
(1993) - et al.
Assessing the nation's dietlimitations of the food frequency questionnaire
J Am Diet Assoc.
(1992) Statistical issues related to semiquantitative food-frequency questionnaires
Am J Clin Nutr.
(1994)Gaps in dietary assessment methodologymeal- vs list-based methods
Am J Clin Nutr.
(1994)- et al.
A cognitive model of children's reporting of food intake
Am J Clin Nutr.
(1994) - et al.
Validation of a self-administered diet history questionnaire using multiple diet records
J Clin Epidemiol.
(1990) - et al.
Comparison of two diet questionnaires validated against multiple dietary records collected during a 1-year period
J Am Diet Assoc.
(1992) - et al.
Comparison of a food frequency questionnaire using reported vs standard portion sizes for classifying individuals according to nutrient intake
J Am Diet Assoc.
(1991) Selection and quantification of typical food portions by young adults
J Am Diet Assoc.
(1984)- et al.
A data-based approach to diet questionnaire design and testing
Am J Epidemiol
(1986)