Factors associated with the reproducibility of specific food items from the southwest food frequency questionnaire

Little is known about the effect on reproducibility of including ethnically unique foods in a food frequency questionnaire. The Southwest Food Frequency Questionnaire (SWFFQ) has a food list that includes foods that are unique to Hispanic populations. 75 adults self-administered SWFFQs twice. Correlation of responses between administrations was calculated for each food item. Several variables were examined as potential predictors of reproducibility. Linear regression models were constructed to relate predictor variables to reproducibility. No significant differences in reproducibility between the gender or ethnic groups or between food types were detected. The regression models constructed included percentage of participants who ate a food and mean frequency with which the food was consumed as significant positive predictors of reproducibility. It was concluded that inclusion of less frequently eaten food items may lead to lower reproducibility. The results also suggested that subject burnout is not an issue with FFQs of moderate length.