Abstract
In this study, the measurement properties of an interview-administered fish consumption frequency questionnaire, used with a pilot study of 20 Vietnamese immigrant women, were described. Reproducibility across two summer interviews and one winter interview for estimates of seasonal and yearly intake of Great Lakes fish was moderate (intraclass correlation coefficients: .51–.61). Detailed questioning, by species, resulted in higher estimates of mean overall consumption (44.6–57.8 meals/y) than did asking about any fresh-water fish consumed (33.5–46.1; differences 5.1–15.7). Estimates based on the fish consumption frequency questionnaire (i.e., 6.2 ± 2.0 meals per winter season) were comparable with those based on extrapolation from a 1-mo calendar (5.8 ± 5.6); however, both estimates of consumption were far less than a weighed record (29.1 ± 22.2). The results of this study suggest that measurement variation in fish consumption estimates should be detailed in research reports and should be discussed with respect to risk assessments.