Estimating the accuracy of dietary questionnaire assessments: validation in terms of structural equation models

The validity and precision of questionnaire assessments of the habitual intake of individuals are usually evaluated by comparison with reference measurements that are supposed to provide a best possible substitute for the individuals' true intake values. In the present paper, a measurement error model is presented, defining different types of error--random or systematic, and within or between individuals--that may occur in dietary intake measurements. It is then discussed how simple latent variable models (structural equation models) can be used to estimate the average magnitude of these various types of error. So far, approaches described for the analysis of dietary validity studies have all been based on the assumption that the random errors of repeat reference measurements, taken by the same method on different occasions, are uncorrelated, so that the average of a sufficiently large number of repeat reference measurements will provide an accurate ranking of individuals by true intake level. In the present paper it is described how, by additional comparison with a third type of measurement such as a biochemical marker, the validity of dietary questionnaire measurements can be evaluated even in situations where the random errors of repeat reference measurements can no longer be assumed to be independent.