Interpretation of Validation Studies

Validation allows for the evaluation of dietary [glossary term:] measurement error. Findings from validation studies can be useful for comparing instruments, informing the choice of the most appropriate one for a given purpose, estimating the loss of [glossary term:] power and the necessary sample size for a study that will include an assessment of dietary [glossary term:] exposures or [glossary term:] outcomes using a given instrument, and providing information about the extent to which diet-outcome [glossary term:] associations based on data from a particular instrument are likely to be affected (e.g., [glossary term:] attenuated) by measurement error.

Often, validation coefficients in the range of 0.4-0.6 are presented as evidence that an instrument is valid. In reality, however, such findings should not be used to answer a "yes" or "no" question with respect to whether or not an instrument is "valid". Further, the study design and whether an unbiased or imperfect reference measure was used to evaluate the instrument should be borne in mind. In the case of imperfect references, validation coefficients may be inflated due to correlated errors in the two instruments (see Imperfect Reference Instruments). The identification of additional unbiased references is needed to allow more extensive evaluation of self-report dietary assessment instruments.