O-91 - In OPEN (n=484) the FFQ under-reported energy by ~31-38% and the 24-hour recall by ~12-20%, FFQ roughly twice as biased
The direct empirical measurement-error base rate for FFQ-derived energy in the OPEN validation sample. Because egg-intake frequency in the major cohorts is FFQ-derived, this quantifies the magnitude of self-report error underlying those exposure variables. The FFQ-vs-24HR gap shows the instrument, not just the respondents, drives much of the bias.
Link to original
Why this is evidence
H-31 reads the large, FFQ-specific (~2x the 24HR) energy under-reporting as evidence the FFQ exposure instrument itself is too degraded to trust; H-41 concedes the magnitude but treats it as non-differential and recoverable by biomarker calibration. Discriminates on how damning the error is.