O-93 - In OPEN the magnitude of dietary under-reporting increased with body mass index (person-specific systematic bias)
The load-bearing structural finding: under-reporting is not just large but person-specific and correlated with BMI, a determinant of many disease outcomes. This is the property that distinguishes OPEN’s error from the classical additive, non-differential model assumed by standard measurement-error correction (see the attached argument).
Link to original
Why this is evidence
Under-reporting rises with BMI - a person-specific component correlated with a determinant of many outcomes. H-31 reads this as violating the non-differential/classical measurement-error model that standard attenuation correction assumes, so the error is not simply correctable; H-41’s manageability claim rests on the error being non-differential. Discriminates.