Generalized Methods-of-Moments Estimation and Inference for the Assessment of Multiple Imperfect Measures of Physical Activity in Validation Studies

Music City Center 

Assessing diet and physical activity in free-living populations is prone to subject to substantial measurement error. Numerous statistical methods have been developed to adjust relative risk estimates for cancer and other chronic diseases to account for bias due to measurement error in long-term dietary and physical activity data. One widely used method for this adjustment is regression calibration, which involves estimating a de-attenuation factor. In this work, we develop semi-parametric generalized method of moments estimators for the de-attenuation factor and other quantities of interest, including the correlation of each surrogate measure with the unobserved truth and intra-class correlation coefficients characterizing the random within-person variation around each measurement. This method relies only on assumptions about the first two moments of the multivariate distribution of the measures. A robust variance is derived to enable asymptotic inference. The performance of the proposed method has been evaluated in extensive simulation studies and data analysis in the Harvard Women's Lifestyle Validation Study.

Measurement error

Regression Calibration