Figure 1.

A) Data generation scenario for single-time-interval initial simulation studies, in which exposure is conditional on error-free covariate L. Measurement error is represented by δ. B) Data generation scenario for 2-interval simulation studies, in which exposure depends on error-free covariate L_j. Measurement error is represented by δ. C) Data generation scenario for secondary 2-interval simulation studies, in which exposure depends upon mismeasured covariate $L_j{}^{\ast }$. Measurement error is represented by δ.