Model (1.8) has greater power than a standard Cox model since in WLW women contribute multiple time to event endpoints (one for each HPV type) that are incorporated into the overall effect estimate. Greater efficiency is also obtained in the type-specific analyses using WLW in model (1.9), relative to using separate Cox regression models for each type, since we reduce the number of covariate coefficients by assuming a common effect (i.e., γ) for nuisance variables. Further, model (1.9) allows testing of equality in exposure effects across HPV types.
