Fig. 2From: Negative log-binomial model with optimal robust variance to estimate the prevalence ratio, in cross-sectional population studiesComparison of the standard errors of estimation of PR adjusted for numerical age, between models with robust varianceBack to article page