Fig. 3
From: Estimating risk ratio from any standard epidemiological design by doubling the cases
Estimated RR from expanded data logistic regression across 2000 simulations scenarios with different levels of prevalence and true RR values, compared to estimates from a naive logistic regression and the (true) log-binomial model. Estimates are presented for the full cohort data (panel A) and for matched case-control samples (panel B) with 1:1 (clear boxes) and 1:2 (shaded boxes) case:control ratio