Fig. 2From: Modeling perinatal mortality in twins via generalized additive mixed models: a comparison of estimation approachesTrue and estimated curves of the estimated nonparametric functions based on 1000 replications (\( {\hat{f}}_1\left({x}_1\right) \) and \( {\hat{f}}_2\left({x}_2\right) \) in the upper panel) and smoothed pointwise coverage probabilities of the 95% confidence intervals (f1 (x1) and f2 (x2) in the lower panel) for 1000 replicated datasets. These results are for the data generation scenario with m = 1000, ni = 2, \( {\sigma}_{int}^2 \) = 0.75, \( {\rho}_{x_1.{x}_2} \) = 0.7 and event probability = 0.05. The curves estimated by Bayesian-IG were almost similar to those obtained by Bayesian-HC and have not been displayed here to make other fits more visibleBack to article page