Results of Bayesian mixed trajectory models of PCA1 with age, best models.

LCI and UCI are 95% Bayesian credibility intervals based on the posterior distribution. Betas are based on the posterior mode. Autocorr: minimum autocorrelation between iterations after thin.
*: This is a cross-sectional likelihood-based glm model, not a Bayesian glmm model, because each individual appears just one time and the above Bayesian model thus converges poorly, unable to estimate individual effects. The two models nonetheless give nearly identical parameter estimates.
Results of Bayesian mixed trajectory models of PCA1 with age, best models.