Skip to main content
. Author manuscript; available in PMC: 2011 Jun 7.
Published in final edited form as: Am Stat. 2011 Feb 1;65(1):16–20. doi: 10.1198/tast.2011.10170

Figure 1.

Figure 1

Plots of estimated marginal posterior densities for the component mean in the 8–component univariate mixture, simulated data example. Plots for μ1:8 show results based on the raw MCMC samples, clearly and strongly evidencing the label switching issues via multimodal margins. Plots for μ̃1:8 show results under the relabeling strategy, resulting in unambiguous and accurate (since the ground-truth is known in this synthetic example) identification of posteriors.