Credible Interval for Repeated Measures ANOVA

rohanp16

Hello,

I have conducted a 3 way Bayesian RM ANOVA. Now, in my analysis, I have the “Model Averaged Posterior Summary”. This is similar to this table from the ANOVA tutorial paper by van den Bergh et al. 2019,

Now, it says in the text that I can get the posterior means by adding/subtracting from the intercept to get the condition specific means. Thus, the posterior mean for O,Yes would be 1.724 + 0.265 = 1.989. Great.

Now, I would like an uncertainty measure (credible interval) as well. However, footnote 10 in the paper states that “This calculation is valid only for the posterior means, not for the other posterior summaries”. Thus, it seems that [1.680, 2.295] is not the appropriate 95% CI for the O,Yes condition.

In light of this, besides just a point estimate, how do I get a measure of uncertainty?

EJ

I'll forward your question to Don!

E.J.

rohanp16

Cheers EJ, appreciate it.

I have been in touch with Don outside of the forum and should have posted the result of the communication for completion sake. He suggested to open a GitHub feature request, which I have below,

https://github.com/jasp-stats/jasp-issues/issues/669

vandenman

Hi Rohan,

See GitHub for an example of how to get uncertainty on the level of the data (the code is too long for the forum)

Cheers,

Don

rohanp16

Thanks Don, appreciate it :)