Examining parameter credibility across two experimental conditions

classic Classic list List threaded Threaded
1 message Options
Reply | Threaded
Open this post in threaded view
Report Content as Inappropriate

Examining parameter credibility across two experimental conditions


I'm new to Bayesian Data Analysis and teaching myself how this works, so please let me know if I have misspoken or need to offer more information.

I am examining the variability of predictor-criterion parameter estimates across two experimental conditions. I have prior data (meta-analytic) on these parameters. Here is what I would like to do:

1. Use the meta-analytic data to inform my prior. In this case, the observed correlation coefficients vary from .11 to .39 (95%CIs from .02 to .44), depending on the coefficient in question. How do I allocate credibility across these values prior to my analysis? How do I use this information to inform my prior?

2. Given that I am examining the same predictor-criterion parameters across two experimental conditions (in this case, it involves two separate survey designs), how would I test for practical equivalence in these parameter estimates? My first thought was that I needed to test for moderation, but I think I am mistaken.

I have John Kruschke's book in case someone has a reading recommendation for me.