Bayesian multiple linear regression calculation

Klodin

Hello, 

I would like to calculate a Bayesian multiple linear regression to predict questionnaire scores based on attention deficiency symptoms variable (between 0-12), on dummy variable with values 0 and 1, and the interaction between the two (which stands for the moderation effect).

I wonder if it is possible to do it and if so how to do it? I would appreciate any help on this. 

Thanks a lot

--

EJ

Hello Klodin,

Our Bayesian multiple linear regression routine is theoretically not suited for this problem, because:

1. The error term is not normally distributed, as the questionnaire scores are bounded and discrete. Some kind of ordinal model seems more appropriate. BUT: in practice this may not matter too much.
2. The 0-1 dummy variable suggests this is a factor. In this case you should, ideally, consider an ANCOVA.

Cheers,

E.J.