Comparing coefficients and DICs

Welcome to the forum for MLwiN users. Feel free to post your question about MLwiN software here. The Centre for Multilevel Modelling take no responsibility for the accuracy of these posts, we are unable to monitor them closely. Do go ahead and post your question and thank you in advance if you find the time to post any answers!

Remember to check out our extensive software FAQs which may answer your question: ... port-faqs/
Post Reply
Posts: 5
Joined: Mon May 03, 2010 3:18 pm

Comparing coefficients and DICs

Post by IngarB »


I have run 4 level models with binary response variable using mcmc.
I have added context variables - one at a time - for each level.

Some questions:
1. Is it straght forward to change the coefficients from odds to probability in such a model?

2. To be able to compare the strenght of the different explanatory context variables that are
introduced at the same level one at a time (in separate models): do I have to standardize the variables in advance?
And how do I do this in such a model? And what do comparsion of the DICs actually tell me?

I have tried to make a direct comparsion between the coefficents in the different models to be able to say which of them explain most. I did that by comparing the size of the coefficients (the ones who were significant) - but I tested it without standardising
and without turning them into probabilities. But I'm not sure if this is the right way to do it?
I have also tried to interpret the DICs (I'm using mcmc) for each model - again to try to say something about which
variable explains the most of the variation in the binary outcome variable (lower DIC = better fitted model and less complexity = explain most?). But I have a feeling that this it not right?

I hope someone have any ideas on how to do this, or mayby where I may find the answer.

All the best,
Post Reply