Hello all,

I wanted to know if anyone has any good resources that they would recommend for learning how to make design matrices/contrasts for GLMs such as those implemented in NBS. I've read through lots of posts on these forums and the FSL FEAT webpage for many specific examples . While this has been very helpful, and I get the gist of what to do in most cases, I still am not 100% confident on how to make my own for GLMs that are different than those mentioned within these two resources.

Thank you!

Paul

Hi Paul, 

I found Statistical Analysis with The General Linear Model (by Miller and Haden) a useful and practical resource for formulating GLMs.

Here is a link to the PDF of the booklet: 

https://www.google.com/url?sa=t&rct=j&q=...

Andrew,

Thank you for the suggestion, I will give this a read. Quick question - as I understand, there is more than one (non-trivial) way to encode a particular GLM using a design matrix/contrast. Is there anything specific to the way GLMs are implemented in the NBS software that restrict these parameters? Or should, in principle, any design matrix/contrast work for NBS as long as it is generally acceptable as defined by the booklet you shared, for example?

Best,

Paul

Hi Paul, 

it is difficult to respond without further context. So many different kinds of GLMs can be formulated. There is nothing peculiar to the way that they are formulated in NBS. 

Originally posted by Paul Thomas:

Andrew,

I understand. From browsing these forums and reading the FSL GLM documentation I got the feeling that there were some formatting features of the design/contrast specific to these programs. Thanks again for your help.

Paul