Dear experts,

I have three questions relating to BPM and would be really grateful for any input for any of the questions:
1. Please could you explain to me the benefits/costs of using model II regression over robust regression?  Or do you know where I could find an answer to that question which is accessible to someone with little mathematical understanding.
2. If I choose to use robust regression, does anyone have any recommendations about which method to choose?
3. In the regression contrast manager, when picking a T contrast, the mean comes up in the contrast window.  I have not been selecting this.  But what is it the mean of?

My apologies if these are naive questions.
Many thanks for your help,
with best wishes
jo

Originally posted by jo archer: I like this paper by Wilcox:
1.    Wilcox, R.R., How many discoveries have been lost by ignoring modern statistical methods? American Psychologist, 1998. 53(3): p. 300-314.
Our papers have the technical details of rBPM, BnPM, etc.. If our models are appropriate, then I would use them. To quote Tukey: "just which robust/resistant methods you use is not important—what is important is that you use some." Is this questions specific to rBPM? I am not exactly sure how you are setting up your design matrix.Perhaps you could ask again while providing a bit more detail.
Happy holidays,
bl

Thank you for your reply.  That is most helpful and has help me make my decision.

In answer to your question regarding the design matrix.  I am using BPM Extended and carrying out parametric regression. Task related fMRI is the dependent modality, then I have one imaging covariate (grey matter probability map - GM) and one non imaging covariate (age).  The design matrix gives three parameters [mean age GM].  I suspect there is rarely cause to include the mean when specifying the contrast, but am interested in clarifying what the mean is the mean of and if we would ever need to include it in the model (the mean is not included in the manual example for rBPM, p4).  I hope this makes things clearer, I would be most grateful for your input.

Thanks again for your quick and helpful response.
With best wishes,
jo

Hi Jo,
Thank you for your interest in our toolbox. The mean and other covariates thereis used to  show you the column of the design matrix. The mean column issimilar to the white column in the spm design matrix, which is all ones.Usually we are not interested in the mean which is the hyper plane of theregression. But we need to set a zero contrast to the mean column.

Thanks,
Xue