multiple comparisons with multiple regressions?

#1
Hi there, I'm hoping that someone can answer my query or point me in the direction of a reliable source to find out more.

My question is - Is the issue of multiple comparisons relevant to a multiple regression? I've read one article suggesting a bonferoni correction, but a journal article I read stated the authors didn't use a correction as a regression analyses data simultaneously.

Any help would be gratefully received, thanks :)
 
#2
Incase this helps anyone with a similar question to mine - I've found a reference to a book which explains how to conduct an extension of fisher's protected t-test for a multiple regression (unfortunately I can't get hold of this book!). Cohen definately advises taking the issue of multiple comparisons seriously in MRC:

Cohen and Cohen (1983) Applied MRC analysis for the behavioral sciences
 
#3
It's very sad that i keep replying to my own posts ;)

But I'm now having a hard time finding out on the internet how to calculate Fisher's test for a multiple regression. Any ideas?
 
#4
Multiple Regression

As I understand multiple regression involves one dep variable and several (multi) indep variables. So we have:

y=a1x1+a2x2+a3x3+...+amxm

where:
a1,a2,...,am are the coefficients; and
z1,x2,...,xm are the 'm' indep variables of length, say, n.
Here we are assessing the contributions x1,x2,...,xm to y.

I am not sure what you mean by "multiple comparisons". Perhaps this is the reason
no responded to your post. You seem to assume everyone knows what the problem is.

More info please

bbkaran
www.global-i-tech.co.uk/9.html