Dear Wuzzy The answer is yes they can.
Consider the case where x1 is highly correlated to x2 and Y is correlated with X1-X2. I on a simulation of this for 100 cases got a R-squared for the regression model of .998 but the individual correlations x1 vs y = -.072 p=.476 x2 vs y=-.105 p=.301 x1 vs x2=.999 p<.001 This was a simulation and this situation is a minority case in my experience but the answer is yes they can. However this case is of course a nonsense if you have perfectly correlated cases as you would have if only the units had been changed. Jean M. Russell To: [EMAIL PROTECTED] Date sent: 5 Feb 2002 18:15:00 -0800 From: [EMAIL PROTECTED] (Wuzzy) Organization: http://groups.google.com/ Subject: Re: can multicollinearity force a correlation? > In my own defense: > > I was asking a simple question: > > will highly correlated cause an irregularly high R^2. > > My answer to my own question is "no" it can't.. > No-one here was able to give me this answer and I believe it is > correct: if your sample is large enough,(as mine is) then "no", > multicollinearity cannot affect your R^2, it will only affect the > coefficients and their signs and errors. > > > ================================================================= > Instructions for joining and leaving this list, remarks about the > problem of INAPPROPRIATE MESSAGES, and archives are available at > http://jse.stat.ncsu.edu/ > ================================================================= ------------------------------------------------------ Jean M. Russell M.A. M.Sc. [EMAIL PROTECTED] Corporate Information & Computing Services, University of Sheffield 285 Glossop Road Sheffield S10 2HB United Kingdom Phone: 0114-222-3098 Fax : 0114-222-3040 ================================================================= Instructions for joining and leaving this list, remarks about the problem of INAPPROPRIATE MESSAGES, and archives are available at http://jse.stat.ncsu.edu/ =================================================================
