An interaction is modelled in the following way:<br><br>Y=Ax1+Bx2+Cx1x2<br><br>If you don&#39;t have x1 times x2 in the model, then you can&#39;t assess the interaction. If they are highly correlated, then it will be hard to assess whether the interaction is is due to the two variables or if the data would be fit better by a quadratic (e.g. Y=Ax1+Cx1^2).<br>
<br>If you have two models, you could compare either the slopes OR the correlation coefficient. This analysis would tell you if the correlation or slope of the relationship is different between the models. This is not an interaction though. For the later, you could convert the T-statistic to an Z-score and compare the Z-scores. Comparing the betas would be slightly more complicated because you&#39;d have to compute the error by hand based since the difference of the betas does not have an error term. If they are so highly correlated, as you suggest, then the correlation coefficients will not be different. The slopes could be different because the variance of the measures could be different.<br>
<br><br clear="all">Best Regards, Donald McLaren<br>=================<br>D.G. McLaren, Ph.D.<br>Postdoctoral Research Fellow, GRECC, Bedford VA<br>Research Fellow, Department of Neurology, Massachusetts General Hospital and <br>
Harvard Medical School<br>Website: <a href="http://www.martinos.org/%7Emclaren" target="_blank">http://www.martinos.org/~mclaren</a><br>Office: (773) 406-2464<br>=====================<br>This e-mail contains CONFIDENTIAL INFORMATION which may contain PROTECTED <br>
HEALTHCARE INFORMATION and may also be LEGALLY PRIVILEGED and which is <br>intended only for the use of the individual or entity named above. If the <br>reader of the e-mail is not the intended recipient or the employee or agent <br>
responsible for delivering it to the intended recipient, you are hereby <br>notified that you are in possession of confidential and privileged <br>information. Any unauthorized use, disclosure, copying or the taking of any <br>
action in reliance on the contents of this information is strictly <br>prohibited and may be unlawful. If you have received this e-mail <br>unintentionally, please immediately notify the sender via telephone at (773) <br>
406-2464 or email.<br><br>
<br><br><div class="gmail_quote">On Thu, Mar 1, 2012 at 11:33 AM, Rodrigo Perea <span dir="ltr">&lt;<a href="mailto:[log in to unmask]"><a href="/cgi-bin/wa-jisc.exe?LOGON=A3%3Dind1203%26L%3DSPM%26E%3Dquoted-printable%26P%3D220528%26B%3D--f46d041826bc8b2f2b04ba36e97f%26T%3Dtext%252Fhtml%3B%2520charset%3DISO-8859-1%26pending%3D" target="_parent" >[log in to unmask]</a></a>&gt;</span> wrote:<br><blockquote class="gmail_quote" style="margin:0pt 0pt 0pt 0.8ex;border-left:1px solid rgb(204,204,204);padding-left:1ex">
<div style="font-family:Tahoma,sans-serif;font-size:13px">
Hello,<br>I&#39;ve been asked about this question and I would like some guidance on how to proceed. Let say I have two models with a contrast on each (for example, the first model has age and the second one has weight). I would like to know how to proceed  to find the interactions between these contrasts in SPM.  And if I create another model and I include these two contrasts together, due to it high correlation I am afraid that the model will be incorrect. In other words I want to compare two contrasts in two different models or a way around this. <br>
Any help?<br>Thanks in advance,<br>Rodrigo<br><br><div style="clear:both">Rodrigo Dennis Perea<br>Graduate Research Assistant<br><a href="mailto:[log in to unmask]" target="_blank"><a href="/cgi-bin/wa-jisc.exe?LOGON=A3%3Dind1203%26L%3DSPM%26E%3Dquoted-printable%26P%3D220528%26B%3D--f46d041826bc8b2f2b04ba36e97f%26T%3Dtext%252Fhtml%3B%2520charset%3DISO-8859-1%26pending%3D" target="_parent" >[log in to unmask]</a></a><br>Bioengineering Program<br>
The University of Kansas</div><br></div>
</blockquote></div><br>