On Fri, Jun 28, 2013 at 1:37 PM, Irene Giannoylis <span dir="ltr">&lt;<a href="mailto:[log in to unmask]" target="_blank"><a href="/cgi-bin/wa-jisc.exe?LOGON=A3%3Dind1306%26L%3DSPM%26E%3Dquoted-printable%26P%3D5612328%26B%3D--e89a8f64700317ed9404e03ad514%26T%3Dtext%252Fhtml%3B%2520charset%3DISO-8859-1%26pending%3D" target="_parent" >[log in to unmask]</a></a>&gt;</span> wrote:<br><div class="gmail_quote"><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">
Thanks for the prompt reply.<br>
<br>
Does this have something to do with over modeling/fitting the data?<br></blockquote><div><br>&gt;&gt; Yes. You need some points that represent baseline to get a good estimate of the constant term in the model. <br></div>
<blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">
<br>
Also, to model noise I am using &#39;conn&#39; to extract PCA components of WM and CSF signal which I regress as covariates of no &#39;interest&#39; in spm.  To do this, task information is entered in &#39;conn&#39; at setup so that it is not included in the estimation of the PCA components.  Do we include the rest information here?<br>
</blockquote><div><br>&gt;&gt; I&#39;ve never used the toolbox for task data. Usually, WM and CSF regressors are not included as covariates in task analysis.<br> </div><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">

<span class="HOEnZb"><font color="#888888"><br>
Irene<br>
<br>
</font></span></blockquote></div><br>