Hi,<div><br></div><div>There are probably multiple ways to do this, but here&#39;s what I recommend for your first design (I believe yours is rank deficient)</div><div><br></div><div>EV1:  1 for G1 Cond1, -1 for G1 Cond2, 0 for all G2</div>
<div>EV2:  1 for G2 Cond1, -1 for G2 Cond2, 0 for all G1</div><div>next 20 EV&#39;s are the subject-specific means (your S1-S20)</div><div><br></div><div>Then the contrast [1 -1 0....0] tests the group differences in the paired differences (i.e. your interaction effect).</div>
<div><br></div><div>For the second design, I&#39;m extrapolating from what was proposed in <a href="https://www.jiscmail.ac.uk/cgi-bin/webadmin?A2=ind1109&amp;L=FSL&amp;D=0&amp;1=FSL&amp;9=A&amp;J=on&amp;d=No+Match%3BMatch%3BMatches&amp;z=4&amp;P=74308">this FSL list email</a>, which apparently was summarizing from an earlier email you would have:</div>
<div><br></div><div><div>EV1:  1 for G1 Cond1, -1 for G1 Cond2, 0 for all G2</div><div>EV2:  1 for G2 Cond1, -1 for G2 Cond2, 0 for all G1</div><div>EV3:  avg estrogen for this subject in G1 Cond1, -1*avg estrogen for this subject G1 Cond2, 0 for G2</div>
<div>EV4:  avg estrogen for this subject in G2 Cond1, -1*avg estrogen for this subject G2 Cond2, 0 for G2</div><div>EV&#39;s 5&amp;6, similar to 3&amp;4, but for progesterone</div><div>next 20 EV&#39;s are the subject-specific means (your S1-S20)</div>
<div><br></div><div>This will control for whether the within-subject difference is a function of estrogen or progesterone levels.  </div><div><br></div><div>Hope that helps!</div><div><br></div><div>Jeanette</div><div><br>
</div><br><div class="gmail_quote">On Tue, Jul 24, 2012 at 2:22 PM, Klara Mareckova <span dir="ltr">&lt;<a href="mailto:[log in to unmask]" target="_blank"><a href="/cgi-bin/wa-jisc.exe?LOGON=A3%3Dind1207%26L%3DFSL%26E%3Dquoted-printable%26P%3D2660290%26B%3D--f46d042dfdc547b4a304c5992498%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:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">
Hi,<br>
<br>
I have a question about repeated measures anova and covariates design.<br>
I was reading the FSL manual and forum but since none of the examples<br>
seem to fit my situation, I was wondering whether some of you<br>
experienced FSLers could confirm whether I got it right/ what do I<br>
need to change.<br>
<br>
Design # 1: Repeated measures anova - 2 groups (medication, control),<br>
each participants scanned twice =  two different conditions (cond1,<br>
cond2). I want to have a look at the main effect of group, condition,<br>
and the group by condition interaction. I have 40 subjects all<br>
together; 10 in group1 cond1, 10 in group1 cond2, 10 in group2 cond1,<br>
10 in group2 cond2. I&#39;m combining the F test with the repeated measure<br>
design - can you please check list 1 of the attached spreadsheet and<br>
see whether the design # 1 is correct?<br>
<br>
Design # 2: Repeated measures anova with 2 continuous variables. The<br>
same dataset as above, but this time I want to look at the effect of<br>
two hormone levels - estrogen and progesterone and just correct for<br>
the fact that each participant was scanned twice (I believe that the<br>
group and condition are described by the two hormone levels). I&#39;m<br>
combining the covariate analysis with the repeated measure design -<br>
can you please check list 2 of the attached spreadsheet and see<br>
whether design # 2 is correct?<br>
<br>
<br>
Thanks a lot for your feedback and advice,<br>
<br>
Klara<br>
</blockquote></div><br></div>