04-18-2012 08:09 AM
I tried to use the ttest procedure while using the WEIGHT option. In the output I got, there was only t-test for the 'equal variances' case.
When I tried to use the COCHRAN option to get the t-test for unequal variances I got a warning message: 'WARNING: The COCHRAN option is ignored when using WEIGHT statement'.
I should note that when I don't weight the cases I receive all of the relevant t-test: t-test for equal variances (pooled), t-test for unequal variances (Satterthwaite) and t-test for unequal variances (Cochran).
Is anyone familiar with this problem?
How can I weight my cases but still get the unequal variances t-test in the output (as some of the Levin tests indicate that I should reject the equal variances null assumption).
04-18-2012 10:43 PM
Did you try to use FREQ statement?
I think the best way is to talk with Technical Support .
From your log , It seems that you can get the t-test for unequal variances.