Hi there,
I am doing a meta analysis about one therapy effect on patients'brain. So I found all researches which include before and after treatment neuroimging. However, when I countered an important issue when I abstract data from each paper, which is some papers just used comparison of before and after only on patients' brain with t value, while other papers adopted an two way ANOVA (both patients and controls were included) and reported an interaction of group by time with F value (the post hoc based on the regions in interaction results maybe also reported, maybe not).
I just wondering that whether I can pool these two type papers together or should just abandon one? I was bewilder because these two tape of research focus on same question but applied two different statistic methods.
if I should include them all, how to input t value when they reported F value for interaction results from two way ANOVA?

Many thanks
Hongru

Dear Dr Zhu

This is an interesting question.

On the one hand, I think that these papers report different things: some report the changes observed with treatment, while the other report the differences between the changes observed with treatment and the changes observed in controls. To combine them, you should assume that the the changes in controls are negligible.

On the other hand, SDM does not accept F-values. That said, you could consider converting them to approximate t-values with the appropriate sign, e.g., as if the papers had conducted a t-test between patients and controls in which the images were the differences between post and pre.

With the best wishes,

Joaquim

Dear Hongru 
Hello, I'm facing the same problem as you. I wonder whether your type of the neuroimaging is fmri . And after I  clicked in the Preprocessing button, I don't konw which kind of correlation template to choose. When you completed the SDM table editor, what kind of data did you input?

Looking forward to your reply
Lageny