in my meta-analysis I have studies with three kinds of data reports of RCTs which I want to integrate into one group with standardized effect sizes.
- studies reporting baseline and post-treatment values
- studies reporting raw change values (change score) and their standard deviation
- studies reporting adjusted change values from ANCOVA models
I understand that I can use the baseline standard deviation to standardize the effect size in case 2 (or use the change score standard deviation or transforming the change score standard deviation to post-treatment standard deviation).
My question: can I also use baseline standard deviations for standardization in cases 1 and 3 to make it consistent across studies? If not, why? Because the variance is reduced as opposed to post-treatment SDs there I overestimate my effect?
This question has been bugging me for a while, I appreciate any input. Thank you!