Dear all,

As I extracted the data for conducting the meta-analysis, I found that some studies used one-tailed t-tests and others used two-tailed t-tests. I think that putting the results of studies using these two statistical methods in one meta-analysis would introduce bias. Because in this case the results of some studies have only one direction (experiment > control or experiment < control), while others have the both. This worries me considering that SDM does a meta-analysis of the results from both directions together.

Can I calculate the meta-analysis separately according to the direction (experiment > control or experiment < control)? Or is there another way to solve this problem？

Best wishes!

Zoey