# How to run meta-analysis for simple linear regression coefficient

I have run several independent studies to look at the relationship between a continuous predictor variable X (ranging from 0 to 100) and a continuous outcome variable Y (ranging from 0 to 10). (essentially, several different Y = bX + e linear regressions)

Some of the studies found significant effects and some didn't

I am now trying to figure out what the average effect size (size and significance) is regarding the relationship between X and Y, but am having trouble figuring out how to do this

I have experience doing something similar using a categorical predictor variable and a continuous outcome variable using the Metafor package in R, but the current setup is something new to me (as you might be able to tell I'm not well versed in meta-analyses)

I've looked at the following posts: Link 1 Link 2 but I still wasn't sure how to do this

I'm not even sure what the effect size I'm trying to calculate is (f^2?).. Any tutorial websites or instruction websites that give example codes would also be greatly appreciated!

• I think Link 1 would be helpful (but since I wrote the answer I would say that). I would not use $r^2$ as an effect size as it is directionless. Commented Mar 22, 2021 at 14:38
• Thanks for the reply! Not sure why you said r^2, but maybe you misread f^2 from my post? :) Do you think using the correlation coefficient as the effect size would make sense? Commented Mar 23, 2021 at 4:12
• I have no idea what $f^2$ is I am afraid. Why do you not want to use the estimated regression coefficients? Commented Mar 23, 2021 at 9:39
• I saw this link and thought f^2 was the one to use: ncbi.nlm.nih.gov/pmc/articles/PMC3328081/…. I definitely want to use the estimated regression coefficients, if that's possible. It's just that I'm not sure how to do that. Ugh sorry for being such a noob at this. Commented Mar 24, 2021 at 5:39