Follow@

Hi Qikai,

mixing path coefficients with correlations in a meta-analytical data base creates the same problems as mixing regression coefficients with correlation coefficients and both make no sense.

I would stick to correlations and look whether there is a correlation matrix in the respective paper. If these are correlations among latent variables I would reflect that in a dummy coded variable (i.a., "latent vs. obsered correlation) that can be used later as a moderator.

There is a different approach that calculates semi partial correlations or partial correlations (which I would prefer) from rsquare. From this perspective, a zero-order correlation could be perceived as a partial correlation with nothing being partialled. Hence, all coefficients belogn to the same class and thus can be meta-analysed. Regression coefficents and correlations in contrast are different things, see

Aloe, A. M. (2015). Inaccuracy of regression results in replacing bivariate correlations. Research Synthesis Methods, 6(1), 21-27.

Aloe, A. M., & Becker, B. J. (2012). An effect size for regression predictors in meta-analysis. Journal of Educational and Behavioral Statistics, 37(2), 278-297.

Hope that helps,

Holger

Hi, have a look on these courses on CMA:

https://www.youtube.com/watch?v=xvtuc3BE-GI&list=PLX-KmXTn6p4i_Ng5yf1d5-mWgtGsxQzjt&index=4