[R-meta] Interpreting the overall effect adjusted for publication bias, in the absence of publication bias

[Daniel Foster]

Hi all,

I am carrying out a multi-level meta-analysis in which I have conducted a FAT-PET to ascertain whether or not there is evidence of publication bias, as well as the overall effect accounting for publication bias.

The results, uncorrected for publication bias, indicate that that there is a significant association between the two variables I am looking at. The FAT was not significant, suggesting the absence of publication bias. However, the results from PET indicate that after accounting for publication bias, the relationship is no longer significant.

Given that I did not find evidence of publication bias, can I conclude that there was a significant effect (i.e., using the findings from the model uncorrected for publication bias)? Or should I emphasize the PET findings in my discussion? How is this commonly dealt with?

Thank you!

Daniel

	[[alternative HTML version deleted]]