It seems you can utilize a meta-regression from binary data, where differences in events vs. non-events stemming from treatment vs. control groups can be computed to an odds ratio (Thompson, Sharp 1999). But can you take aggregated continuous data on your covariate, calculate the covariate's effect size, and utilize this calculated effect size in a meta-regression when the covariate data is not reported in your manuscript of interest?
Thanks
Hello Nic,
If I understand your query, the answer is no. You'll not be able to estimate the impact of a covariate in the absence of all relevant summary statistics (means, SDs, correlations with DV by group and overall). If the F-ratio for the covariate is given, you can estimate the pooled within-groups correlation with the DV (for that data set).
Good luck with your work.
- Badges
- Science topic
- Similar topics
- Papers