[R] Meta-Analysis using RVE (robumeta)

Janina Steinert steinert.janina at gmail.com
Wed Nov 30 10:55:18 CET 2016


I'm runnig a meta-analysis, correcting for clustering of standard errors
with the robumetacommand. R gives the following warning:

If df < 4, do not trust the results

In some cases my df is in fact <4. What does this tell me then? Does it
simply mean that the number of included studies is likely too small (often
aroung 5-6, but with more than 10 single effect sizes)?

Is there anything else I could do for these cases?
Thank you

	[[alternative HTML version deleted]]

More information about the R-help mailing list