[R-meta] Can you please help me?

James Pustejovsky jepu@to @end|ng |rom gm@||@com
Sat Jun 29 18:40:48 CEST 2019


This is a really good question. Unfortunately, I don’t know if power analyses for more complicated scenarios have been worked out—and even if they have been, they would likely require making some pretty specific assumptions about the studies (like the number of ES per study and the correlation between ES estimates from the same study).

For “ballpark” power analysis, I think it would be pretty reasonable to use the existing resources for univariate meta-analysis. If you have multiple ES estimates that are correlated, you’ll gain a little bit more power than if there were only one per study—but the difference might be pretty minor.

As an aside, Terri Pigott has some nice resources for determining power for subgroup analysis based on random effects models:


> On Jun 29, 2019, at 9:29 AM, Samantha Cooper <samantha.cooper90 using yahoo.com> wrote:
> Hi there, 
> I’m a PhD student in need of some advice... can you please help me?
> I’m currently writing a meta-analysis and I found a link that describes how to compute power for a meta-analysis. The issue with this method is that it only considers the old way of doing meta-analysis (i.e., one effect size per study). I do not know how to compute power for a meta-analysis using a multi-level modeling framework.
> I reached out to Daniel Quintana and he suggested I reach out to all of you for help with this.
> https://towardsdatascience.com/how-to-calculate-statistical-power-for-your-meta-analysis-e108ee586ae8 
> I look forward to hearing from you!
> Samantha Cooper 
> Sent from my iPhone
>    [[alternative HTML version deleted]]
> _______________________________________________
> R-sig-meta-analysis mailing list
> R-sig-meta-analysis using r-project.org
> https://stat.ethz.ch/mailman/listinfo/r-sig-meta-analysis

	[[alternative HTML version deleted]]

More information about the R-sig-meta-analysis mailing list