[R-meta] equivalent random formulas but different results in rma.mv

[Michael Dewey]

Dear Timothy

With a complex model like that I would not call a difference in the 
fourth significant figure very important

Michael

On 11/10/2021 22:24, Timothy MacKenzie wrote:
> Dear Wolfgang,
> 
> Per https://www.metafor-project.org/doku.php/analyses:konstantopoulos2011,
> I thought the following random formulas are equivalent (save for correlated
> effects) leading to the same fixed effects (in my case intercept-only).
> 
> But I wonder why they produce different fixed effects?
> 
> Thanks,
> Tim M
> 
> random1 = ~ 1 | research_gr / study / outcome / time / esID
> 
> estimate      se    zval    pval   ci.lb   ci.ub
>    0.3061  0.0905  3.3830  0.0007  0.1288  0.4835  ***
> 
> random2 = list(~ 1 | research_gr, ~outcome | interaction(research_gr,
> study),  ~time | interaction(research_gr, study, outcome), ~1| esID) )
> 
> estimate      se    zval    pval   ci.lb   ci.ub
>    0.3069  0.0908  3.3817  0.0007  0.1290  0.4849  ***
> 
> 	[[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
> 

-- 
Michael
http://www.dewey.myzen.co.uk/home.html