[R-meta] Identifying studies in influence analyses output

[Michael Dewey]

Dear Daniel

Comments in-line

On 04/12/2019 11:05, Daniel Mønsted Shabanzadeh wrote:
> Hey
> 
> I am performing a meta-analysis on proportions of complications following a
> surgical intervention and analyses are with high heterogeneity. I therefore
> explore influential studies in order to exclude them and use this code:
> 

Please do not do that, they are the most interesting studies.

> b<-rma(xi=compl_treat, ni=total, measure = "PAS", data=s)

At this point you did not tell rma what the studies were called (using 
the slab parameter) so it cannot label them. Try supplying slab and see 
what happens.

Michael

> b.1<-influence(b)
> options(max.print=999999)
> print(b.1)
> 
> Which gives me this list:
> 
>   rstudent  dffits cook.d  cov.r tau2.del     QE.del    hat weight    dfbs inf
> 1     1.4286  0.0960 0.0092 0.9992   0.0200 31272.5619 0.0044 0.4369
> 0.0960
> 2    -1.0338 -0.0726 0.0053 1.0048   0.0201 30956.0289 0.0049 0.4912
> -0.0726
> 3    -0.4100 -0.0300 0.0009 1.0095   0.0202 31277.1418 0.0049 0.4891
> -0.0300
> 4     1.0644  0.0744 0.0055 1.0039   0.0201 31114.4810 0.0048 0.4827
> 0.0744
> 5    -0.4970 -0.0336 0.0011 1.0080   0.0202 31315.6235 0.0043 0.4285
> -0.0336
> 6    -0.1984 -0.0148 0.0002 1.0095   0.0202 31311.0548 0.0046 0.4574
> -0.0148
> 7     1.0481  0.0680 0.0046 1.0036   0.0201 31294.6471 0.0042 0.4171
> 0.0680
> 8    -0.3738 -0.0273 0.0008 1.0096   0.0202 31300.1405 0.0049 0.4852
> -0.0273
> 9    -0.9698 -0.0683 0.0047 1.0054   0.0201 31087.1853 0.0049 0.4912
> -0.0683
> 10    0.8417  0.0561 0.0032 1.0058   0.0201 31285.9234 0.0045 0.4483
> 0.0561
> 11    1.0238  0.0657 0.0043 1.0037   0.0201 31297.4093 0.0041 0.4091  0.0657
> 
> 
> How can I obtain the study identifications in order to identify them in my
> dataset?
> 
> Regards,
> Daniel
> 
> 	[[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