[R-meta] double arcsine transformation

Hi James,

You are passing 'transf.ipft.hm' to the second argument of predict.rma():

args(predict.rma)

function (object, newmods, intercept, tau2.levels, gamma2.levels, 
    addx = FALSE, level, digits, transf, targs, vcov = FALSE, ...)

So, you are setting 'newsmods = transf.ipft.hm' which doesn't make sense. It should be:

pes <- predict(pes.dat, transf=transf.ipft.hm, targ=list(ni=dat$total))

Best,
Wolfgang

-----Original Message-----
From: R-sig-meta-analysis [mailto:r-sig-meta-analysis-bounces at r-project.org] On Behalf Of james meyer
Sent: Wednesday, 29 January, 2020 2:16
To: r-sig-meta-analysis at r-project.org
Subject: [R-meta] double arcsine transformation

I am performing a univariate meta-analysis on sensitivity and specificity.   I
fit the model in metafor and when trying to transform the double-arcsine I get
the following error:

 pes<-predict(pes.dat,transf.ipft.hm,targ=list(ni=dat$total))
Error in predict.rma(pes.dat, transf.ipft.hm, targ = list(ni = dat$total)) : 
  Cannot specify new moderator values for models without moderators.

Is this fixable or do you need more information.

James C Meyer DVM DACVIM

Dear James

I know this is not what you asked but are you sure you want to do 
univariate analyses here? Although sensitivity and specificity are 
independent within study in general they are not independent between 
study. It may be that you have too few studies to do the bivariate 
analysis of course.

Michael

I am performing a univariate meta-analysis on sensitivity and specificity.   I
fit the model in metafor and when trying to transform the double-arcsine I get
the following error:

  pes<-predict(pes.dat,transf.ipft.hm,targ=list(ni=dat$total))
Error in predict.rma(pes.dat, transf.ipft.hm, targ = list(ni = dat$total)) :
   Cannot specify new moderator values for models without moderators.

Is this fixable or do you need more information.

James C Meyer DVM DACVIM

_______________________________________________
R-sig-meta-analysis mailing list
R-sig-meta-analysis at r-project.org
https://stat.ethz.ch/mailman/listinfo/r-sig-meta-analysis

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

Good point. And speaking of unsolicited advice -- concerns can be raised about the back-transformation itself:

https://onlinelibrary.wiley.com/doi/abs/10.1002/jrsm.1348

Best,
Wolfgang

-----Original Message-----
From: R-sig-meta-analysis [mailto:r-sig-meta-analysis-bounces at r-project.org] On Behalf Of Michael Dewey
Sent: Wednesday, 29 January, 2020 10:36
To: james meyer; r-sig-meta-analysis at r-project.org
Subject: Re: [R-meta] double arcsine transformation

Dear James

I know this is not what you asked but are you sure you want to do 
univariate analyses here? Although sensitivity and specificity are 
independent within study in general they are not independent between 
study. It may be that you have too few studies to do the bivariate 
analysis of course.

Michael

I am performing a univariate meta-analysis on sensitivity and specificity.   I
fit the model in metafor and when trying to transform the double-arcsine I get
the following error:

  pes<-predict(pes.dat,transf.ipft.hm,targ=list(ni=dat$total))
Error in predict.rma(pes.dat, transf.ipft.hm, targ = list(ni = dat$total)) :
   Cannot specify new moderator values for models without moderators.

Is this fixable or do you need more information.

James C Meyer DVM DACVIM