[R] metafor: using mixed models

[Viechtbauer Wolfgang (STAT)]

Dear Joris,

When you call rma() with the (default) argument "intercept=TRUE", then the intercept is added to the X matrix (a column of 1s is added). So, if you use:

fac <- c(1,1,2,3,3,4,4,5,5,5)
X <- model.matrix(~factor(fac))[,2:5]

and then:

rma(ai, bi, ci, di, mods=X, data=testdata, measure="OR")

the final X matrix has 1s in the first column (the intercept), and then the 4 dummies to code levels 2, 3, 4, and 5 in the other columns (with level 1 being the "reference" level, which is not included in the X matrix). So, if you want to test the factor, use:

rma(ai, bi, ci, di, mods=X, data=testdata, measure="OR", btt=2:5)

Or, you can use:

fac <- c(1,1,2,3,3,4,4,5,5,5)
X <- model.matrix(~factor(fac))
rma(ai, bi, ci, di, mods=X, intercept=FALSE, data=testdata, measure="OR", btt=2:5)

which should give you the same results.

I hope this helps!

Best,

--
Wolfgang Viechtbauer                        http://www.wvbauer.com/
Department of Methodology and Statistics    Tel: +31 (0)43 388-2277
School for Public Health and Primary Care   Office Location:
Maastricht University, P.O. Box 616         Room B2.01 (second floor)
6200 MD Maastricht, The Netherlands         Debyeplein 1 (Randwyck)


----Original Message----
From: r-help-bounces at r-project.org
[mailto:r-help-bounces at r-project.org] On Behalf Of Joris Meys Sent:
Monday, January 04, 2010 11:54 To: R mailing list
Subject: [R] metafor: using mixed models

> Dear all,
>
> I'm currently applying a mixed model approach to meta analysis using
> the package metafor. I use the "model.matrix()" function to create
> dummy variables. The option btt gives me the combined test for the
> dummies. Problem is, I don't know which indices I have to use, and
> can't really figure it out from the help file and the examples. I use
> following code :
>
> X <- model.matrix(~factor, data=testdata)[,2:5] # to exclude the
> intercept. Otherwise I get singular issues in the rma model
>
> rma(ai, bi, ci, di, mods=X, data=testdata,btt=c(2:5),measure="OR")
>
> Now I'm not sure whether btt reads the model intercept of rma as
> index 1, or the first dummy as index 1. So I'm not sure if I should
> use :
>
> rma(ai, bi, ci, di, mods=X, data=testdata,btt=c(1:4),measure="OR")
>
> from the examples in the help files, I should use the first option.
> But there I see no option used to exclude an intercept from the model
> matrix. All help greatly appreciated. Thank you in advance Joris
>
>
> --
> Joris FA Meys
> Statistical Consultant
>
> Ghent University
> Faculty of Bioscience Engineering
> Department of Applied mathematics, biometrics and process control
>
> Coupure Links 653
> B-9000 Gent
>
> tel : 09/ 264 59 87
> Joris.Meys at Ugent.be
> -------------------------------
> Disclaimer : http://helpdesk.ugent.be/e-maildisclaimer.php
>
>       [[alternative HTML version deleted]]
>
> ______________________________________________
> R-help at r-project.org mailing list
> https://stat.ethz.ch/mailman/listinfo/r-help
> PLEASE do read the posting guide
> http://www.R-project.org/posting-guide.html
> and provide commented, minimal, self-contained, reproducible code.