R help - Apr 2009 - meta regression in R using lme function

Dear all,

We are trying to do a meta regression in R using the lme function. The 
reason for doing this with lme function is that we have covariates and 
studies within references. In S-Plus this is possible by using the 
following command:

lme(outcome ~ covars, random = ~1 | reference/study, weights = 
varFixed(~var.outcome), data = mydata, control = lmeControl(sigma = 1))

This means that the residual variance is equal to var.outcome times sigma, 
where sigma is fixed to 1.

In R we do not know how to fix sigma = 1 in the lmeControl. Has anyone 
experience with this? If so, how can we solve this problem?

We can't stand that we can do it in S-Plus, but not in R!

Jan van de Kassteele
Maarten Schipper



Disclaimer RIVM
	[[alternative HTML version deleted]]

Hello Jan,

As far as know, fixing sigma to 1 is not possible in R with lme. That is
why I started to write my own functions to allow me to fit mixed-effects
models in R. Quite some time ago, I put one of those functions on my
website, which can be downloaded here:
http://www.wvbauer.com/downloads.html. Since some people have indicated
to me that they have found the function useful, I have been expanding
the function into a full package, which I will be relasing shortly. It
includes a much-improved version of the function that is currently on my
website and methods for printing, ploting, residual extraction,
influence diagnostics, and so on. The Mantel-Haenszel and Peto's method
are also implemented. In a second stage, I will be adding functions to
fit the bivariate mixed-effects model and then a multivariate model
function, which is useful for more complex syntheses (such as when doing
a meta-analysis of mixed-treatment comparisons). So maybe the upcoming
package will be useful to you.

Best,

-- 
Wolfgang Viechtbauer
 Department of Methodology and Statistics
 University of Maastricht, The Netherlands
 http://www.wvbauer.com/



----Original Message----
From: r-help-bounces at r-project.org [mailto:r-help-bounces at r-project.org]
On
Behalf Of Jan van de Kassteele Sent: Wednesday, April 29, 2009 15:02
To: r-help at r-project.org
Subject: [R] meta regression in R using lme function
> Dear all,
> 
> We are trying to do a meta regression in R using the lme function. The
> reason for doing this with lme function is that we have covariates and
> studies within references. In S-Plus this is possible by using the
> following command:
> 
> lme(outcome ~ covars, random = ~1 | reference/study, weights >
varFixed(~var.outcome), data = mydata, control = lmeControl(sigma 1))
> 
> This means that the residual variance is equal to var.outcome times
sigma,
> where sigma is fixed to 1.
> 
> In R we do not know how to fix sigma = 1 in the lmeControl. Has anyone
> experience with this? If so, how can we solve this problem?
> 
> We can't stand that we can do it in S-Plus, but not in R!
> 
> Jan van de Kassteele
> Maarten Schipper