Using ranef() (package nlme, version 3.1-75) with an 'lme' object I can
obtain random effects for intercept and slope of a certain level (say:
1) - this corresponds to (say level 1) "residuals" in MLWin. Maybe
I'm
mistaken here, but the results are identical.
However, if I try to get the standardized random effects adding the
paramter "standard=T" to the specification of ranef(), the results
differ considerably from the results of MLWin (although MLWin defines
"standardized" in the same way as "divided by its estimated
(diagnostic)
standard error").
Why do the results differ although the estimates (random effects and
thus their variances) are almost identical? I noticed that lme() does
not compute the standard errors of the variances of the random effects -
for several reasons, but if this is true, how does ranef() calculate the
standardized random effects (the help says: '"standardized" (i.e.
divided by the corresponding estimated standard error)').
Is there a way to obtain similar results as in MLWin (or: should I
prefer the results of ranef() for certain reasons)?
Dirk
-----------------------------
R version: 2.3.1 Patched (2006-06-21 r38367)
*************************************************
Dr. Dirk Enzmann
Institute of Criminal Sciences
Dept. of Criminology
Edmund-Siemers-Allee 1
D-20146 Hamburg
Germany
phone: +49-(0)40-42838.7498 (office)
+49-(0)40-42838.4591 (Billon)
fax: +49-(0)40-42838.2344
email: dirk.enzmann at uni-hamburg.de
www:
http://www2.jura.uni-hamburg.de/instkrim/kriminologie/Mitarbeiter/Enzmann/Enzmann.html
Have you tried RSiteSearch("MLWin")? I just got 29 hits. I wonder
if any one of these might relate to your question?
If you would like more help on this issue from this listserve, please
submit another post, preferably illustrating your question with the
simplest possible self-contained example that illustrates your question,
perhaps like the following:
fm1.16 <- lme(distance~age, data=Orthodont[1:16,],
random=~age|Subject)
Hope this helps.
Spencer Graves
p.s. PLEASE do read the posting guide
"www.R-project.org/posting-guide.html" and provide commented, minimal,
self-contained, reproducible code.
Dirk Enzmann wrote:> Using ranef() (package nlme, version 3.1-75) with an 'lme' object I
can
> obtain random effects for intercept and slope of a certain level (say:
> 1) - this corresponds to (say level 1) "residuals" in MLWin.
Maybe I'm
> mistaken here, but the results are identical.
>
> However, if I try to get the standardized random effects adding the
> paramter "standard=T" to the specification of ranef(), the
results
> differ considerably from the results of MLWin (although MLWin defines
> "standardized" in the same way as "divided by its estimated
(diagnostic)
> standard error").
>
> Why do the results differ although the estimates (random effects and
> thus their variances) are almost identical? I noticed that lme() does
> not compute the standard errors of the variances of the random effects -
> for several reasons, but if this is true, how does ranef() calculate the
> standardized random effects (the help says: '"standardized"
(i.e.
> divided by the corresponding estimated standard error)').
>
> Is there a way to obtain similar results as in MLWin (or: should I
> prefer the results of ranef() for certain reasons)?
>
> Dirk
>
> -----------------------------
> R version: 2.3.1 Patched (2006-06-21 r38367)
>
>
> *************************************************
> Dr. Dirk Enzmann
> Institute of Criminal Sciences
> Dept. of Criminology
> Edmund-Siemers-Allee 1
> D-20146 Hamburg
> Germany
>
> phone: +49-(0)40-42838.7498 (office)
> +49-(0)40-42838.4591 (Billon)
> fax: +49-(0)40-42838.2344
> email: dirk.enzmann at uni-hamburg.de
> www:
>
http://www2.jura.uni-hamburg.de/instkrim/kriminologie/Mitarbeiter/Enzmann/Enzmann.html
>
> ______________________________________________
> R-help at stat.math.ethz.ch 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.
> > Why do the results differ although the estimates (random > effects and > > thus their variances) are almost identical? I noticed that > lme() does > > not compute the standard errors of the variances of the > random effects > > - for several reasons, but if this is true, how does > ranef() calculate > > the standardized random effects (the help says: > '"standardized" (i.e. > > divided by the corresponding estimated standard error)'). > > > > Is there a way to obtain similar results as in MLWin (or: should I > > prefer the results of ranef() for certain reasons)?I think there are two different issues here. The lme function does not produce a standard error of the variance component as some other multilevel packages do. It is often recommended in the multilevel literature to consider the p-value of the variance components and "fix" or retain the variance if p < .05. There are good reasons not to follow this practice. If you were using lmer(), you still wouldn't get this statistic, but you could use the MCMCsamp() function to examine the distribution of the random effects. The second issue (I think) is that the conditional variance of the random effect is not the same as the standard error of the variance of the random effect. From the definition in the help of ranef.lme, I believe it is the random effect divided by its conditional standard error. I don't know how to get the posterior variance of the random effects in lme, but I do in lmer, so we can experiment a bit. It has been a while since I have really used lme and I do not think there is an extractor function to get these and I didn't see the variances in the model object. Let's work through an example to see what we get. Here is what I see library(nlme) data(Orthodont) detach(package:nlme) library(Matrix) fm1 <- lmer(distance ~ age + (age|Subject), data = Orthodont) # equivalent to # fm1 <- lme(distance ~ age, data=Orthodont) # Extract the variances of the random effects qq <- attr(ranef(fm1, postVar = TRUE)[[1]], "postVar") # divide the random effects by their standard error of "age" Sranef_lmer <- ranef(fm1)[[1]][,2]/ sqrt(qq[2,2,]) library(nlme) # Now, run the lme model fm2 <- lme(distance ~ age, Orthodont) # get the standardized random effects from lme Sranef_lme <- ranef(fm2, standard=T)[2] cor(Sranef_lmer, Sranef_lme) [1] 1 Notice the perfect correlation. But, the actual values in Sranef_lme and Sranef_lmer are a bit different and I cannot see why just yet. I need to go eat lunch, but I'll think about this. Maybe somebody else sees something.
OK, I see how the standardized random effects are calculated. Here is what I now see library(nlme) fm2 <- lme(distance~age, Orthodont) # unstandardized age_ranef <- ranef(fm2)[,2] #standardized age_Sranef <- ranef(fm2, standard=TRUE)[,2] # We can use these to solve for the standard error, because the formula according to help for ranef.lme is # Standardized_randomEffects = random_effects/standard error age_ranef/age_Sranef # OK, now note the values are exactly the same. Now, look at VarCorr(fm2) You can see the value used to standardize is the standard deviation of the random effect for age. Now, the help function does say "divided by the corresponding standard error". I've copied Doug Bates because the values in the stdDev column are the standard deviations of the variance components and not standard errors of those variance components. So, I'm not sure why the help says that the standardized random effects are divided by the corresponding SE. Maybe he can clarify if he has time. I hope that helps Harold> -----Original Message----- > From: r-help-bounces at stat.math.ethz.ch > [mailto:r-help-bounces at stat.math.ethz.ch] On Behalf Of Doran, Harold > Sent: Sunday, July 30, 2006 3:40 PM > To: Spencer Graves; Dirk Enzmann > Cc: r-help at stat.math.ethz.ch > Subject: Re: [R] standardized random effects with ranef.lme() > > > > > Why do the results differ although the estimates (random > > effects and > > > thus their variances) are almost identical? I noticed that > > lme() does > > > not compute the standard errors of the variances of the > > random effects > > > - for several reasons, but if this is true, how does > > ranef() calculate > > > the standardized random effects (the help says: > > '"standardized" (i.e. > > > divided by the corresponding estimated standard error)'). > > > > > > Is there a way to obtain similar results as in MLWin (or: > should I > > > prefer the results of ranef() for certain reasons)? > > I think there are two different issues here. The lme function > does not produce a standard error of the variance component > as some other multilevel packages do. It is often recommended > in the multilevel literature to consider the p-value of the > variance components and "fix" > or retain the variance if p < .05. There are good reasons not > to follow this practice. > > If you were using lmer(), you still wouldn't get this > statistic, but you could use the MCMCsamp() function to > examine the distribution of the random effects. > > The second issue (I think) is that the conditional variance > of the random effect is not the same as the standard error of > the variance of the random effect. From the definition in the > help of ranef.lme, I believe it is the random effect divided > by its conditional standard error. I don't know how to get > the posterior variance of the random effects in lme, but I do > in lmer, so we can experiment a bit. It has been a while > since I have really used lme and I do not think there is an > extractor function to get these and I didn't see the > variances in the model object. > > Let's work through an example to see what we get. Here is what I see > > library(nlme) > data(Orthodont) > detach(package:nlme) > > library(Matrix) > fm1 <- lmer(distance ~ age + (age|Subject), data = Orthodont) > # equivalent to # fm1 <- lme(distance ~ age, data=Orthodont) > > # Extract the variances of the random effects qq <- > attr(ranef(fm1, postVar = TRUE)[[1]], "postVar") > > # divide the random effects by their standard error of "age" > Sranef_lmer <- ranef(fm1)[[1]][,2]/ sqrt(qq[2,2,]) > > library(nlme) > # Now, run the lme model > fm2 <- lme(distance ~ age, Orthodont) > > # get the standardized random effects from lme Sranef_lme <- > ranef(fm2, standard=T)[2] > > cor(Sranef_lmer, Sranef_lme) > [1] 1 > > Notice the perfect correlation. But, the actual values in > Sranef_lme and Sranef_lmer are a bit different and I cannot > see why just yet. I need to go eat lunch, but I'll think about this. > > Maybe somebody else sees something. > > ______________________________________________ > R-help at stat.math.ethz.ch 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. >