dear all,
the solution to the problem is a slash.
the model which takes into account cluster randomisation then writes:
emm.statc.lme <- lme(data = emm.stat, random = ~weeks | tn.mainc / code, fixed
= oq.tot ~ weeks, na.action = na.omit)
summary(emm.statc.lme) then has
Formula: ~ weeks | code %in% tn.mainc and
Number of Groups:
tn.mainc code %in% tn.mainc
39 294
as compared to the model which does not take into account cluster randomisation
(which writes: emm.stat.lme <- lme(data = emm.stat, random = ~weeks | code,
fixed = oq.tot ~ weeks, na.action = na.omit), standard errors and subsequently
t-values of fixed effects increase while coefficients remain more or less
unchanged. perfect.
thanks so much to klaas for the valuable help.
best wishes
bernd
>-----Ursprüngliche Nachricht-----
>Von: s-news-owner@lists.biostat.wustl.edu
>[mailto:s-news-owner@lists.biostat.wustl.edu] Im Auftrag von
>Puschner, Bernd
>Gesendet: Freitag, 27. Juli 2007 11:44
>An: s-news@lists.biostat.wustl.edu
>Betreff: [S] cluster in lme
>
>dear all,
>
>i am analyzing data of a cluster-randomized trial via lme. when calling
>
>a.lme <- lme(data = emm.stat, random = ~weeks | code, fixed =
>v6.a ~ weeks * as.factor(group.fi), na.action = na.omit)
>
>everything is fine, but cluster structure is ignored; so I call
>
>b.lme <- lme(data = emm.stat, random = ~weeks | code, fixed =
>v6.a ~ weeks * as.factor(group.fi), cluster =~
>as.factor(cs.group), na.action = na.omit)
>
>but get this error message
>
>> b.lme <- lme(data = emm.stat, random = ~ weeks | code,
>fixed = v6.a ~
>weeks *
> as.factor(group.fi), cluster = ~ as.factor(cs.group),
>na.action =
> na.omit)
>Problem in lme: argument cluster= not matched: lme(data =
>emm.stat, random = ~ weeks | code, fixed = v6.a ~ weeks *
> as.factor(group.fi), cluster = ~ as.factor(cs.group),
>na.action =
> na.omit)
>Use traceback() to see the call stack
>
>any help would be highly appreciated.
>
>best wishes
>
>bernd----------------------------------------------------------
>----------
>This message was distributed by
>s-news@lists.biostat.wustl.edu. To unsubscribe send e-mail to
>s-news-request@lists.biostat.wustl.edu with the BODY of the
>message: unsubscribe s-news
>
|