[R-sig-ME] question about the lme result

[ONKELINX, Thierry]

Both fit the same model but the parametrisation is different. Hence the hypotheses tested by summary() or anova() can be different as well. It's up to you to determine which parametrisation is the most useful.

Essentially, all models are wrong, but some are useful (George Box)



Van: Lao Meng [mailto:laomeng.3 at gmail.com] 
Verzonden: woensdag 27 juli 2011 9:57
Aan: ONKELINX, Thierry
CC: r-sig-mixed-models at r-project.org
Onderwerp: Re: [R-sig-ME] question about the lme result

Thanks Thierry.It works well!

According to your suggestion,I find that the following command has the same effect:
lme(sqrt(CD4) ~ group + group:time-1,random = ~time|id,data=data_ori)

>From the result,can I say that maybe the opinion that "a*b = a + a:b" is not "right"?

My best


2011/7/27 ONKELINX, Thierry <Thierry.ONKELINX at inbo.be>
Dear Lao,

Rewrite your model as lme(sqrt(CD4) ~ 0 + group + group:time,random = ~time|id,data=data_ori)

Best regards,

Thierry

> -----Oorspronkelijk bericht-----
> Van: r-sig-mixed-models-bounces at r-project.org [mailto:r-sig-mixed-models-
> bounces at r-project.org] Namens Lao Meng
> Verzonden: woensdag 27 juli 2011 9:26
> Aan: r-sig-mixed-models at r-project.org
> Onderwerp: [R-sig-ME] question about the lme result
>
> Hi all.
> My data is a longitudinal data,which measure everyone(id)'s CD4 level at some
> specific time point.All persons are grouped into 3 groups(0,1,2,3).
>
> y:sqrt(CD4)
> x:time,GROUP(3levels: 0,1,2,3)
>
> data_ori is my original data,which inclues:CD4,group,time,id.
>
> I performed a linear mixed model using lme:
> result <- lme(sqrt(CD4) ~ group*time,random = ~time|id,data=data_ori)
>
> >From the summary(result),I can only get the intercept and slope of
> GROUP0(which is described as "intercept" and "time").
> If I wanna get intercepts and slopes of GROUP1,GROUP2,GROUP3,how can I
> do?
>
> Simply,I can do this manually by the following steps(GROUP1 for instance):
> "intercept" + GROUP1 = intercept_GROUP1; "time" + time:GROUP1 =
> slope_GROUP1;
>
> But is there any "shortcut" method to choose?Also,I cann't get the
> corresponding SE and p value for each parameter.So the above
> "manually"method is not effective to some degree.
>
> Any help from you are welcome.
>
> Many thanks!
>
>       [[alternative HTML version deleted]]
>
> _______________________________________________
> R-sig-mixed-models at r-project.org mailing list
> https://stat.ethz.ch/mailman/listinfo/r-sig-mixed-models