[R-sig-ME] lme, groupedData, random intercept and slope

[David Atkins]

On 9/10/10 4:00 PM, Andrew Robinson wrote:
> Hi Dave,
>
> on the topic of correlation between random slopes and intercepts, it's
> not necessarily bad news.  It may (probably does) indicate that the
> data are remote from the origin.  Under those circumstances, an
> adjustment to the slope, for example, leads inevitably to adjustment
> in the intercept.

Fair enough, though in looking at the original post (most of which I had 
snipped out), the data have 10 groups and 30 observations.  Thus, in 
this particular case, I'm inclined to think it's over-parameterized.

More generally, 9 times out of 10 with my own data, a correlation of 1 
in the random-effects means I'm over-fitting.  Though as you note, it's 
certainly worth wondering whether there is an extreme correlation 
induced by radically different scale.

[Almost sort of vaguely remember an example of this early on in P&B...]

cheers, Dave

>
> Best wishes
>
> Andrew
>
> On Fri, Sep 10, 2010 at 03:24:34PM -0700, David Atkins wrote:
>>
>> John--
>>
>> Focusing on the random-effects output (snipped from your post):
>>
>> Random effects:
>>   Formula: ~time | Subject
>>   Structure: General positive-definite
>>              StdDev       Corr
>> (Intercept) 6.054897e+00 (Intr)
>> time        4.160662e-05 1
>> Residual    9.775954e-04
>>
>> That is reporting random intercept, random slope (for time), and
>> residual error (and that the correlation between intercepts and slopes
>> is 1, which usually isn't a good sign...).
>>
>> Note that those are all on the SD scale (sqrt of variance terms).
>>
>> If you don't have a copy of Pinheiro and Bates (2000) and plan to use
>> nlme, I would strongly suggest tracking one down.
>>
>> Hope that helps.
>>
>> cheers, Dave
>>
>> --
>> Dave Atkins, PhD
>> Research Associate Professor
>> Department of Psychiatry and Behavioral Science
>> University of Washington
>> datkins at u.washington.edu
>>
>> Center for the Study of Health and Risk Behaviors (CSHRB)		
>> 1100 NE 45th Street, Suite 300 	
>> Seattle, WA  98105 	
>> 206-616-3879 	
>> http://depts.washington.edu/cshrb/
>> (Mon-Wed)	
>>
>> Center for Healthcare Improvement, for Addictions, Mental Illness,
>>    Medically Vulnerable Populations (CHAMMP)
>> 325 9th Avenue, 2HH-15
>> Box 359911
>> Seattle, WA 98104
>> http://www.chammp.org
>> (Thurs)
>>
>> _______________________________________________
>> R-sig-mixed-models at r-project.org mailing list
>> https://stat.ethz.ch/mailman/listinfo/r-sig-mixed-models
>