[R] Variance Component/ICC Confidence Intervals via Bootstrap or Jackknife

[Spencer Graves]

<see in line> 

Rick Bilonick wrote:
> On Sun, 2006-10-29 at 11:06 -0800, Spencer Graves wrote:
>   
>>       I can think of two ways to get confidence intervals on intraclass 
>> correlations (ICCs) and more accurate intervals for variance 
>>     
<snip>
>>       
>>     
>
>
>
> The way I've done the bootstrapping (sampling at each level) sounds the
> same as using a simulation. But articles and references I've found
> indicate that only the highest level (a if c is nested in b and b is
> nested a) should be sampled.
>
>   
Correct:  Bootstrapping with mixed effects hurts my brain just to think 
about it, because if you are not careful, you effectively assume away 
the variance components structure, and that's rarely what you want to 
do.  The difference between "simulate.lme" and "mcmcsamp" is that the 
first produces Monte Carlo confidence intervals, while the latter 
produces "Bayesian posterior" intervals, so the interpretation is 
slightly different.  I do not recall having seen any direct comparison, 
so I'd be interested in the results, if you try both.  However, I'd be 
extremely surprised if the answers were very different.  If I did it 
myself and found substantive differences, I would suspect that I'd done 
something wrong with at least one of the two.  If I couldn't find any 
errors in my work, I'd start looking for help. 

      Hope this helps. 
      Spencer Graves
> Rick B.
>
>
>
>