Bookmark and Share

Notice: On March 31, it was announced that Statalist is moving from an email list to a forum. The old list will shut down on April 23, and its replacement, is already up and running.

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: st: how to handle clustering

From   Jeph Herrin <>
Subject   Re: st: how to handle clustering
Date   Thu, 29 Jul 2010 14:18:47 -0400

I agree with Michael, if you have 37 classrooms you should start
with a multi-level model. In my experience power estimates for ML
models are rudimentary at best (unless you did a simulation), and
assuming you have the data on hand, you can do better by simply
estimating the model.


On 7/29/2010 1:52 PM, Michael N. Mitchell wrote:
Dear Kate

My feeling is that a multi-level model is still an appropriate model,
because you have data at the child and classroom level, with the
predictor (intervention) at the classroom level and the outcome at the
child level.

In any case, I do not feel you will derive more power from using
-regress- with -vce(cluster classroom)- than you would from using
-xtmixed .... || classroom:- . In both cases, the standard errors
reflect the correlation of observations within classrooms, but with the
multilevel model I feel you would have the opportunity to more clearly
formulate the model in terms of child level and classroom level predictors.

I hope that helps,

Michael N. Mitchell
Data Management Using Stata -
A Visual Guide to Stata Graphics -
Stata tidbit of the week -

On 2010-07-29 8.36 AM, Kate Welti wrote:

I am analyzing data in which I have treatment and control classrooms
(intervention was at the classroom level) and child-level test scores.
Ideally, I would use a multi-level model, however, I don't have
enough classrooms and children to do so (according to a power
analysis). I need to find a way to control for the clustering of
children in classrooms. Can Stata's cluster option be used in this



* For searches and help try:
* For searches and help try:

*   For searches and help try:

© Copyright 1996–2016 StataCorp LP   |   Terms of use   |   Privacy   |   Contact us   |   Site index