[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]

st: clogit cluster concepts

From   "Timothy Dang" <>
To   stata <>
Subject   st: clogit cluster concepts
Date   Wed, 1 Aug 2007 13:40:38 -0700


I've got time-series data from an economics experiment. I'm using
-clogit- to estimate how buyers in the experiment choose which product
to buy each period, including a lagged variable. There are six
separate experimental runs, so the errors will be independent between
runs, but it's likely there will be dependence within a run.

I'd been simply running -clogit-, until I couldn't convince myself
that was acceptable any more. From Stata docs (I'm using Stata 8
student version, so I have partial docs), it looks like the -cluster-
option is created for just this kind of problem.

I'm wondering if anyone can point me towards good descriptions of what
exactly -cluster- does. I suspect it's some kind of bootstrapping
approach? (although if so, it seems surprisingly fast).

Also, the results when including the -cluster- option have the same
coefficients as without, although the standard errors are different.
Does this just mean I'm lucky in my data, or does -cluster- only
address differences in errors?


Timothy O'Neill Dang / Cretog8
One monkey don't stop no show.
*   For searches and help try:

© Copyright 1996–2015 StataCorp LP   |   Terms of use   |   Privacy   |   Contact us   |   What's new   |   Site index