On Tue, 7 Apr 2009, Brian Quistorff wrote:
When running a regression, is there a way to cluster the errors
around a particular variable? I want to allow the
variance-covariance matrix of the errors to have a
block-diagonal setup. In Stata, it's called "clustered robust".
Is this option available? Thanks very much.
When you estimate a model on panel data in gretl you have the
option of two robust estimators of the covariance matrix
-- Arellano and Beck-Katz -- both of which are "clustered" on the
panel unit/group. See chapter 14 of the Gretl User's Guide for
details.
But at present gretl doesn't support "clustered robust" estimation
outside of the case of panel data.
Allin Cottrell