st: RE: IVREG2 and Standard Errors when Number of Clusters is Small

["Schaffer, Mark E" <M.E.Schaffer@hw.ac.uk>]

Jessie,

With only 12 clusters, the cluster-robust approach is not the way you
want to go.  See, e.g., Austin Nichol's presentation at the 2007 UK
Stata User Group meeting:

http://www.stata.com/meeting/13uk/abstracts.html

The small-number-of-clusters problem comes up pretty often on the list,
and you should have a look at some of these earlier postings.  One that
might be relevant for you is from last November:

http://www.stata.com/statalist/archive/2011-11/msg01381.html

Including separate dummies for your 12 clusters might be a better
approach - see the discussion in the 2nd posting above.

HTH,
Mark

> -----Original Message-----
> From: owner-statalist@hsphsun2.harvard.edu 
> [mailto:owner-statalist@hsphsun2.harvard.edu] On Behalf Of Jessie C
> Sent: 25 January 2012 15:38
> To: statalist@hsphsun2.harvard.edu
> Subject: st: IVREG2 and Standard Errors when Number of 
> Clusters is Small
> 
> I am running a cross-country type regression using ivreg2.  
> The amount of clusters is less than 42.  For example, you can 
> think of the unit of observation as country and the cluster 
> as continent.  Though it is a slightly different context and 
> I have 12 clusters.  How exactly should standard errors be adjusted?
> *
> *   For searches and help try:
> *   http://www.stata.com/help.cgi?search
> *   http://www.stata.com/support/statalist/faq
> *   http://www.ats.ucla.edu/stat/stata/
> 


-- 
Heriot-Watt University is a Scottish charity
registered under charity number SC000278.

Heriot-Watt University is the Sunday Times
Scottish University of the Year 2011-2012



*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/