I have been doing linear regressions using the _cluster_ option.
The first few lines of results would show something like:
*************************************************************************
Regression with robust standard errors Number of obs = 4247
F(130, 3413) = 10.50
Prob > F = 0.0000
R-squared = 0.2319
Number of clusters (personid) = 3414 Root MSE = .47171
*************************************************************************
Now you can see that Stata says there are 3414 unique individuals in the data
set. I happened to have counted the number of individuals and found it to be
different (my numbers are always smaller). I found that the number of
clusters/individuals specified in the Stata results to be the (degrees of
freedom + 1) for all the regressions I have performed.
Could someone tell me whether I have spotted something strange about the cluster
command or am I missing something?
