I've read the faq and the section in the manual on a negative R2 in
-reg3- but I'm still confused about it. The example in the manual
shows non-zero, statistically significant coefficients for the
qDemand equation, yet the R2 is negative. The negative R2 means that
the model predicts the dependent variable worse than a constant-only
model. Okay, I can understand how that could happen but only if most
coefficients are near-zero and non-significant. That's how it would
be in ordinary regression, an R2 of zero means that the coefficients
of all independent variables are zero. Here, there are significant
effects but the predicted values are way off, they predict quantity
more poorly than a constant-only model would. Well, I could
understand that if it meant that something was seriously wrong with
the model, that crucial predictors had been omitted perhaps or that
the instrumental variables were unsuitable for predicting the
endogenous variables. But the note states that the model on the whole
can be considered satisfactory, negative R-squared and all. Can
anyone clarify this for me or point me to a good reference?
Advance thanks,
John Hendrickx
__________________________________________________
Do you Yahoo!?
Yahoo! Mail Plus - Powerful. Affordable. Sign up now.
http://mailplus.yahoo.com
*
* For searches and help try:
* http://www.stata.com/support/faqs/res/findit.html
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/
