'aliasing' is how the computer program decides what your specification will
be by dropping out variables or putting parametric restrictions on the
coeff's (setting 2 equal for example). if you're cool with letting the
computer choose your theory, then this is a good thing :-). or to be
serious, this usually only makes sense if you're doing a prediction and
don't care about much else. for other purposes, i'd respecify; perhaps
regions or a hierarchical model will help.
Gary
--
*Gary King* - Albert J. Weatherhead III University Professor - Director,
IQSS - Harvard University
GKing.Harvard.edu <http://gking.harvard.edu/> - King(a)Harvard.edu -
@kinggary<http://twitter.com/kinggary>- 617-500-7570 - Asst 495-9271 -
Fax 812-8581
On Sun, Apr 24, 2011 at 1:57 AM, Leslie Finger <lfinger(a)fas.harvard.edu>wrote;wrote:
With regard to my last email, if it helps at all, the error that I'm
getting
when I try to use fixed effects with ivreg is
Error in linearHypothesis.default(object, Rmat, vcov. = vcov., test =
ifelse(df
:
there are aliased coefficients in the
model
It looks like aliased coefficients result when there's a bimodal
likelihood.
Anybody know how to fix this?
_______________________________________________
gov2001-l mailing list
gov2001-l(a)lists.fas.harvard.edu
http://lists.fas.harvard.edu/mailman/listinfo/gov2001-l
_______________________________________________
gov2001-l mailing list
gov2001-l(a)lists.fas.harvard.edu
http://lists.fas.harvard.edu/mailman/listinfo/gov2001-l