All,
I have estimated a nested logit model explaining schooling choice, with all schooling options in one branch, and the outside option in the (degenerate) second branch. I have estimated the model sequentially because of the computational burden.
Question: the nesting parameter that measure the dissimilarity among the schooling options comes out slighly negative (-.0459305) and insignificant. Does the insignificance imply that the nesting is inappropriate?
The output (the estimates for the lower model were fixed to their first-step estimates):
Nested logit estimates
Levels = 2 Number of obs = 546468
Dependent variable = chosen LR chi2(14) = 11816.01
Log likelihood = -7198.8489 Prob > chi2 = 0.0000
----------------------------------------------------------------------------
| Coef. Std. Err. z P>|z| [95% Conf. Int.]
-------------+--------------------------------------------------------------
option |
LSIZE | .9400799 . . . . .
VOCL | -.6005096 . . . . .
VOCLmale | .2947062 . . . . .
VOCLfor | .4408798 . . . . .
[cut]
ctrav | -5.145888 . . . . .
-------------+--------------------------------------------------------------
STUDY |
NONE | 1.169447 .2569964 4.55 0.000 .6657433 1.673151
NONEmale | .1646824 .1107066 1.49 0.137 -.0522985 .3816633
[cut]
NONEcath | -.4957498 .1248174 -3.97 0.000 -.7403875 -.2511122
-------------+----------------------------------------------------------------
(incl. value |
parameters) |
STUDY |
/YES | -.0459305 .073946 -0.62 0.535 -.1908621 .099001
/NO | 1 . . . . .
------------------------------------------------------------------------------
LR test of homoskedasticity (iv = 1): chi2(-164)= -26.94 Prob > chi2 = .
----------------------------------------------------------------------------
Many thanks,
Stijn Kelchtermans
Katholieke Universiteit Leuven
Belgium
[email protected]
*
* 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/
