[Limdep Nlogit List] Positive Log likelihood from Maximize
Mattias Boman
mattias.boman at ess.slu.se
Thu Jun 11 00:13:37 EST 2009
Dear all,
I am analyzing payment card valuation data, so wtp is only known by
interval. I'm estimating valuation functions using the MAXIMIZE command in
Limdep. The WTP variable used is ln of the lower (lnYlow) and upper
(lnYhigh) wtp bounds in the dataset. Below is the LIMDEP output of a simple
estimation of mean (A) and standard deviation (K) of the ln WTP variable. As
can be seen, LIMDEP reports a positive Log likelihood value. Since the WTP
data is continuous and not discrete, I know that it is possible for the log
likelihood to be positive, especially since the standard deviation of ln WTP
is quite small.
My question is therefore if this output seems normal, or if there is
something I should look further into?
I am deeply grateful for your advise!
Best regards,
Mattias Boman
maximize; labels=a,K ? Parameter estimation
; start=8.50,0.94
; alg=newton
; fcn=log(phi((a-lnYlow)/K)-phi((a-lnYhigh)/K))$
Normal exit from iterations. Exit status=0.
+---------------------------------------------+
| User Defined Optimization |
| Maximum Likelihood Estimates |
| Model estimated: Jun 10, 2009 at 03:58:05PM.|
| Dependent variable Function |
| Weighting variable None |
| Number of observations 911 |
| Iterations completed 3 |
| Log likelihood function 2995.573 |
| Number of parameters 0 |
| Info. Criterion: AIC = -6.57645 |
| Finite Sample: AIC = -6.57645 |
| Info. Criterion: BIC = -6.57645 |
| Info. Criterion:HQIC = -6.57645 |
| Restricted log likelihood .0000000 |
| Chi squared 5991.146 |
| Degrees of freedom 2 |
| Prob[ChiSqd > value] = .0000000 |
+---------------------------------------------+
+--------+--------------+----------------+--------+--------+
|Variable| Coefficient | Standard Error |b/St.Er.|P[|Z|>z]|
+--------+--------------+----------------+--------+--------+
A | 8.49955870 .03128239 271.704 .0000
K | .94259630 .02243364 42.017 .0000
More information about the Limdep
mailing list