[Limdep Nlogit List] Latent Class
Kristy Wallmo
Kristy.Wallmo at noaa.gov
Wed May 30 04:08:15 EST 2007
Dr. Greene - Thank you for your response. I am obviously doing
something wrong in my code because when I restrict the 'sizesq'
coefficient in class 3 to zero it continues to estimate a parameter. I
also tried ordering the betas in the restriction line from 1 to 21 and
got the same results as below. Any further advice would be
appreciated. Thanks, Kristy Wallmo
-> NLOGIT; Lhs = chosen
; Choices = altA, altB, current
; Rhs = size, sizesq,notake,science,leisure,fishing,cost
; lcm
; pds = count
; pts = 3
; parameters
; prob = nestt
; rst =
b1,b2,b3,b4,b5,b6,b7,b1,b2,b3,b4,b5,b6,b7,b1,0,b3,b4,b5,b6,b7
; keep = pred $
Normal exit from iterations. Exit status=0.
+---------------------------------------------+
| Discrete choice (multinomial logit) model |
| Maximum Likelihood Estimates |
| Model estimated: May 29, 2007 at 11:02:30AM.|
| Dependent variable Choice |
| Weighting variable None |
| Number of observations 5143 |
| Iterations completed 4 |
| Log likelihood function -5285.888 |
| R2=1-LogL/LogL* Log-L fncn R-sqrd RsqAdj |
| No coefficients -5650.1630 .06447 .06238 |
| Constants only -5520.5685 .04251 .04036 |
| Response data are given as ind. choice. |
| Number of obs.= 5143, skipped 0 bad obs. |
+---------------------------------------------+
+---------+--------------+----------------+--------+---------+
|Variable | Coefficient | Standard Error |b/St.Er.|P[|Z|>z] |
+---------+--------------+----------------+--------+---------+
SIZE|1 .02069397 .00739216 2.799 .0051
SIZESQ|1 -.00035148 .00016052 -2.190 .0286
NOTAKE|1 -.38113667 .04058207 -9.392 .0000
SCIENC|1 .33909621 .03676529 9.223 .0000
LEISUR|1 .29362099 .03795363 7.736 .0000
FISHIN|1 .27530156 .03698722 7.443 .0000
COST|1 -.00627851 .00043009 -14.598 .0000
Normal exit from iterations. Exit status=0.
+---------------------------------------------+
| Latent Class Logit Model |
| Maximum Likelihood Estimates |
| Model estimated: May 29, 2007 at 11:02:43AM.|
| Dependent variable CHOSEN |
| Weighting variable None |
| Number of observations 15429 |
| Iterations completed 55 |
| Log likelihood function -4604.229 |
| Restricted log likelihood -5650.163 |
| Chi squared 2091.868 |
| Degrees of freedom 23 |
| Prob[ChiSqd > value] = .0000000 |
| R2=1-LogL/LogL* Log-L fncn R-sqrd RsqAdj |
| No coefficients -5650.1630 .18512 .18329 |
| Constants only -5520.5685 .16599 .16412 |
| At start values -5285.9330 .12897 .12701 |
| Response data are given as ind. choice. |
+---------------------------------------------+
+---------------------------------------------+
| Latent Class Logit Model |
| Number of latent classes = 3 |
| ------------------------------------------- |
| LCM model with panel has 1034 groups. |
| Variable number of obs./group =COUNT |
| Discrete parameter variation specified. |
| ------------------------------------------- |
| Number of obs.= 5143, skipped 0 bad obs. |
+---------------------------------------------+
+---------+--------------+----------------+--------+---------+
|Variable | Coefficient | Standard Error |b/St.Er.|P[|Z|>z] |
+---------+--------------+----------------+--------+---------+
Utility parameters in latent class -->> 1
SIZE|1 .05228005 .01011869 5.167 .0000
SIZESQ|1 -.00096002 .00021866 -4.390 .0000
NOTAKE|1 -.45964188 .06219600 -7.390 .0000
SCIENC|1 .79979990 .05428568 14.733 .0000
LEISUR|1 .55581804 .05484539 10.134 .0000
FISHIN|1 .69392822 .05449280 12.734 .0000
COST|1 -.01197388 .00071690 -16.702 .0000
Utility parameters in latent class -->> 2
SIZE|2 .06450757 .01455872 4.431 .0000
SIZESQ|2 -.00061082 .00030998 -1.971 .0488
NOTAKE|2 .40826357 .07550812 5.407 .0000
SCIENC|2 .45207220 .07217526 6.264 .0000
LEISUR|2 .23724388 .07410923 3.201 .0014
FISHIN|2 -.40714987 .08523645 -4.777 .0000
COST|2 .00340385 .00084422 4.032 .0001
Utility parameters in latent class -->> 3
SIZE|3 -.12378399 .01715690 -7.215 .0000
SIZESQ|3 .00194102 .00039078 4.967 .0000
NOTAKE|3 -.52175536 .09535550 -5.472 .0000
SCIENC|3 -.28264202 .09035623 -3.128 .0018
LEISUR|3 .36204525 .07706081 4.698 .0000
FISHIN|3 .40946390 .07646876 5.355 .0000
COST|3 -.00713881 .00099379 -7.183 .0000
Estimated latent class probabilities
PrbCls_1 .58079430 .02526794 22.985 .0000
PrbCls_2 .18011035 .02286834 7.876 .0000
PrbCls_3 .23909535 .01419923 16.839 .0000
William Greene wrote:
> Kristy. You can do it, but it's going to require a lot of care.
> What you have to do is fit the model with 1 class first to
> see how the coefficients are arranged in the base model.
> NLOGIT reorders things, so you need to get the ordering
> right. After you see how the model is arranged by the program,
> you can use ;RST = list1, list2, list3 ... $ where each list specifies
> the coefficients in the utility function, either free, constrained
> or fixed. Note, you can only specify different utility functions
> through the use of zeros placed in a global specification.
> /B. Greene
>
> ************************************************
> Professor William Greene
> Department of Economics
> Stern School of Business
> New York University
> 44 West 4th St., Rm. 7-78
> New York, NY 10012
> Ph. 212.998.0876
> Fax. 212.995.4218
> URL. http://www.stern.nyu.edu/~wgreene
> Email. wgreene at stern.nyu.edu
> ************************************************
>
> ----- Original Message -----
> From: Kristy Wallmo <Kristy.Wallmo at noaa.gov>
> Date: Friday, May 25, 2007 12:16 pm
> Subject: [Limdep Nlogit List] Latent Class
>
>
>> I am running a latent class model, and would like to know if it's
>> possible in LIMDEP to specify different utility functions for
>> different
>> classes.
>>
>> Thanks, Kristy
>> _______________________________________________
>> Limdep site list
>> Limdep at limdep.itls.usyd.edu.au
>> http://limdep.itls.usyd.edu.au
>>
>>
>
> _______________________________________________
> Limdep site list
> Limdep at limdep.itls.usyd.edu.au
> http://limdep.itls.usyd.edu.au
>
More information about the Limdep
mailing list