Statalist


[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

st: [Fwd: Heteroskedastic probit]


From   mmolina@uniroma3.it
To   statalist@hsphsun2.harvard.edu
Subject   st: [Fwd: Heteroskedastic probit]
Date   Fri, 1 May 2009 15:33:58 +0200 (CEST)

Dear Statlist,
Here the regular Probit. Thanks a lot. As you both suggested, these are
the results:

probit newproc normscapab n5a e6 d12b Le8c if a4b==1

Iteration 0:   log likelihood = -105.70439
Iteration 1:   log likelihood = -102.18133
Iteration 2:   log likelihood = -101.90586
Iteration 3:   log likelihood = -101.19701
Iteration 4:   log likelihood = -100.49243
Iteration 5:   log likelihood = -100.46574
Iteration 6:   log likelihood = -100.46567

Probit regression                                 Number of obs   =       
165
                                                  LR chi2(5)      =     
10.48
                                                  Prob > chi2     =    
0.0628
Log likelihood = -100.46567                       Pseudo R2       =    
0.0496

------------------------------------------------------------------------------
     newproc |      Coef.   Std. Err.      z    P>|z|     [95% Conf.
Interval]
-------------+----------------------------------------------------------------
  normscapab |   1.208397   .6709446     1.80   0.072    -.1066302   
2.523424
         n5a |   9.75e-09   1.41e-08     0.69   0.490    -1.79e-08   
3.74e-08
          e6 |   .0228493   .3045747     0.08   0.940    -.5741061   
.6198047
        d12b |  -.0028756    .005975    -0.48   0.630    -.0145863   
.0088352
        Le8c |   4.50e-06   2.75e-06     1.64   0.101    -8.76e-07   
9.89e-06
       _cons |  -.1643681   .6537495    -0.25   0.801    -1.445694   
1.116957
------------------------------------------------------------------------------
Note: 0 failures and 1 success completely determined.

At 11:45 AM 4/30/2009, jverkuilen wrote:

I second Rich's advice. 165 observations is a pretty small N for
heteroscedastic probit. But I suspect that even regular probit won't find
much either Look at the initial set of iterations to get starting
values---the log-likelihood only changes a little.


Good point. Based on the LLs, the probit model should have a model
chi-square of about 6 with 3 d.f. -- but that might be enough for at least
one variable to sneak in as significant.

-------------------------- Messaggio originale ---------------------------
Oggetto: [Fwd: Heteroskedastic probit]
Da:      mmolina@uniroma3.it
Data:    Gio, 30 Aprile 2009 5:27 pm
A:       statalist@hsphsun2.harvard.edu
--------------------------------------------------------------------------

Dear Statilist,
Richard, normscapab is not dichotomy. I constructed it as a score so it
varies from 0 and 1.You are right that then it has small value compared
with the other variables...
I tried Marteen's advice and obtained these results:

hetprob newproc normscapab e6 d12b if a4b==1, het(normscapab)

Fitting probit model:

Iteration 0:   log likelihood = -105.70439
Iteration 1:   log likelihood = -102.70201
Iteration 2:   log likelihood = -102.69757
Iteration 3:   log likelihood = -102.69757

Fitting full model:

Iteration 0:   log likelihood = -102.69757
Iteration 1:   log likelihood = -102.51681
Iteration 2:   log likelihood = -102.41164
Iteration 3:   log likelihood = -102.39453
Iteration 4:   log likelihood = -102.39175
Iteration 5:   log likelihood = -102.39174

Heteroskedastic probit model                    Number of obs     =
165
                                                Zero outcomes     =
 56
                                                Nonzero outcomes  =
109

                                                Wald chi2(3)      =
1.11
Log likelihood = -102.3917                      Prob > chi2       =
0.7758

------------------------------------------------------------------------------
             |      Coef.   Std. Err.      z    P>|z|     [95% Conf.
Interval]
-------------+----------------------------------------------------------------
newproc      |
  normscapab |   3.722963   3.615273     1.03   0.303    -3.362842
10.80877
          e6 |  -.2859454   .6005927    -0.48   0.634    -1.463085
.8911945
        d12b |  -.0033544   .0098363    -0.34   0.733    -.0226332
.0159244
       _cons |  -.1114548    1.10271    -0.10   0.919    -2.272727
2.049817
-------------+----------------------------------------------------------------
lnsigma2     |
  normscapab |   1.320455   1.518283     0.87   0.384    -1.655325
4.296234
------------------------------------------------------------------------------
Likelihood-ratio test of lnsigma2=0: chi2(1) =     0.61   Prob > chi2 =
0.4342

The model now converged but the prob is very high. Should I suppose the
presence of Heteroskedasticity now and before?
Best,
Alejandra


-------------------------- Messaggio originale ---------------------------
Oggetto: Heteroskedastic probit
Da:      mmolina@uniroma3.it
Data:    Gio, 30 Aprile 2009 4:19 pm
A:       statalist@hsphsun2.harvard.edu
--------------------------------------------------------------------------

Dear Statalist,
I run this model:
hetprob newprod normscapab n5a e6 d12b Le8c if a4b==5, het(normscapab)
and obtained this result: convergence not achieved.
Is tis anyway to rescue the model?
Best,
Alejandra Molina

Heteroskedastic probit model                    Number of obs     =
165
                                                Zero outcomes     =
 56
                                                Nonzero outcomes  =
109

                                                Wald chi2(5)      =
10.61
Log likelihood = -97.15464                      Prob > chi2       =
0.0596

------------------------------------------------------------------------------
             |      Coef.   Std. Err.      z    P>|z|     [95% Conf.
Interval]
-------------+----------------------------------------------------------------
newproc      |
  normscapab |  -.1146763   .0552553    -2.08   0.038    -.2229748
-.0063779
         n5a |   4.15e-10   4.03e-10     1.03   0.303    -3.75e-10
1.20e-09
          e6 |   .0098537   .0189325     0.52   0.603    -.0272533
.0469607
        d12b |  -.0004825   .0004277    -1.13   0.259    -.0013208
.0003557
        Le8c |   7.37e-07   4.60e-07     1.60   0.110    -1.66e-07
1.64e-06
       _cons |   .0818565   .0498956     1.64   0.101     -.015937
.1796501
-------------+----------------------------------------------------------------
lnsigma2     |
  normscapab |  -5.064362          .        .       .            .
  .
------------------------------------------------------------------------------
convergence not achieved











*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/



© Copyright 1996–2014 StataCorp LP   |   Terms of use   |   Privacy   |   Contact us   |   What's new   |   Site index