Bookmark and Share

Notice: On March 31, it was announced that Statalist is moving from an email list to a forum. The old list will shut down at the end of May, and its replacement, statalist.org is already up and running.


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

Re: st: Problem with margins after logit on a person period data


From   Richard Williams <richardwilliams.ndu@gmail.com>
To   statalist@hsphsun2.harvard.edu, statalist@hsphsun2.harvard.edu
Subject   Re: st: Problem with margins after logit on a person period data
Date   Wed, 08 Jun 2011 22:52:40 -0500

The error message isn't terribly informative, but I believe the problem is that mfx does not work with factor variables.

Also, if durat1-durat11 are mutually exclusive (i.e. you will have a score of 1 on one of them and scores of 0 on everything else) then I think you want something like i.durat. Otherwise the margins command won't realize that the values of the variables depend on each other.

Also, if you want margins to produce the same results that mfx would have produced, you should add the -atmeans- option (but many feel that the default -asobserved- is better anyway).

At 09:31 PM 6/8/2011, Urmi Bhattacharya wrote:
Hi,

I dropped one of the duration dummies durat1 and ran the following
logit regression

logit school_left childage i.childfemale i.urban i.scstobc
i.casteother i.dadp i.dadm i.momp i.momm wagep wage5 wage8 wage9 distp
dist
> m disth percapcons durat2 durat3 durat4 durat5 durat6 durat7 durat8 durat9 durat10 durat11, nolog

I then used margins as follows
margins , dydx(_all)

I got results.

However, when i used the lame logit regreesion and then used mfx, I
got the following error message

default predict() is unsuitable for marginal-effect calculation

Why would i get error with mfx and not with margin?

Best

Urmi


On Wed, Jun 8, 2011 at 8:17 PM, Richard Williams
<richardwilliams.ndu@gmail.com> wrote:
> I don't think margins likes it when you use the noconstant option. Is it
> really necessary for your model?
>
> At 05:08 PM 6/8/2011, Urmi Bhattacharya wrote:
>>
>> Dear Statalisters,
>>
>> I am running the following logit on a person-period data
>>
>> logit school_left childage i.childfemale i.urban i.scstobc
>> i.casteother i.dadp i.dadm i.momp i.momm wagep wage5 wage8 w
>> > age9 distp distm disth percapcons durat1 durat2 durat3 durat4 durat5
>> > durat6 durat7 durat8 durat9 durat10 durat11, nocon
>> > s nolog
>>
>> Logistic regression                               Number of obs   =
>>  47569
>>                                                  Wald chi2(28)   =
>> 14601.42
>> Log likelihood = -14502.393                       Prob > chi2     =
>> 0.0000
>>
>>
>> ------------------------------------------------------------------------------
>>  school_left |      Coef.   Std. Err.      z    P>|z|     [95% Conf.
>> Interval]
>>
>> -------------+----------------------------------------------------------------
>>    childage |  -.1413654    .006069   -23.29   0.000    -.1532605
>> -.1294703
>> 1.childfem~e |   .0212387   .0306462     0.69   0.488    -.0388266
>>  .0813041
>>     1.urban |   .0124911   .0368072     0.34   0.734    -.0596497
>>  .0846319
>>   1.scstobc |   1.972676   .1234074    15.99   0.000     1.730802
>> 2.21455
>> 1.casteother |   1.883233    .125722    14.98   0.000     1.636822
>>  2.129643
>>      1.dadp |    .680585    .044327    15.35   0.000     .5937057
>>  .7674643
>>      1.dadm |   .3872552   .0506982     7.64   0.000     .2878886
>>  .4866217
>>      1.momp |   1.351053   .0835161    16.18   0.000     1.187364
>>  1.514741
>>      1.momm |   .9494066   .0918146    10.34   0.000     .7694532
>> 1.12936
>>       wagep |  -.0281962   .0061845    -4.56   0.000    -.0403176
>> -.0160748
>>       wage5 |  -.0040992   .0045604    -0.90   0.369    -.0130373
>> .004839
>>       wage8 |    .021027   .0043869     4.79   0.000     .0124289
>>  .0296251
>>       wage9 |   .0159427   .0022335     7.14   0.000     .0115652
>>  .0203203
>>       distp |  -.0355589   .0162329    -2.19   0.028    -.0673748
>> -.0037431
>>       distm |  -.0024083   .0112658    -0.21   0.831    -.0244888
>>  .0196722
>>       disth |   .0186932   .0039384     4.75   0.000      .010974
>>  .0264124
>>  percapcons |  -.0001459   .0000278    -5.26   0.000    -.0002003
>> -.0000915
>>      durat1 |  -5.771466   .1406919   -41.02   0.000    -6.047217
>> -5.495715
>>      durat2 |  -4.947835   .1222363   -40.48   0.000    -5.187414
>> -4.708256
>>      durat3 |  -4.690019   .1193602   -39.29   0.000    -4.923961
>> -4.456078
>>      durat4 |  -4.054464   .1132586   -35.80   0.000    -4.276447
>> -3.832481
>>      durat5 |  -3.055883   .1082449   -28.23   0.000    -3.268039
>> -2.843727
>>      durat6 |  -3.560284   .1130896   -31.48   0.000    -3.781936
>> -3.338633
>>      durat7 |  -2.825943   .1099477   -25.70   0.000    -3.041436
>> -2.610449
>>      durat8 |  -2.238741   .1094063   -20.46   0.000    -2.453173
>> -2.024308
>>      durat9 |  -1.427979   .1099217   -12.99   0.000    -1.643421
>> -1.212536
>>     durat10 |  -.3967904   .1152271    -3.44   0.001    -.6226313
>> -.1709495
>>     durat11 |  -2.168164   .1486318   -14.59   0.000    -2.459477
>> -1.876851
>>
>> ------------------------------------------------------------------------------
>>
>> .
>> end of do-file
>>
>> Since I am interested in the marginal effects of the variables on the
>> probability of hazard,
>>
>> I do
>>
>> margins,dydx(*)
>>
>> But this gives me the following output
>>
>>  margins,dydx(*)
>>
>> Average marginal effects                          Number of obs   =
>>  47569
>> Model VCE    : OIM
>>
>> Expression   : Pr(school_left), predict()
>> dy/dx w.r.t. : childage 1.childfemale 1.urban 1.scstobc 1.casteother
>> 1.dadp 1.dadm 1.momp 1.momm wagep wage5 wage8
>>               wage9 distp distm disth percapcons durat1 durat2 durat3
>> durat4 durat5 durat6 durat7 durat8 durat9
>>               durat10 durat11
>>
>>
>> ------------------------------------------------------------------------------
>>             |            Delta-method
>>             |      dy/dx   Std. Err.      z    P>|z|     [95% Conf.
>> Interval]
>>
>> -------------+----------------------------------------------------------------
>>    childage |  (not estimable)
>> 1.childfem~e |  (not estimable)
>>     1.urban |  (not estimable)
>>   1.scstobc |  (not estimable)
>> 1.casteother |  (not estimable)
>>      1.dadp |  (not estimable)
>>      1.dadm |  (not estimable)
>>      1.momp |  (not estimable)
>>      1.momm |  (not estimable)
>>       wagep |  (not estimable)
>>       wage5 |  (not estimable)
>>       wage8 |  (not estimable)
>>       wage9 |  (not estimable)
>>       distp |  (not estimable)
>>       distm |  (not estimable)
>>       disth |  (not estimable)
>>  percapcons |  (not estimable)
>>      durat1 |  (not estimable)
>>      durat2 |  (not estimable)
>>      durat3 |  (not estimable)
>>      durat4 |  (not estimable)
>>      durat5 |  (not estimable)
>>      durat6 |  (not estimable)
>>      durat7 |  (not estimable)
>>      durat8 |  (not estimable)
>>      durat9 |  (not estimable)
>>     durat10 |  (not estimable)
>>     durat11 |  (not estimable)
>>
>> ------------------------------------------------------------------------------
>> Note: dy/dx for factor levels is the discrete change from the base level.
>>
>> Can someone explain what am I doing wrong? How do I get the marginal
>> effects after running the logit?
>>
>> Best
>>
>> Urmi
>> *
>> *   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/
>
> -------------------------------------------
> Richard Williams, Notre Dame Dept of Sociology
> OFFICE: (574)631-6668, (574)631-6463
> HOME:   (574)289-5227
> EMAIL:  Richard.A.Williams.5@ND.Edu
> WWW:    http://www.nd.edu/~rwilliam
>
> *
> *   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/
>

*
*   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/

-------------------------------------------
Richard Williams, Notre Dame Dept of Sociology
OFFICE: (574)631-6668, (574)631-6463
HOME:   (574)289-5227
EMAIL:  Richard.A.Williams.5@ND.Edu
WWW:    http://www.nd.edu/~rwilliam

*
*   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   |   Site index