Stata The Stata listserver
[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]

st: AW: Problem with logit


From   "Jann, Ben" <ben.jann@soz.gess.ethz.ch>
To   <statalist@hsphsun2.harvard.edu>
Subject   st: AW: Problem with logit
Date   Thu, 7 Aug 2003 12:04:37 +0200

Hi Hervé

The model almost perfectly explains your data (only 1 failure and 2 successes are not "perfectly" determined). Because of that, the coefficients are huge (!). That the standard errors are huge as well, is probably due to very high collinearity between 'pa' and 'na' (what are 'pa' and 'na'?). Such a situation is likely to happen, if casenumbers are very low. 

(Furthermore, note that your logit estimation may be tremendously biased in general because of insufficient N)

The reason for the missing CIs in the first table, is evident if looking at the CIs in the second table. For example, exp(-17028.46) practically equals zero and exp(16912.24) practically equals infinity.
(the CI's are so large becuase of the huge standard errors)

Try estimating a model including only one regressor. (or think about not using logistic regression at all)

ben 


> -----Ursprüngliche Nachricht-----
> Von: Hervé CACI [mailto:hcaci@wanadoo.fr] 
> Gesendet: Mittwoch, 6. August 2003 23:56
> An: STATALIST
> Betreff: st: Problem with logit
> 
> 
> Dear statalisters,
> 
> I ran a couple of logit/logistic regressions to predict a dichotomous
> variable "y" (0 = Control subject, 1 = Suicide Attempter) 
> using different
> sets of personality traits. The sample size is very limited 
> N=2*15 subjects.
> 
> Can someone explain the following output, and diagnose the problem ?
> 
> Thank you very much in advance.
> Hervé.
> -- 
> Hervé CACI, MD, PhD
> Child and Adolescent Psychiatry
> Service de Pédiatrie
> Hôpital de l'Archet 2
> 151, route de Saint Antoine de Ginestičre
> 06202 Nice Cedex 3 -- FRANCE
> Tel: 04 92 03 60 74
> Fax: 04 92 03 60 81
> email: caci.h@chu-nice.fr (at work)
>        hcaci@wanadoo.fr   (at home)
> Web: http://perso.wanadoo.fr/herve.caci
> 
> 
> . logistic SAMPLE pa na
> 
> Logit estimates                                 Number of obs 
>   =         30
>                                                 LR chi2(2)    
>   =      41.59
>                                                 Prob > chi2   
>   =     0.0000
> Log likelihood = -1.689e-07                     Pseudo R2     
>   =     1.0000
> 
> --------------------------------------------------------------
> --------------
>    SAMPLE_T0 | Odds Ratio   Std. Err.      z    P>|z|    [95% 
> Conf.Interval]
> -------------+------------------------------------------------
> --------------
>           pa |   5.81e-26   5.03e-22    -0.01   0.995         
>    0         .
>           na |   3.90e+57   7.64e+61     0.01   0.995         
>    0         .
> --------------------------------------------------------------
> --------------
> 
> note: 14 failures and 13 successes completely determined.
> 
> . logit SAMPLE pa na
> 
> Iteration 0:   log likelihood = -20.794415
> Iteration 1:   log likelihood = -8.9274011
> Iteration 2:   log likelihood = -6.3319885
> Iteration 3:   log likelihood = -4.9662853
> Iteration 4:   log likelihood = -4.1380091
> Iteration 5:   log likelihood = -3.5844148
> Iteration 6:   log likelihood = -2.9739135
> Iteration 7:   log likelihood = -2.2594929
> Iteration 8:   log likelihood = -1.6925544
> Iteration 9:   log likelihood = -1.0983049
> Iteration 10:  log likelihood =  -.5188107
> Iteration 11:  log likelihood = -.21385608
> Iteration 12:  log likelihood = -.07698551
> Iteration 13:  log likelihood = -.02772706
> Iteration 14:  log likelihood = -.01011442
> Iteration 15:  log likelihood = -.00370927
> Iteration 16:  log likelihood = -.00136299
> Iteration 17:  log likelihood = -.00050121
> Iteration 18:  log likelihood = -.00018435
> Iteration 19:  log likelihood = -.00006782
> Iteration 20:  log likelihood = -.00002495
> Iteration 21:  log likelihood = -9.178e-06
> Iteration 22:  log likelihood = -3.376e-06
> Iteration 23:  log likelihood = -1.242e-06
> Iteration 24:  log likelihood = -4.569e-07
> Iteration 25:  log likelihood = -1.578e-07
> Iteration 26:  log likelihood = -1.553e-07
> Iteration 27:  log likelihood = -1.541e-07
> Iteration 28:  log likelihood = -1.538e-07
> 
> Logit estimates                                 Number of obs 
>   =         30
>                                                 LR chi2(2)    
>   =      41.59
>                                                 Prob > chi2   
>   =     0.0000
> Log likelihood = -1.689e-07                     Pseudo R2     
>   =     1.0000
> 
> --------------------------------------------------------------
> --------------
>    SAMPLE |      Coef.   Std. Err.      z    P>|z|     [95% 
> Conf. Interval]
> ----------+---------------------------------------------------
> --------------
>        pa |  -58.10766   8658.502    -0.01   0.995    
> -17028.46    16912.24
>        na |   132.6071   19606.43     0.01   0.995    
> -38295.29     38560.5
>     _cons |  -510.9476    75810.8    -0.01   0.995    
> -149097.4    148075.5
> --------------------------------------------------------------
> --------------
> 
> note: 14 failures and 13 successes completely determined.
> 
> 
> 
> *
> *   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/
> 

*
*   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/



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