RE: st: difficulty with aflogit command

[Nick Cox <n.j.cox@durham.ac.uk>]

Just to underline that Richard's advice here is nearer the mark than my earlier comments. 

Nick 
n.j.cox@durham.ac.uk 

Richard Williams

With user-written postestimation commands, you often need to run the 
estimation command under version control. This is because ereturned 
results can change across versions of Stata, and the post-estimation 
command was written for an older format. Also, it may be that aflogit 
doesn't like variable names that are more than 8 characters, which I 
am guessing was Stata's limit way back when. The following code seems to work:

sysuse auto, clear
version 10: logit foreign mpg rep78
aflogit
clonevar displace = displacement
poisson price weight displace
aflogit


At 12:54 AM 5/29/2011, Ari Samaranayaka wrote:

>I am having difficulty with the user-written command aflogit.
>Although it is expected to work after unconditional logit or poisson 
>it does not work for me.
>
>sysuse auto,clear
>logit foreign mpg rep78
>aflogit
>
>this produce
>
>Logistic regression                               Number of obs
>=         69
>                                                   LR chi2(2)
>=      33.10
>                                                   Prob > chi2     =
>0.0000
>Log likelihood = -25.851693                       Pseudo R2       =
>0.3903
>------------------------------------------------------------------------------
>      foreign |      Coef.   Std. Err.      z    P>|z|     [95% 
> Conf. Interval]
>-------------+----------------------------------------------------------------
>          mpg |   .1328672   .0735776     1.81   0.071    -.0113423
>.2770767
>        rep78 |    1.73607   .4928761     3.52   0.000     .7700508
>2.70209
>        _cons |  -10.03775   2.407455    -4.17   0.000    -14.75628
>-5.319226
>------------------------------------------------------------------------------
>.
>. aflogit
>Can only be used after unconditional logistic or poisson regression
>last estimates not found
>r(301);
>
>Then I used the following.
>
>poisson price weight   displacement
>aflogit
>
>This produces
>
>Poisson regression                                Number of obs
>=         74
>                                                   LR chi2(2)      =
>29652.15
>                                                   Prob > chi2     =
>0.0000
>Log likelihood = -29162.101                       Pseudo R2       =
>0.3370
>
>------------------------------------------------------------------------------
>        price |      Coef.   Std. Err.      z    P>|z|     [95% 
> Conf. Interval]
>-------------+----------------------------------------------------------------
>       weight |   .0003235   4.52e-06    71.55   0.000     .0003147
>.0003324
>displacement |   .0000446    .000037     1.20   0.228     -.000028
>.0001172
>        _cons |   7.708671   .0081077   950.79   0.000      7.69278
>7.724561
>------------------------------------------------------------------------------
>
>. aflogit
>
>Population attributable fraction from poisson regression
>Cross-sectional / cohort data (n=74)
>
>
>Term      Ref.     A.F.        s.e.      [95% Conf. Int.]*
>----------------------------------------------------------
>weight     0      0.6355     11.9561   -3.03e+27   1.0000
>ment not found
>r(111);
>
>Can someone help me.
>I am using stata SE 11.2.
>my version of aflogit is
>c:\ado\plus\a\aflogit.ado
>*! version 1.0.6   16 Sep 1997  STB-42 sbe21
>

*
*   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/