Statalist


[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]

st: RE: -xtlogit- produces different results under different versions


From   "Zhiqiang Wang" <z.wang@uq.edu.au>
To   <statalist@hsphsun2.harvard.edu>
Subject   st: RE: -xtlogit- produces different results under different versions
Date   Fri, 26 Sep 2008 16:03:24 +1000

While trying to update on an old ado file to include -xtlogit- command for my Australian colleague, we noticed that using the same data, -xtlogit- produces different results. 
 
. webuse union
. keep in 1/1000
. version 8: xtlogit  not_smsa age black, nolog

Random-effects logistic regression          Number of obs      =      1000
Group variable: idcode                      Number of groups   =       175

Random effects u_i ~ Gaussian               Obs per group: min =         1
                                                         avg =       5.7
                                                         max =        12

                                           Wald chi2(2)       =     20.97
Log likelihood  = -117.94789               Prob > chi2        =    0.0000

...

. version 9: xtlogit  not_smsa age black, nolog

Random-effects logistic regression         Number of obs      =      1000
Group variable: idcode                     Number of groups   =       175

Random effects u_i ~ Gaussian              Obs per group: min =         1
                                                           avg =       5.7
                                                           max =        12

                                           Wald chi2(2)       =     12.74
Log likelihood  = -130.83111               Prob > chi2        =    0.0017

...
. version 10: xtlogit  not_smsa age black, nolog

Random-effects logistic regression         Number of obs      =      1000
Group variable: idcode                     Number of groups   =       175

Random effects u_i ~ Gaussian              Obs per group: min =         1
                                                          avg =       5.7
                                                          max =        12
                                           Wald chi2(2)       =     12.80
Log likelihood  =   -115.103               Prob > chi2        =    0.0017
...


I wonder if I missed any thing. Can we check all older versions of -xtlogit- to identify the causes? The -which xtlogit- only shows the location of the current version of xtlogit.   

Thanks

Zhiqiang Wang, PhD
School of Medicine, University of Queensland
Email: z.wang@uq.edu.au


*
*   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