st: OR results of mhodds vs. logistic

[Suzy <scott_788@wowway.com>]

Dear Statalisters:
I think this might be more of a statistical question rather than a Stata 
question. I would like to know why the  odds ratio estimate, as an 
approximation to the odds ratio
for a one unit increase in x1 (x1 is continuous) for the mhodds command  
is so different than the logistic output from the command below.  FYI, 
x2 is continuous as well. Both x1 and x2 are linear in the logit.
Suzy


OR results for x1 controlling for x2 are different for the mhodds 
command and the logistic command below:


mhodds Y x1 x2 by(x3)

-------------------------------------------------------------------------------
    Y    | Odds Ratio        chi2(1)         P>chi2       [95% Conf. 
Interval]
----------+--------------------------------------------------------------------
       1 |   1.154021          21.50         0.0000         1.08622    
1.22606
       2 |   1.095894           5.87         0.0154         1.01764    
1.18017
       3 |   1.073977           3.63         0.0566         0.99799    
1.15574
       4 |   1.308015           5.38         0.0203         1.04258    
1.64103
       5 |   1.004919           0.00         0.9552         0.84669    
1.19271
-------------------------------------------------------------------------------

 
by x3, sort : logistic Y x1 x2 [the output right here is my summary of 
the results obtained, not the actual output]

-------------------------------------------------------------------------------
    Y    | Odds Ratio        chi2(1)         P>chi2       [95% Conf. 
Interval]
----------+--------------------------------------------------------------------
       1 |   1.187428           5.32         0.000          1.114544   
1.265078
       2 |   1.141416           3.56         0.000          1.061168   
1.227733
       3 |   1.127073           3.46         0.001          1.053328   
1.205982
       4 |   1.198436           2.16         0.031          1.01679    
1.412532
       5 |   1.055348           0.76         0.445           .9191339  
1.211748
-------------------------------------------------------------------------------


by x3, sort : logistic Y x1 x2 [this is the actual output for x3]

------------------------------------------------------------------------------
-> x3 = 1

Logistic regression                               Number of obs   
=       2862
                                                 LR chi2(2)      =     
130.25
                                                 Prob > chi2     =     
0.0000
Log likelihood = -493.86326                       Pseudo R2       =     
0.1165

------------------------------------------------------------------------------
      Y     | Odds Ratio   Std. Err.      z    P>|z|     [95% Conf. 
Interval]
-------------+----------------------------------------------------------------
        x1  |   1.18209    .039921     4.95   0.000      1.10638    
1.262982
        x2  |   1.051794   .0060071    8.84   0.000     1.040086    
1.063634
------------------------------------------------------------------------------





*
*   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/