I have a questionnaire where the dependent variable has 10 ordered responses and there are various independent variables such as gender, age and ethnic group, and I am trying to use ologit to analyse the data:
Code:
. ologit q1 i.ethnic, nolog

Ordered logistic regression                     Number of obs     =        383
                                                LR chi2(6)        =       7.74
                                                Prob > chi2       =     0.2580
Log likelihood = -252.71891                     Pseudo R2         =     0.0151

-----------------------------------------------------------------------------------------------
                           q1 |      Coef.   Std. Err.      z    P>|z|     [95% Conf. Interval]
------------------------------+----------------------------------------------------------------
                       ethnic |
                       White  |          0  (base)
            Mixed / Multiple  |   .1759967   .7712277     0.23   0.819    -1.335582    1.687575
           Asian / A-British  |  -.4441779   .4089195    -1.09   0.277    -1.245645    .3572895
Black AfroCarib/B-AC-British  |   .3807691   .5542749     0.69   0.492    -.7055898    1.467128
                        Arab  |   12.83302   1504.048     0.01   0.993    -2935.048    2960.714
                       Other  |   12.83302   1063.523     0.01   0.990    -2071.633    2097.299
               Not disclosed  |  -1.910208   .7850429    -2.43   0.015    -3.448864   -.3715524
------------------------------+----------------------------------------------------------------
                        /cut1 |  -3.936211   .3725274                     -4.666351    -3.20607
                        /cut2 |  -3.520482   .3119141                     -4.131822   -2.909141
                        /cut3 |  -3.359486   .2921114                     -3.932014   -2.786958
                        /cut4 |  -3.034537   .2572624                     -3.538762   -2.530312
                        /cut5 |  -2.829487   .2384413                     -3.296823    -2.36215
                        /cut6 |  -2.367124   .2037146                     -2.766398   -1.967851
                        /cut7 |  -1.798812   .1726757                      -2.13725   -1.460374
-----------------------------------------------------------------------------------------------
Note: 3 observations completely determined.  Standard errors questionable.
I am happy that
Prob > chi2 = 0.2580 indicates that the regression coefficient for ethnic is not significantly different from 0, but what does P>|z| = 0.015 for 'Not disclosed' mean?
Is it that overall there is no effect of ethnic but 'Not disclosed' is significantly different from 'White'?
Or am I completely misinterpreting the output?

Thank you.
Janet