Puzzling results from -listcoef- command

[Louis Boakye-Yiadom]

I have been running logit regressions and making use of the -listcoef- command, which I got through -findit SPost9-. I am, however, puzzled by some of the results I'm getting. I've noticed that for some of the regressors, the outputs for b, z, and P>|z| are not the same as those of the -logit- outputs for Coeff, z, and P>|z|, respectively. In the results shown below, the coefficients for oip1, dr, and urbrur1 are examples of my puzzle. Any help will be appreciated. Thank you.


. logit ds1 `plist4' if sex==1;

note: emp5 != 0 predicts failure perfectly
emp5 dropped and 4 obs not used

note: emp6 != 0 predicts success perfectly
emp6 dropped and 1 obs not used

Iteration 0: log likelihood = -718.12999
Iteration 1: log likelihood = -673.06324
Iteration 2: log likelihood = -670.23666
Iteration 3: log likelihood = -670.20789
Iteration 4: log likelihood = -670.20777
Iteration 5: log likelihood = -670.20777

Logistic regression Number of obs = 1423
LR chi2(18) = 95.84
Prob > chi2 = 0.0000
Log likelihood = -670.20777 Pseudo R2 = 0.0667

------------------------------------------------------------------------------
ds1 | Coef. Std. Err. z P>|z| [95% Conf. Interval]
-------------+----------------------------------------------------------------
age | -.0234963 .0058438 -4.02 0.000 -.0349498 -.0120427
edu2 | .5577593 .262392 2.13 0.034 .0434805 1.072038
edu3 | .5054113 .2648162 1.91 0.056 -.013619 1.024442
edu4 | .401732 .2025544 1.98 0.047 .0047327 .7987312
edu5 | 1.026113 .2660829 3.86 0.000 .5045998 1.547626
edu6 | -.171827 1.090377 -0.16 0.875 -2.308927 1.965273
ms1 | -1.089054 .2926991 -3.72 0.000 -1.662734 -.5153742
ms3 | -.0935468 .2778263 -0.34 0.736 -.6380762 .4509827
emp2 | .7404608 .3505719 2.11 0.035 .0533525 1.427569
emp3 | 1.034576 .3613939 2.86 0.004 .3262569 1.742895
emp4 | -.9776898 1.081074 -0.90 0.366 -3.096555 1.141176
emp5 | 0 (omitted)
emp6 | 0 (omitted)
haqf2 | -.3332583 .2666087 -1.25 0.211 -.8558018 .1892853
haqf3 | .0224107 .2661995 0.08 0.933 -.4993308 .5441521
haqf4 | .2263022 .2729632 0.83 0.407 -.3086958 .7613001
haqf5 | .1463086 .2876302 0.51 0.611 -.4174362 .7100535
oip1 | -.2268579 .1724274 -1.32 0.188 -.5648094 .1110936
dr | .0810507 .0996962 0.81 0.416 -.1143502 .2764517
urbrur1 | -.1358947 .1642383 -0.83 0.408 -.4577959 .1860064
_cons | -1.290111 .5393789 -2.39 0.017 -2.347274 -.2329474
------------------------------------------------------------------------------

. listcoef, std help;

logit (N=1423): Unstandardized and Standardized Estimates

Observed SD: .40244191
Latent SD: 1.9611839

Odds of: Owes vs No_debt

-------------------------------------------------------------------------------
ds1 | b z P>|z| bStdX bStdY bStdXY SDofX
-------------+-----------------------------------------------------------------
age | -0.02350 -4.021 0.000 -0.3788 -0.0120 -0.1932 16.1232
edu2 | 0.55776 2.126 0.034 0.1651 0.2844 0.0842 0.2960
edu3 | 0.50541 1.909 0.056 0.1511 0.2577 0.0770 0.2989
edu4 | 0.40173 1.983 0.047 0.1985 0.2048 0.1012 0.4941
edu5 | 1.02611 3.856 0.000 0.3321 0.5232 0.1693 0.3236
edu6 | -0.17183 -0.158 0.875 -0.0144 -0.0876 -0.0073 0.0836
ms1 | -1.08905 -3.721 0.000 -0.3595 -0.5553 -0.1833 0.3301
ms3 | -0.09355 -0.337 0.736 -0.0266 -0.0477 -0.0136 0.2842
emp2 | 0.74046 2.112 0.035 0.3463 0.3776 0.1766 0.4676
emp3 | 1.03458 2.863 0.004 0.4164 0.5275 0.2123 0.4024
emp4 | -0.97769 -0.904 0.366 -0.1285 -0.4985 -0.0655 0.1314
haqf4 | -0.33326 -1.250 0.211 -0.1384 -0.1699 -0.0706 0.4153
haqf5 | 0.02241 0.084 0.933 0.0097 0.0114 0.0049 0.4308
oip1 | 0.22630 0.829 0.407 0.1110 0.1154 0.0566 0.4906
dr | 0.14631 0.509 0.611 0.1075 0.0746 0.0548 0.7345
urbrur1 | -0.22686 -1.316 0.188 -0.1073 -0.1157 -0.0547 0.4732
-------------------------------------------------------------------------------
b = raw coefficient
z = z-score for test of b=0
P>|z| = p-value for z-test
bStdX = x-standardized coefficient
bStdY = y-standardized coefficient
bStdXY = fully standardized coefficient
SDofX = standard deviation of X

[Richard Williams]

Unless you are using a hopelessly antiquated version of Stata I suggest you uninstall spost9 (-ado uninstall spost9_ado-) and install spost13 (-findit spost13_ado-). Despite its name spost13 generally workers with older versions of Stata. If the problem persists, contact Scott Long directly.

Alternatively, just make sure that you have the most current version of spost9.

Incidentally, output is much easier to read if you use the code option, available as part of the advanced text editor. Click on the underlined A that appears on the top right of the message you are writing.

[Louis Boakye-Yiadom]

Richard, many thanks for your suggestions! After following your first suggestion, I don't see the puzzle any more. Everything is fine now.