# st: heckprob mfx

 From "Jiang, Lan" To Subject st: heckprob mfx Date Fri, 23 Jun 2006 13:48:10 -0400

Hi experts,

I am new in stata and even newer in statalist.  I am having a heckprob problem for your help.  As below, I have hospital (dummy var) insignificant in the 2nd step modeling insignificant, but when I use mfx to calculate the marginal effects, it becomes significant.  Here is some of the output.

Thanks!

Lan

heckprob v29424  hospital mh othorg numclient_100 svc_breadth invol_10pct, sel(v29422=hospital mh othorg sanicotine_10pct svc_breadth phyimp) nolog

Probit model with sample selection              Number of obs      =       494
Censored obs       =       254
Uncensored obs     =       240

Wald chi2(6)       =     18.75
Log likelihood = -452.8194                      Prob > chi2        =    0.0046

------------------------------------------------------------------------------
|      Coef.   Std. Err.      z    P>|z|     [95% Conf. Interval]
-------------+----------------------------------------------------------------
v29424       |
hospital |   .6798989   .3715279     1.83   0.067    -.0482823     1.40808
mh |   .2269317   .3036784     0.75   0.455    -.3682671    .8221305
othorg |  -.0007292   .2278594    -0.00   0.997    -.4473254    .4458669
numclien~100 |   .0235725    .010234     2.30   0.021     .0035143    .0436308
svc_breadth |    .052458   .0230032     2.28   0.023     .0073725    .0975435
invol_10pct |  -.0561788    .025818    -2.18   0.030    -.1067811   -.0055764
_cons |  -1.420717    .874694    -1.62   0.104    -3.135086    .2936517
-------------+----------------------------------------------------------------
v29422       |
hospital |   .5296687   .2100133     2.52   0.012     .1180503    .9412872
mh |  -.0741861   .1824389    -0.41   0.684    -.4317597    .2833876
othorg |  -.1386789   .1368669    -1.01   0.311    -.4069331    .1295752
sanicotine~t |   .0505129   .0209759     2.41   0.016      .009401    .0916248
svc_breadth |   .0342173    .009774     3.50   0.000     .0150606    .0533739
phyimp |  -.0611262   .0253105    -2.42   0.016    -.1107339   -.0115186
_cons |  -.5084769   .2220102    -2.29   0.022     -.943609   -.0733449
-------------+----------------------------------------------------------------
/athrho |   .0993535   .8370807     0.12   0.906    -1.541294    1.740001
-------------+----------------------------------------------------------------
rho |   .0990279   .8288718                     -.9123376    .9402268
------------------------------------------------------------------------------
LR test of indep. eqns. (rho = 0):   chi2(1) =     0.01   Prob > chi2 = 0.9047
------------------------------------------------------------------------------

mfx compute, eqlist(v29424) predict(pcond) at (hospital=1, svc_breadth=12.6, numclient_100=7.1, invol_pct=3.9)

Marginal effects after heckprob
y  = Pr(v29424=1|v29422=1) (predict, pcond)
=  .48923093
------------------------------------------------------------------------------
variable |      dy/dx    Std. Err.     z    P>|z|  [    95% C.I.   ]      X
---------+--------------------------------------------------------------------
hospital*|   .2403324      .10073    2.39   0.017   .042913  .437752         1
mh*|   .0919905      .11726    0.78   0.433  -.137838  .321819   .143725
othorg*|   .0026067      .08942    0.03   0.977  -.172645  .177859   .425101
numc~100 |   .0090431      .00395    2.29   0.022   .001307  .016779       7.1
svc_br~h |   .0201245      .00859    2.34   0.019   .003293  .036956      12.6
inv~0pct |  -.0215519      .00987   -2.18   0.029  -.040897 -.002207    3.5481
------------------------------------------------------------------------------
(*) dy/dx is for discrete change of dummy variable from 0 to 1

*
*   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/