From "Scott Merryman" To Subject Re: st: beginnerXs ask about Xtlogit probabilities Date Sat, 15 Nov 2003 08:15:14 -0600

```----- Original Message -----
From: "Nick Varian" <nk7_br@yahoo.com.br>
To: <statalist@hsphsun2.harvard.edu>
Sent: Friday, November 14, 2003 5:51 AM

> Scott,
> thanks for your help. Could you help me once more? I
> am a lit bite confused, because using xtlogit i got
> some p-value that means that may parameter is
> significant different from zero. After calculing mfx
> compute, they all turn to insignificant. What does its
> means? In which one should I believe?
>
>

I can't really help you with that, however you may find reporting the percentage
change in odds rather than the marginal effects to be insightful.  -listcoef-
will conveniently provide you with the odds ratio and the percentage change in
odds.

For example:

. webuse union
(NLS Women 14-24 in 1968)

. xtlogit union age grade south year, i(id) fe nolog
note: multiple positive outcomes within groups encountered.
note: 2744 groups (14165 obs) dropped due to all positive or
all negative outcomes.

Conditional fixed-effects logistic regression   Number of obs      =     12035
Group variable (i): idcode                      Number of groups   =      1690

Obs per group: min =         2
avg =       7.1
max =        12

LR chi2(4)         =     68.46
Log likelihood  = -4515.9536                    Prob > chi2        =    0.0000

------------------------------------------------------------------------------
union |      Coef.   Std. Err.      z    P>|z|     [95% Conf. Interval]
-------------+----------------------------------------------------------------
age |   .0758677   .0960711     0.79   0.430    -.1124282    .2641637
grade |   .0857237   .0418685     2.05   0.041     .0036629    .1677845
south |  -.7469976   .1249048    -5.98   0.000    -.9918065   -.5021887
year |   -.059335   .0967972    -0.61   0.540     -.249054    .1303839
------------------------------------------------------------------------------

. mfx compute, predict(pu0)

Marginal effects after clogit
y  = Pr(union|fixed effect is 0) (predict, pu0)
=  .16861097
------------------------------------------------------------------------------
variable |      dy/dx    Std. Err.     z    P>|z|  [    95% C.I.   ]      X
---------+--------------------------------------------------------------------
age |   .0106352      .02044    0.52   0.603  -.029435  .050705    30.538
grade |   .0120169      .03911    0.31   0.759  -.064646  .088679   12.7934
south*|   -.099099      .32063   -0.31   0.757  -.727525  .529327   .381388
year |  -.0083177      .01301   -0.64   0.522  -.033809  .017174   79.6184
------------------------------------------------------------------------------
(*) dy/dx is for discrete change of dummy variable from 0 to 1

. listcoef, p

clogit (N=12035): Percentage Change in Odds

Odds of: 1 vs 0

--------------------------------------------------
union |      b         z     P>|z|      %
-------------+------------------------------------
age |   0.07587    0.790   0.430      7.9
grade |   0.08572    2.047   0.041      9.0
south |  -0.74700   -5.981   0.000    -52.6
year |  -0.05934   -0.613   0.540     -5.8
--------------------------------------------------

With this, the interpretation is
For each additional grade the odds of being in a union increase by 9%
holding     all other variables constant.

or,
Working in the south reduces the odds of being in a union by 53%

Hope this helps,
Scott

*
*   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/
```