Re: st: Reason for mfx error is 'BECAUSE' ?

[May Boggess <mboggess@stata.com>]

On Thursday morning, Sam asked about an error message issued by -mfx-:

> Using Stata 8.2 I estimated a multinomial logit model as follows:
>
> . mlogit marstat tvfact1, basecategory(6) ;
>
> I then ran the mfx compute command, using several different versions of
> syntax I found, obtaining the following result:
>
> . mfx compute, predict(outcome(3)) ;
> invalid 'BECAUSE'
> r(198);
A close inspection of Sam's output reveals the likely cause. Notice that 
Sam's variables must have values labels: they are being used for the 
equation names.

> ------------------------------------------------------------------------------
>      marstat |      Coef.   Std. Err.      z    P>|z|     [95% Conf. Interval]
> -------------+----------------------------------------------------------------
> MARRIED      |
>      tvfact1 |   2.755563   .1691352    16.29   0.000     2.424064    3.087062
>        _cons |   -1.07647    .088055   -12.22   0.000    -1.249055   -.9038857
> -------------+----------------------------------------------------------------
> NOT MARRIE~  |
>      tvfact1 |  -.8979021   .2608498    -3.44   0.001    -1.409158    -.386646
>        _cons |  -.7464122    .122721    -6.08   0.000     -.986941   -.5058833
> -------------+----------------------------------------------------------------
> WIDOWED      |
>      tvfact1 |   2.361517   .8398351     2.81   0.005     .7154704    4.007564
>        _cons |  -5.034124   .4667658   -10.79   0.000    -5.948968    -4.11928
> -------------+----------------------------------------------------------------
> DIVORCED     |
>      tvfact1 |   1.020558   .2798024     3.65   0.000     .4721552     1.56896
>        _cons |  -1.888785   .1431811   -13.19   0.000    -2.169415   -1.608155
> -------------+----------------------------------------------------------------
> SEPARATED,~  |
>      tvfact1 |   1.697536   .3699212     4.59   0.000     .9725039    2.422568
>        _cons |  -2.915832   .1966681   -14.83   0.000    -3.301295    -2.53037
> ------------------------------------------------------------------------------
> (Outcome marstat==NEVER BEEN MARRIED is the comparison group)
Now notice that the equation name on the last equation contains a , . 
While this is not an "illegal" character to have in a value label, it is
a character that is worth avoiding. (Other characters worth avoiding
are ' and spaces). Some characters can cause unexpected errors as they
can interfere with the parsing of arguments.

Sam can most likely solve his problem by dropping the values labels as 
follows:

 label drop _all
 mlogit marstat tvfact1, basecategory(6)
 mfx compute, predict(outcome(3))

 -- May
mboggess@stata.com

*
*   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/