Re: st: Cramer Ridder test for multinomial logit model - syntax error

[Tirthankar Chakravarty <tirthankar.chakravarty@gmail.com>]

Sangamitra,

Here is my output from running the above example:

************************************
. crtest

**** Cramer-Ridder test for combining outcome categories

Ho: Candidates for pooling have the same regressor coefficients
    apart from the intercept

                       ln L     ln Lr        LR    P>chi2
 Indemnit:Prepaid  -534.362  -549.988    31.252     0.000
Indemnit:Uninsure  -534.362  -539.756    10.790     0.056
 Prepaid:Uninsure  -534.362  -542.431    16.139     0.006

degrees of freedom for chi-square distribution:  5
************************************

Please ensure that your edited and saved version of -crtest- is being
called by Stata. You can check by typing:
. viewsource crtest.ado
 and looking for the changes you made. Make sure all instances of
"e(k_cat)" have been replaced.

You should also try closing and restarting Stata and then check if it
works. If all else fails, I will send you my edited version of the
"crtest.ado" file off-list which should work.

T

2010/11/21 Sangamitra Ramachander <sangamitra@gmail.com>:
> Tirthankar,
>
> Thanks very much. I got the ado file from the same source that you
> mention below:
>
> *! version 1.2 (25 November 2003)               Joao Pedro Wagner de Azevedo
>
> I tried what you have suggested: changed all instances of "e(k_cat)"
> to "e(k_out)" on the ado file, but I get the same invalid syntax
> error. I'm not sure if this should be implemented in a specific way
> other than manually, which is what I tried. Thanks again.
>
> Best,
> Sangamitra
>
>
> On Sun, Nov 21, 2010 at 10:22 AM, Tirthankar Chakravarty
> <tirthankar.chakravarty@gmail.com> wrote:
>> Sangamitra,
>>
>> You need to tell the list where you get user-written ado-files from. In my case:
>> . ssc install crtest
>> . which crtest
>> <output omitted>
>> *! version 1.2 (25 November 2003)               Joao Pedro Wagner de Azevedo
>>
>> This routine will not work with a fully updated Stata 10.1 because the
>> scalars returned by -mlogit- have been renamed since Stata 8 (my
>> guess), viz., the number of categories used to be returned in
>> "e(k_cat)", but is now returned in "e(k_out)". So you need to change
>> all instances of the occurrence of "e(k_cat)" in the file "crtest.ado"
>> to "e(k_out)" (on line #s 28, 43, 44 and 81). Then, as in the
>> following example, -crtest- should work fine:
>>
>> *********************************************
>> webuse sysdsn3, clear
>> mlogit insure age male nonwhite site2 site3
>> crtest
>> *********************************************
>>
>> T
>>
>> On Sun, Nov 21, 2010 at 1:42 AM, Sangamitra Ramachander
>> <sangamitra@gmail.com> wrote:
>>>
>>> Dear All,
>>>
>>> I'm trying to use the Cramer Ridder test to see whether I can combine
>>> two of the categories used in a multinomial logit model.
>>>
>>> I have used this test in the past with an old version of stata (Stata
>>> 8). I am now using Stata 10 and have looked at the help and it says
>>> the command is, after the mlogit, 'crtest' - this should tell you
>>> pairwise whether categories are different.
>>>
>>> However, when I do this, I get an invalid syntax error. I am not sure
>>> if this test ceases to work in subsequent versions of stata or if I
>>> should be doing something different. I've updated my version of Stata
>>> 10 and downloaded the ado file for this test.
>>>
>>> I've scanned the archives of this list but wasn't able to find a
>>> similar discussion. Any help would be very much appreciated! Thank you
>>> very much.
>>>
>>> Best,
>>> Sangamitra
>>> *
>>> *   For searches and help try:
>>> *   http://www.stata.com/help.cgi?search
>>> *   http://www.stata.com/support/statalist/faq
>>> *   http://www.ats.ucla.edu/stat/stata/
>>
>>
>>
>> --
>> To every ω-consistent recursive class κ of formulae there correspond
>> recursive class signs r, such that neither v Gen r nor Neg(v Gen r)
>> belongs to Flg(κ) (where v is the free variable of r).
>>
>> *
>> *   For searches and help try:
>> *   http://www.stata.com/help.cgi?search
>> *   http://www.stata.com/support/statalist/faq
>> *   http://www.ats.ucla.edu/stat/stata/
>>
>
> *
> *   For searches and help try:
> *   http://www.stata.com/help.cgi?search
> *   http://www.stata.com/support/statalist/faq
> *   http://www.ats.ucla.edu/stat/stata/
>



-- 
To every ω-consistent recursive class κ of formulae there correspond
recursive class signs r, such that neither v Gen r nor Neg(v Gen r)
belongs to Flg(κ) (where v is the free variable of r).

*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/