Re: st: Margins error

[Amy Hsin <hsin.amy@gmail.com>]

Issa,

I have essentially abandoned the margins command and gone back to
using the adjust command.  I am trying to calculate predicted values
so the adjust command is sufficient for my purposes.  What it lacks is
the ability to support factors (i.e. ## functions) but after several
days and no luck, I went the brute force approach and generated all my
interactions by manually.

If what you wish to accomplish can be done by adjust, I would suggest
that you use it instead. I found the STATA help for adjust to be more
useful.  Otherwise, perhaps other who are more advanced programmers
than I can offer you more help.

Good luck,
Amy



On Mon, Jul 2, 2012 at 2:58 PM, Issa Kohler-Hausmann <ikh210@nyu.edu> wrote:
> Thanks Amy, I am suddenly having luck with the following syntax but I am now trying to figure out conceptually what the right one would be to use:
>
> margins r.arrestnum_cat@p_misdcon_cat, at(arr_re_combi=2 male =1 p_felcon=0 MisdArr=1)
>
> margins arrestnum_cat, at(p_misdcon_cat=(0 1)) at(arr_re_combi=2 male =1 p_felcon=0 MisdArr=1)
>
> Are you having any luck? I guess I am still confused about what the different forms mean.
>
> Thanks much,
>  - issa
>
>
> On Jul 1, 2012, at 12:16 PM, Amy Hsin wrote:
>
>> You may want to read the thread "generating predicted values for
>> growth models" that was posted yesterday.  I had similar problems with
>> the margins command using xi.
>>
>> Maarten and William's suggest was to drop the xi (see the response for
>> details). That worked to a degree.
>>
>> However, I am still having problems with the margins command to
>> generate predictions and am hoping to receive some help.
>>
>> Amy
>>
>> On Sun, Jul 1, 2012 at 5:39 PM, Issa Kohler-Hausmann <ikh210@nyu.edu> wrote:
>>> Hello -
>>>
>>> I am sorry if this is a dumb question but I have having trouble with the "margins" command after running a logistic regression and I cannot figure out what I am misunderstanding.
>>>
>>> I am trying to graph the effect of the number of misdemeanor convictions (p_misdcon_cat) on the likelihood of conviction to a misdemeanor crime (outcome --> convict_Misd). The problem (it seems to me) is that STATA does not recognize the variables for purposes of the margins command that were generated using xi. logit and I don't know any other way of doing this. It also seems from all of the examples online that this should work.
>>>
>>> I ran the following logic model:
>>>
>>> . xi: logit convict_Misd i.race_eth_cat male p_felcon i.p_misdcon_cat arrestnum_cat MisdArr, cluster(my_case_id)
>>>
>>> I then get the following errors when I try the margins command evaluated at specified levels of my covariates.
>>>
>>> . margins, dydx(p_misdcon_cat) at (_Irace_eth__2=1 male=1 p_felcon=0 arrestnum_cat=4 MisdArr=1)
>>> 'p_misdcon_cat' not found in list of covariates
>>> r(322);
>>>
>>> .
>>> . margins p_misdcon_cat, at (_Irace_eth__2=1 male=1 p_felcon=0 arrestnum_cat=4 MisdArr=1)
>>> 'p_misdcon_cat' not found in list of covariates
>>> r(322);
>>>
>>> Any help would be greatly appreciated!
>>>
>>> Thank you
>>>
>>>
>>>
>>> *
>>> *   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/
>>
>>
>>
>> --
>> Amy Hsin
>> Assistant Professor of Sociology
>> Queens College, City University of New York
>> http://qcpages.qc.cuny.edu/~ahsin
>>
>> *
>> *   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/



-- 
Amy Hsin
Assistant Professor of Sociology
Queens College, City University of New York
http://qcpages.qc.cuny.edu/~ahsin

*
*   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/