Bookmark and Share

Notice: On March 31, it was announced that Statalist is moving from an email list to a forum. The old list will shut down on April 23, and its replacement, statalist.org is already up and running.


[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: st: Fractional logit model interaction effect


From   Richard Williams <richardwilliams.ndu@gmail.com>
To   "statalist@hsphsun2.harvard.edu" <statalist@hsphsun2.harvard.edu>
Subject   Re: st: Fractional logit model interaction effect
Date   Mon, 24 Dec 2012 11:23:51 -0500

I never understand the requests to get marginal effects of interaction
terms. If you have x1, x2, and x1*x2, then x1*x2 can't change unless
x1 and/or x2 changes. If you run a model with these terms in Stata
using factor variable notation, you can get marginal effects for x1
and x2 but not for their interaction term. Which, to me, is the way
things should be. Am I missing something?

Sent from my iPad

On Dec 24, 2012, at 3:59 AM, Ebru Ozturk <ebru_0512@hotmail.com> wrote:

> Dependent variable is bounded between 0 and 1. I wanted to use -glm with -logit. But my concern is that in logit models we have some difficulties in order to get marginal effects of interaction terms. In fractional logit models do we have the same problem? or do we just need to add the interaction terms and interpret the results as they are in the outcome table?
>
> Kind regards
> Ebru
>
> ----------------------------------------
>> Date: Fri, 21 Dec 2012 08:59:43 -0500
>> Subject: Re: st: Fractional logit model interaction effect
>> From: jvverkuilen@gmail.com
>> To: statalist@hsphsun2.harvard.edu
>>
>> On Fri, Dec 21, 2012 at 8:09 AM, Ebru Ozturk <ebru_0512@hotmail.com> wrote:
>>>
>>> Dear All,
>>>
>>> How do we test interaction effect in fractional logit model?
>>
>> This question is not well-formed. It's very important to specify what
>> procedure you are using and what you mean. If it's the -fraclogit-
>> procedure that can be downloaded (-findit fraclogit-) the answer is
>> that you specify the interactions in the usual way (i.e., with # and
>> ## between terms you want to interact) and can get a Wald test via
>> -testparm- in the usual way. -lrtest- does not appear to work directly
>> because this is a quasi-likelihood model and the log-likelihood term
>> is not saved.
>> *
>> * For searches and help try:
>> * http://www.stata.com/help.cgi?search
>> * http://www.stata.com/support/faqs/resources/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/faqs/resources/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/faqs/resources/statalist-faq/
*   http://www.ats.ucla.edu/stat/stata/


© Copyright 1996–2014 StataCorp LP   |   Terms of use   |   Privacy   |   Contact us   |   Site index