Re: st: Interpretation of interaction term in nonlinear models

[Maarten Buis <maartenlbuis@gmail.com>]

It often helps to include the reference category for one of the
categorical variables:

M.L. Buis (2012) "Stata tip 106: With or without reference", The Stata
Journal, 12(1), pp. 162-164.


Hope this helps,
Maarten

On Thu, May 30, 2013 at 4:52 PM, Suryadipta Roy <sroy2138@gmail.com> wrote:
> Dear Maarten and fellow Statalisters,
>
> I actually had a related question as to whether there might be a
> similar (one-sentence) interpretation in case of a three-way
> interaction between the same categorical variable, the continuous
> variable with another categorical variable (0/1). Of course, I can and
> have used -margins - with -marginsplot- to show how the interaction
> effects differ in the presence of this categorical variable, but I was
> wondering if I could get some help with an easier interpretation. The
> coefficient of the three-way interaction term (standardized continuous
> by categorical by categorical) in the fixed effects Poisson regression
> with the -irr- option is 0.74 and in the probability metric form is
> -0.21. Once again, thank you very much for the help!
>
> Sincerely,
> Suryadipta.
>
> On Thu, May 30, 2013 at 7:53 AM, Suryadipta Roy <sroy2138@gmail.com> wrote:
>> Maarten,
>> This is very helpful, thank you very much! For some reason, I thought
>> that the z_phd variable in your example is a categorical variable as
>> well when I had previously read the thread. I guess that I would need
>> to standardize my continuous variable for a similar interpretation.
>>
>> Sincerely,
>> Suryadipta.
>>
>> On Thu, May 30, 2013 at 7:19 AM, Suryadipta Roy <sroy2138@gmail.com> wrote:
>>> Dear Statalisters,
>>>
>>> I am studying the effect of an interaction between a categorical
>>> variable (0/1) and a continuous variable (0-6) on the dependent
>>> variable in a nonlinear model (using -xtpoisson-). The value of the
>>> coefficient using the -irr- option is 0.90, while the size of the
>>> interaction term in the probability metric form is, of course, -0.11
>>> (exp(-0.11) = 0.90). My basic question is, if it might be possible to
>>> have a one-sentence interpretation of the value of the coefficient in
>>> the multiplicative form (0.90), e.g. something in the lines of "the
>>> effect of the categorical variable increases by a factor of 0.9 (i.e.
>>> a 10% reduction of the dependent variable) due to an 1% increase in
>>> the continuous variable"? Any suggestion in this regard will be highly
>>> appreciated.
>>>
>>> Sincerely,
>>> Suryadipta Roy.
>>> *
>>> *   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/



-- 
---------------------------------
Maarten L. Buis
WZB
Reichpietschufer 50
10785 Berlin
Germany

http://www.maartenbuis.nl
---------------------------------
*
*   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/