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: Count data interaction terms


From   DA Gibson <DAGibson1@sheffield.ac.uk>
To   statalist@hsphsun2.harvard.edu
Subject   Re: st: Count data interaction terms
Date   Tue, 23 Aug 2011 10:40:12 +0100

Hi Maarten

Thanks for that, i do have one further question however. In the
example the IRR on the c.persons#c.child interaction term is positive
at .77 however in your email you say that the interpretation is that
the effect is reduced. Intuitively the extra person being a child
reducing the effect of the number of persons makes sense but im
slightly confused by the sign on the IRR value, could you clarify that
for me please?

Many thanks

Danny

On 23 August 2011 10:20, Maarten Buis <maartenlbuis@gmail.com> wrote:
> On Mon, Aug 22, 2011 at 5:50 PM, DA Gibson wrote:
>> Im running an investigation into how unhealthy lifestyles (obestiy,
>> smoking etc) effect health service usage and ideally id like to use a
>> count analysis to do this. However im not sure whether as with OLS
>> analyses you can include interaction terms e.g. y = B0
>> +B1(X1)+B2(X2)+B3(X1*X2) where X1 and X2 are binary variables. Is this
>> possible using count data?
>
> Yes, the tricky part is interpretation, but that is not a big deal as
> long as you interpret the incidence rate ratios rather than marginal
> effects. Consider the example below. The baseline tells you that a
> visitor to the park without camper, who visited the park alone and has
> no kid with him/her is expected to catch .39 fish per visit(*). An
> extra person added to the group increases that rate by a factor 3.07,
> i.e. (3.07-1)*100%=207 % (the main effect of person). If this extra
> person happens to a child, this effect of number of persons decreases
> by a factor .78, i.e. -22%.
>
> *------------------ begin example -----------------
> webuse fish, clear
> gen byte baseline = 1
> poisson count i.camper c.persons##c.child baseline, ///
>        irr nocons
> *----------------- end example --------------------
> (For more on examples I sent to the Statalist see:
> http://www.maartenbuis.nl/example_faq )
>
> Also see: M.L. Buis (2010) "Stata tip 87: Interpretation of
> interactions in non-linear models", The Stata Journal, 10(2), pp.
> 305-308.
>
> Hope this helps,
> Maarten
>
> (*) I still have Stata 11, so I need this trick with a "variable"
> baseline which is secretly a constant. I understand that in Stata 12
> the display of the baseline rates, odds, hazards, etc. are no longer
> suppressed, so there this trick is no longer necessary.
>
> --------------------------
> Maarten L. Buis
> Institut fuer Soziologie
> Universitaet Tuebingen
> Wilhelmstrasse 36
> 72074 Tuebingen
> Germany
>
>
> http://www.maartenbuis.nl
> --------------------------
> *
> *   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/


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