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 at the end of May, and its replacement, statalist.org is already up and running.


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

Re: st: collecting interaction terms after estimation


From   Maria Ana Vitorino <vitorino@wharton.upenn.edu>
To   <statalist@hsphsun2.harvard.edu>
Subject   Re: st: collecting interaction terms after estimation
Date   Sat, 5 Nov 2011 13:34:12 -0400

Dear Richard,
Thanks for your suggestions.
I'm actually using asclogit and it seems that factor variables and time series operators are not allowed...
Ana


On Nov 5, 2011, at 2:20 PM, Richard Williams wrote:

At 11:48 AM 11/5/2011, Maria Ana Vitorino wrote:
Dear statalist users,
Suppose I have the following model in which I regress people's
decision to buy or not to buy something on people's characteristics
interacted with distance.
(the Xs refer to interactions between distance to the store and
other people's characteristics).

logit choice dist dist2 ageXdist sexXdist incXdist

If you have Stata 11 or 12, you are making this too hard. Use factor variable notation instead.

clonevar xdist = dist
logit choice i.sex xdist age inc c.xdist#c.xdist c.age#c.xdist i.sex#c.xdist c.inc#c.xdist

Follow that up with

replace xdist = 10 if !missing(xdist)
predict p1, p

Change the last line to

predict p1, p if e(sample)

if you only want to use the same cases that were used in the estimation.

Then,

drop xdist
clonevar xdist = dist
[repeat process with remaining models; call predictions p2, p3, or whatever)

Note that you run the logistic regressions using the original values of distance. You then set distance = 10 and do your predictions. By using the factor variable notation, Stata will understand how the value of distance affects both the squared term and the interaction terms.

Also note that in your original example you did not have the main effects of sex, age, income or inc. I have added them, but if for some reason you really really really don't want them you should drop the main effects from the model.

If you are condemned to using Stata 10 or earlier, you will need to tweak the approach so that you run the regression, and then recompute all the terms setting distance to 10 and then doing the prediction.


And I would like to obtain predicted probabilities for each person
when distance is set to 10.

This would be easy enough if I only had this model specification. But
I have several possible models (and may keep adding and changing the
current ones).

This is what I have so far:


local chars "sex age inc"
foreach varname in `chars'{
       gen distX`varname' = dist*`varname'
}
gen dist2= dist*dist

*model 1
logit choice dist dist2 ageXdist sexXdist incXdist

*model 2
logit choice dist dist2  sexXdist incXdist

*model 3
logit choice dist ageXdist  incXdist

My question is then: Is it possible to write code general enough that
would calculate predicted probabilities for when distance is set to 10
for *any* of the above models?


Any insights are appreciated.
Thanks,
Ana
*
*   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/

-------------------------------------------
Richard Williams, Notre Dame Dept of Sociology
OFFICE: (574)631-6668, (574)631-6463
HOME:   (574)289-5227
EMAIL:  Richard.A.Williams.5@ND.Edu
WWW:    http://www.nd.edu/~rwilliam

*
*   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