Notice: On April 23, 2014, Statalist moved from an email list to a forum, based at statalist.org.
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: st: poisson regress probabilities
From
Austin Nichols <[email protected]>
To
[email protected]
Subject
Re: st: poisson regress probabilities
Date
Fri, 16 Apr 2010 14:57:15 -0400
Wolney Conde <[email protected]> :
ps.
You might or might not find this kind of comparison useful:
sysuse auto, clear
qui reg foreign mpg, r
predict p1
la var p1 "Linear Prob Model"
qui logit foreign mpg, r
predict p2
la var p2 "Logit Model"
qui poisson foreign mpg, r
predict p3
la var p3 "Poisson Model"
locpr foreign mpg, plot(||sc p1 p2 p3 mpg)
(-locpr- is on SSC)
On Fri, Apr 16, 2010 at 2:40 PM, Austin Nichols <[email protected]> wrote:
> Wolney Conde <[email protected]> :
> On the contrary, Stata does allow you to use -predict-, but of course
> using -poisson- to predict probabilities can result in predicted
> probabilities in excess of 1:
>
> sysuse auto, clear
> qui logit foreign mpg, r
> predict p1
> qui poisson foreign mpg, r
> predict p2
> sc p2 p1
>
> On Fri, Apr 16, 2010 at 2:31 PM, Wolney Conde <[email protected]> wrote:
>> Dear listers,
>> I'd like to express my output Poisson regression as probabilities.
>> Stata doesn't allow to this using <predict> command instead of after
>> <logit> command for example. After logit regression I can get
>> probabilities directly using <p> option in <predict> command or
>> calculating by exp(xb)/(1+exp(xb)). Can I do the same to Poisson
>> regress?
>> I'm afraid it's a silly question once probability distribution
>> estimation is different for Poisson and binomial distribution but I'd
>> like to some advice.
>> thanks in advanced,
>> Wolney.
>
*
* 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/