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: Overdispersed poisson regression


From   Nick Cox <njcoxstata@gmail.com>
To   statalist@hsphsun2.harvard.edu
Subject   Re: st: Overdispersed poisson regression
Date   Mon, 5 Sep 2011 12:33:07 +0100

Overdispersion means higher variance than expected on some model. With
a Poisson, variance is equal to the mean.

I think you need advice from experts in models for counts (not me).

Nick

On Mon, Sep 5, 2011 at 12:16 PM, Lynsey Patterson
<Lynsey.Patterson@hscni.net> wrote:

> Sorry Nick,
>
> I am new to this list and I got several bounces saying the message had not been posted. I tried to find it to check and thought to be safe I would try again.
>
> I was using the term over dispersed as I know this has been applied to data from other administrations.
>
> Basically, I am trying to look at the trend in MRSA rates - hence the choice of Poisson. I then need to decide which model best fits the data - sorry for the confusion, evidently I'm not a statistician!!
>
Nick Cox

> You posted this earlier. As your variance is much less than the mean,
> why do you call the data overdispersed?
>
> There are several possible reasons why your earlier mail did not get a
> reply, ranging from many people being on vacation to the possibility
> that this is not enough information to provide well-grounded advice on
> modelling. Perhaps many of your zeros belong in a different group.
>
> Nick
>
> On Mon, Sep 5, 2011 at 11:55 AM, Lynsey Patterson
> <Lynsey.Patterson@hscni.net> wrote:
>
>> I am interested in analysing the trend in rates of meticillin resistant S. aureus (MRSA) for my region. I have several variables to include in the model – time (the year (1) and the quarter (2)), the number of episodes (3), the number of occupied bed days (the denominator (4)) and the Trust (a measure of geography (5)).
>>
>> When I summarize the rate I get a mean of 0.127/1000 bed days with a variance of 0.008. So I think a normal poisson model cannot be fitted and I need to use an over dispersed? I have read about a command for negative binomial regression – does anyone know if this is correct and what syntax I should use?
>>
>> E.G. nbreg number of episodes (3) year(1) quarter(2) trust(5), exp(occupied bed days (4))

*
*   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