My feeling is: There's no particular reason we can only use Poisson or
NBin on count data. Surely the important thing is that the distribution
matches, right? In Poisson or NBin regression, we express results in terms
of Incidence Rate Ratio, which I guess only makes sense if you're thinking
of events happening. But what about calling it 'mean ratios', as
effectively they are just that?
I have no backing from any reference or anything, but just thinking
logically (I feel), that is what I would conclude. Richard, you don't
agree with using count-type regression techniques on non-count data. Why
is that?
But concerning Matthew's situation, my feeling is: even NegBin or Poisson
may not give a very good approximation of the distribution. (Thanks
Matthew for sending me the references, BTW). Often weird distribution is
as a result of a mixture of distributions, which is why people have come
up with zero-inflated models, so that would be what I would suggest
Matthew to do first, to see whether the data can be modelled separately,
first with 0 vs greater than 0, and then model the 'greater than 0' data
using whatever means.
On a slightly separate note, has anyone considered the use of median
regression for this sort of data? Using bootstrapping with median
regression does not require any assumption of the distribution of the
dependent variable. Apart from the difficulty in convergence, I can't
think of any disadvantage. Does anyone know of any work that has made use
of it (on questionnaire type data)?
Tim
Richard Williams <Richard.A.Williams.5@ND.edu>
Sent by: owner-statalist@hsphsun2.harvard.edu
10/10/2006 06:10
Please respond to
statalist@hsphsun2.harvard.edu
To
statalist@hsphsun2.harvard.edu
cc
Subject
Re: st: NBREG for ordinal scales
At 10:00 PM 10/9/2006, Matthew C. Johnson wrote:
>The measure is a composite scale. One is a four-item property offending
>scale: "How often in the past 12 months did you ..." Response options:
>0-never; 1-once or twice; 2-three or four times; 3-five or more times.
The
>other scale is a five-item violent offending scale with the same format.
I
>have been wary to use a count regression procedure because there is
>nothing in the statistics literature that mentions, one way or the other,
>using negative binomial regression for this type of data. However, there
>are several prominent articles (within the criminological literature)
that
>do this. As a mere grad student, I am a bit confused.
So, you add these items up, and the property offending scale goes
from 0-12 and the violent offending scale goes from 0-15? Given the
coding of these items, it seems shaky to use count models. If I was
bound and determined to use count models, I think I would at least
recode the items first, so they were coded 0, 1.5, 3.5, and 5, or
something like that, so they would at least be my best guess as to
what the count actually was. The articles just do this stuff, with
no justification offered whatsoever, no seminal source cited??? As a
mere Sociology statistics professor, I am a bit confused. :)
-------------------------------------------
Richard Williams, Notre Dame Dept of Sociology
OFFICE: (574)631-6668, (574)631-6463
FAX: (574)288-4373
HOME: (574)289-5227
EMAIL: Richard.A.Williams.5@ND.Edu
WWW (personal): http://www.nd.edu/~rwilliam
WWW (department): http://www.nd.edu/~soc
*
* For searches and help try:
* http://www.stata.com/support/faqs/res/findit.html
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/
*
* For searches and help try:
* http://www.stata.com/support/faqs/res/findit.html
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/