Re: st: Multinomial Logistic - ordinal variables and Odds Ratios

[Richard Williams <richardwilliams.ndu@gmail.com>]

At 09:33 PM 11/14/2013, Imogen Jones wrote:
> Hi Everybody
>
> I was hoping somebody could shed some light on 
> the differences between using an ordinal and 
> nominal variable in multinomial logistic regression in Stata.
>
> I understand the processes are different, but 
> I?m more concerned about the model fitting.  My 
> interpretation of Hosmer and Lemeshow (2000) was 
> that there isn?t a reliable test to estimate model fit.  Is this still correct?

First off, I am curious as to why you are not 
using an ordered logit (ologit) model. Have you 
already determined that the assumptions of such a 
model are violated with your data.

Second, I am not sure by what you mean when you 
say "the processes are different." If you use an 
ordinal DV in an mlogit model you are basically 
ignoring the ordering of the categories and treating the variable as nominal.

> Also, when using an ordinal variable as a 
> dependent variable in multinomial logistic, are 
> there any assumptions specific to this analysis 
> that need to be met or parameter estimates that may vary from binary logistic?

The assumptions will be the same whether the DV 
is ordinal or categorical with more than 2 
categories. Again, the fact that the variable is 
ordinal is irrelevant since you are ignoring the 
ordered nature of the data when you use mlogit.

> Another question ? Stata only seems to produce a 
> coefficient or relative risk ratio for 
> multinomial logistic.  Is there a specific 
> reason why it won?t produce an odds ratio or is 
> it just not a function of Stata?  I assumed it 
> was the former since Stata is so comprehensive 
> it probably wouldn?t leave something like that out.

No matter what you call it, you are 
exponentiating the coefficient, i.e. it is the 
same mathematical transformation. There was a 
thread some years ago about why Stata used terms 
like relative risk ratios when other programs 
called them odds ratios. I'll see if I can find it.


-------------------------------------------
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/faqs/resources/statalist-faq/
*   http://www.ats.ucla.edu/stat/stata/