0

I runned several ordered logistic regression using the polr function from the MASS package and interpreted the odds ratios for each model. However, I'm in doubt how to use the intercepts/tau-cuts from the regression summary output. Can I use them to say that the distance between each category is far or close to the higher order? I have 7 independent variables and each variable have many categories, so I don't think it make sense to calculate each value using the intercepts and estimate coeff and insert them in the regression model to get values (hope this make sense).

rr19
  • 65
  • You can try to do the division analysis I recommend here, though that makes more sense for continuous variables. – dimitriy Dec 30 '22 at 00:32
  • @dimitriy thanks for your comment! Sorry I didn't specified in my post but my dependent variable is ordinal discrete scaled from 0 to 10, that's why I used ordered logistic regression. – rr19 Dec 30 '22 at 00:39
  • 1
    I understood that. I am talking about continuous covariates and the cutpoints, not the outcome. – dimitriy Dec 30 '22 at 02:03
  • Oh okay thanks, I will have a look at it. – rr19 Dec 30 '22 at 02:41
  • @dimitriy did I understood correctly, that you mean to do binary logistic regression? Actually I don’t know if it would be a good idea because I have in all 10 ordered logistic regressions and the ordered dependent variable has 11 categories 0-10.. so it will be too many binary logistic regressions. – rr19 Dec 30 '22 at 07:43
  • 1
    I do not. Here is what I have in mind. Take the differences of adjacent cut values and divide by the slope coefficient of covariate x. There are estimated by your ordered logit model. This tells you the max change in x necessary to move out to the next bin. – dimitriy Dec 30 '22 at 15:11

0 Answers0