how do I get the intercept in PLUM ordinal regression?

#1
I am running a ordinal regression model with spss plum. I would like SPSS to print me the intercept, but somehow I can't find the synax command for it. Anyone familiar with the problem?

Thanks in advance!
 
#4
Yes. You need those to figure out the predicted probabilities for each outcome category.

Scott Long has a really good book on regression for categorical outcomes, including ordinal regression. You may be able to get it in your library.

I also did a webinar on logistic regression, including ordinal and nominal. It was a while ago, but you can download the recording for free: http://www.theanalysisfactor.com/binary-ordinal-multinomial-logistic/.

Karen
 
#5
I see now: probability for highest category or lower=1, so basically just convert and substract thresholds for all the other categories...thanks that was helpful! I had only seen models with 1 constant/intercept so far and got pretty confused.
Btw:i just subscribed to your webinars!
 
#6
Yes, if you're trying to compute predicted probabilities for each category, just remember that the equation is giving you cumulative probabilities, so as you surmised, you'll have to subtract.

Also, be careful b/c SPSS uses a strange formula (it makes all betas but the intercept negative). It makes the results more intuitive, but can mess up predicted probabilities if you're not careful.

http://www.theanalysisfactor.com/ordinal-logistic-regression-mystery/

And I look forward to seeing you on the webinars! :)

Karen