Hi everyone,
I have a parameter estimation table (ordinal logistic regression) obtained from SPSS. How could convert estimated parameters table into equations (i.e. y1=b0+b1x1+b2x2, ...., y2=b0+b1x1, .....)?
I attached my table for more information
Hello dear brother, you can simply follow the attached pic,
Best wishes
Bo is the estimate given for threshold when all the X's are zero ; there is one for Y = 0 and Y =1 . As this is an ordinal model, while the intercepts are allowed to be different ( that is two B0s) the slopes are constrained to be the same irrespective of the outcome (proportional odds assumption).The slope terms are for continuous variables (eg 0.002 for X7) and for each categorical variable, one category is dropped out and is the base or referrent category. So X2 =1 is 0.75 higher than the base category X2 = 2. The big effect is for X1. Your threshold intercepts are estimated with a large uncertainty, you may want to recode the dummies or center the continuous variables to get more meaningful values. I would strongly recommend doing a set of meaningful predictions on the probability scale.
Have a look at
Ordinal responses: multiple categories ordered
http://www.bristol.ac.uk/cmm/software/mlwin/mlwin-resources.html#discrete
I'm agree with professor Kelvyn Jones about uncertainty of thresholds (it's so clear through SE and Sig)
For equation reporting all ordered levels constrained to take the same equation, the difference is in estimating of appartenance probability. As showed in my previous commitment
- Badges
- Science topic
- Similar topics
- Mathematical Sciences
- Statistics
- Statistical Software
- SPSS