I have done some analysis in ordinal regression, but I am un sure about whitch I should use acending or decending, becouse this turns around the sing of the coëfficient.
I would like to have ordinal regression 2 outcomes, but I do not know if this is valid.
Kind regards,
Jochem
Jochem Groot Jebbink, you may benefit from the following link although the output is from different software:
https://stats.idre.ucla.edu/stata/dae/ordered-logistic-regression/
Jochem Groot Jebbink , I think your ordinal regression response a variable (ASQ) with four level 0, 1, 2, and 3. Is that right? Let see that two figure above yield same result, but in different sign, since the "reference category" may be different.
You can interpret the effect of independent variables by examine the hypothesis test, which similar as "t-test for beta" in linear regression, as well as the confidence interval. Recall that the "null value" is 1, which means that the coefficient is not significant if 1 is in the confidence interval.
For the ASQ, the sig. test whether the respective order/level is differ from the reference category. In the first figure, the ASQ = 3 is not present, means that it set as the reference category. While in the second figure, the ASQ = 0 was set as the reference category.
Ezra Putranda Setiawan so when I want to depict a positive increase in ASQ as ordinal variable from 0 to 3, I should have 0 as my baseline. Do I not?
What do the negative values from the thresholds mean?
- Badges
- Science topic
- Similar topics
- Mathematics
- Statistics
- Regression