I am currently doing an ordinal regression for my dissertation. However, although my tests are statistically significant my parameter estimates are negative.
I have attached a photo below. My scale is 1-5 (Very Unlikely, Unlikely, Neutral, Likely, Very Likely)
Could anybody offer assistance?
Hello Gabrielle,
Generally, the parameters of interest involve levels of the IV, which appears to be the variable, SustainablePractices ("SP") in your given output. By default, the highest level (=5) has been set as the reference category. Relative to this, you could conclude that (as an example):
For cases falling in the lowest level of the SP variable (=1), the odds of ending up in a higher level of the Attitudes variable decline to 1 - exp(-5.486) = 99.6% lower (or less) compared to those in the reference category (SP=5) [e.g., odds of going to higher attitude value for cases in SP = 1 are .004 of what the odds are for cases in SP = 5].
Had your base/reference category been set at SP = 1, the other parameter estimates (e.g., for SP = 2, SP = 3...) would have been positive.
Here's a helpful overview of spss ordered logistic regression output: https://stats.oarc.ucla.edu/spss/dae/ordinal-logistic-regression/
Good luck with your work.
- Badges
- Science topic
- Similar topics
- Mathematics
- Statistics
- Regression