I have a question regarding the interpretation of odds ratios. Take a case where I have a logistic regression model with a continuous IV and a binary DV. My understanding is that the odds ratio I receive in Mplus reflects the average change in the DV for a unit change in the IV.
Is there anyway to obtain an odds ratio that reflects the average change in the DV for a standard deviation change in the IV? My independent variable is individuals’ total score on a scale that measures a theoretical construct. Therefore a unit change in my IV is not intrinsically meaningful as it would be with other continuous IVs (e.g., age). If I can not obtain or calculate an odds ratio to serve this purpose, how would you recommend presenting the results from my model?
The log odds (logistic regression coefficient) reflects the average change in the DV for a unit change in the IV. You can multiply the logs odds by the standard deviation of x and then exponentiate it. I think that will give you what you want.
Yusef K posted on Saturday, October 10, 2009 - 7:06 am
Jim: You're absolutely correct when you say that the odds ration in Mplus will reflect the average amount of change in the DV of 1 unit of change in IV.
As you had pointed out, however, it is not entirely clear as to whether or not the odds ration reflects the average change in DV.
Dr. Muthen...could you please elaborate on the answer you had listed above? It would be greatly appreciated.