I'm interested in performing a beta regression in which the outcome is a proportion bounded between 0 and 1. I will want to interpret the coefficients of the model; however, I cannot find any resource that tells me how to interpret the coefficients. How might I interpret a coefficient from a beta regression such that the interpretation is meaningful and useful to me and the reader?
http://stats.stackexchange.com/questions/63350/how-to-interpret-the-coefficients-from-a-beta-regression
I found the link from UCLA (see below) very helpful. It directs you to an appropriate statistical analysis based on the nature of your dependent variable as well as independent variable. Moreover, the resource also provides codes for several commonly used statistical package and interpretation of the results, such as regression coeffients.
http://stats.idre.ucla.edu/other/mult-pkg/whatstat/
Jayden,
For logistic/logit models, the coefficient associated with a variable indicates the change in log-odds of the target outcome ("success," "retention," "survival," etc.) per unit change in the independent variable (IV). If you exponentiate the coefficient, that converts the result to the change in odds of the target variable per unit change in the IV.
Example: If mother's age (IV) as a predictor of whether mother will or will not breast feed a newborn child (DV: Yes or No) yields a regression coefficient of 0.157, that tells you that (for this sample):
Each additional year of mother's age increases:
1. The estimated log-odds that mother will breast feed by 0.157.
2. The odds that mother will breast feed by exp(0.157) = 1.170 times...which is to say that the estimated odds of mother breast feeding increase by 17%.
If you have multiple IVs, then you need to qualify the statements by adding, "...while holding all other IVs constant."
Good luck with your work.
- Badges
- Science topic
- Similar topics
- Mathematics
- Statistics
- Regression