Despite the ordinal nature of Self-Rated Health (SRH) variable, logistic regression models or regression models have been used without adequate justification for these applications. It is shown that ordered-logit regression model is the appropriate statistical strategy to estimate SRH, whereas the Ordinary LeastSquares model leads to biased conclusions.
"Ordered Logit Regression Modeling of the Self-Rated Health in Hawai‘i, With Comparisons to the OLS Model,"
Journal of Modern Applied Statistical Methods:
2, Article 23.
Available at: http://digitalcommons.wayne.edu/jmasm/vol12/iss2/23