Binary logistic regression is a well-known and widely used regression technique in psychology and health sciences. This technique allows the introduction of all types of regressors and is very flexible in terms of assumptions; it requires a random sample of n participants evaluated on k predictor variables that show low collinearity and on a dichotomous qualitative variable that is the predicted variable. A practice that is found with relative frequency in this field of research is to dichotomize a quantitative variable (by the cut-off point to define the case) to apply logistic regression and thus take advantage of the usefulness of the logistic regression. However, it is not a recommended procedure, since a lot of information (variance) of the predicted variable is lost, when there is a much better alternative, namely, quantile regression. This is a little-known and rarely used regression technique in psychology. It requires a quantitative variable as a predicted variable, accepts all kinds of predictor variables, and is free from the restrictive assumptions of ordinary least squares linear regression. This methodological article aims to present quantile regression in its theoretical aspects and shows an example applied to the area of health psychology to promote its knowledge and use.
Key words: Quantile regression, logistic regression, multiple linear regression, multivariate statistics, psychology.
Copyright © 2023 Author(s) retain the copyright of this article.
This article is published under the terms of the Creative Commons Attribution License 4.0