Logistic Regression

Written by Scott Martin
Bookmark and Share

Logistic regression, or logistic discriminant analysis, is a variation of standard regression that is used when the dependent variable is a dichotomy (that is, yes or no). Moreover, like regression analysis, the predictor variables can be either continuous or dichotomous. It is one of the generalized linear models, which includes ANOVA, ordinary regression, as well as loglinear regression.

Logistic regression is used to predict the category of outcome. A model is created using predictor variables related to the outcome. In a stepwise regression, the fit of the model is tested after each addition or deletion of a coefficient.

Use of Logistic Regression

This analytical tool is used in order to determine whether a result has occurred or not. Additionally, it can provide a comparison of the relationships and strengths among the variables that are used in the analysis. For example, how is scoring in the upper quintile of the math SAT influenced by a student's grade in algebra and his/her family income?

Logistic regression can be calculated using a variety of statistical software packages such as SPSS, SAS, or STATA. If you are unsure how to conduct the analysis, a statistical consultant can provide a tutorial for you or conduct the analysis herself. Furthermore, a consultant can help you interpret the odds ratios that are calculated.

Bookmark and Share