What we can do with logistic regression
Questions we can answer using logistic regression
Ways to report results
Using Stata
Getting ready
Opening the dataset
Exploring the data
Labeling values for categorical variables
Saving the edited dataset
Conventional ordinary least-squares regression versus logistic regression
What OLS regression can tell us
What logistic regression can tell us
Interpreting an odds ratio
What is an odds ratio?
Interpreting ORs as a percentage difference for binary predictors
What is wrong with ordinary least-squares regression for a binary outcome?
Hypothetical data
How does logistic regression fit better than ordinary least-squares linear regression?
Fitting and interpreting logistic regression models
Interpreting coefficients and odds ratios
Fitting logistic regression models with multiple predictors
Interpreting ORs for quantitative predictors
Selecting the right base level for categorical predictors
How well does the model fit the data?
Pseudo-R² measures of fit
Information criteria
Identifying cases that the model fits poorly
Sensitivity and specificity
Criteria for evaluating an analysis
Estimation of sensitivity and specificity
Receiver operating characteristic curves and cutpoints for screening tests
ROC curves
Comparing tests
Predictions using the margins command
What is better than reporting coefficients and odds ratios?
Data preparation
Estimating the ORs
The margins command
Graphic presentation using the marginsplot command
When and why
Graphs of categorical predictors that include three or more categories
Graphs with one quantitative predictor
Graphs with one quantitative and one categorical predictor
Graphs of a pair of categorical predictors
Graphs of a pair of categorical predictors
Curve fitting with quadratic models
A hypothetical example of a quadratic model using OLS regression
Estimating the curve (uncentered predictor)
Centering, collinearity, and nonessential collinearity
Estimating the curve (centered x)
Compare centered and uncentered models
Use of a quadratic with logistic regression
Interaction
Introduction
Interaction of a categorical and a quantitative variable using logistic regression
Estimating and interpreting probabilities (uncentered)
Interaction of categorical variables
Interaction of quantitative variables
Running nestreg and postestimation commands
Nested logistic regression
Selected postestimation commands
Special topics
Collinearity and multicollinearity
Evaluating multicollinearity
Sample size
Small-sample bias
Relative risk
A Appendix
References
Biography
Alan Acock is a University Distinguished Professor Emeritus in the School of Social and Behavioral Health Sciences at Oregon State University. Among other awards, he was an Alumni Distinguished Professor for his teaching. He is the author of A Gentle Introduction to Stata and Discovering Structural Equation Modeling Using Stata. He has authored more than 175 articles in leading journals across many fields, including Structural Equation Modeling, Psychological Bulletin, Multivariate Behavioral Research, American Journal of Public Health, American Sociological Review, Educational and Psychological Measurement, and American Journal of Preventive Medicine.
