Session Ready
Exercise

Multiple logistic regression

Linear regression is appropriate for continuous outcomes. For binary outcomes, we can use logistic regression. Here we will use the Acupuncture study data to analyze the binary response variable to assess whether a patient's decrease in headache score from baseline exceeded 35%. We will use multiple logistic regression to include multiple explanatory variables. The Acupuncture dataset has been preloaded for this session.

Instructions
100 XP
  • Use the relevel() function to set "Control" as the reference for treatment.group and set "less than or eq to 35%" as the reference for resp35.month12.
  • Use the glm() function to fit a multiple logistic regression model for resp35.month12 with treatment.group, sex and score.baseline.4 as covariates.
  • Display a summary of the model.
  • Display the treatment effect and its 95% CI as odds ratios by taking the exponentials.