Predicting Binary Outcomes using Logistic Regression (QCIF) [webinar]

This is a QCIF course and places are highly limited as Graduate School is paying the fee. If you are unable to attend, please de-register (or email graduateschool@uq.edu.au if you are unable to do so). An unexplained absence could result in all your future QCIF registrations becoming Waitlisted. You may only attend the workshop once. Repeat bookings will be removed and placed on the Waitlist.

This practical half-day workshop will help participants to develop an understanding of the principles, methods, and interpretation of logistic regression, a statistical technique to generate binary (yes/no) outcomes from complex input data.

Recommended Participants

Researchers who are dealing with binary outcome data, and wish to understand how to analyse them effectively. The workshop is relevant for all disciplines, although examples and exercises will focus on biological datasets.

Prior expertise with R and the command line interface is required to a level equivalent to that provided by the R for Reproducible Scientific Analysis workshop, as the basics of R will not be covered. Participants are also expected to have a basic familiarity with the concepts of statistical hypothesis testing and regression analysis.

Syllabus

• Introduction to the analysis of categorical variables
• The principles of logistic regression
• Performing univariate and multivariate logistic regression in R
• Assessing the fit of a logistic regression model
• Assumptions, errors, and what can go wrong in logistic regression

Learning Objectives

• Recognise datasets suitable for logistic regression and formulate appropriate research questions
• Understand the principles of logistic regression methods
• Carry out logistic regression analysis using R
• Interpret and evaluate logistic regression output from R