Analysis of Categorical Data
To be scheduled
This course deals with the analysis of categorical data, with emphasis on binary outcomes and commonly used methods. The course covers the following topics; not all in the same depth:
- a brief review of the basic concepts of inference
- the distinction between categorical and numerical data
- visual representations for categorical data
- binomial distribution
- approximate inference for binomial proportions: one proportion, and the difference between two independent proportions
- idea of exact confidence intervals and application to one proportion
- Pearson's x2 test
- Fisher’s exact test
- alternative ways of comparing proportions: risk ratio and odds ratio, plus approximate inferences
- exact inference for an odds ratio
- concept of maximum likelihood
- logistic regression
- ordinal logistic regression for an ordinal outcome
- nominal logistic regression
- multi-level modelling for proportions
The presenter is Professor Ian Gordon, the Director of the Statistical Consulting Centre. Ian has had extensive experience over two decades in the practical application of these methods and has delivered many statistics courses to participants coming from a wide variety of backgrounds.
The four days are deliberately arranged so that there is a weekend break during the course. Each day will consist of four approximately equal-length sessions; the first session of the day will commence at 9:15 a.m. and the final session will end at approximately 4:45 p.m. The sessions will mix lecture presentations with practical work; tutorial help will be available.
All participants have access to a PC. The statistical package MINITAB will be used in this course; the notes will also provide information about the application of the methods in SPSS. Most, but not all, of the methods covered in the course can be implemented in Minitab or SPSS.
Participants will need to be comfortable with mathematical notation and reasoning, and have used statistical methods at a basic level; the course is not suitable for someone who has not studied statistics at an introductory level. For example, participants should know about hypothesis tests and confidence intervals. The course “Statistics for Research Workers” would be suitable preparation.
Who should take this course?
The course is suitable for researchers who need to analyse data that are categorical in nature, especially binary data. Models for such data are required in every discipline. This course is open to the public.
Cost and enrolment details:
The cost of the course is $1210 (incl. $110 GST).
We have a discounted rate for University of Melbourne postgraduate students of $990 (incl. $90 GST, GST does not apply if paying through your University school.)
$30 Cancellation fee applies
The fee includes a comprehensive set of notes, and morning and afternoon tea. Lunches are not provide.
School of Mathematics and statistics
Peter Hall building
Wilson computer lab
T: +61 3 8344 6995
$30 cancellation fee applies