This course, delivered online through Wattle, focuses on models using two or more variables. The aim of the course is to introduce participants to the most common modelling techniques and to raise awareness of some problems that can arise when using them. The emphasis is on concepts and the use of formulae has been minimised. There are practice exercises to support the concepts; there are also examples of analyses of data sets with an emphasis on interpretation of output from statistical packages.
The course content
- Evaluate association between two numerical variables using plots and correlation coefficients;
- Describe and quantify the relationship between two variables using simple linear regression;
- Recognise when a relationship is nonlinear and apply an appropriate transformation;
- Describe, detect and present relationships between two categorical variables;
- Assess the independence of two categorical variables;
- Use multiple linear regression to model the effect of several variables on a numerical response variable;
- Use logistic regression to model the effect of several variables on a categorical response variable;
- Avoid reaching the wrong conclusions from not including enough variables in the analysis.
- On completion of this course participants will be able to:
- Organise their data into a form suitable for analysis using a statistics package;
- Specify a standard multiple linear regression model or logistic regression model in a statistics package;
- Interpret the output from a statistics package used for fitting linear or logistic regression models;
- Examine diagnostics and other indicators to determine whether the fitted model is appropriate;
- Judge when they need to seek specialised help from a statistical consultant and conduct a meaningful discussion about the analysis of their data.
Knowledge of the material taught in Introductory Statistics Online is assumed.