HAD5307H

Introductory R helpful but not required.

This course gives students in clinical epidemiology and health systems research the knowledge and skills to apply core statistical methods to health and health-care data. Students gain hands-on experience analysing datasets, interpreting findings, and communicating results in writing and in presentations to support future research and publication.

For each method, the course emphasizes both what it is and how to do it, with a consistent focus on estimation and uncertainty, i.e. reporting effect sizes and confidence intervals, and interpreting p-values correctly, rather than significance testing alone. Topics include data types and measurement; exploratory analysis and data visualization; summarizing data and quantifying uncertainty; comparing groups on means, proportions, and rates using parametric and nonparametric methods, with measures of association; sample size and statistical power; analysis of variance within the linear-model framework; simple and multiple linear regression, including model assumptions, diagnostics, and flexible modelling of continuous predictors; logistic regression; and an introduction to survival analysis. The course also introduces multivariable model building, distinguishing analyses aimed at description, prediction, and causal explanation. Computing is integral: all analysis is carried out in R using a reproducible workflow introduced from the first session. Prior statistical software experience is recommended but not required.

1. Classify data by type and measurement scale and produce appropriate exploratory summaries and visualizations.

2. Summarize data and quantify uncertainty using confidence intervals and interpret p-values and their limitations correctly.

3. Compare two or more groups using parametric and nonparametric methods for means, proportions, and rates, reporting effect sizes with confidence intervals.

4. Calculate sample size and statistical power for common study designs.

5. Fit, interpret, and check simple and multiple linear regression models, and distinguish analyses aimed at description, prediction, and causal explanation.

6. Fit and interpret logistic regression models for binary outcomes.

7. Apply introductory survival analysis, including Kaplan–Meier curves, the log-rank test, and interpretation of the Cox proportional hazards model.

8. Carry out analyses in R using a reproducible workflow and communicate results clearly in writing and in presentations.