Longitudinal data and mixed-effects models

Measure the same person twice and the rows stop being independent. The model has to say so.

Taught by

Michael Crowther, PhD

CEO & Director of Statistical Methodology

Formerly Associate Professor of Biostatistics at the University of Leicester, and the author of the merlin framework.

Research survival analysis · joint models · multi-state models · statistical software

PhD in Medical Statistics, University of Leicester, 2014

Register your interest

What you will be able to do

• Recognise longitudinal data, lay it out correctly, and describe it before modelling it
• Say what a regression that ignores the clustering gets wrong, and in which direction for which coefficient
• Fit random-intercept and random-slope models with merlin, and choose the covariance structure and the functional form of time on evidence
• Predict at population and at subject level, and say which of the two answers the question asked

Curriculum

15 steps: 12 lectures and 3 practicals.

Provisional — this course is still being written, and the outline will change.

Week 1 · what longitudinal data is, and how to look at it

• 01WatchWhat makes data longitudinal22 min
• 02WatchWide, long, and the reshape between them20 min
• 03WatchDescribing repeated measurements24 min
• 04PracticalLab I · getting to know the data40 min

Week 2 · the regression that ignores the clustering

Opens 7 days after you start.

• 05WatchTime, treatment, and their interaction24 min
• 06WatchWhat ignoring the correlation costs26 min
• 07PracticalLab II · the model that ignores the clustering45 min

Week 3 · linear mixed-effects models

Opens 14 days after you start.

• 08WatchThe random intercept model26 min
• 09WatchRandom intercepts and random slopes26 min
• 10WatchChoosing a covariance structure22 min
• 11WatchThe functional form of time26 min
• 12PracticalLab III · adding a random intercept55 min

Week 4 · prediction after a mixed model

Opens 21 days after you start.

• 13WatchPopulation-level predictions24 min
• 14WatchSubject-specific predictions and the BLUPs26 min
• 15ReadResiduals, and which model predicts best22 min