Practical, applied statistical training for research teams — built around your data, your software and the questions your group actually needs to answer.
Formats
Designed from scratch around a specific need — a method your team is about to use, a software transition, or a recurring problem in your analyses. Content, examples and datasets are built for your context rather than adapted from a generic course.
Typical length: half day to two days.
Structured multi-session courses covering a topic in depth — for example an introduction to R across four half-day sessions, or a regression course running weekly across a term. Suits departments and doctoral training programmes.
Typical length: three to eight sessions.
Training a whole research group together, so that the team develops shared conventions for analysis, documentation and reporting. Particularly useful when a group is standardising on new software or preparing for a large study.
Typical length: one to three days, often spread out.
Focused individual sessions for a researcher working through a specific analysis — common for doctoral researchers, early career academics and clinicians returning to statistics after a gap.
Typical length: one to two hours per session.
Pre-conference workshops, invited methods seminars and departmental talks on applied statistical topics for a mixed-expertise audience.
Typical length: one to three hours.
A recurring block of time — for example a half day each month — during which members of your organisation can bring statistical questions. An efficient way to give a whole department access to statistical advice.
Typical length: recurring, by arrangement.
Curriculum
The topics below are frequently requested. Any of them can be combined, shortened or extended, and new topics can be developed on request.
Getting started in R and RStudio: projects, importing data, the tidyverse, data wrangling, summarising and plotting — for researchers with no programming background.
Fitting, checking and interpreting models in R, from linear regression through mixed-effects and survival models, with reproducible reporting via Quarto.
Working effectively in SPSS, including using syntax rather than menus so that analyses are documented, repeatable and easy to correct.
Data management, modelling and do-file workflows in Stata, with emphasis on reproducible pipelines for health and epidemiological data.
Linear and logistic regression in practice: model specification, confounding, interaction, diagnostics, and interpreting coefficients and odds ratios correctly.
Analysing clustered and hierarchical data: random intercepts and slopes, variance partitioning, centring decisions and common estimation problems.
Repeated measures in practice: mixed models for repeated measures, GEE, growth curves, and handling dropout and irregular measurement times.
Time-to-event methods: Kaplan–Meier estimation, Cox regression, checking proportional hazards, competing risks and correct interpretation of hazard ratios.
How to calculate and justify a sample size, what assumptions drive it, and how design choices such as clustering change what is required.
Evaluating and developing measurement instruments: factor analysis, reliability, validity evidence and measurement invariance.
Understanding missingness mechanisms, why complete-case analysis is often the wrong default, and how to run and report multiple imputation properly.
Building analyses that can be re-run and audited: project structure, scripted workflows, version control with Git, and literate reporting.
A grounding for clinical and health services teams: choosing the right test or model, reading statistical sections critically, and avoiding common errors.
Propensity score methods, interrupted time series and difference-in-differences — what each assumes and when it can support a causal claim.
Where prediction methods genuinely help, how to validate and calibrate them honestly, and how they compare with well-specified regression.
Process
A short conversation about who is attending, what they already know, what software they use and what they need to be able to do afterwards.
An outline is proposed covering topics, timings and exercises, with realistic expectations about what can be achieved in the time available.
Materials, code and example datasets are built for your context, using scenarios your participants will recognise.
The session runs hands-on, and participants keep all materials. A short follow-up window for questions is included as standard.
Audiences
Tell JH Stat who is attending, what they need to be able to do, and roughly when — and you will receive an outline proposal with topics, timings and cost.