Practical workshops for researchers
Audience: University researchers, research institutes, funded research teams, and higher-education departments, including PIs and ECRs, PhD cohorts, and research-development teams.
I design and deliver focused workshops that help researchers make stronger claims from data. Each workshop is built around real research problems, not abstract theory, and can be delivered remotely or in-person, for half-day or full-day sessions.
I gave regular talks at international conferences in causal inference, Bayesian statistics, and research methods. I spoke at events including PyData Global and the BP Causal Inference Symposium, and taught research methods, experimental design, and quantitative reasoning to undergraduate and postgraduate students across 15 years in UK higher education. These workshops are built on the same principle: rigorous ideas explained clearly, with practical takeaways researchers can use immediately.
Causal Inference for Researchers
The problem
Most researchers are trained in statistics (regression, factor analysis, structural equation modelling), but never formally taught causal inference. As journals, reviewers, and funding bodies raise the bar on causal identification, that gap is increasingly costly.
What I cover
- Why correlation-based statistical methods fall short for causal questions
- Directed acyclic graphs (DAGs) as a tool for study design and confound control
- Quasi-experimental methods for when randomisation isn't feasible
- Interpreting results causally and flagging unjustified causal claims
- Building causal narratives that hold up to methodological scrutiny
Outcome
Researchers leave with a practical toolkit for designing stronger studies, writing more defensible papers, and producing impact narratives that can withstand expert review.
Best for
Quantitative researchers across disciplines: social scientists, health researchers, psychologists, and others. Suitable for PIs, ECRs, and PhD cohorts.
REF evaluates on three pillars and causal inference training speaks to all of them:
- Impact: Impact case studies must demonstrate that research caused a change in policy or practice. Reviewers are increasingly sophisticated about this distinction. Researchers who can reason causally write stronger, more defensible impact narratives.
- Environment: Structured methodological training in genuinely underserved areas demonstrates investment in research quality and staff development.
- Outputs: Journals in economics, epidemiology, psychology, and social science are requiring stronger causal identification. Researchers without exposure to DAGs or quasi-experimental methods face harder scrutiny at top venues.
Using AI Constructively in Research
The problem
Most academic discussion of AI splits into two unhelpful camps. One is uncritical adoption that produces faster but weaker work. The other is blanket avoidance after a bad experience, such as a chatbot inventing a plausible-looking citation. The more useful approach is to understand where AI genuinely strengthens research, where it cannot be trusted, and how to use it while meeting the data-handling and disclosure obligations of academic work.
What I cover
- Where AI genuinely strengthens research: stress-testing the logic behind your causal claims, spotting variables that could be skewing your results, and pressure-testing study design before data collection locks you in
- Navigating the AI ecosystem: an overview of graphical chat tools, code-based workflows, and local LLMs, helping researchers understand the range of ways to interact with AI and choose tools that fit their goals, privacy needs, and workflow
- Using AI as an adversarial reviewer: getting it to argue the "reviewer 2" case against your own paper before reviewer 2 does
- The slop problem, and why it's a use-case failure, not a reason to abstain
- Data privacy and confidentiality: what can and can't safely go near a cloud model, given ethics approval terms, funder data policies, and unpublished-data confidentiality
- Local, private LLMs for the workflows that need to stay on your machine: email triage, early-draft review, sensitive or unpublished data
- Practical safeguards against fabricated citations and confidently wrong facts: checking outputs against primary sources, keeping interpretation and substantive writing with the researcher, and using AI to critique, educate, and support research rather than generate research text
- Journal and funder disclosure requirements, and how to write an AI-use statement that holds up
Outcome
Researchers leave with a working framework for where AI earns its place in the research pipeline, hands-on practice using it to critique, educate, and support their research rather than generate research text, a private low-cost setup for the workflows that shouldn't touch a cloud model, and a draft disclosure statement they can adapt for their own use or their department's.
Best for
Faculty, PIs, and PhD cohorts who are AI-curious. No coding required; supporting technical documentation is provided for groups who want to explore implementation.
Tailored to your research context
The workshop can be adapted for your discipline, research methods, and cohort's experience. Combinations and custom topics are welcome.
Flexible format
Half-day, full-day, or multi-day. Remote or in-person. I'll work with you to find the format that fits your research group's schedule.
For any level
From PhD cohorts and early-career researchers to experienced PIs, each workshop is pitched to the audience.
Research rates
Preferential research rates are available. Contact me to discuss your department, cohort, funding context, and the format that would work best.
Workshops are priced per group, with group size, customisation, and repeat bookings factored into the final quotation.
Enquire about a research workshop
Tell me about your research group and what you're looking for. I'll respond within two business days.