Analysis plan
Questions, hypotheses, variables, population, methods, sensitivity analyses, and interpretation criteria.
Statistical consultingFrom the question to a defensible result.
We plan and execute statistical analyses for research, healthcare, and business. The work begins with the decision the data must support and ends with reproducible results, explicit limits, and communication suited to the audience.
Design before testingQuestion, population, and outcome first.
Responsible inferenceEffect, uncertainty, and sensitivity.
ReproducibilityCode and decisions documented.
Clear communicationFrom technical teams to executives.
Where we help
Problems that require a method, not just a tool.
A trustworthy analysis does not begin with the automatic selection of a test. It depends on the question, data-generating process, sources of bias, and how the result will be used.
Research and healthcare
- Analysis planning and protocols
- Observational studies and clinical data
- Survival and time-to-event analysis
- Longitudinal and repeated-measures models
- Review of methods, results, and tables
Business and product
- Experiments and A/B testing
- Customer segmentation and behavior
- Indicators, uncertainty, and monitoring
- Forecasting and time series
- Validation of models and decision rules
Deliverables
A scope proportional to the decision.
A focused review does not need to become a long engagement. An analysis headed for production receives additional controls.
Analytical dataset
Consistency rules, transformations, missing data, dictionary, and a trace of preparation decisions.
Reproducible analysis
R or Python code, versioned results, diagnostic checks, and an explicit record of limitations.
Report and presentation
Tables, charts, interpretation, practical implications, and material designed for the decision-makers.
Method
Four decisions before the conclusion.
The goal is to avoid a formally correct analysis that answers the wrong question.
Define the estimand
Which effect, association, prediction, or difference actually matters to the decision?
Examine the data
How was it generated, what is missing, and where are selection, dependence, confounding, or measurement error?
Validate the method
Assumptions, diagnostics, alternatives, robustness, and sensitivity to analytical choices.
Communicate the limit
What the data supports, what it does not, and which decisions still depend on context.
Frequently asked questions
Scope and collaboration
Do you only run the analysis, or can you also help with planning?
Both formats are possible. Before data collection, we can align outcomes, sample size, and the analysis plan. With existing data, we begin with a feasibility and limitations review.
Do you work with academic research?
Yes, with a focus on methodology, analysis, reproducibility, and communication. Scientific authorship and responsibility remain with the researchers.
Can you review an analysis that is already complete?
Yes. The review may cover alignment between question and method, assumptions, code, interpretation, tables, and consistency between narrative and results.
Next step
Start with the question.
Send a summary of the objective, population, available data, and timeline. We will respond with feasibility and the smallest useful next step.