From a question to a traceable analysis
The method depends on measurements and the question. You need not prescribe a test. Consulting can support planning, review existing results or deliver an agreed analysis.
Describe the question and purpose first
Is the goal description, comparison, association or prediction? Explain the subject and decision. Association alone does not prove causation; existing data may not answer the intended question.
Explain data and collection
More rows do not necessarily mean more independent observations. Identify repeated measurements, missing values, selection and collection changes.
| Input | Why it matters |
|---|---|
| Unit | Explain person, order or measurement and repetitions. |
| Variables | Meaning, units, coding and missing values. |
| Selection | Who or what was recorded and what is absent. |
Distinguish planning and analysis
Before collection, assess measurement, comparison and required sampling. Afterwards, missing evidence cannot simply be created. Feasibility, cleaning and modelling should serve the question without forcing a preferred result.
Agree advice, analysis and handover
Methods advice differs from programming and full analysis. Agree data preparation, explanations, files and follow-up.
| Output | Specify |
|---|---|
| Recommendation | Method, assumptions and remaining limits. |
| Report | Question, evidence, uncertainty and interpretation. |
| Code / files | Cleaning, versions and traceable steps. |
Respect confidentiality and result limits
Agree secure exchange and authorized access before sharing protected data. Keep uncertainty and assumptions visible. Academic work must follow institutional rules and retain your own responsibility.
Prepare a short request
Raw or confidential files are unnecessary for initial scoping.
- Question. Explain the context and intended decision.
- Stage. Identify planning, collection, data or results.
- Handover. Specify advice, report or traceable code.
Work in traceable stages
Unsuitable evidence may require revising the question.
- Clarify. Understand the question, evidence and responsibilities.
- Agree. Record the plan, scope and access.
- Deliver. Explain results with assumptions and limits.
Frequently asked questions
Must I know the right test?
No. The question, collection and evidence guide method assessment.
Can I request advice before collection?
Yes. Measurement design and required sampling can be assessed early.
Is programming included?
Only if agreed. A methods discussion is not a full commissioned analysis.
Do many rows mean a large sample?
Not necessarily. Repeated measurements and dependencies must be considered.
Is a significant result guaranteed?
No. A sound analysis also reports uncertainty and inconclusive results.
Must I upload raw data immediately?
No. Describe the evidence; agree access and secure exchange later.
