The fastest responsible remote analysis starts with a clear question, a usable dictionary, an untouched raw copy and the minimum data necessary. Good preparation reduces avoidable fees and prevents analysts from guessing what the variables mean.
1. Send the decision context first
Write the research question, objectives, study design, population, sampling method, data-collection period and intended output. Include the protocol or methods section when it governs coding or analysis.
Identify the primary outcome and any pre-specified comparison. A request to analyse everything invites multiple-testing problems and an unfocused deliverable.
2. Pair the data with a codebook
List each field name, label, type, units, allowed codes, missing-value meaning, skip rule and derived-variable definition. Supply the questionnaire version that created the export.
Use stable, unique record identifiers that do not contain names or contact details. Explain one-to-many structures such as household members, visits or repeat groups.
- Readable variable names and labels
- Choice codes and reference groups
- Units and date formats
- Missing and not-applicable states
- Repeat or linkage keys
3. Preserve the source before cleaning
Keep one read-only raw export and note the source, export date and form version. Put attempted corrections in a separate working copy or cleaning log. Never colour a cell as the only record of a decision.
Do not delete outliers or incomplete records simply to make the dataset look tidy. Mark the concern and provide field context so the analyst can apply an explicit rule.
4. Remove identifiers and agree transfer controls
Remove names, phone numbers, email addresses, national identifiers, exact addresses and unnecessary free text. Review combinations such as precise location, rare occupation and exact dates that may still identify someone.
Before transfer, agree the authorised files, named users, storage, access period, output disclosure, retention and deletion. Begin with a de-identified dictionary or structural sample when discussing fit.
5. Package one clear handover
Use unambiguous filenames and include a short README with the authoritative file, software format, known issues and contact for coding questions. Avoid sending several files all labelled final.
Ask the statistician to confirm the received row count, fields, primary outcome and planned deliverables before analysis. That checkpoint catches the most expensive misunderstandings early.
Start a remote analysis fit review
We can review your actual form, dataset and reporting requirement, then scope the smallest service that solves the risk.
Explore the service ↗