Research data preparation checklist

Study label: ____________________ Working-file version: ____________________ Review date: ____________________

Use a non-identifying study label. Check each item only after reviewing it. Record questions rather than guessing. Consultation and ordering on Statikali are currently closed; this checklist can be used independently.

Preserve the source and explain the structure

• Keep an unchanged original and a separately named working copy.
• Describe the research aims, questions and hypotheses, and note which data have already been collected.
• State what one row represents. Identify the column or combination of columns that should distinguish a record; keep any person-to-code lookup separately.
• Document groups, repeated measurements, waves, clusters and weights where relevant. Repeated participant codes across waves are not automatically duplicates.
• List each file and sheet, its role and any planned joining keys. Check matches before joining files.

Document the variables and instrument

• Fill one codebook row per variable using the exact column name, meaning, storage type, measurement level, units and permitted values.
• Record category labels and missing codes, including the meaning of a blank if known. Keep a valid zero distinct from a missing response.
• Keep the questionnaire/instrument version, item wording and scoring manual with the project materials, subject to its use permissions.
• Identify scale membership, item direction and whether any reverse scoring has already occurred. Do not reverse items from correlations alone.
• Describe each derived score and its source: contributing items, direction, minimum answered items and missing-item rule. Mark unknown rules for clarification.

Check and document decisions

• Use unique column names and one header row; document dates, decimal conventions and units. Do not rely on cell color or merged cells to convey data.
• Investigate duplicate keys, out-of-range values and inconsistent dates against their source; keep a list of unresolved discrepancies.
• Log every change with version/date, affected variable or coded record, old/new value, reason and evidence. Keep the log within the project's permitted storage.
• Do not delete records, impute missing values, recode categories or drop items without an agreed, documented reason. Flag uncertainty before changing data.

Review what may be shared

• Remove unnecessary direct identifiers from any copy intended for transfer, including free text, file names and sheet names. Do not include the person-to-code lookup by default.
• Check whether rare combinations or sensitive information can identify a person even after names are removed. Follow study permissions and institutional restrictions; do not upload restricted or customer data to a public tool.
• Assemble the permitted working data, codebook, study description, questionnaire/manual references, change log and open questions. Keep originals separate.

Open questions / next decisions

Completion documents preparation; it does not establish anonymity, analytical suitability or approval of the study design. No data are uploaded or automatically checked by this checklist.