Freeze the input and output versions before QA. A reviewer cannot reliably sign off a moving collection of files. Identify the objectives, analysis plan, transformations and draft report included in the candidate, plus any unavailable evidence.
Define the review candidate
Freeze the input and output versions before QA. A reviewer cannot reliably sign off a moving collection of files. Identify the objectives, analysis plan, transformations and draft report included in the candidate, plus any unavailable evidence.
Review decisions before formatting
Check whether the sample, outcome definition and analysis population support each objective. Then inspect methods, assumptions, uncertainty and interpretation. A perfectly formatted table can still answer the wrong question or hide an eligibility change.
Worked example: paired observations
Illustrative example: a before-and-after study reports an independent-group comparison. QA should ask whether the same participants were measured twice and whether missing follow-up changed the eligible pairs. The correction is a design and population decision before it becomes a software command.
Original review record
Record Finding ID | Candidate version | Objective/output | Evidence | Severity | Required correction | Analyst response | Reviewer disposition. A return must contain actionable reasons. A resubmitted candidate gets a new version reference; prior findings should not be overwritten.
Separate QA from release
A QA pass does not itself authorise client release. Commercial and release prerequisites remain with the authorised lead. State any unresolved issue and whether the reviewer withheld a conclusion. Do not describe a revision or screening run as final approval.
Review this research decision
We can review your actual form, dataset and reporting requirement, then scope the smallest service that solves the risk.
Explore the service ↗