Governing evidence and applicability rules (take precedence over aggregate bands): establish the document purpose, design, stage and available evidence before rating. A fit-for-purpose single-method design can earn the highest applicable rating; do not require complexity, mixed methods or two sources for every finding. Distinguish applicable, not applicable with reason, not evidenced, and observed inadequate. Report applicable coverage and the raw denominator. If any dimension is N/A or not evidenced, report the profile and gaps without a 25-point band or an unvalidated rescaled equivalent. Never interpret missing documentation as proof no activity occurred. Use source-linked historical facts only; keep proposed work and illustrative examples separate. Preserve unresolved source conflicts. Do not invent baseline values, quotations, fieldwork, pilots, validation, consent or approvals to improve a score. A serious consent, confidentiality, safeguarding, measurement-validity or unsupported-inference failure is a blocking issue regardless of the total, including 22/25. High document-quality scores do not establish legal compliance, instrument readiness, field authorization or donor approval. Accountable human review against the applicable protocol is separate. A well-documented failed pilot may score 25/25 for documentation while remaining not ready for fielding.
Use This Rubric
Bring: the evaluation scope, deliverables, effort-day budget, team availability, dependencies, review windows, and approval requirements.
Stage: State whether this is a proposed design, a draft, or a report of completed work. Assess planned procedures as plans; require records before claiming they occurred. A complete submitted document can have an observed documentation defect even when the underlying activity is unknown.
Limits: Apply requirements to the stated purpose, methods, population, and risk. Do not impose extra methods or sensitive data collection solely to increase a score. These editorial anchors are provisionally checked, not practitioner-calibrated.
Scoring Criteria
All four major phases (inception, data collection, analysis, reporting) are shown as distinct blocks of work with start and end dates. Sub-phases within each (e.g., tool development inside inception, transcription inside analysis) are visible. The decomposition lets a reviewer track progress phase by phase.
All four phases visible. Sub-phases partial or only shown for some phases.
Three of four phases visible as distinct blocks. One phase collapsed into another (e.g., analysis folded into reporting).
Phases not distinguished. Workplan reads as a single block from start to end.
No phase decomposition. Workplan is a single line on a calendar.
Milestones are concrete, dated, and tied to specific deliverables a reviewer can verify (inception report submitted, tools pretested, fieldwork complete, draft report due, validation workshop held, final report submitted). Each milestone has a date, owner, and acceptance criterion.
Milestones dated and tied to deliverables. One or two missing owner or acceptance criterion.
Milestones listed and dated but not tied to verifiable deliverables. "Analysis complete" without a deliverable that proves completion.
A few milestones mentioned without dates or deliverables. "Inception phase" and "fieldwork phase" without specific dates.
No milestones defined. Workplan has no checkpoints.
Dependencies between activities are shown explicitly (e.g., enumerator training cannot start until tools are pretested; analysis cannot start until transcription is complete; draft report cannot be circulated until DQA is complete). A Gantt, network diagram, or dependency table makes the critical path visible.
Most major dependencies shown. One or two implicit. Critical path identifiable but not labelled.
Some dependencies shown. Sequential ordering implies dependency but no explicit map. Critical path not visible.
Activities listed in order but no dependency logic shown. Reviewer cannot tell what blocks what.
No dependencies shown. Activities appear to run independently when they cannot.
Realistic slack is built in for review cycles (donor review of inception, draft report review by program staff, validation workshop scheduling), ethics approvals (IRB or donor ethics review with typical turnaround), ground-truth delays (weather, access, holidays, religious observances), and contingencies (illness, equipment failure). No activity sized at best-case-only duration.
Buffer time included for most risk points. One or two underbudgeted (e.g., donor review window too short, or no holiday buffer).
Some buffer time included. Review cycles allotted but ethics or ground-truth delays ignored.
Buffer time absent or token. Every activity sized at best case. Schedule will slip on first delay.
No buffer time. Workplan assumes no delays anywhere, which is unrealistic for any evaluation.
Team member assignments are visible per phase. A reviewer can see who is on inception, fieldwork, analysis, and reporting and at what effort level. Double-booking is avoided across concurrent activities. Key roles (team leader, qualitative lead, quantitative lead, enumerator supervisor) are loaded consistently with their effort days in the budget.
Team assignments visible for most phases. One phase or role unclear.
Team listed at a high level. Phase-by-phase loading partial. Possible double-booking not flagged.
Team members listed without phase assignments. Effort days not visible. Concurrent activities suggest double-booking.
No team loading shown. Workplan does not connect to budget personnel days.
Score Interpretation
Report blockers, the dimension profile, and evidence coverage first. Use these aggregate bands only when all five dimensions are applicable and sufficiently evidenced with no unresolved evidence conflict. Otherwise give no total, band, or rescaled equivalent. High documentation quality does not establish readiness, legal compliance, or approval.
| Total (out of 25) | Band | Next Step |
|---|---|---|
| 22-25 | Strong | Strong documentation profile only. Resolve blocking issues and obtain the applicable human review; this score does not authorize use. |
| 17-21 | Adequate | Prioritize source-linked revisions in the weakest dimensions. Request missing records before proposing additional activity; confirm the need for any new work from evidence. |
| 11-16 | Needs Revision | Use located findings as a revision brief. Distinguish reporting gaps from demonstrated design or implementation failures and label proposed work for confirmation. |
| 5-10 | Substantial Revision | Rework the evidence-backed weaknesses with accountable review. A low documentation profile alone does not establish that the underlying activity failed or must be repeated. |