# Illustrative output: prospective DQA plan

**Plan status:** Draft DQA plan; checks not yet performed. **Intake:** Attached fictional input. This plan does not establish that either reported value is accurate.

## Purpose and boundary

The planned desk assessment concerns May reporting for Indicator A, used for repair prioritization and funder reporting, and Indicator B, used for water-quality monitoring. Site visits, household contact, identifiable household records, and independent laboratory retesting are not authorized in the supplied information. No discrepancy or completeness threshold was supplied.

## Evidence register

| ID | Supplied information | Limitation |
|---|---|---|
| `input.md#Indicator A` | May report lists 84 functioning points out of 110 registered | “Functioning” has two inconsistent draft interpretations; definition not approved |
| `input.md#Indicator B` | May summary lists 14 samples; laboratory export has two results without matching IDs in the summary | Cause and effect of the unmatched results are unknown |
| `input.md#Data flow` | Technicians record logs; supervisor enters workbook; monitoring officer updates dashboard and report | No supplied change log or submission deadlines |
| `input.md#Known risks` | Three repeated IDs were flagged; values support repair prioritization and funder reporting | No DQA or corrective-action history supplied |
| `input.md#Access and constraints` | Desk records can be supplied; two staff workdays stated | Record counts, access to original logs and permissions for field checks need confirmation |
| `input.md#Standards and tolerances` | No approved tolerance or acceptance rule supplied | Do not assign pass/fail status using an invented percentage |

The supplied Indicator A counts give `84 / 110 × 100 = 76.4%` when rounded to one decimal place. This is arithmetic on the reported counts only; it does not verify the numerator, denominator, or meaning of “functioning.”

## Indicator and risk selection

| Indicator | Planned priority | Basis and selection limit |
|---|---|---|
| A: functioning registered water points | High, proposed | Funder reporting and repair decisions use the value; definition ambiguity and repeated IDs are known risks. Review the full May asset list if supplied, but do not infer field condition from desk records. |
| B: samples with result and identifier | Medium, proposed | Two unmatched IDs are a known reconciliation issue. Trace available records and report exceptions; no lab retesting or sample-quality conclusion is authorized. |

Priorities are qualitative planning recommendations, not an approved score. Confirm the scope and priority with the program manager before field work or use in a funder response.

## Planned check matrix

| Check and quality lens | Planned procedure | Evidence needed | Limit |
|---|---|---|---|
| Completeness, both indicators | Establish the expected sites, reporting fields, periods, and sample events from approved lists or schedules; compare with received records. | Approved asset/sample list, expected reporting schedule, logs, summary | Expected denominator and completeness rule are not yet supplied. Do not call missing files zero. |
| Validity, Indicator A | Resolve the approved “functioning” definition; compare the reporting rule with field log options and indicator population/period. | Approved definition, asset list, logs, report | The current mixed definitions prevent an unqualified interpretation of the reported value. |
| Reconciliation, both indicators | Recalculate totals using the agreed rules and trace site-level totals through workbook, dashboard, and report. Preserve versions and every difference. | Original logs, workbook versions, dashboard export, report | Agreement across these stages does not prove that observations reflect actual site conditions. |
| Source verification, Indicator A | After the asset list and access are confirmed, select records for authorized tracing. Include flagged duplicate IDs and ordinary records. Record the selection frame and selection method. | Source logs and the corresponding asset records | Risk-based selection can locate exceptions but cannot estimate a population error rate. No site verification is included without separate approval. |
| Reconciliation/validity, Indicator B | Match sample identifiers between source log, workbook, summary, and laboratory export; determine whether the two unmatched results are omitted, duplicated, late, or otherwise unresolved. | Sample log, lab export, workbook, summary | No acceptance threshold or lab retest is supplied; do not infer sample validity from ID matching alone. |
| Reliability, both indicators | If time permits after scope and record volume are known, have a second authorized reviewer independently re-abstract selected records using the same instructions; compare entries and document disagreements. | Shared extraction instructions and selected records | Workdays are capped at two; exact sample volume and reviewer availability are unresolved. Any small risk-based repeat is nonrepresentative. |
| Precision, both indicators | Confirm units, numerator/denominator rules, rounding, and detail required for the intended repair/reporting decisions. | Approved indicator definitions and reporting rules | No precision margin or statistical inference is planned without a suitable design and inputs. |
| Integrity, both indicators | Review available version history, roles, approvals, and transfer controls. Ask which change log or access evidence is maintained. | Workbook versions, role descriptions, available audit records | No supplied change log is not evidence of manipulation. Do not infer intent from discrepancies. |
| Timeliness, both indicators | Record collection, entry, review, and report dates and compare them with the approved reporting calendar. | Dated logs, version metadata, reporting calendar | No due dates were supplied, so timeliness cannot receive a pass/fail conclusion until confirmed. |

## Safeguards, feasibility, and decisions

- Limit desk access to the listed records and the authorized staff. Do not copy identifiable household data unless a documented purpose and authorization require it; use de-identified extracts where sufficient.
- The two-workday constraint is supplied, but records counts and reviewer availability are not. Finalize record volume only after an inventory; reduce scope or seek more time through an explicit decision, not by silently omitting high-risk checks.
- Confirm approved definitions, expected lists, completeness rules, reporting dates, access to original records, second-reviewer availability, and any tolerance before assessing against them.
- Site visit, household contact, and independent lab retesting remain outside this plan unless separately authorized with privacy and safety arrangements.

## Conditional escalation

If a check reveals a privacy or safety risk, pause that check and use the program's approved safeguarding or data-protection contact. If the supplied records materially conflict, preserve both versions and notify the designated data owner before changing a report. If suspected deliberate alteration arises, do not accuse or investigate beyond authority; secure the evidence and follow the supplied escalation route. No escalation contact was supplied, so confirm the route before work begins.

## Completion record for the future assessment team

- Actual dates, assessors, and approved scope: **Not completed**
- Records/sites selected and selection rationale: **Not completed**
- Procedures performed and evidence IDs: **Not completed**
- Exceptions, reconciliations, and limitations: **Not completed**
- Decisions, approvals, and escalation: **Not completed**

This is a plan only. It makes no finding about the quality, accuracy, integrity, or completeness of the program's data.