The Importance of Data Quality in M&E

The Importance of Data Quality in M&E

Reliable monitoring and evaluation depends on reliable data. This practical guide explains the dimensions of data quality, the risks of weak evidence, and the systems, roles and checks organisations can use to improve decisions, accountability and learning.

Monitoring and evaluation (M&E) helps organisations understand whether their activities are being implemented as planned, whether they are reaching the intended people and whether they are producing useful results. However, the value of an M&E system depends heavily on the quality of the data it collects. A well-designed indicator cannot support a sound decision if the information used to measure it is incomplete, inaccurate, late or misunderstood.

Data quality is therefore not a technical concern reserved for researchers or database managers. It is a management responsibility. Programme leaders, field officers, partners, finance teams and decision-makers all influence how data is defined, collected, checked, stored and used. In a Kenyan community health project, an agricultural programme in Uganda or a skills initiative in Ghana, trustworthy data can help an organisation direct limited resources where they are most needed.

What Data Quality Means in M&E

Data quality refers to how well data meets the requirements for its intended use. High-quality data is not necessarily data that is perfect in every respect. It is data that is sufficiently accurate, complete, timely, consistent, relevant and accessible for the decision being made.

For example, a programme may need monthly information on the number of young people attending entrepreneurship sessions. If attendance records arrive six months late, they may be of little use for adjusting current activities, even if the figures are accurate. Similarly, data submitted on time may still be unreliable if facilitators count repeat attendances as different participants when the indicator requires unique individuals.

The first question should therefore be: What decision will this data support? The answer helps determine the level of quality required, the collection method, the review process and the resources that should be invested.

The Main Dimensions of Data Quality

Accuracy

Accuracy means that the data correctly represents the situation it is intended to describe. An accurate figure for households receiving drought-resistant seed, for instance, should correspond to households that actually received the specified seed during the reporting period.

Accuracy can be weakened by misunderstanding an indicator, poor record keeping, data-entry mistakes, duplicate entries or pressure to report better results. It can be improved through clear definitions, well-designed tools, staff training, supervision and verification against source documents.

Completeness

Completeness means that all required data has been provided. Missing data can occur when some field teams fail to submit reports, when forms contain unanswered questions or when important categories are omitted from a database.

Missing information does not always mean that an activity did not happen. A blank field might represent zero, an error, a question that was not applicable or a value that was forgotten. These meanings must not be confused. Data collection tools should distinguish between zero, not applicable, unknown and not recorded where those distinctions matter.

Timeliness

Timely data is available when it is needed. A county-level programme manager may need weekly information to respond to stock-outs, while a donor report may require quarterly results. Data that arrives after a decision has already been made has reduced practical value.

Timeliness should be realistic. Requiring field staff to submit detailed information every day may create rushed reporting and reduce accuracy. The reporting schedule should match the pace of the activity, the capacity of the team and the decisions the data must inform.

Consistency

Consistent data is collected and interpreted in the same way across people, locations and reporting periods. If one project officer defines a trained participant as someone who attended one session while another requires attendance at three sessions, their figures cannot be meaningfully compared.

Consistency depends on common indicator definitions, standard operating procedures, stable forms and communication about changes. When a definition or method changes, the change should be documented so that users understand why figures before and after the change may not be directly comparable.

Validity

Validity concerns whether the data actually measures what the indicator claims to measure. Counting the number of leaflets distributed may show the reach of a communication activity, but it does not by itself show whether people understood or acted on the information.

A valid measurement begins with a clear indicator. The indicator should identify what is being measured, the unit of measurement, the population or location covered, the time period and, where relevant, the method of calculation. Vague indicators create confusion even when the collection process is efficient.

Integrity and security

Data integrity means that information remains complete and unaltered except through authorised, documented changes. Security protects data from loss, unauthorised access, accidental disclosure or misuse.

These dimensions are particularly important when M&E systems contain personal information, such as names, phone numbers, health details, disability information or household circumstances. Organisations should collect only information they genuinely need, restrict access according to roles, protect stored records and follow applicable privacy requirements.

Why Data Quality Matters

It improves management decisions

Managers use M&E data to decide where to deploy staff, whether to change an activity, which sites require additional support and whether resources are being used as intended. Weak data can make an underperforming activity appear successful or hide a problem until it becomes expensive to address.

Suppose a vocational training programme reports high completion rates. Further review may show that the figures include participants who attended only the first session. A decision based on the original figure could lead to more investment in an approach that is not producing the expected result. Better attendance definitions and follow-up records would support a more honest assessment.

It strengthens accountability

Organisations are accountable to communities, funders, boards, government authorities and their own staff. Reliable data allows them to explain what was delivered, who benefited and what was learned. It also makes it easier to identify gaps and acknowledge when targets were not achieved.

Accountability is weakened when figures are produced mainly to satisfy reporting requirements. Data should be collected for genuine management and learning purposes, not simply copied into templates at the end of a reporting period.

It supports learning and adaptation

M&E is more useful when it helps an organisation learn what works, for whom and under which conditions. High-quality data enables teams to compare approaches, notice patterns and investigate unexpected results.

For example, an agricultural project may find that participation in demonstration plots is high near trading centres but low in remote areas. This pattern could lead to practical changes such as decentralising sessions or adjusting meeting times. Such learning is possible only when location and participation data are recorded consistently.

It protects credibility and resources

Inaccurate reporting can damage relationships with partners and communities. It may also cause organisations to purchase unnecessary supplies, overlook underserved groups or allocate funding based on a misleading picture of performance. Good data quality reduces these risks and helps demonstrate responsible stewardship of resources.

Common Causes of Poor Data Quality

  • Unclear indicators: Staff interpret the same measure differently because the numerator, denominator, target group or reporting period is not defined.
  • Weak collection tools: Forms contain confusing questions, duplicate fields, missing response options or no space for essential information.
  • Insufficient training: Data collectors understand the programme but not the measurement method, or new staff begin work without proper orientation.
  • Heavy reporting burdens: Teams complete too many forms, leading to rushed entries, copying and delayed submission.
  • Poor supervision: Managers check whether reports arrived but do not examine the underlying registers or question unusual figures.
  • Manual errors: Transcription from paper forms to spreadsheets introduces omissions, duplicated records and calculation mistakes.
  • Pressure to meet targets: Staff may consciously or unconsciously change records when performance is closely tied to approval, recognition or funding.
  • Disconnected systems: Different departments use different definitions, identification codes or reporting periods, making the data difficult to combine.

Building a Data Quality System

1. Define indicators before collecting data

For every indicator, prepare a short reference sheet. It should state the indicator name, purpose, definition, calculation, source, frequency, responsible person and disaggregation required. If the indicator is the percentage of trainees completing a course, specify exactly who belongs in the denominator and what counts as completion.

Also identify whether the indicator measures an input, activity, output, outcome or longer-term change. This prevents teams from treating the number of activities delivered as proof that meaningful change occurred.

2. Select practical data sources

Use the most appropriate source for the question. Possible sources include attendance registers, service records, stock cards, financial documents, surveys, interviews, observation checklists and administrative databases. No single source is automatically best.

Consider reliability, cost, burden on participants and staff, frequency of collection and the sensitivity of the information. A short, well-designed register used consistently may be more valuable than a complex survey that the team cannot maintain.

3. Train and support data collectors

Training should cover more than how to complete a form. Data collectors need to understand the purpose of each question, the meaning of key terms, how to handle unusual cases and when to ask for clarification. Use examples and practice exercises, then review early records before full implementation.

When working through community organisations or local partners, agree on definitions and reporting responsibilities at the beginning. A shared understanding is more effective than correcting inconsistencies months later.

4. Build checks into the process

Quality checks should happen at several levels. At the point of collection, a field officer can check whether required fields are complete. A supervisor can compare totals with registers and investigate large changes from the previous period. A data manager can run validation rules to identify impossible dates, duplicate identification numbers or percentages above 100.

Verification does not mean assuming that staff are dishonest. It means creating a routine process for finding errors while records are still available and memories are fresh.

5. Document corrections and changes

When a value is corrected, retain an appropriate audit trail showing what changed, who made the change, when it was made and why. Do not silently overwrite historical information. Similarly, document changes to indicator definitions, tools, sampling methods or reporting boundaries.

6. Use data quality assessments proportionately

A data quality assessment may include reviewing source documents, retracing a sample of reported figures, checking calculations, interviewing data collectors and examining the flow of information from the service point to the final report.

The process should focus on indicators that are important, high-risk, frequently used or difficult to measure. A small organisation may begin with a quarterly review of a few priority indicators rather than attempting a costly assessment of every data point.

Roles and Responsibilities

Data quality improves when responsibility is shared but clearly assigned. Field staff are responsible for recording information accurately and promptly. Supervisors review records, provide feedback and address recurring errors. M&E officers maintain indicator definitions, tools, training and analysis. Programme managers ensure that data is used and that reporting expectations are realistic. Senior leaders provide resources and create an environment where reporting problems leads to improvement rather than automatic punishment.

Communities and participants also have an important role. Organisations should explain why information is being collected, avoid unnecessary questions and provide appropriate ways for people to raise concerns about inaccurate records or harmful use of their information.

Using Technology Without Assuming It Solves Everything

Digital forms, mobile data collection and dashboards can reduce some transcription errors and speed up reporting. They can also include required fields, range checks, location information and automatic calculations. However, technology does not correct a poorly defined indicator or a deliberate misreport. It may simply make bad data arrive faster.

Before adopting a digital tool, assess connectivity, device access, electricity, user skills, maintenance, data protection and the ability to export or analyse records. In areas with unreliable internet access, offline functionality and clear procedures for synchronising data may be essential. Paper systems can still work well when they are simple, securely stored and regularly reviewed.

Applying This in Practice

Consider a youth enterprise programme that wants to track the number of participants who start or expand a small business after training. A weak approach would ask trainers to report successful businesses using an undefined checklist. A stronger approach would proceed as follows:

  1. Clarify the result: Define what counts as starting or expanding a business, and specify the period in which the change must occur.
  2. Identify the source: Combine training records with a follow-up survey or structured interview, rather than relying only on the trainer’s impression.
  3. Set a realistic schedule: Collect baseline information before training and follow up at an interval that allows participants time to apply their learning.
  4. Use consistent questions: Ask all participants about business status, activity, changes and relevant barriers using the same wording.
  5. Check unusual results: Investigate a sudden increase in success rates, a very high number of missing responses or identical answers across many records.
  6. Protect participants: Explain how information will be used, limit access to personal details and report aggregated findings where individual identification is unnecessary.
  7. Use the findings: Discuss whether the training content, mentoring, market access or follow-up support should change.

Teams can use the same logic for other sectors. A health programme might verify service figures against facility registers. A school-support project might compare activity reports with lesson plans and observation records. A water project might distinguish between infrastructure constructed, infrastructure functional and households actually using the service. The appropriate checks depend on the result being measured.

Questions to Consider

  • What decision will each important indicator support?
  • Could two trained staff calculate or record this indicator differently?
  • What is the original source of the reported figure?
  • How are missing values, zero values and not-applicable cases distinguished?
  • Which figures would cause the greatest harm if they were wrong?
  • Can staff complete the tool accurately within the time and resources available?
  • How will the organisation respond when data reveals weak performance?

Key Takeaways

  • Data quality means that information is fit for its intended use, not merely that it has been collected.
  • Accuracy, completeness, timeliness, consistency, validity, integrity and security all affect the usefulness of M&E data.
  • Clear indicator definitions and practical collection tools prevent many errors before they occur.
  • Routine checks should compare reported figures with source records and investigate unusual or missing information.
  • Technology can support quality, but it cannot replace sound indicators, trained staff and responsible supervision.
  • Reliable data helps organisations make better decisions, remain accountable and adapt programmes to real conditions.

Comments

Learner discussion on this EduHub resource.

No comments yet.