Understanding Data Triangulation

4.2.1 What is data triangulation?

Data triangulation is the process of comparing and interpreting information from different sources, methods or perspectives to understand a question more reliably.

Think of three people describing the same event. One person saw what happened at the beginning. Another observed what happened later. A third has a written record. Comparing their accounts may provide a fuller understanding than relying on one account alone. Health data triangulation follows a similar principle, although datasets require systematic examination rather than simply choosing the most convincing account.

4.2.2 Triangulation does not mean adding numbers

Can we calculate 800 + 300 + 950? No. The figures represent different measures, and the sources may overlap. The 800 facility contacts and 300 outreach contacts may include repeat visits. Some outreach activity may already be included in facility reporting. The survey estimates unique people rather than service contacts. Adding these figures would combine incompatible measures.

4.2.3 The five questions to ask

1. Are we measuring the same thing? One source may count visits, another doses administered, and another unique people.
2. Are we looking at the same period? A report covering January to March should not be compared directly with one covering April to June without accounting for the difference.
3. Are we looking at the same population or geographical area? One dataset may cover three communities while another covers an entire local government area.
4. How was the information collected? Routine facility records, surveys and programme reports use different methods, and each has different strengths and limitations.
5. What could explain the differences? Missing reports, duplicate records, different reporting periods, different indicator definitions, inaccurate population estimates, differences in geographical coverage, and sampling or measurement uncertainty are possible explanations to investigate, not conclusions to assume.

4.2.4 What if the sources agree?

Suppose facility records and a programme report both show that a service increased over three months. Their agreement may strengthen confidence that the pattern deserves attention, especially if the sources are independent and measure the same thing. Agreement does not automatically prove that the figures are accurate. The sources might rely on the same original data or share the same underlying error. Independent evidence can be more informative than two reports repeating the same numbers.

4.2.5 What if the sources disagree?

Disagreement is not necessarily a failure. It may reveal an important issue. One dataset may contain records from outreach services that are missing from another report. Alternatively, the datasets may measure different populations or use different definitions. The analyst's responsibility is to investigate the discrepancy and explain what can and cannot be concluded.