Preparing Data for Comparison

Before comparing datasets, we must prepare them. Imagine receiving one Excel file from a hospital, another from an NGO and a third from a community survey. The files may have different column names, date formats and geographical labels. Comparing them immediately could produce incorrect conclusions.

4.3.1 Check the reporting period

The first two sources cover a three-month reporting period. The survey was conducted later. The survey may ask respondents about an earlier period, or it may measure their current status. Examine the questionnaire and reporting definitions before comparing the results.

4.3.2 Check geographical coverage

Dataset A covers Central and Riverside communities. Dataset B covers Central, Riverside and Hilltop communities. Their totals do not cover the same area. Where possible, align the geographical coverage or explicitly document the difference.

4.3.3 Check the population

Suppose one dataset covers children aged 0–11 months and another covers children aged 12–23 months. Even if both datasets concern immunisation, their populations differ. Confirm the target population before comparing the figures.

4.3.4 Check the indicator definition

Consider two possible measures: the number of vaccination doses administered, and the number of unique children who received at least one vaccination dose. One child may receive several doses. The two measures cannot be treated as equivalent.

4.3.5 Standardise column names

The column names differ, but the underlying concepts might be similar. Before standardising them, verify that the fields genuinely represent equivalent concepts. Then, where appropriate, create a consistent structure for comparison.

4.3.6 Prepare a data comparison checklist

• Source and collection method.
• Reporting period.
• Geographical coverage.
• Target population.
• Indicator definition.
• Unit of measurement.
• Data structure and formats.
• Missing values.
• Potential duplicate records.
• Possible overlap between sources.
• Known limitations.