Types of Data

Qualitative vs Quantitative Data

Quantitative data is numerical — GDP figures, population counts, survey percentages. It can be measured, compared, and analyzed statistically. Qualitative data describes qualities or characteristics — interview transcripts, ethnographic observations, case study narratives.

Both types are essential. Quantitative data tells you what is happening at scale; qualitative data helps explain why. A drop in voter turnout (quantitative) needs context from interviews with non-voters (qualitative) to be properly understood. Be wary of analyses that rely exclusively on one type.

Primary vs Secondary Data

Primary data is collected firsthand for a specific purpose — a government census, a clinical trial, original survey research. Secondary data reuses data collected by others — a think tank analyzing census data, a journalist using WHO statistics.

Primary data is more reliable for its intended purpose but expensive and time-consuming to collect. Secondary data is accessible and allows comparison across studies, but may have been collected using different methodologies or with different definitions. Always check how data was originally collected before drawing conclusions from it.

Worked Example: Classifying Real Indicators

Before you can judge a statistic you must know what kind of data it is. Consider a report on national wellbeing that cites four sources. We can classify each along two axes at once:

Notice that the two axes are independent: data can be quantitative and primary, or qualitative and secondary. A single number can also hide a definitional choice. Take unemployment. The headline U-3 measure counts only people actively searching for work. A broader U-6 measure adds discouraged and marginally attached workers plus involuntary part-timers. In a typical month U-3 might read 4.2% while U-6 reads about 7.7% for the same economy — nearly double. Neither is wrong; they answer different questions. If a writer quotes the lower figure to argue the labour market is healthy, the choice of definition is doing the persuasive work, not the underlying reality.

The literacy skill is to ask three questions of every figure: What exactly was measured? Who collected it and for what purpose? And what plausible alternative definition would change the story? A statistic is only as trustworthy as the answers to those questions.

Levels of Measurement

Beyond these two axes, quantitative data comes in levels that determine what maths is even legal. Nominal data are unordered labels (country, party) — you can count them but not rank them. Ordinal data have an order but unequal gaps (a 1-5 satisfaction scale); averaging them is common but technically dubious, because the distance from 'poor' to 'fair' need not equal 'good' to 'excellent'. Interval data have equal gaps but no true zero (temperature in Celsius), so you can add but not sensibly multiply. Ratio data have a true zero (income, population, distance), so every operation is valid. When you see someone compute an average of ordinal survey ratings, or claim 20 degrees is 'twice as warm' as 10, they have applied arithmetic the data cannot support. Knowing the level of measurement tells you which claims a dataset can honestly bear.