Counting Society: How Sociologists Turn Numbers Into Insight

In 1897, the sociologist Emile Durkheim did something radical: he studied suicide not as a private tragedy but as a public pattern. By pulling government death records from across Europe, he found that suicide rates stayed remarkably stable year after year within the same country, but varied sharply between Protestant and Catholic regions, between married and unmarried people, and between peacetime and wartime. No single case explained this. Only the numbers, gathered across thousands of lives, revealed the pattern. That was one of the first major uses of quantitative methods in sociology — using counts, rates, surveys, and statistics to find patterns in how humans behave in groups.

- What quantitative methods are and how they differ from qualitative methods - How sociologists design surveys and use variables - How to read a correlation without assuming it proves causation - How sample size and sampling bias can distort results

Quantitative sociology relies on data that can be counted or measured: age, income, hours worked, number of children, survey answers on a 1-5 scale. Researchers define 'variables' — things that can differ from person to person, like education level or voting behavior — and look for relationships between them. The U.S. General Social Survey (GSS), running since 1972, has asked the same core questions to thousands of Americans every year or two, letting researchers track shifts in attitudes about work, family, and trust in institutions across five decades.

A famous quantitative finding: cities with more ice cream sales also tend to have more drownings. Does ice cream cause drowning? No — both rise in summer heat. This is why sociologists distinguish correlation (two things moving together) from causation (one thing actually producing the other). Serious quantitative studies use techniques like controlling for other variables, longitudinal tracking (following the same people over years), or natural experiments to rule out these hidden 'confounding' factors before claiming one thing causes another.

A survey is only as good as its sample. In 1936, the magazine Literary Digest predicted Alfred Landon would beat Franklin Roosevelt for president, based on 2.4 million responses — the largest poll ever conducted at that point. Roosevelt won in a landslide. The problem was who was sampled: the Digest mailed surveys to its own subscribers and car and telephone owners, a wealthier slice of the population during the Great Depression, not a random cross-section of voters. Modern sociologists use random sampling — where every person in a population has an equal chance of being selected — specifically to avoid this trap.