Making Sense of Science Stats

Introduction

By Elisabetta Tola / 2 minute read

For many people, numbers convey a sense of specificity, precision, and objectivity. But numbers are merely one description of facts — one interpretation of phenomena. They are a means to compare things, to see patterns, and to evaluate risks. They are not “truth.”

Consider the following: climate change deniers point to the average global temperature recorded in 1998 — 58.3° Fahrenheit — and compare it with that of more recent years, pointing out a decrease in average temperatures for more than a decade that followed. After all, in 2012, the average temperature was lower, at 58.2° Fahrenheit. This, they claim, is proof that global warming isn’t real.

The numbers are true, but the conclusions are deeply flawed. The year 1998 was unusually hot because of a surprisingly strong El Niño. But when the data are considered as a trend, the reality of climate change is incontrovertible.

This cherry-picking of convenient figures is just one example of how scientific data can be misinterpreted. Other ways include:

• when figures are provided without context
• when references to margins of error, methodologies, sample sizes, or compositions are missing
• when correlation and causation are conflated
• when conclusions are drawn from insufficient data

How can you avoid falling into these traps? It is important that as an editor, you approach them with the same sort of skepticism and care that you take with words. What do statistics say and not say? What figures are truly relevant to the story you are telling? Which ones mean something to your readers?

The key, of course, is to understand what a given number or statistic means. Take the statistical concept of “risk.” There are about 95 fatal car accidents a day in the United States. To put it another way, for all practical purposes there is a 100-percent chance of dozens of fatal car crashes in the United States today. But what risk is there to you when you get behind the wheel? How nervous should you be? Not very. Your individual risk of dying in a car crash is less than 1 in 4,000 in any given year.

It is incumbent upon you, the editor, to ensure that numbers and statistics are properly framed and contextualized to honestly represent reality, and that your readers properly understand them.