Why numbers deserve a second look
A figure in an article carries an air of authority. It looks exact, and it is hard to argue with. But a number is only meaningful with context, and the same number can support very different impressions depending on what surrounds it. You do not need advanced mathematics to read figures carefully. You need a handful of questions.
This article reports no study of its own. Its numbers are all invented examples, labeled as hypothetical, to show how a figure can mislead without being false.
Ask what the number is compared with
A figure alone rarely says much. “Sales were 500 units” tells you little until you know whether that was higher or lower than last month, last year or a competitor. Look for the comparison. If an article gives only a bare total, a rise or a record, ask: compared with what, and how far back does the comparison go?
Choosing the starting point can change the story. A hypothetical measure that fell sharply and then partly recovered can be described as “up” if you start at the bottom, or “down” if you start at the peak. Both are accurate in a narrow sense.
Watch the difference between a percentage and an amount
Percentages are easy to quote and easy to misread.
- A large percentage of a small number may be tiny in real terms. Suppose, hypothetically, something rose from 2 cases to 4. That is a 100% increase, yet only two additional cases.
- A small percentage of a large number can be big in practice. A change of 1% in a hypothetical population of 10 million is 100,000.
- Percentage points and percent are different. If a hypothetical rate moves from 10% to 12%, that is a rise of 2 percentage points, and a relative increase of 20% (2 divided by 10). Articles sometimes use these loosely, and the wording can mislead.
When you meet a dramatic percentage, look for the underlying counts. A good article will give both.
Averages hide the spread
“Average” can mean different things. The mean adds everything up and divides by the count. The median is the middle value when the numbers are lined up in order. When a few values are very large, the mean can be pulled up while the median stays closer to the typical case. In hypothetical terms, if five people earn 20, 22, 24, 26 and 108 units, the mean is 40 (since 20 + 22 + 24 + 26 + 108 = 200, and 200 divided by 5 is 40), but the median is 24. Neither is wrong; they answer different questions. If an article says “average” without saying which, and the figures could be skewed, that is worth noting.
Averages also say nothing about how spread out the values are. Two groups can share an average and look very different.
Ask who and how many were counted
Behind many figures is a sample: a group of people or items that were measured. Useful questions include:
- How big was the group? Small groups can produce striking results that do not hold up more widely.
- Who was included? Results from volunteers, readers of one site or people in one region may not apply to everyone.
- How was it measured? A survey depends on how questions are worded and who answers. A count depends on what counts.
- Is it a total, a rate or an estimate? An estimate typically carries some uncertainty even when it is presented as a single number.
If an article does not say, the information may be in the original source, and tracing a figure back is where source checking helps.
Correlation is not the same as cause
Articles often report that two things go together: people who do one thing also tend to do another. That alone does not show that one causes the other. A third factor might drive both, the direction could run the other way, or the link might be a coincidence. Look for language such as “linked to,” “associated with” or “tend to,” which signals a pattern and not a proven cause. Be cautious when a headline turns that into “causes” or “leads to.”
Reading charts
Charts compress a lot into an image, and how they are drawn affects what you see.
- Read the title and labels first. What is being measured, in what units, over what period?
- Check the axes. Does the vertical axis start at zero? A bar chart that starts well above zero can make small differences look huge. A line chart may reasonably start elsewhere, but you should notice that it does.
- Look at the scale. Stretching or squeezing an axis changes how steep a line appears.
- Check the time span. A short window can exaggerate a trend or hide a longer one.
- Count the categories and see how they are grouped. Combining or splitting groups changes the picture.
- Look for the source and date in the small text under the chart.
Be cautious with charts where the pictures themselves are scaled, such as icons of different sizes. Area grows faster than height, so a symbol drawn twice as tall can look much more than twice as large.
Pie charts show parts of a whole, so the slices should add up to 100%. Maps shaded by color can overstate the importance of large, sparsely populated areas.
Other things worth noticing
- Rounding. Figures may not add exactly because of rounding. That is usually harmless.
- Currency and units. Check whether amounts are adjusted for inflation or expressed in different units or periods.
- Forecasts. A projection is an estimate based on assumptions, not a fact. Treat it as one possibility.
- Selective examples. A single vivid number may have been chosen because it is striking. Ask whether it is typical.
A practical routine
When a figure matters to you, try three quick steps. First, restate it in plain words: “Out of how many, over what time?” Second, ask what would make it look different, such as another comparison or a different starting point. Third, look for the source. You will not always find answers, and then the right response is to hold the claim lightly.
Careful reading of numbers is not about distrusting data. Good figures, honestly presented, are among the best tools we have for understanding the world. The aim is to read them with enough care to tell the difference.