How the Media Gets Percentages Wrong

Percentage statistics appear in the news every day. They are used to convey scale, urgency, and trend — but they are also one of the most frequently misused tools in journalism. Here are six patterns that should make you pause.

1. Cherry-picking the base year

A percentage change always depends on the starting point. A company that had a terrible year and then recovers can honestly claim a “200% increase in profits” — but if the starting point was unusually low, that figure is deeply misleading.

What to ask: What was the base value, and was it a typical year? Pandemic-year comparisons are a common example: a 150% rise in restaurant visits compared with 2020 lockdown levels tells you almost nothing useful.

2. Confusing percent change with percentage points

When an unemployment rate drops from 8% to 6%, some headlines say “unemployment falls by 2 percentage points”. Others say “unemployment falls 25%.” Both are correct — but they sound completely different.

2 percentage points = the absolute arithmetic difference
25% = the relative change ((8 − 6) ÷ 8 × 100)

Politicians in favour of the policy choose “25% fall.” Critics may choose “only 2 percentage points.” Neither is wrong, but the choice is rarely neutral.

3. Large relative percentages from tiny absolute numbers

“Shark attacks doubled this year!” If there were 2 attacks last year and 4 this year, that is technically a 100% increase. Expressed in absolute terms — two more incidents globally — it sounds much less alarming.

The rule: Always find the absolute number behind a percentage. A 100% increase in a rare event is not the same as a 100% increase in a common one.

4. Percentages that exceed 100% without explanation

“Traffic to our site grew by 400% in one month.” This is possible — it means the site got five times as many visitors as before. But the phrasing implies something implausible happened when the reality may be mundane (the site was featured by a large publication once and the spike has since reversed).

What to ask: Does this reflect a sustained trend or a one-time spike? Is the base period representative?

5. Not disclosing the sample size

“80% of dentists recommend this toothpaste.” How many dentists were asked? If the survey included five dentists and four agreed, the claim is technically true but statistically meaningless. Sample sizes below a few hundred are generally too small to draw reliable percentage conclusions about a wider population.

6. Averaging percentages directly

A company has two divisions: Division A grew 10%, Division B grew 90%. The “average growth” is 50%, right? Not necessarily — it depends on the size of each division. If Division A is ten times larger, the combined growth is actually much closer to 10% than 50%.

Percentages should be weighted by the underlying values they represent, not added and divided like raw numbers.

A quick checklist for reading percentage statistics

What is the base value? Is it typical?
Is this a relative change (percent) or an absolute difference (percentage points)?
What is the absolute number behind the percentage?
What is the sample size?
Is the trend sustained or a one-off spike?
Are the percentages being weighted correctly?