10 · Statistical Manipulation
Real numbers, chosen and displayed to mislead
Every number in this section can be technically accurate and still create a false impression —
the manipulation lives in which number gets reported, what it's compared against, and how it's
displayed, not in the arithmetic itself.
How it works
A few recurring patterns do most of the work: reporting a relative change ("risk doubled")
without the absolute numbers behind it (from 1 in a million to 2 in a million), choosing a
convenient start or end date for a trend line that flatters one narrative, truncating a chart's
y-axis so a small difference looks enormous, or citing a correlation as though it demonstrates
the causation a reader will naturally infer.
Why it works on us
Most readers don't have the base rate, the full trend line, or the original chart in front of
them to check the framing against — they're relying entirely on the number and visual as
presented. Numbers also carry an air of objectivity that words don't, so a misleading statistic
often faces less skepticism than an equivalently misleading sentence would.
Signs to watch for
- A relative percentage change ("up 200%") with no absolute numbers given anywhere in the piece.
- A chart or graph whose axis doesn't start at zero, exaggerating the visual size of a gap.
- A trend described over a suspiciously specific date range that starts or ends at a convenient outlier.
- Correlation reported in language that implies causation ("linked to," "associated with") without any discussion of confounding factors.
Illustrative example
A headline states a policy "doubled" a rare outcome. The article's own numbers show the
underlying rate moved from 0.001% to 0.002% — a real change, but one that reads very
differently once the absolute figures are visible.