Definition
A stereotype is a fixed, oversimplified mental image of a group that is applied uniformly to every member of that group, ignoring the variation that actually exists inside it.
Rosling's standard example is the assumption that any African country's roads, schools, or fertility rates can be inferred from the label "African." In reality the continent spans every income level Rosling defines; the only thing the label tells you reliably is geography. Stereotypes feel efficient because they collapse a complex group into a single cached picture, but the picture is almost always built from outdated, dramatic, or unrepresentative examples.
Why it matters
How it works
A stereotype starts as a category — a useful shorthand for a group — and hardens into a fixed image when one or two memorable examples become the template for the whole. Once the template exists, the mind uses it as a default lookup: ask about a country and you fetch the cached picture rather than the actual numbers. Updates require new evidence, and the evidence usually has to be vivid to compete with the template.
This is why stereotypes lag reality so badly. Most of the data Rosling presents — child mortality, fertility, school enrollment, electricity access — has moved dramatically over the last fifty years. The cached images in most heads are roughly 1965 vintage. The world moved; the mental picture did not.
The corrective is to test the picture before relying on it. Ask: who inside this group is the exception, and how big is the exception? What is the actual distribution on the dimension I care about? When the picture turns out to span every income level — as "Africa" or "the developing world" does — you have not refined the stereotype, you have retired it.