“Sampling error” is the concept that a sample might not be representative of a population, because the sample is too small.
We might encounter sampling error more often than we think: whenever we read a news article, talk to a few people, hear a story on the radio, we risk making the wrong inference if we extrapolate what’s in the sample to the whole population.
For example, if we visit a school and meet five left-handed individuals, we might then think the whole school is left-handed. Except we actually won’t, in this example, because we know that being left-handed is a minority trait for the world at large, and we encountered an unrepresentative group.
However, there might be many other situations where we don’t know what the full population looks like – where five left-handed individuals looks normal to us – and the danger is we draw an incorrect reference.
What we need to do, then, is be careful of judging anything from “too few” a number of observations and be ready to shift our view when we collect more evidence.
The danger of extrapolating from “too few” – when discussing stakeholders advocating their views in the media – has been described as a small group having a “disproportionate influence on the climate of opinion”.