The idea
A sample can represent its respondents accurately while misrepresenting the population of interest. Selection bias arises when the process that brings observations into a dataset is connected to the characteristic being estimated. More responses do not automatically repair that process.
Worked example
A library surveys visitors at its upstairs information desk. Ninety percent say stairs are convenient. The librarian initially recommends removing a ground-floor service point. A colleague asks who could not reach the desk. They add telephone interviews with homebound members and invitations near the entrance. The new responses reveal access needs that the first survey could hardly capture. The original answers were sincere; the route into the sample concealed an important group.
Use it carefully
No survey reaches everyone. The practical task is to describe coverage, nonresponse, and plausible differences between participants and missing people. Weighting helps only when its assumptions and relevant population information are credible. A large biased sample can provide a very precise answer to the wrong population question. Report recruitment methods alongside the headline result.
FROM UNDERSTANDING TO ACTION
Put it to work.
Design two recruitment routes for a survey about public transport. Identify who each route misses.
RETRIEVE. REASON. REMEMBER.
Knowledge check.
Choose one answer for each question, then check your reasoning.
Sources & further reading.
Links support the core concepts or identify the book. Examples, exercises, and critical analysis are original Institute material. Source checks are not an independent expert review of every claim.
- OpenStax: sampling methods and sampling bias ↗
OpenStax: sampling methods and sampling bias; supports the core concept. Scenarios, calculations, and practice questions are original. Source reviewed 2026-10-19.