Glossary
What is Response bias?
How it works
Social desirability bias makes people over-report good behaviour (exercise, recycling) and under-report bad. Acquiescence bias makes people agree with statements regardless of content. Order effects mean an earlier question colours the answers to later ones.
Non-response bias is related: when the people who don’t answer differ from those who do, results tilt towards the people who chose to respond.
Because bias pushes answers in one direction, it doesn’t average out. The usual defence is design: neutral wording, anonymity where it matters, balanced scales and a sensible order, checked by reading the questions as a respondent would before the survey goes out.
Example
Asked in front of a manager whether they follow every safety procedure, nearly all staff say yes. Asked anonymously, a different picture emerges. The procedure didn’t change; the conditions of answering did.
Tips
- Make sensitive surveys anonymous, and say so clearly.
- Use neutral wording and balanced scales.
- Mix statement directions in long agree–disagree sets, or ask direct questions instead.
- Randomise the order of long option lists where order doesn’t matter.
Response bias in FormsAI
FormsAI doesn’t store IP addresses or set tracking cookies on public forms, which helps when you promise anonymity; just don’t ask for names or emails in those surveys. Logic lets you show questions only to the people they apply to, and filters in results let you check whether some groups answered very differently.
FAQ
Response bias, answered
What are the main types of response bias?
Social desirability bias, acquiescence (agreeing with everything), extreme or middle responding, and order effects. Non-response bias is a related problem about who answers at all.
Can a bigger sample fix response bias?
No. Bias pushes answers in a consistent direction, so more responses only repeat it. Fix the questions and the conditions of answering instead.