Social Desirability Bias
Social desirability bias occurs when people report answers that present them favorably rather than accurately describing their views or behavior.
Download Markdown (.md)Soroush’s Product Wiki · https://wiki.soroush.uk/topic/social-desirability-bias/
01What it means
Social desirability bias is a response bias in which perceived social expectations influence what someone says. Participants may overstate good habits, hide mistakes, or soften criticism to appear competent or considerate. It is one specific source of research reactivity, which also includes changes caused by observation and other study conditions.[1][2]
02Why it matters
Pleasant interviews can conceal serious problems. If users feel judged by the researcher or believe their employer may see their answers, the resulting evidence can misrepresent actual work and lead to ineffective product decisions.
03When to use it
Consider this bias when researching sensitive behavior, workplace practices, mistakes, or feedback on a design presented by its creator. It also affects surveys. Use methods suited to the question rather than assuming that anonymity or observation completely removes it.
04How it works
- Identify questions where respondents may perceive a socially preferred answer.
- Explain confidentiality accurately and avoid collecting unnecessary identifying details.
- Use neutral wording and make it safe to describe ordinary difficulties.
- Ask about specific recent events rather than ideal behavior.
- Compare self-reports with appropriate behavioral evidence and document remaining uncertainty.
05A practical example
A team asks employees how consistently they follow an approval process. Rather than praising compliance, the interviewer asks them to describe the last request and any workarounds. The team compares those accounts with consented workflow observations before redesigning approval steps.
06Where it breaks
A reassuring script cannot eliminate power differences or fear of consequences. Anonymous surveys still have framing and recall problems. A discrepancy between reported and observed behavior does not prove dishonesty; context and memory may also explain it.
Common mistakes
- Introducing the designer and asking participants to praise the design.
- Promising anonymity when responses remain identifiable.
- Treating every discrepancy as intentional deception.
07Sources
- Rating Scales in UX Research: Likert or Semantic Differential?Nielsen Norman Group · Professional guidanceRating Scales in UX Research: Likert or Semantic Differential?
- User Interviews 101Nielsen Norman Group · Professional guidanceUser Interviews 101