Description
Checks whether asking respondents to recall few or many examples changes their later global judgements, because people use the felt difficulty of recall as information.
A recall task yields two things: the recalled content and the experience of how easy or hard recall was. Listing many examples is hard, so people infer that the behaviour or attitude is rarer or weaker, reversing what content alone would predict: people asked for twelve examples of their own assertiveness rate themselves less assertive than those asked for six. The effect is discounted when the experience is attributed to something else, and is used more for judgements of low personal relevance. The same mechanism undermines debiasing: listing ten alternative outcomes raised hindsight bias (68.0%) compared with listing two (54.3%). In questionnaires and interviews, prompts such as 'list the problems you had' before a global satisfaction rating can therefore shift that rating. The check identifies such sequences and tests them with split samples or reordering.
Attributes
- Discipline
- Survey & Psychometrics
- Loss type
- Unmigrated
- Stage
- Discovery, Evaluation
- Data
- Quantitative
- Effort
- Days
Steps
- Identify prompts that ask for examples, reasons or incidents before a global rating, in surveys, interviews or diary studies.
- Where the order cannot be avoided, split the sample by the number of examples requested (for example two versus ten).
- Alternatively place the global rating before the recall prompt in one version.
- Compare global ratings across versions; a difference indicates an ease-of-retrieval effect.
- Keep debiasing and reflection prompts short (few reasons) to avoid backfire, and fix the safer order for future waves.
Sources
- Heuristics and Biases (Gilovich et al., 2002)
Also known as
- Ease-of-retrieval check for recall-based questions
- Ease of retrieval paradigm
- Few-versus-many examples check
Categories
- Analytics & Behaviour
- Survey & Psychometrics