Description
Compares what users predict they will do with what they actually do in the same situation, quantifying the say-do gap and its drivers.
People are poor predictors of their own behaviour: Sheeran's meta-analysis found intentions explain only about 28% of variance in behaviour. In UX, interviews and surveys capture predictions such as 'I would use this weekly', which are then treated as evidence of demand. The behavioural prediction gap test pairs a prediction with a later measured behaviour for the same individuals and computes the gap per behaviour and segment. It distinguishes inclined abstainers (said yes, did not) from disinclined actors, which point to different interventions: the former need ability or prompts, the latter suggest habitual or social drivers. Rigour comes from individual-level pairing, precise behavioural definitions and pre-specifying the observation window.
Attributes
- Discipline
- Behavioral Science & Economics
- Loss type
- Latent
- Stage
- Evaluation, Discovery
- Data
- Mixed
- Effort
- Weeks
Steps
- Define 3 to 8 target behaviours with observable criteria and an observation window, for example 'submits at least one digital form in the next 14 days'.
- Recruit 30 to 200 participants whose subsequent behaviour you can observe through logs or diaries, with consent to link predictions to behaviour.
- Collect predictions using probability scales (0% to 100%) rather than yes or no, plus the reasons and anticipated barriers.
- Observe actual behaviour over the window via analytics, diary or field observation, blind to individual predictions where feasible.
- Compute per-person and per-behaviour gaps, the share of inclined abstainers and disinclined actors, and the correlation between prediction and behaviour.
- Interview 6 to 12 inclined abstainers to identify which barrier (motivation, ability, prompt, habit, social context) prevented the predicted behaviour.
Sources
- Heuristics & Biases (Kahneman ed.)
- Judgment Under Uncertainty
- Choices Values & Frames (Kahneman/Tversky)
- The Inquiry Series, Ch 21 — Heuristics, Biases & the Architecture of Judgment
Also known as
- Intention-behaviour gap test
- Say-do gap test
- Stated versus revealed behaviour comparison
Categories
- Analytics & Behaviour
- Behavioral Science & Economics