Description
Elicit each person's indifference threshold for a consequential step, compute the chance the step is needed under each design, and estimate the share for whom each option is rational.
A decision tree with averaged preferences answers what the typical person should do, but people's attitudes to outcomes vary widely. Pauker and colleagues, evaluating a prenatal screening programme, elicited from 338 prospective parents the burden of each bad outcome on a 0-100 scale, expressed as the lowest probability at which they would choose an intervention. They modelled each testing sequence, used Bayes' rule to compute the probability of the condition at the decision point, and for each person compared that probability with their threshold. The result is the share of people for whom each policy is the logical choice, plotted against uncertain parameters such as test accuracy. Because attitudes were very spread out (mean 33, SD 31), the programme's value depended on the distribution, not on the average.
Attributes
- Discipline
- Game Theory & Decision Science
- Loss type
- Unmigrated
- Stage
- Strategy, Definition
- Data
- Mixed
- Effort
- Weeks
Steps
- Model the process or programme as a sequence of steps or tests with their known accuracies.
- For each respondent, elicit a threshold: the lowest probability of the bad state at which they would choose the intervention, anchored on a 0-100 burden scale.
- For each design, compute the probability of the bad state at the decision point with Bayes' rule.
- Compare each respondent's threshold with that probability to decide whether the option is their logical choice.
- Tally the proportion for whom each option is right under each design, and repeat under pessimistic or dependent assumptions.
- Plot acceptance proportions against the most uncertain parameters.
Sources
- Decision Making: Descriptive, Normative, and Prescriptive Interactions (Bell et al., 1988)
Also known as
- Per-respondent decision analysis
- Distribution of individual decision thresholds
- Attitude-integrated policy analysis
Categories
- Strategy & Collaboration
- Game Theory & Decision Science