Conditioned Assessment of Subjective Probabilities: Identifying the Benefits of Decomposition

Conditioned assessment of subjective probability decomposes an assessment task by using a related event to define a series of conditional probabilities. These component assessments are subsequently aggregated mathematically. Two experiments compared conditioned and unconditioned (direct) assessments of 32 general knowledge questions by graduate business students. Reliability was estimated through repeat administration of the questionnaire and accuracy was measured with a quadratic scoring rule. Conditioned assessments were more reliable and more accurate than direct assessments, confirming previous research on other decomposition approaches. Conditioned assessment appears to be effective because the mathematical combination of conditional probabilities reduces random response errors and reduces the tendency to produce overly extreme probability judgments. Implications for further research and application are considered.