Multiple-Ranked Choice Identification in Exploded Logit Systems
Single-choice discrete demand models cannot distinguish whether an unexpected selection stems from unobserved individual preference weights or random idiosyncratic error terms. Without observing sequential preference orderings, researchers cannot reliably estimate the population distribution of taste parameters.
Picture this
If a person picks a distant restaurant once, it might be a random fluke. But if that person is asked to rank their top three choices and repeatedly picks distant restaurants featuring high ratings, it proves a consistent preference for quality over travel distance.
What the evidence says
Between 27.4% and 50.9% of parents who listed their guaranteed home-school first or second submitted subsequent choices beyond the default, providing within-person substitution patterns that isolated preference parameters from random unobserved utility shocks.
- Who
- N = 36,887 human students in Charlotte-Mecklenburg Schools, North Carolina, submitting up to three ranked choices.
- How
- Exploded multinomial mixed logit estimation where each submitted rank sequentially removes the chosen alternative from the decision set, generating within-individual choice set variation.
What to do
Elicit ranked preferences in choice surveys to dynamically modify individual choice sets and separate systematic utility weights from extreme-value error terms.
From the source
"The multiple-ranked responses provided by each parent creates variation in the choice set by effectively removing the prior chosen school from the subsequent choice set. This choice-set variation allows us to estimate the distribution of values parents place on school characteristics from observed substitution patterns for each individual"
Heterogeneous Preferences and the Efficacy of Public School Choice