Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Experimental School Choice with Parents

Mikhail Freer, Thilo Klein, Josué Ortega

Abstract

We conduct the first laboratory school choice experiment in which parents-the relevant decision makers in the field-are the experimental subjects. We compare Deferred Acceptance (DA) with two manipulable but potentially more efficient alternatives: Efficiency-Adjusted Deferred Acceptance (EADA) and the Rank-Minimizing mechanism (RM). We find that all mechanisms are frequently manipulated, with no significant differences in truth-telling rates. Parents and students manipulate at similar rates, supporting the external validity of student-based experiments, though students make significantly more obvious errors, suggesting parents' deviations are more deliberate. Despite widespread manipulation, the predicted welfare-stability tradeoff largely survives: DA never produces Pareto-efficient allocations yet generates little justified envy; whereas RM delivers substantial efficiency gains at a meaningful stability cost. EADA occupies a middle ground: its efficiency gains over DA are modest and imprecisely estimated yet double justified envy. Higher cognitive ability is associated with more deviations, and under EADA with worse outcomes. While DA does not induce truth-telling, it is the only mechanism in which manipulation never pays off and rarely changes outcomes.

Experimental School Choice with Parents

Abstract

We conduct the first laboratory school choice experiment in which parents-the relevant decision makers in the field-are the experimental subjects. We compare Deferred Acceptance (DA) with two manipulable but potentially more efficient alternatives: Efficiency-Adjusted Deferred Acceptance (EADA) and the Rank-Minimizing mechanism (RM). We find that all mechanisms are frequently manipulated, with no significant differences in truth-telling rates. Parents and students manipulate at similar rates, supporting the external validity of student-based experiments, though students make significantly more obvious errors, suggesting parents' deviations are more deliberate. Despite widespread manipulation, the predicted welfare-stability tradeoff largely survives: DA never produces Pareto-efficient allocations yet generates little justified envy; whereas RM delivers substantial efficiency gains at a meaningful stability cost. EADA occupies a middle ground: its efficiency gains over DA are modest and imprecisely estimated yet double justified envy. Higher cognitive ability is associated with more deviations, and under EADA with worse outcomes. While DA does not induce truth-telling, it is the only mechanism in which manipulation never pays off and rarely changes outcomes.

Paper Structure

This paper contains 52 sections, 13 equations, 6 figures, 27 tables.

Table of Contents

  1. Introduction
  2. Results preview.
  3. Related Literature
  4. Experimental School Choice.
  5. External Validity and Subject Pools.
  6. EADA and RM Properties.
  7. Model
  8. Three Mechanisms of Interest
  9. Deferred Acceptance (DA).
  10. Efficiency-Adjusted Deferred Acceptance (EADA).
  11. Rank-Minimizing (RM)
  12. Experimental Design
  13. Preferences and Priorities
  14. Information and Interface
  15. Timeline
  16. ...and 37 more sections

Figures (6)

  • Figure 1: Information table displayed to subjects.
  • Figure 2: Ranking interface.
  • Figure 3: Frequency of Pareto-efficient allocations by mechanism (recombinant estimates with 95% CI)
  • Figure 4: Average assigned rank by mechanism (recombinant estimates with 95% CI)
  • Figure 5: Share of students with true justified envy by mechanism (recombinant estimates with 95% CI)
  • ...and 1 more figures