Patrick Egan New York University

“Don’t Just Stand There: Do Something!”

Empirical Evidence for Policy Preferences that Violate Single-Peakedness

with Implications for Social Choice

Patrick EganNew York University

• Question: Under what conditions are policy preferences truly “single-peaked?”

• Design: Ask respondents to rank their preferences over policies on four issues:

– Education– The U.S. prison in Guantanamo Bay– Illegal immigration– The nation’s reliance on foreign oil

• For each policy, respondents ranked preferences over the status quo (Q) and alternatives to its right (R) and left (L).

Question and Design

Issue

Single-peaked preferences Non-single-peaked preferences

Totals

L>Q>R R>Q>L Q>R>L Q>L>R L>R>Q R>L>Q% with non-

single-peaked

preferences

Immigration 35.5 19.7 5.1 4.0 18.3 17.5 35.7 100.0

Foreign Oil 28.4 30.6 10.7 9.2 14.4 6.7 21.1 100.0

Guantanamo 24.7 42.4 9.2 5.2 9.4 9.0 18.4 100.0

Education 34.7 34.9 5.9 9.9 8.7 5.9 14.5 100.0

Results: ranked preferences

Results: pairwise votes derived from rankings

IssueL vs R L vs Q R vs Q

% L % R % L % Q % R % Q

Immigration 57.8 42.2 71.2 28.8 55.4 44.6

Foreign Oil 52.0 48.0 49.4 50.6 51.7 48.3

Guantanamo 39.3 60.7 43.2 56.8 60.9 39.1

Education 53.3 46.7 49.3 50.7 49.4 50.6

Both L and R are preferred to Q

Q is preferred to both L and R

Cycling occurs Only issue on which aggregate ranked

preferences correspond to aggregate marginals

• Hypothesis: a heightened sense of the problem leads people to abandon a moderate status quo

– Challenge: assessments of problem seriousness are endogenous to policy preferences

• Experiment: randomly expose subjects to a reading passage and image that makes the problem more salient

– The treatment exogenously raises subjects’ assessment of problem seriousness on the issue.

• Result: On issues where the treatment successfully raised problem seriousness, the share of voters ranking the status quo last rose significantly.

Identifying a mechanism