Valuing police alternatives

We will use a contingent valuation survey to estimate the WTP for a police alternative.

We will use an iterative multiple price list to measure the subject's WTP for the police alternative using an approach similar to Allcott and Kessler (2019) and Holz et al. (2024). Each subject's answers to a series of pairwise decisions maps into a WTP interval: (-\infty,3], (3,12], (12,21], (21,34], (34,46], (46,56], (56,69], (69,81], (81,94] (94,108], and (108,\infty).

We assign one unique WTP for each range. For the eight interior ranges, we assign the mean of the endpoints. For example, we assign a WTP of $7.50 for all responses on (3,12] and a WTP of $75 for all responses on (69,81]. For the unbounded ranges, we assume that the conditional distribution of the WTP is triangular.

The survey randomizes the initial value the subject considers in their pairwise decisions among $21, $46, and $81. We also randomize subjects into conditions in a list experiment. The list experiment will estimate the portion of subjects who truthfully report their WTP in the iMPL. As a robustness check, we will assume that subjects who do not truthfully report their WTP have a $0 WTP and conduct robustness checks using that assumption. We measure the WTP for each component of HEART as the amount of the WTP that the subjects allocate to HEART.

2025-06-05

2025-12-01

The willingness to pay for a police alternative.

See intervention.

The portion of subjects who truthfully reported their WTP.

We will conduct a contingent valuation survey.

The respondents receive a description of the HEART program and are required to correctly answer four comprehension questions to proceed.

After passing the comprehension check, the respondents are randomized into three initial WTP brackets using an iterative multiple price list (iMPL) design. In this task, the survey asked subjects to make a series of binary voting decisions about paying a tax to continue HEART for another year, with values ranging from $3 to over $108. We randomized the initial value each respondent considered between-subjects to be either $21, $46, or $81. Each response adjusted the next offered amount up or down, placing participants into one of twelve WTP ranges. After we pinned down the respondent's WTP we had subjects allocate their stated WTP across the three HEART programs.

Next, we used a list experiment to assess the truthfulness of the respondents' self-reports of their WTP. In this section, we randomly assigned respondents between a control group that received four neutral statements and a treatment group that received the same four statements plus a sensitive fifth item: their stated willingness to pay to fund the HEART program. Participants indicated how many of the statements applied to them but not which ones, allowing them to express support for HEART without directly revealing it. The difference in mean responses between the groups provides an estimate of the share of respondents who truthfully support the program at that price point.

The survey ended with questions on demographics and attitudes toward public safety. These included preferences for police vs. nonpolice responses in crisis scenarios.

Randomization is done by a computer.

Individual

No

Our target sample is 700-1000 depending on recruitment from yougov

Our target sample is 700-1000 depending on recruitment from yougov

Approximately 1/3 of subjects receive each initial value in the contingent valuation survey. About 1/2 subjects will be in each of the conditions in the list experiment.