Understanding Demand for Police Alternatives: Preferences for Police Engagement and Alternatives

Respondents are randomly assigned to one of three video treatment arms: (1) contol (2) government hotlines, (3) Don’t Call the Police. We test whether treatments change respondents' level of demand for police or police alternatives.

The control group will serve as a placebo, providing information about unemployment, a topic unrelated to police. The second treatment will offer information on alternative governmental services (211, 311, and 988), referred to as an active control group to help disentangle the effects of priming from genuine belief updating. The final treatment will present information about vetted nongovernmental options prioritizing minimizing police involvement.

This online survey uses a 3x1 design to measure baseline demand for police and for police alternatives in specific issues and scenarios among a representative sample of U.S. adults, and evaluates how demand is affected by information treatments. We test whether treatments change respondents' level of demand for police or police alternatives.

We assign respondents to one of three video treatment arms: (1) Control on unemployment, (2) government hotlines, which provides information on 311,988, and 211 -- government resources that can serve as viable alternatives to a police response, (3) Don't Call the Police, which presents the Control video information along with information about a vetted database of community-based resources that can serve as viable alternatives to a police response.

2024-07-30

2024-08-13

1) Police index z-score
2) Violent index z-score
3) Nonviolent index z-score
4) Likelihood of calling the police for each of the five scenarios
5) Dial 988/911 for the suicidal ideation scenario

1) Police index equals an equally weighted average of z-scores of the five scenarios
2) Violent index equals an equally weighted average of z-scores of the violent crime and domestic violence scenarios
3) Nonviolent index equals an equally weighted average of z-scores of the homelessness and mental health scenarios
4) Respondent's willingness to call the police (0 to 100) for scenarios involving:
a violent crime; domestic violence; homelessness; mental health (erratic behavior in public); mental health (threats of suicide)
5) Binary variable (0: no, 1:yes) if dial 988/911 for a suicidal ideation scenario

see explanations

1) Respondent's confidence in previous response for likelihood of calling the police for each scenario (scale: 0-100)
2) Respondent's preference for who should respond (police, social worker, no one) to each of the scenarios (violent crime, domestic violence, homelessness, mental health)
3) Interest in learning about website j={Dontcallthepolice.com, 911alternatives.com, not intested} ---binary variable (0: no, 1:yes) if for website j

Respondents are randomly assigned to one of three video treatment arms: (1) Control (2) government hotlines, (3) Don’t Call the Police.

Participants will complete two surveys, separated by one week. We will recruit participants for the first survey in the first week using Prolific. We aim to recruit 3000 participants in the first survey. A week later, we will advertise the survey only to those who completed the first survey. We expect at least 30\% of the participants who complete the first survey to complete the second survey. To maximize the number of participants that complete the second survey, we will leave the second survey open for one week. The second survey will have a bonus payment for correct answer. The main analysis only considers respondents who answered both surveys.

Randomization takes place in the first survey. In the second survey, respondents will be asked which governmental number they would call in a suicidal ideation scenario

To conduct randomization, we use Qualtrics.

individual

No

cluster at the individual level.

3000 respondents - final sample will be a function of the second survey.

(1) About 1000 individuals in Control, (2) About 1000 in Government hotlines (3) About 1000 individuals in Don’t Call the Police.

N/A