More information about the study will be available after the completion of the trial.
Intervention
Information on the intervention is hidden until the end of the trial.
Experimental design
Information on the experimental design is hidden until the end of the trial.
Design details
Randomization method
After the user installs the browser extension (and agrees to the data collection), the extension assigns the user one of the experimental groups. The following example visualizes our approach well. The extension generates a random number between 0 and 1. If the number exceeds 0.4, the user is assigned one of the treatments: if it is between 0.4 and 0.7 to the treatment that uses Perspective API as classifier, if it is above 0.7 to the treatment that additionally uses Alienation as classifier. If it is below 0.4, the user is assigned to the control group: if it is between 0 and 0.2 it is assigned to the control group with no hiding, while if it is between 0.2 and 0.4 then it is assigned to the control group with random hiding.
Between the midline and the endline, we are planning to cross-randomize the respondents in both control groups into two groups that will either experience an engagement intervention or not. Randomization will be performed by the extension.
Randomization unit
Individual
Clusters
N/A
Observations
The number of observations depends on how successful we will be in promoting the browser extension (it will be promoted through advertisement on social media). We estimate to be able to recruit approximately 2,500 users.
Arms
We will randomly assign individuals to treatment groups with the following probabilities: 40% to the control group (equal split between pure control with no hiding and control with random hiding), 30% to a treatment group that uses Perspective API classifier, and 30% to a treatment group that uses Alienation classifier.
Primary outcomes
1. Dehumanization
2. Support for partisan violence
3. Affective polarization
4. Political knowledge
Outcomes 1-4 will be measured in all surveys (baseline, midline, endline).
Heterogeneity:
We will look at heterogeneity with respect to: political affiliation and minority Details are provided below.
Secondary outcomes
1. Spiral of silence
2. Moral disengagement
3. Intergroup empathy
4. Meta-perceptions
5. Committing partisan violence
6. Behavioral outcomes measured by the extension
Outcomes 1-4 will be measured in all surveys (baseline, midline, endline).
Heterogeneity:
We will look at the same angles of heterogeneity as for the primary outcomes.
Sponsors
Name: Civic Health Project
Location: Palo Alto, CA
Url: https://www.civichealthproject.org/
Name: Jigsaw LLC
Location: New York City, NY
Url: https://jigsaw.google.com/
Cited by
No citation data available for this registration yet.