Data for: Public opinion on police misconduct

Published Mar 22, 2026 on Dryad. https://doi.org/10.5061/dryad.gtht76j0q

Data files

Mar 22, 2026 version files 1.11 MB

analysisdata_vino_abridged.csv

874.31 KB
analysisdata_vino_abridged.RData

236.41 KB
README.md

4.05 KB

Abstract

High-profile cases of police violence have given rise to contentious debate across the globe, which has often erupted into protest and unrest. These debates hinge on disagreements about whether police use of force is wielded appropriately. Yet scholars know little about why individuals interpret violent police encounters differently. To address this gap, we fielded a survey experiment in Germany which is designed to probe whether the victim's identity matters to public opinion (N=15,941). Our novel design mimics the way news events unfold sequentially over time, allowing us to measure whether eventual discriminatory beliefs persist even in the face of mounting evidence. We show that respondents are less likely to consider the actions of the police officers to constitute misconduct when the victim has an immigrant name compared to a traditionally German name. Importantly, these discriminatory beliefs remain even when respondents are informed about various forms of evidence confirming that misconduct

Dataset DOI: 10.5061/dryad.gtht76j0q

Description of the data and file structure

This dataset contains an abridged version of the survey data of the VINO project "Variation in Institutional Oversight of Police Miconduct". We excluded a subset of the variables to avoid that individuals can be identified via indirect identifiers. The survey was implemented by YouGov's online panel with a representative sample of the German population in 2024. YouGov recruited the sample with quotas on gender, age, education, household income, and state of residence, which were adjusted to the average distributions in the German population.

Files and variables

File: analysisdata_vino_abridged.csv and analysisdata_vino_abridged.RData

Variable Name	Description
ahmed	Binary treatment variable coded as 1 if respondents in vignette 1 receive “ahmed”-treatment
scale_misconduct_v1_1	Numeric variable of respondents’ evaluations of police misconduct on a scale from 0=no misconduct to 9=severe misconduct (vignette 1, stage 1)
scale_misconduct_v1_2	Numeric variable of respondents’ evaluations of police misconduct on a scale from 0=no misconduct to 9=severe misconduct (vignette 1, stage 2)
approval_v1_3	Binary variable coded as 1 if respondents approve with the decision of the police oversight body that misconduct occurred (vignette 1, stage 3)
rural	Binary variable coded as «1» for all respondents that indicated “village” in the question about the place of residence
employed	Binary variable coded as “1” for all respondents that indicated that they have a paid work
difpay_often	Binary variable coded as “1” if respondents have indicated that they “usually” have difficulties to pay their bills at the end of the month
left_right	Numeric variable from 1 to 10 indicating the self-identification on a scale from 1=left to 10=right
religious	Binary variable coded as 1 if respondents answer that they belong to a religious group or confession
level_edu	Ordinal variable ranging from 1=not finished school to 4=abitur
manipass_name	Binary variable coded as 1 if respondents passed the manipulation test of whether the resident has a name that indicates a migration background
mean	Value on the latent left-right scale derived through Aldrich McKelvey scaling
weight	Weight indicator

Code/software

R code is used to prepare and analyze the data.

Human subjects data

This study obtained ethical approval from Aalborg University's Research Ethics Committee on August, 2024. Survey participation was based on informed consent. No individuals are identifiable in the dataset, all information is fully anonymized with unique case IDs. We also ensured that individuals cannot be indirectly identified through combinations of variables. Therefore, the dataset published on Dryad excludes the individual identifiers "age", "gender", "children", and "born_in_germany".