Introduction
This is an HTML rendering of an R Markdown script for the paper under review: Public Perceptions of Police Use-of-Force: Legal Reasonableness and Community Standards.
The report was created on 2019-08-01. The underlying data is taken from the General Social Survey (GSS). The data repository for the GSS is located at https://gss.norc.org/get-the-data and the original dataset can be retrieved there. The truncated dataset used here is available from the authors upon request.
The code below produces plots using the ggplot and plotly packages. The formated tables were produced with stargazer.
Abstract
How do public expectations of police use-of-force align with the strict professional and legal guidelines under which police officers train and operate? This is a largely unexamined but salient question in the use-of-force literature, and is important given the ongoing public discourse regarding police use-of-force, community standards, and perceived gaps between the two. This study focuses on two main research questions: Are substantial portions of the public predisposed to disapprove of legally reasonable police use-of-force? If so, what are the principal correlates of those disapproving attitudes? We analyze responses (n=20,781) to General Social Survey (GSS) questions from 1990 to 2018 entailing police use-of-force scenarios that are prima facie legally reasonable. We find a substantial proportion of GSS respondents have expressed their disapproval of legally reasonable, justifiable police uses of force over the entire period, and such disapproval has increased over time. Causes and policy implications of this misalignment are discussed.
Keywords: use-of-force; police; citizen perceptions; procedural justice; General Social Survey; negative binomial regression
Dependent Variable
We are primarily interested in “no” answers from the following questions in the General Social Survey. These are the scenarios for which there are legally reasonable justification for police use-of-force under the Graham doctrine:
- POLHITOK: Are there any situations you can imagine in which you would approve of a policeman striking an adult male citizen?
- POLESCAP: Would you approve of a policeman striking a citizen who was attempting to escape from custody?
- POLATTAK: Would you approve of a policeman striking a citizen who was attacking the policeman with his fists?
There are two other questions from this block of GSS questions. These two questions highlight scenarios where use-of-force would not be legally justified. We do not deeply analyze responses to these questions, but they are sometimes displayed in figures, so are defined here:
- POLABUSE: Would you approve of a policeman striking a citizen who had said vulgar and obscene things to the policeman?
- POLMURDR: Would you approve of a policeman striking a citizen who was being questioned as a suspect in a murder case?
Summary Table of Dataset
| Statistic | N | Mean | St. Dev. | Min | Pctl(25) | Pctl(75) | Max |
| Year | 20,781 | 2,003.739 | 8.879 | 1,990 | 1,996 | 2,012 | 2,018 |
| REALINC | 20,781 | 29,891.280 | 31,860.040 | 0 | 8,595 | 38,610 | 155,140 |
| POLATTAK | 20,781 | 0.100 | 0.300 | 0 | 0 | 0 | 1 |
| POLESCAP | 20,781 | 0.271 | 0.444 | 0 | 0 | 1 | 1 |
| POLMURDR | 20,442 | 1.894 | 0.308 | 1.000 | 2.000 | 2.000 | 2.000 |
| POLABUSE | 20,524 | 1.914 | 0.281 | 1.000 | 2.000 | 2.000 | 2.000 |
| POLHITOK | 20,781 | 0.301 | 0.459 | 0 | 0 | 1 | 1 |
| POLVIEWS | 20,045 | 2.087 | 0.787 | 1.000 | 1.000 | 3.000 | 3.000 |
| SIZE | 20,781 | 335.012 | 1,161.479 | 0 | 7 | 111 | 8,175 |
| RACE | 20,781 | 1.284 | 0.592 | 1 | 1 | 1 | 3 |
| SEX | 20,781 | 1.546 | 0.498 | 1 | 1 | 2 | 2 |
| DEGREE | 20,746 | 1.546 | 1.190 | 0.000 | 1.000 | 3.000 | 4.000 |
| AGE | 20,735 | 46.522 | 17.228 | 18.000 | 33.000 | 59.000 | 89.000 |
| ID | 20,781 | 1,326.048 | 876.323 | 1 | 616 | 1,918 | 4,510 |
| BALLOT | 20,781 | 2.499 | 0.500 | 2 | 2 | 3 | 3 |
| DV | 20,781 | 0.671 | 0.928 | 0 | 0 | 1 | 3 |
Negative Binomial Model Results
| Dependent variable: | |
| DV | |
| SEX | 0.368*** |
| (0.020) | |
| RACE | 0.406*** |
| (0.014) | |
| POLVIEWS | -0.120*** |
| (0.013) | |
| DEGREE | -0.148*** |
| (0.009) | |
| AGE | 0.001 |
| (0.001) | |
| SIZE | 0.0001*** |
| (0.00001) | |
| REALINC | -0.00000*** |
| (0.00000) | |
| Year | 0.010*** |
| (0.001) | |
| Constant | -21.156*** |
| (2.196) | |
| Observations | 19,972 |
| Log Likelihood | -21,063.780 |
| theta | 3.635*** (0.286) |
| Akaike Inf. Crit. | 42,145.570 |
| Note: | p<0.1; p<0.05; p<0.01 |
Data Visualizations
Scroll through the left pane to see the available figures. Click on the name of the figure and it will load the visualization. Most of the visualizations are interactive, so play around!
Line graph all five answers by color
Scatter and line graph of three legally unreasonable responses
Line graph 2 reasonable responses by color
Scatter and line graphs of each response with years
Combined Scatter & Line Plot (Not Interactive)
Stacked density of the dependent variable (not included yet)
Different Linear Fits
Negative binomial model or poisson?
Though not a formal test, the plots show how a negative binomial versus a poisson fit on the data appear. This is not an interactive graph.
## Warning: Ignoring unknown parameters: family## Warning in theta.ml(Y, mu, sum(w), w, limit = control$maxit, trace =
## control$trace > : iteration limit reached
## Warning in theta.ml(Y, mu, sum(w), w, limit = control$maxit, trace =
## control$trace > : iteration limit reached
Plotting Legally Unreasonable Answers with a Negative Binomial Method
This plot repeats an earlier plot, but here we repeat it with a more appropriate "glm.nb" method. This is not an interactive plot. 