Panel Paper: Humans or Machines: Implications for Representative Bureaucracy

Saturday, November 10, 2018
8216 - Lobby Level (Marriott Wardman Park)

*Names in bold indicate Presenter

Lael Keiser, University of Missouri and Susan Miller, University of South Carolina

Governments all over the globe increasingly rely on technology to provide services and perform functions that were traditionally provided by public employees. The use of technology to automate government decision-making has raised a number of questions and concerns, some of which are related to the implications of technology use for biases in decision-making. The move toward the use of technology to replace or enhance the activities of public employees has the potential to change radically how citizens interact with their government and how they view service delivery, accountability, and responsiveness. For example, in some cases, technological solutions remove human bias from service delivery; thus, we might expect that the public would view such efforts as leading to greater fairness in service delivery. However, public response to the use of technology often reflects a preference for human interaction and discretion on the part of street-level bureaucrats. In this paper, we draw from the theories of representative bureaucracy, street-level bureaucracy, and e-government to develop hypotheses about citizens’ response to technology and automated decision-making in service delivery.

For our case, we focus on the use of red light cameras to help enforce traffic laws. Enforcement of traffic safety laws has often been controversial, with some concerns centering on potential biases in the decision-making of police officers and other concerns focused on the use of a camera, instead of a police officer, to catch violations. Using a survey experiment, we use this case to test hypotheses regarding the use of technology and fairness in government decision-making. In our survey experiment, we have a mock local news story that explains that a city is considering two options to deal with an intersection where people keep running the red light and causing accidents: 1) stationing an officer there 24 hours a day or 2) installing a red light camera. The vignette explains that the vignette explains that the cost is equivalent for each option.We then ask respondents to rate the fairness of each of these approaches. The treatment is a picture of the city’s police force that varies by racial diversity that is embedded in the news story. The control group receives the same news story without the picture of the police force. We expect that the degree to which the racial makeup of the police force differs from the race of the respondent will be positively related to viewing the red light camera as more fair. We expect this result to be particularly pronounced for African American respondents, for whom race may be particularly salient when interacting with police officers. The results of our experiment will provide insight into perceptions of fairness associated government decision-making in an era of automation. They will also be informative for policy-makers seeking to improve public perceptions of fairness in service delivery.