To profile or not?

Yael Deutsch, Arieh Gavious

Research output: Contribution to journalArticlepeer-review

Abstract

Developing effective inspection processes at border crossings in order to identify violators within large groups of mostly innocent people is an important and difficult task. Passenger profiling is a tool used to deal with this task, but it raises many public concerns and ongoing debates about its usefulness. In this article, we study whether profiling is helpful, how it should be used to maximize its effectiveness, and how social utility is affected by its use. We consider two game models that take place at a crowded border crossing, where passengers are divided into different groups based on their risk profile. The models involve three players: the defender, the attacker, and the passenger recruited to engage in a violation. The defender decides on an inspection policy for every group of passengers. The attacker decides from which group to recruit a single passenger as a violator. The recruited passenger has private knowledge about the likelihood that she/he will engage in the violation or not. We solve the game models and compare their solutions with those of a no-profiling inspection policy. We then study a model with four players, including a social planner who is concerned about social utility, and chooses the defender's inspection resources and policy. We show that the announced profiling policy provides the highest payoffs to the defender. Regarding social utility, profiling is always better than no profiling. However, there are cases where the unannounced profiling policy provides the most social utility.

Original languageEnglish
Pages (from-to)650-665
Number of pages16
JournalRisk Analysis
Volume44
Issue number3
Early online date23 Jun 2023
DOIs
StatePublished - Mar 2024

Bibliographical note

Publisher Copyright:
© 2023 Society for Risk Analysis.

Keywords

  • counterterrorism
  • game theory
  • incomplete information
  • inspection
  • risk analysis

Fingerprint

Dive into the research topics of 'To profile or not?'. Together they form a unique fingerprint.

Cite this