The United States Supreme Court uses the wrong approach to analyze reasonable suspicion. The Court asks whether, if criminal activity were afoot, an officer would be likely to see what she saw (e.g. furtive gestures or flight from police). The ultimate question we are interested in, however, is whether, given what the officer saw, her conclusion about criminal activity was reasonable. These two questions are different. Even if it is highly likely that an officer would make a set of observations when criminal activity is afoot, it does not follow that criminal activity is itself highly likely when an officer makes these observations. The answer to this latter question is the one we are most interested in. The Court’s analysis of reasonable suspicion is inaccurate and illogical, and it fails to consider all the relevant information.
In this Essay, I propose that a Bayesian approach could improve judicial assessments of reasonable suspicion. Bayesian analysis is designed to deal with conditional probabilities (e.g., how likely is it that criminal activity is afoot given a furtive gesture) and relative likelihoods (of all the possibilities that might explain the furtive gesture, which is the most likely and by how much). Crucially, Bayesian analysis includes all relevant information — especially information about false positives that result from innocent people displaying the same behavior as those engaged in criminal activity. This essay will demonstrate how Bayesian tools might more easily reveal logical gaps in the analysis of reasonable suspicion and help to estimate the “reasonableness” of a given investigation more coherently, consistently, and accurately.
W. David Ball,
The Plausible and the Possible: A Bayesian Approach to the Analysis of Reasonable Suspicion
, 55 Am. Crim. L. Rev. 511
Available at: https://digitalcommons.law.scu.edu/facpubs/966