§ 5.05. Constraining Discretion

(a) A suspicionless policing policy should be implemented evenhandedly according to criteria developed in advance.

(b) Evenhandedness requires that, within the group targeted for a suspicionless policing action under § 5.04, the procedure be applied to:

  • (1) every person or entity within that group;
  • (2) a subset of that group that is selected on a random or neutral basis; or
  • (3) a subset of that group that there is sound basis for believing is more likely to be engaged in unlawful conduct or pose a greater risk of harm than the rest of the target group.


a. Importance of evenhandedness and generality. This Section works in tandem with the nondiscrimination principle found in the previous Section. It applies after a jurisdiction has identified the group to be subjected to suspicionless policing activity pursuant to §§ 5.03 and 5.04. The core tenet expressed here is that when carrying out the authorized action, government officials in the field should not be able to pick and choose whom to subject to policing actions, but rather should act in an evenhanded manner that avoids arbitrary or discriminatory exercises of discretion. Thus, a policy authorizing inspections of a particular type of business also ought to make clear under what circumstances, if any, subsets of that type of business can be subject to inspection. A policy authorizing sobriety checkpoints ought to specify the criteria for determining the circumstances, if any, under which only a subset of motorists may be stopped at an established checkpoint. The areas to be subject to a camera-surveillance program authorized by a city council should be selected on neutral grounds.

The requirement of evenhandedness furthers a number of important goals. First, it minimizes the risk of arbitrary intrusions by ensuring that the decision of whom to subject to suspicionless policing reflects evidence-based legislative or agency policy, and not the whim or caprice of officers in the field. In the absence of cause, there is a risk that such a decision, if left to individual officers, would be based on impermissible or arbitrary criteria such as the person’s race or physical appearance.

Second, it helps guard against unjustified or ill-conceived programmatic policies by broadly distributing the costs of the program across the target population. The more people or entities that are affected by a program, the greater will be the backlash against one that is irrational or overly intrusive. Otherwise, there is a risk that officials will try to minimize these political costs by concentrating the costs of an authorized program on less-powerful or underrepresented groups, for example, by focusing inspections on less influential business owners within a particular industry, or by locating checkpoints in low-income or minority neighborhoods.

Third, a program that is carried out in an evenhanded manner is likely to be perceived by the target group as both more legitimate and less intrusive. Although airport screenings are time-consuming and annoying, they are not thought to be as intrusive as a stop of a single individual on the airport concourse. Because everyone is subjected to the same screening procedures, individuals are not left to wonder why their bags were searched while others were not. Similarly, sobriety checkpoints are thought to be less intrusive or jarring than a stop of a car singled out by the highway patrol.

b. Implementing evenhandedness. This Section furthers these goals by requiring that either the program apply to the target population as a whole on a universal or neutral basis or there be some legitimate, demonstrable basis for singling out certain groups as opposed to others. Specifically, in the absence of cause, the requirement of evenhandedness should be met in one of three ways: searching every person or entity that falls within the target group (universal application); searching a neutrally selected or random subset of that group (neutral or random application); or—in what amounts to another application of the nondiscrimination principle of § 5.04, but at the implementation stage rather than the policy formation stage—searching a subcategory that is demonstrably more likely to be involved in wrongdoing than the rest of the target population (statistical application).

Suspicionless policing activity should conform to this Section at each stage in the decisionmaking process. For example, in carrying out a legislatively authorized sobriety-checkpoint program, an agency first will need to decide where the checkpoints should be located. Consistent with this Section, an agency may decide to rotate checkpoints along major thoroughfares in different parts of town (neutral application) or to concentrate checkpoints in particular neighborhoods that have experienced higher rates of alcohol-involved crashes or fatalities (statistical application). An agency then will need to decide whether to stop every car that passes through the checkpoint (universal application) or to stop every fifth or tenth car (neutral application).

By conducting suspicionless information gathering and encounters in a universal and neutral manner, agencies can eliminate any possibility that discretion or bias will be used in selecting the individuals or entities to be subjected to such activity. In theory, agencies can do the same by relying on statistical application. However, unless used carefully, the statistical method of reducing bias could actually reinforce it, especially if it relies on data collected from prior discriminatory policing practices. Further, if used to single out a particular person, vehicle, or business, the statistical method could lead to the reintroduction of precisely the sort of discretion that this Section is designed to avoid, and it generally should not be an option unless it can reliably produce the equivalent of individualized suspicion.

However, there are circumstances in which universal or neutral application either would not be possible given the nature of the program or would be appreciably less effective at achieving the program’s goals. For example, an agency may wish to concentrate its inspections on those businesses that have had the highest rates of accidents or consumer complaints; so long as the agency has sufficient information at its disposal to make such a determination and applies its formula according to an administrative plan or algorithm that is transparent, this approach would likewise be consistent with this Section. Similarly, in selecting the site for a sobriety checkpoint, a jurisdiction may wish to focus its resources on neighborhoods with higher rates of alcohol-involved fatalities or a higher concentration of bars and other establishments that serve alcohol. On the other hand, an agency should not limit sobriety checkpoints to certain high-crime areas when instances of impaired driving are just as likely (or more likely) in other neighborhoods, because doing so would not be designed to further the goals of the program and may be perceived as discriminatory.

Of course, this Section would not prohibit suspicionbased policing of individuals or entities that happen to fall within a group targeted by a suspicionless program. Although such actions at times have been analyzed under the “special needs” framework, they should conform to the Principles on suspicion-based policing activities discussed in Chapters 3 and 4.

Reporters’ Notes

This Section addresses two concerns with agency implementation of suspicionless policing activity that courts have routinely recognized. First and most directly, it guards against arbitrary actions by requiring that individuals or entities be selected according to a neutral plan that constrains officer or agency discretion. As courts repeatedly have acknowledged, the “grave danger” when it comes to suspicionless searches and seizures is that, in the absence of suspicion or alternate safeguards, officers may rely on arbitrary or impermissible criteria in deciding whom to search or seize. See the Reporters’ Notes to § 5.01.

Second, the requirement of neutrality and evenhandedness serves as an additional check against unnecessary suspicionless programs by evenly distributing the costs of the programs across the target population—thereby increasing the likelihood of political pushback against unusually intrusive or ill-conceived programs. As Justice Robert H. Jackson stated: “There is no more effective practical guaranty against arbitrary and unreasonable government than to require that the principles of law which officials would impose upon a minority must be imposed generally. Conversely, nothing opens the door to arbitrary action so effectively as to allow those officials to pick and choose only a few to whom they will apply legislation and thus to escape the political retribution that might be visited upon them if larger numbers were affected.” Ry. Express Agency, Inc. v. New York, 336 U.S. 106, 112-113 (1949) (Jackson, J. concurring).

Given the importance of evenhandedness, agencies should ensure that suspicionless policing activities conform to this Section at all stages in the decisionmaking process. Certain categories of such activities, such as subway inspections or sobriety checkpoints, actually involve two stages at which discretion may be exercised. The first stage involves the agency deciding which sites should be subjected to suspicionless action. When, as is often necessary in such situations, policing agencies decide to carry out a suspicionless program in one area as opposed to another, they are unequally distributing benefits (in terms of crime prevention) and burdens (in terms of inconvenience and intrusion). This unequal distribution ought to be justified, ideally by reference to quantified evidence that focusing on the areas chosen for the suspicionless action is more likely to further the goals of the program. Agencies increasingly are relying on data to pinpoint significant public safety problems and make decisions about how best to allocate limited resources. See Lawrence Sherman, The Rise of Evidence-Based Policing: Targeting, Testing and Tracking, 42 Crime & Justice 377, 378, 383 (2013). At the same time, unsophisticated use of such data can perpetuate racialized or otherwise unbalanced policing, without having significant impact on crime rates. See, e.g., William S. Isaac, Hope, Hype and Fear: The Promise and Potential Pitfalls of Big Data, 15 Ohio St. J. Crim. L. 543, 553-558 (2018) (reporting studies); Andrew Ferguson, Crime Mapping and the Fourth Amendment: Redrawing “High Crime Areas,” 63 Hastings L.J. 179, 215-216 (2001).

Courts have acknowledged that jurisdictions can employ a range of criteria in making this determination, so long as the criteria are closely linked to the goals of the program. For example, as the Second Circuit described in MacWade v. Kelly, the New York Police Department selected subway stations for its Container Inspection Program “based on a sophisticated host of criteria, such as fluctuations in passenger volume and threat level, overlapping coverage provided by its counter-terrorism initiatives, and available manpower.” 460 F.3d 260 (2d Cir. 2006). A number of other decisions, from both the lower courts and the U.S. Supreme Court, have highlighted similar types of neutral, evidence-based findings in evaluating suspicionless programs. Compare Lowe v. Commonwealth, 337 S.E.2d 273 (Va. 1985) (upholding sobriety-checkpoint plan developed after extensive research into locations within city where there had been DUI arrest and alcohol-related accidents), with State v. Parms, 523 So.2d 1293 (La. 1988) (sobriety checkpoint unconstitutional because there was “no evidence of [a] basis for the site selection”). See also Indianapolis v. Edmond, 531 U.S. 32, 35 (2000) (drug-interdiction checkpoint locations selected “based on such considerations as area crime statistics and traffic flow”); United States v. Martinez-Fuerte, 428 U.S. 543, 552 (1976) (immigration checkpoints located “on important roads leading away from the border”). A failure to rely on neutral criteria that reflect the stated goals of the checkpoint program can create the impression that decisions are being made on the basis of impermissible criteria, such as race. See, e.g., Angela Caputo, “Chicago Police Sobriety Checkpoints Target Black, Latino Neighborhoods,” Chicago Tribune, May 8, 2015 (alleging that minority neighborhoods in Chicago were targeted with sobriety checkpoints while white neighborhoods with much higher rates of drunk-driving fatalities were not).

Evenhandedness is still more important when it comes to the second stage of a suspicionless program—an officer’s decision to confront a particular individual at the chosen site or a particular entity within the target group. In this context, courts often have required that selection take place either on a universal or neutrally determined basis. For instance, the U.S. Supreme Court has indicated that police at individual checkpoints should stop all vehicles or stop vehicles based on their position in line. Delaware v. Prouse, 440 U.S. 648, 657 (1979) (distinguishing random stops from roadblocks “where all vehicles are brought to a halt or to a near halt”); United States v. Ortiz, 422 U.S. 891, 895-896 (1975) (invalidating checkpoint near the border where officers searched vehicles that “arouse[d] their suspicion,” a procedure that did not limit “to any meaningful extent the officer’s discretion to select cars for search”); see also United States v. Marquez, 410 F.3d 612, 614 (9th Cir. 2005) (upholding referral to secondary checkpoint at airport because it was made on a “completely random” basis); State v. Book, 847 N.E.2d 52 (Ohio Ct. App. 2006) (holding courthouse officer’s security screening was unreasonable when, as a professional courtesy, he did not screen some courthouse visitors he had known for a long time, and there was no objective policy exempting those persons from screening). Courts have made a similar point in drug-testing cases as well. See, e.g., Skinner v. Ry. Labor Executives’ Ass’n, 489 U.S. 602, 622 (1989) (“in light of the standardized nature of the [drug] tests and the minimal discretion vested in those charged with administering the program, there are virtually no facts for a neutral magistrate to evaluate”); Shoemaker v. Handel, 795 F.2d 1136, 1143 (3d Cir. 1986) (approving drug-testing program of jockeys because either “each jockey is required to take a breathalyzer test daily” or urinalysis occurred “by a lottery”). Many police-department policies also recognize the importance of this principle. See New York Police Dep’t, PG-221-16, Vehicle Checkpoints (June 1, 2016) (requiring nonarbitrary stops at vehicle-safety checkpoints, such as stopping every third vehicle); Chicago Police Dep’t, S04-08-05, Sobriety Safety Check Program (Mar. 28, 2003) (requiring nonarbitrary stops at sobriety-safety checkpoints, such as stopping every fifth vehicle; Wash. D.C. Metro. Police Dep’t, GO-PER-100.24, Drug Screening Program (Sept. 11, 2015) (providing that all sworn department members are subject to random drug screening as determined by a computer database); Chicago Police Dep’t, E01-08, Random Drug and Alcohol Testing Program (Feb. 21, 2012) (same).

Despite these precedents, courts in several contexts have dispensed with—or more often, simply ignored—the evenhandedness requirement and have permitted searches and seizures to take place in the absence of any meaningful constraints on officer or agency discretion. This is particularly true in the context of administrative inspections of regulated industries—ranging from coal mines and factories to pharmacies and barbershops—which some courts have upheld despite clear evidence in the record that officers selected a particular location on the basis of a hunch or some other rationale that was unrelated to any systematic administrative plan, and yet may not have qualified as a legitimate suspicion-based search. See, e.g. New York v. Burger, 482 U.S. 691, 694 n.2 (1987) (acknowledging that “it was unclear from the record why, on that particular day, Burger’s junkyard was selected for inspection”); Commonwealth v. Eagleton, 521 N.E.2d 1363 (1988) (inspection of auto-body shop requested by police who were suspicious of the shop because of late-night activity there). Courts have upheld similarly standardless searches of parolees. See Samson v. California, 547 U.S. 843 (2006).

Although courts have emphasized that individuals in these contexts have a “reduced expectation of privacy,” that observation is largely beside the point. As the U.S. Supreme Court itself has recognized, “the ‘grave danger’ of abuse of discretion . . . does not disappear simply because the automobile is subject to state regulation resulting in numerous instances of police–citizen contact.” Prouse, 440 U.S. at 662 (1979); see also Los Angeles v. Patel, 135 S. Ct. 2443, 2456 (noting that the hotel-inspection statute “creates an intolerable risk that searches authorized by it will exceed statutory limits, or be used as a pretext to harass hotel operators and their guests.”). And indeed, there is substantial evidence in the case law that such discretion is occasionally exercised in arbitrary or discriminatory ways. See, e.g., Gordon v. City of Moreno Valley, 687 F. Supp. 2d 930 (C.D. Cal. 2009) (finding that plaintiffs adequately stated a claim against the city for singling out African-American-owned barber shops for “raid-style” inspections); Turner v. Dammon, 848 F.2d 440, 442 (4th Cir. 1988) (noting that police, not state license inspectors, conducted 100 warrantless searches of a particular bar without issuing a single citation).

There may be circumstances in which regulatory prerogatives or individuals’ reduced expectations of privacy counsel in favor of permitting searches on less than probable cause. See, e.g., New Jersey v. T.L.O., 469 U.S. 325 (1985) (permitting searches of students’ belongings at school on the basis of reasonable suspicion); United States v. Knights, 534 U.S. 112 (2001) (upholding search of probationer’s apartment on the basis of reasonable suspicion). But then that lowered suspicion threshold should be the basis for approving the search, not the existence of an administrative scheme.

Table of Contents