Discretion: using judgment fairly and lawfully

In LES-101 the class learned that police discretion is unavoidable: there are far more violations of law than any service could enforce, and officers must choose. This lesson looks at discretion as an ethical problem. Desjardins starts with numbers. In 2025, Port Ardent officers made 54,300 calls for service and observed many thousands of traffic violations, minor drug offences, noise complaints and public disorder incidents that never became calls. Officers chose to act formally on a fraction of them. Each choice was a small exercise of state power, and taken together those choices decide who in the city experiences the law as a ticket, a warning, an arrest or nothing at all. Fair discretion means similar cases are handled similarly, for reasons the officer could state openly. Unfair discretion means the outcome depends on who the person is, how the officer is feeling or what the officer stands to gain.

Why discretion exists

Kenneth Culp Davis, whose 1969 book Discretionary Justice shaped the study of administrative power, argued that discretion is both necessary and dangerous. It is necessary because laws are written in general terms and cannot fit every situation, because resources are limited and because rigid enforcement would often be cruel or absurd. A strict policy of arresting every person found with a small amount of a prohibited drug would fill cells and courts without making anyone safer. A strict policy of ticketing every driver going slightly over the limit would consume patrol time better spent elsewhere. Discretion lets officers fit the response to the situation and the purpose of the law.

It is dangerous because it is largely invisible. When an officer decides not to act, there is usually no record. The person who receives a warning rarely complains, and the person who receives a ticket rarely knows that someone else in the same situation was let go. Davis argued that discretion should be confined, structured and checked: confined by clear limits, structured by policies and guidelines that say what factors matter, and checked by supervision and review. Much of modern police management, from charging policies to data collection on stops, can be understood as an effort to do what Davis recommended.

Legitimate and illegitimate factors

Not every factor that influences a decision is a fair one. Legitimate factors relate to the purpose of the law and the facts of the case: the seriousness of the conduct, the risk to others, the strength of the evidence, the person's history of similar conduct, the wishes and safety of the victim, the person's age and circumstances where the law recognizes them, and the likely effect of each response. Illegitimate factors include the person's race, ethnicity, religion, gender, sexual orientation, disability or social status; the officer's personal feelings about the person; personal gain, including favours or relationships; and organizational pressures such as quotas that have nothing to do with the case. Section 15 of the Charter and provincial human rights codes make some of these factors unlawful as well as unethical.

Some factors sit in a grey zone. Research since the 1960s, beginning with studies by Donald Black and Albert Reiss, has found that a person's demeanour, meaning how cooperative or hostile the person is, strongly predicts whether officers make an arrest. Is that fair? Partly. Hostility can signal risk or an ongoing offence, and cooperation can show that a warning will be enough. But demeanour is also shaped by fear, mental illness, intoxication, past experience with police and cultural differences in communication, and treating disrespect toward an officer as a reason for harsher treatment confuses the officer's ego with the public interest. Desjardins tells the class that being rude to a police officer is not an offence, and that officers who punish rudeness are using public power for private satisfaction.

How discretion is structured

Legislatures, courts and services have structured discretion in several ways. Some laws remove or limit it. Most provinces have charging policies for intimate partner violence that direct officers to lay a charge whenever there are reasonable grounds, regardless of the victim's wishes, because victims are often under pressure to withdraw. Other laws require officers to consider alternatives. The Youth Criminal Justice Act requires officers, before starting judicial proceedings against a young person, to consider whether it would be sufficient to take no further action, warn the young person, administer a caution or refer the young person to a community program. Service policies add guidance on matters such as when to ticket, when to use discretion for minor drug possession and when a supervisor must approve a decision.

Structuring discretion involves trade-offs. Mandatory rules promote consistency and protect vulnerable people, but they can produce unjust results in unusual cases and can discourage some victims from calling. Research on mandatory arrest for domestic violence, starting with the Minneapolis experiment in the early 1980s and its replications, produced mixed results: arrest appeared to deter some offenders and increase violence by others. Flexible guidelines preserve judgment but rely heavily on the officer's fairness. Most services combine both, with firm rules where the risks of discretion are high and guidance elsewhere.

Consistency: the test of fair discretion

Because individual decisions are hard to judge in isolation, fairness is often best assessed by comparing patterns. If two officers patrol similar areas and stop similar drivers for similar reasons, but one gives warnings in most cases and the other almost never does, at least one of them is applying discretion in a way that needs explaining. The difference might be legitimate: one may work a zone with more collisions. It might also reflect personal style, bias or pressure. Port Ardent's traffic unit started reviewing warning rates in 2024 after a complaint that drivers from one neighbourhood were ticketed more often. The code below shows the kind of comparison the unit makes, using stops for the same offence, speeding between 15 and 30 over the limit, during daytime hours.

# Share of daytime stops ending in a warning for speeding 15 to 30 over, by officer (Port Ardent traffic unit, 2025)
from math import sqrt
stops = {"officer A": (118, 52), "officer B": (96, 41), "officer C": (130, 18), "officer D": (88, 39)}
total_stops = sum(n for n, w in stops.values())
total_warn = sum(w for n, w in stops.values())
p_all = total_warn / total_stops
print(f"unit warning rate {p_all:.2f}")
for name, (n, w) in stops.items():
    p = w / n
    se = sqrt(p_all * (1 - p_all) / n)
    z = (p - p_all) / se
    note = "  <- review" if abs(z) > 2.5 else ""
    print(f"{name}: {w:3d} warnings in {n:3d} stops, rate {p:.2f}, z {z:+.1f}{note}")

unit warning rate 0.35
officer A:  52 warnings in 118 stops, rate 0.44, z +2.1
officer B:  41 warnings in  96 stops, rate 0.43, z +1.6
officer C:  18 warnings in 130 stops, rate 0.14, z -5.0  <- review
officer D:  39 warnings in  88 stops, rate 0.44, z +1.9

Officer C warns far less often than colleagues, and the gap is too large to be chance. That does not prove unfairness. The review that followed found that officer C mostly worked a school zone where the unit had agreed to enforce strictly after two children were struck the year before. Once stops in the school zone were separated out, officer C's rate was close to the unit average. The lesson, Desjardins says, is that data raise questions and supervisors answer them; patterns are the start of a conversation, not a verdict. Lesson 9 applies the same idea to differences between groups of people rather than officers.

Quotas, performance measures and pressure

Discretion can be distorted by the organization as well as by the individual. When officers are evaluated on the number of tickets, arrests or street checks they produce, discretion tilts toward formal action whether or not it serves the purpose of the law, and it tilts toward the people and places where formal action is easiest to generate. Several American states have passed laws banning ticket or arrest quotas, and many services in both countries have dropped explicit quotas in favour of broader performance measures. Informal pressure persists, however, in the form of supervisors who compare activity numbers at platoon briefings, as Malhotra's sergeant did in the Lesson 1 example. Desjardins argues that the ethical response is not to ignore performance but to measure the right things: response times, case outcomes, problems solved, complaints, community feedback and the quality of investigations, not raw counts of enforcement.

Officers who feel pressure to produce numbers face a real ethical choice. The decision model from Lesson 3 helps: the pressure is a reason that would not survive the publicity or reversibility tests. An officer can meet a legitimate expectation of activity by focusing enforcement where it matters most, such as collision-prone intersections, rather than by issuing marginal tickets to whoever is easiest to stop. And an officer who believes a numbers target is driving unfair enforcement can raise the concern with a supervisor, the association or, if necessary, the professional standards bureau.

Discretion and the person in front of you

Discretion also has a personal side that numbers miss. Two drivers stopped for the same offence may be in very different situations: one late for a job interview, one fleeing an abusive partner, one impaired. Good discretion pays attention to such differences when they are relevant to the purpose of the law and the risk involved, while resisting differences that should not matter, such as which driver reminds the officer of a friend. Desjardins suggests a simple habit: before deciding, state the reason for the decision in one sentence that refers only to the facts and the purpose of the law. If the sentence would sound wrong read aloud to the person, to the next person stopped or to a supervisor, the reason probably is wrong. He adds that officers should record their reasons for significant discretionary decisions, such as a decision not to charge an assault, because the record protects both fairness and the officer.

Jesse points out that discretion also includes choosing not to use enforcement at all. In his community, he says, the officers people trust most are the ones who use the law last: they mediate disputes, connect people with elders or services and arrest only when they must. Desjardins agrees that the most important discretionary decision is often whether a situation needs the criminal law at all. Diversion to mental health services, crisis lines such as 988, launched in Canada in November 2023, community justice programs and restorative approaches are all exercises of discretion, and their availability shapes how fair discretion can be.