Self-Selection Policing: What Minor Offences Can Reveal
Self-selection policing uses minor offences as lawful triggers to look for wider criminality. The evidence supports the idea that serious offenders are often offence-versatile; applying the same logic to police misconduct is plausible, but still needs testing.
Audio edition
≈ 8 min · narrated
Audio edition
≈ 8 min · on-device voice
Self-selection policing starts from a practical observation: some people involved in serious crime also commit minor offences that bring them lawfully to police attention. The minor offence can therefore be more than a disposal problem. It can be an investigative clue.
Jason Roach and Ken Pease developed the approach over a series of studies and later set it out in Self-Selection Policing: Theory, Research and Practice. Roach’s wider work on practical police psychology is profiled in this site’s influential-people series. Their formulation is deliberately different from profiling. The trigger is something the person has done — a minor offence or infraction — rather than their identity, neighbourhood or an officer’s prior suspicion. Roach & Pease, 2016.
That idea is useful, but easy to overstate. The research does not show that minor offending is a reliable badge of serious criminality in general. It shows that some carefully selected minor offences identify groups with higher levels of wider offending than relevant comparison groups. The distinction is what separates a testable policing tactic from a hunch.
A trigger offence should justify a closer look. It should never be treated as proof of the serious offence you hope to find.
The original idea was behavioural selection
Roach’s 2007 paper described offender self-selection as an additional method for finding active serious offenders. Instead of beginning with a list of known offenders and waiting for grounds to stop them, police begin with a minor offence that has already provided lawful grounds for attention and ask whether further proportionate checks are justified. Roach, 2007.
The method depends on offence versatility. If serious offenders committed only serious offences, minor infractions would tell police very little. If some serious offenders also repeatedly commit driving, acquisitive or regulatory offences, then particular low-level behaviours may provide useful opportunities for detection.
Roach and Pease’s later book reviews this theory and the empirical work behind it, including research on visitors to a young offenders’ institution, driving offences and failure to produce driving documents. The authors describe SSP as complementary to ordinary investigation, not a replacement for witness evidence, CCTV, intelligence or known-offender methods. Roach & Pease, 2016.
That restraint is important. The strongest version of self-selection policing is not “small offenders are really big offenders”. It is “certain observable offences may increase the expected value of lawful further enquiry”.
Driving research shows the mechanism, not a universal rule
Several studies have tested driving-related triggers.
Roach’s work on drivers who failed to produce documents after an HO/RT1 requirement found that non-producers were more likely than producers to have recorded offending histories, to have multiple previous offences and to have offended more recently. Roach, 2007.
His later study of people caught driving while disqualified examined whether that offence could function as a self-selection trigger. The wider literature reviewed in the paper links serious driving offences with broader criminality, and the study asks whether disqualified drivers can help police identify active offenders beyond the motoring offence itself. Roach, Policing: A Journal of Policy and Practice.
These findings make intuitive operational sense. Traffic enforcement already creates lawful encounters in which identity and driving status are checked. Where research shows that a particular offence is associated with wider offending, a proportionate additional check may have a higher yield than the same check applied randomly.
But association is not certainty. Even a useful trigger will identify many people who are not involved in serious crime. That is why a force needs to know not only how often the trigger catches a serious offender, but how often it does not.
The false-positive rate is part of the tactic
Self-selection policing is attractive partly because it sounds efficient: use an encounter that is happening anyway to uncover more serious offending.
Efficiency cannot be judged from the success stories alone. A force needs the denominator.
If 100 people commit a candidate trigger offence and 10 are linked to serious active offending, that may or may not justify the additional police activity depending on what the extra check involves. If the check is a quick intelligence query, the threshold may be low. If it leads routinely to searches, prolonged detention or intrusive investigation, the evidential case needs to be much stronger.
The important measures are therefore familiar ones:
- how common the trigger offence is;
- how often the trigger identifies the target serious offending;
- how many people are checked without anything further being found;
- whether the relationship remains after obvious confounders such as age, offending history or police-contact frequency are considered;
- whether the tactic is applied consistently across groups.
Without those measures, “self-selection” can become a respectable label for selective suspicion.
Why the approach can be less discriminatory than person-based targeting
The conceptual advantage of SSP is that the initial selection criterion is observable offending behaviour.
That does not make the tactic automatically fair. Enforcement itself can be uneven. Police may detect the same minor offence more often in some neighbourhoods or groups because those places receive more patrol, surveillance or traffic enforcement. A behaviour-based trigger can therefore inherit disparities from the policing system that detects the behaviour.
The safeguard is transparency. Forces should publish or internally specify the trigger offence, the evidence supporting it, the additional check authorised by it and the outcomes produced. If one trigger generates poor yield or disproportionate impact, it should be changed or stopped.
This is precisely where evidence-based policing improves the idea. The question is not whether an experienced officer believes that a particular minor offence “often leads to something”. The question is whether the data shows that it does.
The misconduct analogy is interesting — and not yet established
The same mechanism has an obvious appeal inside policing.
If serious offenders are offence-versatile, perhaps officers who later commit serious misconduct also show earlier patterns of minor rule-breaking: dishonesty in routine records, repeated incivility, inappropriate access to systems, poor use-of-force recording, driving breaches or other low-level conduct indicators.
That is a plausible hypothesis. It is not the same evidence base.
The offender SSP studies examine offending behaviour and criminal histories. They do not establish that minor police rule breaches predict later serious misconduct, nor which behaviours would be valid triggers. Importing the concept directly would risk exactly the error SSP is designed to avoid: selecting triggers because they feel intuitively suspicious rather than because they have been shown to predict the outcome of interest. How far that professional intuition deserves trust in the first place is the question examined in The Copper’s Nose – Can We Trust It?
Professional-standards teams have a better option. Test it.
Forces hold historical misconduct, complaint, vetting, use-of-force, driving, sickness and other workforce data. Subject to lawful governance and careful handling of sensitive data, researchers could examine whether specified minor indicators are associated with later serious misconduct and whether they add useful information beyond existing risk factors. The team-level version of that question — whether the same data could identify harmful team cultures before serious misconduct — is examined in Before Serious Misconduct: Can Police Data Identify Harmful Team Cultures Earlier?
A useful study would ask:
- which earlier behaviours are genuinely associated with later serious misconduct;
- how large the association is;
- how many false positives each indicator produces;
- whether the relationship differs across roles or demographic groups;
- whether the indicator is simply a proxy for exposure, such as working in a role that generates more complaints;
- whether intervention after the early indicator changes the later outcome.
Until those questions are answered, “self-selection policing for bad cops” should remain a research proposition rather than an operational rule.
What a force could test now
The original SSP idea is mature enough to support local experimentation without pretending that every trigger is already known.
A force could select one candidate minor offence for which there is published or local evidence of wider offending, define the exact additional check officers will make, and run a prospective evaluation for a fixed period. The outcome should include serious offending uncovered, officer time, disproportionality and false-positive rate.
That is a much stronger implementation model than issuing a briefing telling officers to “look deeper” whenever something feels wrong.
The professional-standards extension can be tested in parallel as retrospective research, without making employment decisions from an unvalidated indicator.
The principle behind self-selection policing is worth keeping because it is modest: pay attention to what people reveal through their own behaviour. The discipline around it matters just as much. The behaviour has to be a demonstrably useful trigger, the extra enquiry has to be lawful and proportionate, and the person committing the minor offence remains a person suspected of a minor offence unless evidence establishes something more.
That is what turns a clever observation into defensible policing.
Sources and further reading
Discussion questions
- 01
Which minor offences in your force area actually predict wider serious offending, rather than merely feeling suspicious?
- 02
What safeguards would stop a self-selection approach drifting from behaviour-based triggers into profiling?
- 03
Could professional standards test the same mechanism using historical workforce data before using it operationally?
