UK policing and the wider criminal justice system is using flawed calculations for ‘disproportionality’, particularly when it comes to racial disparities. It all comes down to inconsistent logic and faulty assumptions (e.g. ‘all sections of society are equally likely to come into police contact’). Policing should be evidence-based, not assumptions-based.
Such logic leads to misinformed conclusions, unfortunately resulting in divisiveness, distraction and reduced public confidence. Divisiveness in the emotionally charged debate on racial disparities leads to entrenchment. As such, very little changing for the better. Chasing faulty statistics and inferring flawed conclusions is distracting policing from resolving the real problems needing reform. The result is reduced public confidence, the foundation of police legitimacy.
This article provides a thesis of the issue (benchmarking police tactics against the general population), proposing better measurement solutions. In turn, we can better understand the real issues in policing diverse communities, in line with the Police Race Action Plan goal to explain or reform disproportionality.
Skip to the end ‘Conclusion’ section for the short version.
(Note: This is a guest blog, onto which I’ve added relevant links. If you’ve got something interesting to say on policing and/or leadership, or if you’d like to write a rebuttal, please get in touch. I’m always happy to share my platform! – Steve)
What is Disproportionality?
What is disproportionality? ‘Disproportionality’ is defined in the English dictionary as follows:
“When something is too large or too small when compared to something else.”
The definition is also used by the Independent Scrutiny and Oversight Board (ISOB). ISOB are a public body set up to scrutinise progress made in the national Police Race Action Plan (PRAP). The PRAP itself defines disproportionality as:
“A group’s representation in a category that exceeds expectations for that group or differs substantially from the representation of others in that category.”
So key facets of disproportionality are “too large/small” or “exceeding expectations” and “compared to something else”.
In the policing context, the additional colloquial understanding of disproportionality relates to whether policing tactics are being used fairly. In relevant reports, it’s used interchangeably with ‘disparity’. Disparity is defined by the dictionary as:
“A lack of equality or similarity, especially in a way that is not fair.”
When disproportionality or disparity occurs in conjunction with someone’s personal demographics, such as their race, sex, or sexual orientation, it becomes synonymous with prejudice and discrimination. The foundation of police legitimacy is severely weakened when police actions aren’t seen as proportionate.
Why is Disproportionality Important?
“When the police use their powers disproportionately – in differing proportions on different ethnic groups – it causes suspicion among some communities that they are being unfairly targeted.” – HMICFRS
So ultimately, the notion of disproportionality comes down to policing’s fundamental legitimacy and the value of fairness. Legitimacy is the 200 year bedrock of British policing. Procedural fairness is a core foundation to maintaining public confidence. If the public are suspicious or feel something is unfair, especially if driven by discrimination, confidence in policing is shattered.
Concerns over whether policing powers are proportionate seem particularly intense on the protected characteristic of Race. For other equally-protected characteristics, like Sex and Age, nobody seems to bat an eyelid. “Police are stop searching young adults disproportionately more than pensioners.” And? “Police use force on males far more often than they do for females.” So what? These headlines and exclamations simply don’t exist.
For context, the Police Race Action Plan is described by the National Police Chiefs Council (NPCC) as “the biggest commitment ever by policing to rid our service of racism, discrimination, and bias”. It was “formed against the backdrop of George Floyd’s horrific murder by police officers in the US and the Black Lives Matter movement”.
Some police forces have been described, or describe themselves, as ‘institutionally racist’. This was a conclusion reached by the Casey Review about the Metropolitan Police, based on an array of evidence. Dorset became the latest to announce their institutional racism just this week, in an open letter to their black communities.
Police Chiefs are divided on the ‘institutional racism’ matter. Certainly some in policing are overtly racist and should unquestionably be rooted out. An increasing number of officers and staff are being dismissed for discriminatory behaviour (54 in 2022/23, 0.02% of the workforce). Some of the more overt and egregious examples include putting bacon in a Muslim officer’s boots (Casey), or bananas by the lockers of Black officers.
Reports of ‘disproportionality’ of police interactions versus the general population amplify concerns over police racism. The media is just one source misinforming the public that ‘disproportionality/disparity equals discrimination’.
“Despite Avon and Somerset Police committing to becoming an anti-racist police force, data on their use of strip search powers on children reveals huge racial disparities.” – Bristol Live
This false notion is widespread, including among senior officers. The self-flagellation and frequent media reporting unnecessarily exaggerates the problems, damaging both public confidence and officer morale. The damage caused by such jumping to conclusions without really explaining the data or investigating it further is unnecessary to say the least. Others argue that the quiet majority are feeling abandoned by misguided political correctness.
Sometimes, buried in the small print of official reports, the truth is neatly outlined and caution warned. But most of the time, these facts are overlooked.
“We wish to make it clear that evidence of disparity isn’t in itself evidence of discrimination. But it should warrant further investigation.” – HMICFRS Race Disparity Inspection 2023
Don’t Explain, Just Reform?
In writing, the PRAP takes an ‘explain or reform’ approach to police disparities, derived from the 2017 Lammy Review. But wherever you look in practice, the demands are for reforms, without properly explaining the issues or investigating their root causes.
The demographic makeup of those subjected to police powers is always compared to the general population. The underlying assumption is the subjects of police tactics must match the population… on ethnicity. Any disparity is denounced as discriminatory.
The PRAP Progress Report 2024 leans into inferring links between disparities and racism.
“A Black person is twice as likely to be arrested, three times as likely to be subject to police use of force and four times as likely to be stop and searched than if you are White. Racism cannot be defined simply through statistics. Behind these figures is the visceral, emotive response of Black communities and how they feel about the police.” – T/DAC Dr Alison Heydari
The NPCC are clear in their objective to eradicate the apparent disproportionality (i.e. reform) when comparing demographics of those subject to police tactics against the general population demographics. The ISOB shares this goal of ensuring police powers match the general public demographic with respect to Race. By doing so, both are clearly bypassing ‘explain’ in the formal ‘explain or reform’ commitment of the PRAP.
“Bedfordshire has done some fantastic work around stop and search scrutiny, really taking it onto the next level. In relation to Asian members of the community being searched, in the last quarter, the figures in Luton show that they’d reduced the disproportionality to zero. I’m not aware of anybody achieving that before.” – CC Gavin Stephens
“In our view, the aim should be the elimination of race disparity.” – ISOB
Many forces in turn have introduced their own local PRAP, along with associated governance hierarchies to implement. In Dorset’s recent PRAP launch, they include the aim to tackle racial disparities versus the general population, without explanation and “regardless of their root causes”.
The PRAP itself is beyond the scope of this focused discussion on statistical methods. It comes from the perspective of Black officers and members of the public. The National Black Police Association however have recently withdrawn their support.
Generally, the PRAP contains decent, well-intentioned goals and activities. Not least given the gruesome findings of the Casey Review or Angiolini Inquiry, resonating with others beyond the Met Police.
Reducing disproportionately is core to the objectives and performance measures by which PRAP “success” is determined. One broad objective is “We only use our powers in a fair, respectful and equitable way”. Fair and respectful is pretty straightforward and clear. But determining “equitable” is where the problems begin: What exactly are we comparing it against?
Drawing Conclusions from Statistics
Before going deeper on the flaws in the logic and assumptions of the disproportionality statistics, it’s worth briefly covering what populations and statistics mean. Most people have the simplistic (mis)understanding that “population = the general public”.
In normal research and statistics, population refers to the entire group from which you want to draw conclusions about. That group could be people, animals, places, or other ‘stuff’. A sample is the specific subset of the population from which you will collect data. You can then infer from the sample conclusions about the wider population.
Think of a cake: The population is the whole cake, a sample is a slice, several slices, or a greedy bite from the top.
You normally take a sample, because it takes too long to study the entire ‘population’ of stuff. If you can measure the whole lot, this is called a census. The bigger your sample, the more confident you can be in your findings and their extrapolation to the wider population.
An entirely hypothetical example: If I wanted to draw conclusions about what England and Wales police officers thought about the IOPC, my ‘population’ is specifically 147,000 England & Wales police officers.
I couldn’t possibly ask all of them to get a precise answer in a reasonable time, so I’d be happy with a sample. If I’m lazy, I’d ask around my colleagues or briefly check Twitter to get a feel. But this is such a small sample that I wouldn’t be confident it was representative.
Instead, I’d send out a survey inviting responses, hopefully getting 500 or more responses to be more confident in drawing wider conclusions from my findings.
Here’s some more handy and simple explainers on these core concepts, if you’re interested. After all, we’re keen on evidence-based policing, aren’t we?
One key concept at this point is to define your population precisely, else the conclusions you draw about things will be completely off the mark.
How is Disproportionality Currently Being Calculated?
The PRAP has two broad objectives relating to disproportionality of police powers and to build trust with Black people: Explain or reform their use. This seems entirely reasonable, but as we’ve seen the reality is fast forward to reform, without understanding what’s going on. Most people are too scared of putting their head above the parapet to explain anything.
Given all disproportionality/disparity statistical comparisons are benchmarked against the general public demographic, the question currently being answered is:
“Are police powers being used on demographics in proportion to how those demographics are represented in the general public?”
The deeply flawed assertion is that the proportion of people subject to police powers should match the proportion of each demographic observed in wider society. The unevidenced assumption underlying this goal is that every demographic in society is as equally likely as each other to encounter or require attention from the police.
The faulty logic does not get applied to all demographics, like Age, Sex, or other protected characteristics, so people know it’s logically wrong. But assessing police in this way for Race feeds distrust in this world where ‘disproportionality = discrimination’ and distracts resources from the real changes needed to improve policing for Black people.
“We have a number of remaining doubts about whether [the PRAP] has improved the experiences of policing for Black people and if it has created a strong enough foundation to change police culture.” – DI Andy George, NBPA President
Consider for example the latest police use of force statistics (example snapshots below). The ‘per 1000 (general) population’ comparisons show the following ‘disproportionality’ statements to be true:
- Males aged 18-34 are 43X more likely than those aged 65+ to be subjected to use of force. Is this disproportionate and ageist?
- Males are 4.5X more likely than females. Does this suggest sexism, with less force being used proportionately against women?
- Black males are 4X more likely than White males. Is this an indicator of racism?
- White females are 4X more likely than Asian females. Does this also indicate racism, with Asian females treated more favourably than White?
- Asian and Mixed ethnic groups are slightly less likely than White. Hang on, isn’t White supposed to be the benchmarking standard, why are minority groups less?
- Once drawn, Tasers are discharged similarly across ethnicities. Does this mean the assumed racism is dependent on specifics tactic used?
- Police firearms are 15X more likely to be involved with Males than Females. Does this mean police are more gung-ho when it comes to policing men?
- 17% of use of force incidents involved people with a mental health condition. Are police disproportionately targeting those with poor mental health?
The questions people want to know are about fairness. But they clearly cannot possibly be answered by benchmarking everything against the general public demographic. It’s a case of ‘apples vs. pears’ and a somewhat meaningless notion of ‘disproportionality’ or fairness. Contrary to the flawed assumption, the general population does not present itself so uniformly to police, with observed disparities arising from wider societal or other factors beyond policing’s influence.
“Black people account for 14% of all missing people in England and Wales, despite only making up 4% of the population.” – ISOB Annual Report 2024
The below is another snapshot from the official police use of force statistics. A more considered assessment of disproportionality of police action might include such details as part of explaining any apparent disparities.
Society’s ails and inequalities are not the police’s responsibility to sort out. The police exist to uphold the law and maintain public order. Would it be proportionate and intelligence-led for example to ensure men and women are represented 50/50 in stop search demographics, to match the public? Pursuing flawed comparisons and “parity targets” only serve to mask the deep-rooted societal, multi-agency problems that need fixing. Worse, it does a disservice to minority groups, the wider public, and police officers alike.
The focus on ‘whiteness’ as the centre of all comparisons is also questionable. It assumes whatever the ‘White’ rate, that should be considered the ‘gold standard’ to aim for. Given the wide variety among ethnic groups, including where some minority ethnic groups have lower rates of the use of police powers than the ‘White’ category, this seems flawed at best. The broad brush also excludes the experiences of other, equally-protected minority ethnic groups, such as Eastern European or Gypsy/Roma.
Policing should be evidence-based and intelligence-led, maintaining safety for the general public by focusing attention on the criminal population. Demands to meet arbitrary or politically motivated ‘parity quotas’ ironically pressures police officers to incorporate people’s personal demographics into their decision-making. This will be at the expense of intelligence and evidence.
Such demands are either unchallenged or even encouraged by senior police leaders and its governing institutions. ‘Courage’ is a core ethical value policing should abide by. The College of Policing define courage as:
“This means making, communicating and being accountable for decisions, and standing against anything that could bring our profession into disrepute… we do the right thing even when the circumstances are difficult… challenge unprofessional behaviour and all forms of prejudice and discriminatory behaviour, and any activity which undermines the impartiality of policing.”
Pretending the criminal population marries up to the general population demographics doesn’t sound like standing up for impartiality. Nor does being selective with logic. Moving away from evidence-based and ‘justice is blind’ policing to chase quotas doesn’t seem like standing against things that could bring our profession into disrepute.
Instead, the PRAP has an agreeable order in which policing should address any disparities found: Explain, or reform. To help explain things (often to determine whether reform is necessary), there are several better measures of success.
Better Measures of Success
“Impartiality should mean that the police have the courage to enforce the law against whosoever breaks it – not because of who they are, or not – and should not be overly concerned about the complaints of racism, or any other ‘ism’, that will inevitably follow.” – Wiltshire PCC Philip Wilkinson
Proposed below are better measures of success when seeking to ‘explain or reform’ police actions towards different demographics of society. These aren’t just relevant for Race; the greater academic rigour and consistent logic applies to other demographics too.
1. Find Rates for Stop and Search
Stop and search powers exists to search people who the police believe on reasonable grounds are carrying for example drugs, weapons, or stolen property. Neither policing nor the public want those ‘reasonable grounds’ influenced by racism or racial profiling.
A far better and more precise measure of success would be the ‘find rates’ of items in stop and search. It would be a more immediate concern and highlight something wrong with police actions if there is not parity in the find rates relating to stop and search.
For example, should Black people be 2x more likely to have nothing found following searches than other groups, this would be a big concern. That’s because it suggests something other than ‘intelligence-led’ or ‘reasonable grounds’ are being used, with racial bias or discrimination becoming obvious possible factors to explain the gap.
As it happens, find rates are already monitored in the national stop and search statistics. Yet this metric is rarely recognised nor features in the attention-grabbing headlines, despite being a far more direct way of determining success in disproportionality. As per the table shown, this metric identifies negligible differences between the ethnic groups to 2024:
2. Qualitative Assessments
It’s easy to get lost in the quantitative statistics and either still not know precisely what’s going on or infer faulty conclusions. Further, the Home Office acknowledge the demographic stats can be skewed significantly by persons repeatedly subject to those powers. This suggests a need for more qualitative information to inform our judgements, understanding what can be explained and where policing needs to reform.
People tend to avoid qualitative data, because numbers either feel somehow more ‘scientific’ or it takes too much work to wade through the detail. Yet policing should be led by intelligence. Here are a few relevant and manageable options:
- Dip-sample and review body-worn video footage of stop searches or use of force, assessing the officer’s behaviours during the interaction.
- Ask a sample of those directly subject to the police tactics on their perspectives and what needs reform.
- Ask a sample of officers to explain their actions relating to the instance (debrief and give feedback).
You could focus sampling from those records flagged for supervisor review or where nothing was found in stop searches. The act of scrutinising things in this depth will also give inappropriate or racist behaviours and officers ‘no place to hide’.
Further, the qualitative commentary behind the public confidence surveys would be valuable here, to explain why confidence rates are lower among Black communities and what specifically should improve.
3. Triangulated Population Benchmarking
If disproportionality is to be measured, the benchmarking against the general public population is demonstrably the wrong comparator. We instinctively know that the police aren’t interacting with a representative sample of the general public and the evidence plays that out.
The demographics of the criminal population is significantly different to the general population. So given policing wants to focus its attention on criminals, it’s a far more meaningful and relevant comparator population for potential ‘disparities’ or ‘disproportionality’. The population the police should be most interested in for benchmarking could be defined as:
“Those who come into contact with the police”, or even more specifically, “Those with who come to into contact with the police AND warrant the use of police powers.”
Unfortunately, no such single dataset exists for this specific question, but there are several which come reasonably close. Suspects, the convicted population, and/or the prison population are some alternative benchmarks to assess the real proportionality of police actions. It’s probably a good idea to triangulate against several populations to better explain the police action, in line with the PRAP goals.
The official police use of force statistics already contain some of this information. The below chart is a good example. Unfortunately, these more relevant comparisons are currently ignored or their importance to the question of disproportionality overlooked.
Incidentally, various reports (such as the Lammy Report) all note the overrepresentation of those from minority ethnic backgrounds in such comparator populations. Clearly, societal issues beyond policing are at play here. They won’t be fixed by pressuring police forces to paper over the cracks with statistics.
4. Control for Offences when Benchmarking
But even that doesn’t adequately explain what’s going on. Different demographics don’t commit different types of offences equally to one another. There’s so much evidence pointing this out across the protected characteristics, including for Race.
So a more sophisticated assessment of apparent disparities in criminal justice decisions, such as search, charge, caution, sentencing, etc.) must control for specific crime types. Only such like-for-like comparisons will indicate fairness, or otherwise unearth the truly unexplainable disparities. These will then require reform by the police.
This gets complex though, because every broad crime category has differing severity of specific offences within it. For example, violent crimes range from homicide to assault without injury. The Cambridge Harm Index however is already available, should organisations wish to use this for a head start. It handily documents the severity (using sentencing guidelines) of each of the thousands of different offences. It might be best to focus first on the top 20 by volume, to both maintain like-for-like comparisons while overcoming natural variability problems encountered when dealing with low numbers.
5. Control for Officer Demographics
A final thing overlooked in all benchmarking metrics is the demographics of the officers involved. Given we are attempting to detect potential bias and racism by officers against the subjects of police powers, the ethnicity of the officers performing the actions seems pretty key, and so should be factored in to any benchmarking. You would hope for example that parity exists in the stop search rate of Black people between officers belonging to different ethnic groups.
Conclusions
The Police Race Action Plan makes it clear in black and white: Policing should either explain or reform its ‘disproportionate’ actions.
Racism remains a thorn in the side of policing, as it is for society, and there’s much to resolve. Not least given the gruesome findings of Casey, other formal reviews, or from anecdotal experiences. But there is faulty logic of how ‘disproportionality’ is currently defined and applied to Race, unhelpfully turning the debate septic.
Benchmarking police actions against the general public creates false comparisons and conclusions. This means policing is inadequately explaining things and assuming guilt, possibly deriving from a lack of strategic courage to do things right in the face of vocal criticism. This self-flagellation does no good, harming public confidence for everyone, wherever they stand on the police trust spectrum and wider DEI debate.
Changes instigated by faulty statistics and conclusions harm policing tactics, moving them away from evidence- or intelligence-based, towards a more demographic and identity-quota driven approach. Without truly understanding where the issues are and what cannot be explained, reforms being made are either unnecessary, mismatched to the problems, or even divert attention away from the real issues in policing, especially relating to Race.
It need not be like this. There are better measures of success to help clear up much misunderstanding, encouraging a more intelligent assessment of police legitimacy. Find rates, better benchmarking and more qualitative assessments are alternatives to help more precisely define where reforms and meaningful changes are required.
Seeking police promotion? Want a MASSIVE head start right now? Hit the ground running with your personal digital promotion toolkit, and/or my market-leading Police Promotion Masterclass. There’s nothing else like it to effectively prepare you for success in your leadership aspirations. You can also contact me to arrange more personal coaching support. Or try my podcast for your ongoing police leadership CPD covering a range of fascinating subjects.