Matt Ashby
@mattashby.com
I help people use data to reduce crime. Associate Professor, Crime Science, UCL. Former police officer.
🌐: mattashby.com
🌐: mattashby.com
Reposted by Matt Ashby
After accounting for other factors, suspects (🇺🇸) were *less* likely to be injured during police use of force if taser was used or if the officer was more experienced. Suspects were *more* likely to be injured when officers attempted 'soft-hand' tactics like wrist locks.
doi.org/10.1177/0032...
doi.org/10.1177/0032...
Sage Journals: Discover world-class research
Subscription and open access journals from Sage, the world's leading independent academic publisher.
doi.org
January 9, 2026 at 3:03 PM
After accounting for other factors, suspects (🇺🇸) were *less* likely to be injured during police use of force if taser was used or if the officer was more experienced. Suspects were *more* likely to be injured when officers attempted 'soft-hand' tactics like wrist locks.
doi.org/10.1177/0032...
doi.org/10.1177/0032...
Reposted by Matt Ashby
I'm sure someone else has suggested this, but it's time for mandatory DOIs in reference lists. This will make it easier to identify hallucinated references
January 9, 2026 at 1:06 PM
I'm sure someone else has suggested this, but it's time for mandatory DOIs in reference lists. This will make it easier to identify hallucinated references
Police officers (🇦🇺) feel body-worn video from responding officers improves the quality of evidence in domestic-abuse cases, but victims tend to be concerned about how the recordings will be used and whether cameras properly capture non-physical abuse.
www.crimrxiv.com/pub/a5qwz2er
www.crimrxiv.com/pub/a5qwz2er
Comparing Police Officers’ and Domestic Violence Victims’ Perspectives on Body-Worn Cameras
Vakhitova, Z., Iliadis, M., Flynn, A., Harris, B., & Tyson, D. (2025). Comparing Police Officers’ and DomesticViolence Victims’ Perspectives on Body-Worn Cameras. Justice Quarterly, 1–35.…
www.crimrxiv.com
January 8, 2026 at 3:01 PM
Police officers (🇦🇺) feel body-worn video from responding officers improves the quality of evidence in domestic-abuse cases, but victims tend to be concerned about how the recordings will be used and whether cameras properly capture non-physical abuse.
www.crimrxiv.com/pub/a5qwz2er
www.crimrxiv.com/pub/a5qwz2er
Useful review of existing evidence on how police can effectively manage missing person cases. Should be a useful starting place if you're working in this area, either as a practitioner or researcher.
doi.org/10.1177/0032...
doi.org/10.1177/0032...
Mapping the evidence on What works for the police to effectively manage missing persons cases: A scoping review - Stavros Chatzisymeonidis, Susan Giles, Sara Waring, Louise Almond, 2025
Background: In most countries, the police are legally responsible for investigating reports of missing persons. However, there is a lack of comprehensive knowle...
doi.org
January 7, 2026 at 3:02 PM
Useful review of existing evidence on how police can effectively manage missing person cases. Should be a useful starting place if you're working in this area, either as a practitioner or researcher.
doi.org/10.1177/0032...
doi.org/10.1177/0032...
New review of evidence finds the most-effective police crime prevention tends to be:
a) focused on very specific places (not people or neighbourhoods), and
b) tailored to preventing a *specific* type of crime.
⚠️ Broad-brush crime prevention generally doesn't work.
doi.org/10.1007/s112...
a) focused on very specific places (not people or neighbourhoods), and
b) tailored to preventing a *specific* type of crime.
⚠️ Broad-brush crime prevention generally doesn't work.
doi.org/10.1007/s112...
The evidence-based policing matrix at 14 years - Journal of Experimental Criminology
Objective To update the 2011 results of the Evidence-Based Policing Matrix, a tool for analyzing, visualizing, and translating research on police crime prevention strategies. Methods We added 177…
doi.org
January 6, 2026 at 3:02 PM
New review of evidence finds the most-effective police crime prevention tends to be:
a) focused on very specific places (not people or neighbourhoods), and
b) tailored to preventing a *specific* type of crime.
⚠️ Broad-brush crime prevention generally doesn't work.
doi.org/10.1007/s112...
a) focused on very specific places (not people or neighbourhoods), and
b) tailored to preventing a *specific* type of crime.
⚠️ Broad-brush crime prevention generally doesn't work.
doi.org/10.1007/s112...
AI-generated feedback on police body-worn video recordings (🇺🇸) halved the likelihood of unprofessional behaviour during encounters in one agency, but not in a second. The results were similar whether the feedback was delivered directly or via a supervisor.
doi.org/10.1111/1745...
doi.org/10.1111/1745...
January 5, 2026 at 3:00 PM
AI-generated feedback on police body-worn video recordings (🇺🇸) halved the likelihood of unprofessional behaviour during encounters in one agency, but not in a second. The results were similar whether the feedback was delivered directly or via a supervisor.
doi.org/10.1111/1745...
doi.org/10.1111/1745...
A frozen Derwentwater just before sunset today. A very nice way to end the holidays.
January 4, 2026 at 6:09 PM
A frozen Derwentwater just before sunset today. A very nice way to end the holidays.
Reposted by Matt Ashby
Our new journal @evidencebasejnl.bsky.social will officially be launching its inaugural issue in April. Evidence assessments, cutting edge methodology papers, research commentaries .. www.tandfonline.com/journals/reb...
December 30, 2025 at 3:00 PM
Our new journal @evidencebasejnl.bsky.social will officially be launching its inaugural issue in April. Evidence assessments, cutting edge methodology papers, research commentaries .. www.tandfonline.com/journals/reb...
🙄 the government is again raising penalties (this time for worrying livestock) “as a deterrent”. This will have zero deterrent effect because there’s almost no chance of prosecution if your dog attacks a sheep.
📢 Increasing the likelihood of punishment can deter, increasing severity probably won’t.
📢 Increasing the likelihood of punishment can deter, increasing severity probably won’t.
December 26, 2025 at 6:14 PM
🙄 the government is again raising penalties (this time for worrying livestock) “as a deterrent”. This will have zero deterrent effect because there’s almost no chance of prosecution if your dog attacks a sheep.
📢 Increasing the likelihood of punishment can deter, increasing severity probably won’t.
📢 Increasing the likelihood of punishment can deter, increasing severity probably won’t.
A lovely Christmas Day walk in some amazing light.
December 25, 2025 at 7:28 PM
A lovely Christmas Day walk in some amazing light.
Happy Christmas!
(This picture is actually from January this year, but we can still dream of a white Christmas.)
(This picture is actually from January this year, but we can still dream of a white Christmas.)
December 24, 2025 at 3:00 PM
Happy Christmas!
(This picture is actually from January this year, but we can still dream of a white Christmas.)
(This picture is actually from January this year, but we can still dream of a white Christmas.)
Useful review of evidence on what works, what doesn’t and what’s promising in using AI in policing. Although the speed of AI development means many emerging technologies haven’t been properly evaluated yet.
Browse all journals
Browse all journals
doi.org
December 23, 2025 at 3:03 PM
Useful review of evidence on what works, what doesn’t and what’s promising in using AI in policing. Although the speed of AI development means many emerging technologies haven’t been properly evaluated yet.
Companies sell AI-based software to automatically review police body-worn video and flag potential misconduct. This study (🇺🇸) found officers are more likely to accept automated flagging if the system also emails supervisors when it finds good conduct.
Does automated feedback impact the acceptability of AI-generated police body worn camera review? An implementation science natural experiment
The diffusion of innovations in policing has often been hindered by barriers to implementation and officer acceptance, which can derail an innovation regardless of its validity or effectiveness.…
www.crimrxiv.com
December 22, 2025 at 3:01 PM
Companies sell AI-based software to automatically review police body-worn video and flag potential misconduct. This study (🇺🇸) found officers are more likely to accept automated flagging if the system also emails supervisors when it finds good conduct.
Reposted by Matt Ashby
Christmas movie? Yes.
Terrorism movie? No.
Thanks for coming to my TED talk.
Terrorism movie? No.
Thanks for coming to my TED talk.
December 20, 2025 at 11:55 PM
Christmas movie? Yes.
Terrorism movie? No.
Thanks for coming to my TED talk.
Terrorism movie? No.
Thanks for coming to my TED talk.
‘Nudge’-based signs encouraging people to lock their bikes at a railway station (🇯🇵) increased how likely people were to do so. This might seem obvious, but some previous research has found signs often aren’t good at changing victim behaviour.
Impact of Antitheft Nudging Signage on Bicycle Owners’ Locking Behavior: A Quasi-Experimental Study in Rural Japan - Ai Suzuki, Takahito Shimada, 2025
This study tested the effectiveness of a nudge-based intervention aimed at encouraging cyclists to lock their bicycles, thereby reducing theft risk in nonurban ...
doi.org
December 19, 2025 at 3:06 PM
‘Nudge’-based signs encouraging people to lock their bikes at a railway station (🇯🇵) increased how likely people were to do so. This might seem obvious, but some previous research has found signs often aren’t good at changing victim behaviour.
Citizens (🇬🇧) greatly overestimate the frequency of bribery in UK businesses: 43% of people think bribery in UK business happens ‘very/fairly often’ but only 3% of businesses were asked for or gave a bribe in the past year.
www.gov.uk/government/p...
www.gov.uk/government/p...
December 18, 2025 at 3:03 PM
Citizens (🇬🇧) greatly overestimate the frequency of bribery in UK businesses: 43% of people think bribery in UK business happens ‘very/fairly often’ but only 3% of businesses were asked for or gave a bribe in the past year.
www.gov.uk/government/p...
www.gov.uk/government/p...
People (in 30 countries) are more likely to co-operate with police if they think police make fair decisions and treat people with respect. In most countries that's more-important in explaining co-operation than police effectiveness or legal duties to obey.
doi.org/10.1111/1745...
doi.org/10.1111/1745...
doi.org
December 17, 2025 at 3:03 PM
People (in 30 countries) are more likely to co-operate with police if they think police make fair decisions and treat people with respect. In most countries that's more-important in explaining co-operation than police effectiveness or legal duties to obey.
doi.org/10.1111/1745...
doi.org/10.1111/1745...
Some students cheat on essays using AI or by paying someone to write for them. But the answer isn’t “severe penalties”, it’s to redesign assessments so it’s harder to cheat.
Too many lecturers set essays that are easy to cheat on and then moan that students cheat.
www.bbc.co.uk/news/article...
Too many lecturers set essays that are easy to cheat on and then moan that students cheat.
www.bbc.co.uk/news/article...
Essay cheating at universities an 'open secret' despite new law
An essay mill owner says he's made millions, while an ex-lecturer says he quit due to rife cheating.
www.bbc.co.uk
December 17, 2025 at 10:08 AM
Some students cheat on essays using AI or by paying someone to write for them. But the answer isn’t “severe penalties”, it’s to redesign assessments so it’s harder to cheat.
Too many lecturers set essays that are easy to cheat on and then moan that students cheat.
www.bbc.co.uk/news/article...
Too many lecturers set essays that are easy to cheat on and then moan that students cheat.
www.bbc.co.uk/news/article...
Training police (🇺🇸) on procedural-justice techniques (e.g. explaining the rationale for police actions) led to improvements in just treatment of people stopped, with the effect not varying by ethnicity of the person.
(Note, though, small sample size)
(Note, though, small sample size)
doi.org
December 16, 2025 at 3:01 PM
Training police (🇺🇸) on procedural-justice techniques (e.g. explaining the rationale for police actions) led to improvements in just treatment of people stopped, with the effect not varying by ethnicity of the person.
(Note, though, small sample size)
(Note, though, small sample size)
Police (🇬🇧) retrospective facial recognition system (e.g. comparing CCTV to mugshots) almost always identifies right suspect but is more often wrong with non-White suspects. Home Office is buying new software that’s equally effective across ethnicities.
Facial recognition technology tests: National Physical Laboratory
Independent testing of facial recognition algorithms commissioned by the Home Office.
www.gov.uk
December 15, 2025 at 3:00 PM
Police (🇬🇧) retrospective facial recognition system (e.g. comparing CCTV to mugshots) almost always identifies right suspect but is more often wrong with non-White suspects. Home Office is buying new software that’s equally effective across ethnicities.
Half of people arrested by police in London who did not have an ADHD diagnosis tested positive on an ADHD screening test after arrest, driven by 60% positive tests among those arrested for drugs offences (for other offence types it was <20% positive).
Neurodiversity in Custody: Screening Results for ADHD and Autistic Traits in Individuals Arrested by the London Metropolitan Police
Background Previous studies have identified high rates of attention-deficit/hyperactivity disorder (ADHD) and autism in incarcerated populations. Despite such findings and the potential benefits of…
doi.org
December 12, 2025 at 3:04 PM
Half of people arrested by police in London who did not have an ADHD diagnosis tested positive on an ADHD screening test after arrest, driven by 60% positive tests among those arrested for drugs offences (for other offence types it was <20% positive).
Reposted by Matt Ashby
“The narrative is ‘rape gangs’ roaming the streets, targeting girls,” he added. “That does happen, but the majority of the crimes that are committed against children are done by people that they know.”
The report also found that child sexual abuse is increasingly happening wholly or partly online
The report also found that child sexual abuse is increasingly happening wholly or partly online
December 11, 2025 at 11:00 AM
“The narrative is ‘rape gangs’ roaming the streets, targeting girls,” he added. “That does happen, but the majority of the crimes that are committed against children are done by people that they know.”
The report also found that child sexual abuse is increasingly happening wholly or partly online
The report also found that child sexual abuse is increasingly happening wholly or partly online
6–15 min police patrols (🇧🇪) every 24h were associated with longest crime-free period after patrols, relative to longer patrols. It’s an observational study, not an experiment, but suggests patrols can have an effect even below 15-min Koper-curve limit.
Short infrequent police stops maximize crime-free time at street segments: Revisiting the Koper curve to establish optimal police dosage - Christophe Vandeviver, Philipp M Dau, Matthew Bland, Stephanie Van De Walle, Maite Dewinter, Frank Witlox, Tom Vander Beken, 2025
Establishing optimal police dosage is central to evidence-based policing. The current consensus is that 15-minute patrols represent the optimal level of such do...
doi.org
December 11, 2025 at 3:01 PM
6–15 min police patrols (🇧🇪) every 24h were associated with longest crime-free period after patrols, relative to longer patrols. It’s an observational study, not an experiment, but suggests patrols can have an effect even below 15-min Koper-curve limit.
This paper introduces an interesting way of comparing violence between countries: the average number of days of life lost to homicide. In the UK it's ~3.4 million total days of life lost to homicide per year, or ~70 minutes off the average British life.
doi.org/10.1177/1088...
doi.org/10.1177/1088...
Sage Journals: Discover world-class research
Subscription and open access journals from Sage, the world's leading independent academic publisher.
doi.org
December 10, 2025 at 3:03 PM
This paper introduces an interesting way of comparing violence between countries: the average number of days of life lost to homicide. In the UK it's ~3.4 million total days of life lost to homicide per year, or ~70 minutes off the average British life.
doi.org/10.1177/1088...
doi.org/10.1177/1088...