An interesting study that highlights the limits of both police data and efforts to use e.g. focused deterrence to tackle non-DV related knife crime in TVP, which is identified to be diffuse. When I was at the Police Foundation, our work with TVP in Slough similarly found that violence was diffuse and largely defied typical problem-solving approaches (http://www.police-foundat...).

The generalisability to other parts of the UK is (acknowledged to be) unclear, but experience suggests the picture in parts of London, for example, is likely to be different, where rival territorial groups exist at higher densities and in closer proximity than I believe is the case in TVP.

I couldn't help thinking that some clarity of definitions might have helped. It is unclear, for example, whether the ‘offenders’ in the study are only those named as suspects (my assumption) or specifically those accused/charged. ‘Gangs’ and ‘OCGs’ are apparently used interchangeably without comment or definition.

The removal of cases where either the victim or offender was unknown may have been necessary for the analysis, but the degree to which it is a key limitation could arguably have been explored further (and I didn't find that it was very clear what the resulting attrition rate was). For example, are there systematic differences in offence types or victim/offender characteristics between cases included and excluded from the analysis? How many ‘offenders’ were victims in crimes with no named suspect/offender, and were there any systematic differences in their characteristics (age, ethnicity, offending history, place of residence etc)? How do police recorded knife violence data compare to health care (ambulance and hospital) data on related injuries and therefore can more be said about the likely completeness of the former?

It would also be helpful to have a more detailed breakdown of the knife crime offences, including whether the network characteristics for all knife crime hold for the most serious kinds of knife crime, especially serious sexual offences (with reference to repeat offending in particular), robberies with injuries, and GBH and homicide offences (including attempts) – which are inevitably the focus for police and others. Similarly, it is unclear how knife crime related to domestic settings/relationships contributed to overall knife crime in TVP, and the exclusion of those cases obscures the possibility that e.g. domestic perpetrators may also be offending or victimised in other settings.

It also strikes me that an analysis of revenge and tit-for-tat violence may need to consider violence in the round, not just knife crime offences, given the potential for other mechanisms for injury to be involved - a reflection that highlights one limitation of viewing violence through the 'knife crime' lens (including at the level of government policy).

A final broad, but I think important, point is that there is no discussion of how the recommendations relate to actual practice in TVP, the force intelligence picture or the understandings of force analysts and officers (which aren't driven solely by quantitative data). E.g. in my own recent qualitative work on the links between drug market and violent crime dynamics (https://bit.ly/2WhfjrP), county lines were characterised in some parts of the country as transitory, with quite high levels of churn of personnel and indeed the lines themselves. It is possible, for example, that the operation of drugs markets contributes to local knife/violent crime dynamics, even while specific individuals or groups aren’t identified as having a recurrent/disproportionate impact, which would arguably shift the emphasis of preventative work towards harm reduction relating to the drugs markets and away from individuals. (That extends to the impact growing market-related violence and criminal wealth can have on wider relationships within communities.) Or indeed, it may be that local drugs markets are well-established with low levels of competition between groups reflecting supply and demand being well matched and limited disruption e.g. from incoming groups or policing - which certainly isn't the case in some other parts of the UK.

Two final minor points.

First, the omission of Ian Brennan’s 2018 BJC paper on ‘weapon-carrying and the reduction of violent harm’ seems like a important gap in the literature review.

Second, the reference to ‘risk’ in the abstract (“white males are at greatest risk”) is misleading – they make up the largest proportion of victims and offenders, but that is much lower than their presence in the population. Contrast, eg, that 56% of victims (62% of offenders) were White vs 84% of the TVP population (ONS 2016 estimate), while 8% of victims (and 18% of offenders) were Black, but only 3% of the TVP population is.

Gavin Hales
Senior Associate Fellow
The Police Foundation (UK)
@gmhales