What the evidence says#

Most schemes like this quote the one study that flatters them. Here is the whole picture, including the parts that don’t.

The optimistic case#

Tilburg, 2015. Across 35 Dutch areas that started WhatsApp prevention groups, monthly burglaries fell by half — from about 60 to about 30. The researchers said themselves they hadn’t expected a fall that large.
A national trend the insurers noticed. Dutch home burglaries fell 18% in 2015 against 2014, and the insurance body linked part of it to the spread of neighbourhood groups.
The systematic review. The Campbell review behind the College of Policing’s toolkit finds Neighbourhood Watch is associated with reduced crime — while admitting it’s hard to say how it works.

The sceptical case#

Later Dutch research undoes much of it. Follow-up studies report that neighbourhood WhatsApp groups do not demonstrably prevent local crime. Across four studied towns, a shared “suspicious situation” led to an arrest in exactly one case.
Participants can feel worse. Rotterdam research (2018–2022) found many groups had no measurable effect on objective safety, and that members often felt less safe than non-members. A constant drip of “suspicious” messages manufactures anxiety, and breeds false accusations and suspicion of outsiders.
Selection effects. Watch schemes cluster in wealthier, lower-crime neighbourhoods — some of the apparent effect is who joins, not what joining does.

The part that isn’t in doubt#

GoodSAM. Thousands of cardiac arrests, measured against survival to hospital discharge, with a clear result: getting a nearby volunteer there before the ambulance saves lives. It’s the one place this field has hard outcome data, and it’s about response, not surveillance.

Alongside that, the softer findings hold up well: faster reporting, deterrence from visible signage, stronger cohesion, and — from the Spanish rural experience — regular contact with isolated elders that reduces both danger and loneliness.

What we conclude#

Crime reduction is a plausible side effect, not the promise. The reliable wins are speed, deterrence, cohesion and less isolation, so those are what Escudo is built and measured on.

The design follows directly from the sceptical half:

Low volume, tightly scoped. Alerts and organising who goes — not a chat channel (why).
No suspect photos, no ethnicity, ever (the rule).
Cheap false alarms, so nobody hesitates (how).
Neighbours checking on neighbours, because that’s the intervention with the best risk-to-benefit ratio of the lot — and the one a village needs no protocol from us to run.

Measure it#

Three numbers, honestly recorded, beat any claim on this page:

How long from alarm to the first neighbour arriving.
How many false alarms, and whether people still press the button.
Whether residents feel safer — asked directly, once a year.

If the second and third go the wrong way, the design is wrong and should change. Full detail and sources are in the landscape survey in the project repository.