What actually deters burglars: the evidence

Verdict: the strongest deterrent on record. In the UNC Charlotte survey of 422 incarcerated burglars, the proximity of other people was a top factor in choosing targets, and most burglars in the Oregon interviews knocked or rang first to make sure nobody was home. If a burglar believes someone is present, most simply leave.

The UNC Charlotte study (Kuhns et al., 2013) asked 422 incarcerated burglars how they picked targets; the proximity of other people topped the list alongside escape routes and visible signs of security. The Oregon material — 86 inmates answering a KGW-TV journalist's questionnaire in 2017 — reads the same way: essentially all of them knocked or rang the doorbell before entry, and a car in the driveway or sound inside the house was among the strongest signals to move on.

The stakes are worth stating plainly. The Bureau of Justice Statistics found a household member was present during roughly 28 percent of burglaries. Burglars try hard to avoid people and still regularly fail, which is exactly why the appearance of occupancy — light, sound, a vehicle, an answered doorbell — does so much work.

Verdict: the best-tested environmental deterrent, with real causal evidence. A randomized experiment across 77 New York City public-housing developments found that intense new outdoor lighting cut nighttime outdoor index crimes by roughly 36 percent, and follow-up work found the effect persisted for about three years. No other common deterrent has evidence this strong.

Because the developments were randomized, the study (Chalfin, Hansen, Lerner & Parker, NBER Working Paper 25798) supports a causal claim rather than a correlation — rare in this field. The follow-up published in Criminology & Public Policy in 2022 confirmed the reduction was not a novelty effect that faded within months.

The caveat: the experiment used intense, constant lighting installations, not a cheap motion flood over a garage. Treat it as strong evidence that visibility deters nighttime crime, and as weaker evidence about any particular fixture. We take that question up in detail in our article on motion lighting.

Verdict: solidly supported, with a disclosure. About 83 percent of surveyed burglars said they would try to determine whether an alarm was present, and roughly 60 percent would pick another target if one was. A Newark study tied higher alarm density to fewer burglaries. Both studies were funded by the alarm industry's research foundation.

In the UNC Charlotte survey, over half of the burglars also said they would abort a burglary already in progress on discovering an alarm — a striking finding, because it means the deterrent works even after the decision to enter. Separately, a 2009 study by Seungmug Lee at the Rutgers School of Criminal Justice found Newark neighborhoods with higher residential alarm density had fewer burglaries, with no displacement of crime to neighboring homes.

Disclosure: both were funded by the Alarm Industry Research and Educational Foundation. That does not make them wrong — the methods are documented, and the direction matches what burglars told an independent journalist in Oregon — but it is why we flag the funding rather than quoting the percentages as settled fact.

Verdict: better than most hardware, by burglars' own account. In the Oregon interviews, big, loud dogs ranked with occupancy and noise among the strongest reasons to walk away, and the UNC Charlotte survey listed dogs inside among the top signs burglars checked for. The deterrent is the noise and unpredictability, not the breed.

Notice what a dog actually is in this framework: a resident that always answers the door. It guarantees the burglar will not work in silence, and it cannot be predicted, distracted, or reasoned with. That is why the interviews specify big and loud — the deterrence is exposure, not injury. There is no credible population-level study of dog ownership versus burglary rates, so the survey testimony is the evidence, and it is consistent across both samples.

Verdict: real but weak, and double-edged. Outdoor cameras appear in the UNC Charlotte survey as one of the signs of security burglars checked for, ranked below alarms and occupancy. In the Oregon interviews, some burglars read cameras as a deterrent — and some read them as a hint that the house held something worth taking.

The structural problem is timing. Every deterrent burglars rank highly threatens a consequence during the crime; a camera mostly threatens one after it — footage that may matter someday, if anyone follows up. A hood and a fast exit shrink that threat to a size many burglars accept. Our full article on cameras walks through this; the short version is that recording is a witness, not a response.

Verdict: unproven on their own. Alarm signs show up in the UNC Charlotte survey only as part of a broader cluster of "signs of security" burglars looked for — never isolated as an independent deterrent. Every popular statistic about yard signs alone traces to sources that cannot be verified. Treat signs as a supporting detail, not a defense.

The logic is easy to reconstruct: a sign claims a system, and a system claims a response. Burglars who case houses test whether the claim is real — the knock on the door is exactly that test — so a sign with nothing behind it fails at the first probe. Put the sign up; it costs almost nothing. Just do not confuse it with a plan.

Security marketing runs on a small set of zombie statistics — numbers that circulate endlessly without a locatable primary source. We looked for the original research behind each of the following and could not find it. If one of these appears on a vendor's page with no author, institution, year, or methodology attached, treat the claim as unsupported:

None of this means deterrence is fake — the verified research above is genuinely encouraging. It means the honest evidence base is smaller than the marketing implies, and the numbers that survive scrutiny are the only ones worth building decisions on.

Verdict: the logical endpoint of everything above, with the least published research. Burglars avoid people who might confront them; active response makes confrontation immediate instead of hypothetical. The category is young — human guard services and AI deterrence products emerged in the last few years — so grade it by the mechanism, not by long-run studies that don't yet exist.

The market has started building toward this conclusion. As of mid-2026, Deep Sentinel stations live human guards in a surveillance center from roughly $100 per month, SimpliSafe's Active Guard tier adds overnight human intervention at around $80 per month, and Ring offers a Virtual Security Guard professional-agent add-on at about $99 per month. Alarm.com announced an AI deterrence product at CES 2025. All of these are monitoring models: people or software watching feeds from somewhere else.

This is the finding an Automatic Property Defense System like PerimeterEye is built on. Instead of adding a distant watcher, it makes the property itself respond: when it verifies a genuine unknown person, a motorized searchlight locks onto them and physically follows — homeowners compare it to a police-helicopter beam — while the AI speaks to them directly, describing the specific person in frame, then escalates to a 130 dB siren and priority alerts with live video if they stay. It knows the household, so family arrivals and mail carriers never trigger any of it. PerimeterEye is not an alarm-monitoring service — alerts and control go directly to you. And to be plain about limits: no security system can guarantee prevention of every incident.