Ask whether cameras deter burglars and you will get a confident yes from almost every source with something to sell. The research answer is narrower, conditional, and in places points the other way. It is still useful, but only if you take it as written.
What the Meta-Analysis Found
The largest synthesis available is a 2019 meta-analysis of 76 evaluations conducted between 1978 and 2017, summarized in the National Institute of Justice’s practice profile on closed circuit television surveillance. The headline numbers are modest: roughly a 14 percent reduction in overall crime, about 16 percent for property offenses and for vehicle crime, and no significant effect on violent offenses. The practice is rated Promising for the first three outcomes and Ineffective for violence.
The subgroup results matter more than the headline. Effects concentrated in car parks, where crime fell about 37 percent. Residential areas showed a smaller but statistically significant reduction of around 12 percent. Housing complexes, city centers, and public transport settings showed no significant reduction at all. And the split by monitoring type is the finding most homeowners never hear: systems that were actively monitored, so that police or security could be alerted in real time, produced significant reductions, while passive record only systems produced nonsignificant effects.
A camera on a house that nobody watches and that only stores footage is, in research terms, closer to the passive category than the active one.
Cameras Can Improve Investigation Without Reducing Crime
A useful illustration is NIJ’s evaluation summary of Milwaukee’s closed circuit television camera program. Intersections with new cameras showed a 14 percent higher clearance rate, meaning more arrests, which was statistically significant. But reported crime at those intersections went up rather than down, including property crime and assaults. Reported crime rising where cameras are installed is hard to interpret cleanly, since better observation can raise reporting as well as reflect real change, but it is a long way from a deterrent effect.
The practical distinction is worth internalizing. Cameras are an evidence collection technology with a modest, setting dependent prevention effect. Those are two different value propositions, and the second one is much weaker than the first.
What Offender Research Says About Target Selection
Studies of how burglars pick targets tend to find that the dominant factors are structural rather than device related. A cross national analysis published in a peer reviewed journal and available through the National Library of Medicine, a discrete choice study of burglar target selection across three countries, modeled 1,835 burglaries by 854 offenders in The Hague, Birmingham, and Brisbane. The consistent predictors were proximity to the offender’s own home, the proportion of single family dwellings (more entry points, more accessibility), and the sheer number of available households in an area.
That study did not examine cameras, alarms, or signage, and it used only cleared offenses, so it describes burglars who were caught. Both limits are worth stating plainly. The point it does support is that a meaningful share of target selection happens at the level of neighborhood and building type, before any individual house’s hardware enters the picture.
The Sign Without a System
Yard signs and window decals advertising a system you do not have are cheap, and there is no credible measured evidence either way on their effect at the individual house level. Reasonable people can argue that a sign borrows the deterrent value of the systems around it. The honest position is that this is an untested assumption, not a finding, and that a sign contributes nothing to detection, delay, or evidence if someone tries the door anyway.
What to Do With This
Buy cameras for what they demonstrably do: give you notification, let you verify what triggered an alert, and preserve a record. If you want the prevention effect the literature associates with active monitoring, the relevant variable is whether anyone actually responds to what the camera sees, not the camera’s resolution. And do not let a camera budget crowd out the boring measures, a graded deadbolt and a reinforced strike among them, that change how long a door holds.