How to Read Area Crime Statistics Before You Buy

The Trap, With Real Numbers

The South African Police Service publishes reported crime by police precinct. There are 1 165 precincts with enough data to rank nationally, and the convention, used by SAPS and Statistics South Africa alike, is to express counts as a rate per 100 000 residents.

Here are two Johannesburg precincts, using the most recent twelve months of reported residential burglary and robbery.

Hillbrow has 152 780 residents and 110 reported break-ins. That is 72 per 100 000, which places it in the 7th percentile nationally. On that number Hillbrow is one of the safest places in South Africa. Randburg has 20 534 residents and 257 break-ins. That is 1 251 per 100 000, the 98th percentile, one of the worst in the country.

No one who lives in Johannesburg would tell you Hillbrow is seventeen times safer than Randburg. The arithmetic is correct and the conclusion is nonsense.

Why the Denominator Is Wrong

Residential burglary does not scale with people. It scales with dwellings, and more precisely with accessible dwellings.

Hillbrow is dense high-rise flats. Many residents share one front door, buildings have controlled access, and there are very few ground-floor windows and back yards to work with. Divide a modest number of break-ins by a very large resident population and the rate collapses.

Randburg’s precinct covers standalone houses on plots with yards, garages and multiple points of entry, and its resident count is small relative to the number of separate targets. Divide a large number of break-ins by a small resident population and the rate explodes.

The same flaw applies to vehicle crime, where the correct denominator is registered vehicles rather than people. A precinct full of car owners will always read badly against one where few households own a car, regardless of how careful either is.

Where Per Capita Is the Right Measure

Violent crime is the exception, and it is the one to trust. Murder, assault and robbery are offences against people, so a rate per 100 000 residents is the correct construction and is the international convention.

Even there, one distortion remains. A precinct with a small resident population and a very large daytime population will read badly, because the crimes are counted where they happen and the residents are counted where they sleep. Johannesburg Central has 37 765 residents against a working-day population many times that. Its violent crime rate is real, but it describes a place people commute to, not a residential risk for someone buying a flat there.

This is why a percentile against every other precinct is more useful than a raw rate. It tells you where an area sits in the national distribution rather than asking you to hold an absolute number in your head.

Provincial Baselines Worth Knowing

Before reading any precinct, it helps to know what normal looks like. Aggregated to province level, per 100 000 residents over twelve months, the pattern is consistent and often surprises people.

Limpopo reports the lowest figures on all three categories, at roughly 181 break-ins, 68 vehicle crimes and 588 violent crimes per 100 000. Gauteng runs at about 233, 313 and 1 122. The Western Cape is the highest of the major provinces at roughly 318, 356 and 1 565.

These are reported crimes, which is the important caveat. Under-reporting is real, and it is generally worse in rural areas where a police station is far away and confidence in the outcome is low. A very low rural figure may reflect fewer reports rather than fewer crimes.

How to Read It Properly

Start with the housing type, then read the burglary number against it. A low burglary rate in a precinct of flats tells you almost nothing about the risk to a house with a garden two streets away.

Use violent crime as your area signal, because the denominator is right and the reporting is more consistent for serious offences.

Compare the precinct against its neighbours rather than against the country. Crime is intensely local, and precinct boundaries can put a quiet suburb and a high-risk area in the same statistic.

Look at the direction of travel as well as the level. A precinct improving from a bad base is a different investment proposition to one deteriorating from a good one, and the year-on-year change is usually more actionable than the absolute number.

• Identify the dominant housing type before reading the burglary rate.
• Trust violent crime per capita; discount burglary and vehicle crime per capita.
• Compare against neighbouring precincts, not the national average.
• Check the year-on-year direction, not only the current level.
• Treat very low rural figures with caution, because reporting rates differ.

Use GemFinder Alongside This Guide

Use GemFinder to find live property auctions, compare reserve prices and auction dates, save watchlists and move promising listings into a structured due diligence workflow.

Browse live auctions or return to the research library.

Sources and Further Reading

This article is general educational information, not legal, tax, conveyancing or financial advice. Confirm the latest law, the specific conditions of sale and your own numbers before bidding.

• South African Police Service crime statistics
• Statistics South Africa