Crime statistics: who are we to believe?

Members of the Anti-Gang unit in Hanover Park, Cape Town. Archive photo by Ashraf Hendricks

Once again, statistics on crime from the country’s two most important sources wildly contradict each other. South African Police Service (SAPS) crime statistics do not match those of the Victims of Crime Survey (VOCS) released by Statistics South Africa at the end of August 2024.

The SAPS murder figures, for example, are almost 25% higher than those of VOCS. SAPS reported cases of sexual offences were about double those of VOCS. Conversely, home burglaries reported by SAPS were about 15% of the figure reported by VOCS.

The VOCS survey produces estimates of the actual number of incidents of crime experienced by households and individuals, while SAPS statistics are derived from crime actually reported at police stations.

Stats SA says the VOCS figures “complement SAPS crime statistics by providing estimates for quantities that cannot be measured at police stations”.

VOCS does indeed add value to the SAPS crime stats. The survey measures indicators such as how safe people feel in certain settings, what people do to protect themselves, their trust in police services, and the way they assess the performance of the courts. SAPS crime stats do not measure these issues and many have argued that they have no interest in doing so.

The idea that VOCS is a “complement” to the official crime statistics arises from the claim that VOCS is able to record incidents of crime that do not appear in the SAPS statistics. But if that is the case, then VOCS stats should be consistently higher than SAPS, and this is not so. Why would a victim of crime take the trouble to report it to SAPS but not reveal it to a Stats SA researcher (assuming of course that VOCS sampling is sound)?

For example, comparing figures for April 2022 to March 2023 from both data sets, we find several inconsistencies.

Murder

The VOCS found that about 22,000 households experienced a case of murder, significantly up from the previous year’s figure of 16,000. But the SAPS stats reflect 27,494 murder cases for 2022/23 – almost 25% higher than the VOCS figure. Murder is one of the most commonly reported crimes, and the VOCS indicates 100% of households claim they reported this crime to the police. This discrepancy is hard to explain.

Sexual offences

According to the VOCS, household experiences of sexual offences virtually doubled from 11,000 cases to 22,000 cases, though Stats SA acknowledges that its sample size was inadequate. Total sexual offences reported to SAPS for 2022/23 were 53,498 – well over double the VOCS figure, despite the fact that sexual offences are known to be heavily under-reported. In fact, VOCS estimates suggest that just over 78% of such cases are reported to the police, so that the figure for sexual offences (using SAPS) may be over 68,000.

Assault

The VOCS survey reports only one category of assault, unlike the SAPS stats which report both common assault and assault with intent to do grievous bodily harm (GBH). VOCS reports that 129,000 households experienced assault. This is nearly double the figure of 68,000 for the previous year. For the same period, SAPS reported 185,374 cases of common assault and 169,374 cases of assault with GBH, yielding a total assault figure of 354,748 cases, nearly three times the VOCS figure. In theory, the SAPS figures should have been much lower than VOCS as the Stats SA survey found that less than 68% of households report such cases to SAPS.

Burglary

For 2022/23, SAPS reported 163,493 burglaries at residential premises, but VOCS reports that about 1.1 million households experienced housebreaking in the same period. If the VOCS figure of about 44% of cases reported to the police is correct, then using the VOCS figure, the police should have been notified of about 484,000 cases.

Robbery

Both VOCS and SAPS define home robbery / robbery at residential premises as involving some form of contact or force between perpetrators and victims. VOCS reports that 209,000 households experienced home robberies in 2022/23. This is about nine times the SAPS reported figure of 23,065. However, SAPS has several other categories of robbery, and classification can be sloppy. If the VOCS figure of 58% of victims reporting to SAPS is correct, then, using the VOCS totals, SAPS should have received about 121,000 complaints of this nature.

Motor theft

The VOCS reports that 87,000 households experienced theft of motor vehicles in 2022/23 of which 74,000 (about 85%) were reported to the police. SAPS however reports only 37,461 cases for the same period (and unlike VOCS, SAPS includes motorcycles in this crime category). This suggests that only about 51% of the cases that VOCS estimates are reported, are in fact reported. Motor vehicle theft typically has a high reporting rate due to the fact that it is an insurance requirement.

What to believe

South Africa is not unique in battling with this data puzzle. Around the world, contradictions between official crime statistics and independent victim surveys are the norm.

There are several reasons for this. Many crimes go unreported, public confidence in the police and the courts is lacking, rumours and media coverage of crime (often fuelled by social media) can lead to panic perceptions that do not match hard data.

Survey data is often skewed by crime victims who understandably have problems recalling the incident and when exactly it occurred. Victims often struggle to match their description of the crime to the definition used by researchers.

But none of these explanations should become a long-term rationale for accepting the wide gap between VOCS data and SAPS crime statistics. Whether through better-aligned formats or more credible methods, VOCS and SAPS data must be made more comparable.

Those working to make communities safer through better policing strategies and crime prevention need good reliable evidence. Dependable data is critical. But who to believe: SAPS crime stats with the acknowledged problems of administration, likely under-reporting and possible manipulation? Or VOCS, which relies on Stats South Africa’s questionable technical competence and the perceptions and memories of possibly traumatised victims?

Glenn Hollands works in the field of community safety

Donate using SnapScan.