The UN estimates 117% rise in killing of women in South Africa since 2015
Source: Al Jazeera's Inside Story (August 2018)
Explainer: 117% is from data now withdrawn by Stats SA
A TV programme on gender-based violence claimed there had been a 117% rise in the killing of women in South Africa since 2015, according to "UN estimates".
We traced the claim to a Statistics South Africa report that compared two different sets of data. The report has since been revised.
South African police data points to a 7.7% rise in femicide.
Researched by Gopolang Makou
A man holds a banner reading "No woman will be killed in my name" during the "One Nation, One Voice" campaign to remove the scourge of child abuse, femicide and other social ills facing all communities in South Africa on June 30, 2018 in Durban. / AFP PHOTO / RAJESH JANTILAL
Crowds of women took to the streets of South Africa in early August to protest against the violence suffered by women, children and gender non-conforming people in the country.
The Al Jazeera Media Network's Inside Story hosted a panel on #TheTotalShutdown protests and what could be done to stop gender-based violence across the world.
During the programme, text was shown on screen: "The UN estimates [a] 117% rise in killing of women in South Africa since 2015." A reader emailed Africa Check asking us to verify the figure.
We looked at the latest data to get a picture of the risk South African women face.
Femicide rate from Statistics South Africa
The 117% figure came from a number of sources, Mohammed Elaichi, a programme editor at Al Jazeera, told Africa Check.
These included articles from the South African Government News Agency, Timeslive, The South African and IOL.
All the articles gave Statistics South Africa (Stats SA) as the source of the figure - not the United Nations, as Al Jazeera's video text said.
'Crimes against women' report
Stats SA released a report on crimes against women in South Africa on 19 June 2018. It lists a number of definitions of "femicide", which here we simply define as "the murder of women".
The report included a table showing that the femicide rate had dropped every year from 2000 to 2015. But it also showed, worryingly, a dramatic increase from 9.6 murders per 100,000 women in 2015 to 20.8 in 2016/17.
This was a 117% increase. The figure was widely reported in South African media, but later withdrawn by the statistical agency.
Report revised to remove femicide estimate
Stats SA removed the femicide estimate from the report the week after it was published, Raphael Kasonga, a director at the statistical agency, told Africa Check.
The dramatic increase shown in the report was the result of comparing two different data sets: the World Health Organisation's (WHO) femicide estimates and Stats SA's victims of crime survey.
The WHO estimated that 9.6 out of every 100,000 women in South Africa were murdered in 2015. The victims of crime survey estimate was 20.8 per 100,000 in 2016/17.
But Stephanie Burrows, a technical officer in the WHO's Violence and Injury Prevention programme, told Africa Check that "it is not correct to compare WHO estimates with other figures as the methodology differs completely".
Stats SA's victims of crime survey is useful as a complement to police crime data, according to Lizette Lancaster, the Institute for Security Studies' Crime and Justice Information Hub manager.
It does have it's limitations, however.
First, it wasn't the "best vehicle for measuring interpersonal violence such as violence against women", Lancaster said. This is because the crimes of femicide and domestic violence take place in the very household being surveyed.
Second, as statistician-general Risenga Maluleke explained in the revised report, while the survey sample was "reasonably large", the number of households that experience these crimes was "often very small". This could lead to high levels of error.
Third, people understand definitions and concepts differently. Violence and assault may not be understood as crimes and so might not be reported to Stats SA in the victims of crime survey.
"The concept of murder is also controversial as some may not regard a killing as murder while others may believe that a legally defined murder was indeed an accident," said Lancaster.
WHO data doesn't show 117% increase
So what do we know about the risk of murder faced by women in South Africa?
The WHO's Burrows provided the organisation's most recent femicide numbers.
To calculate their estimates WHO uses information from various sources. They consider crimes recorded by the police as well as vital registration information, data from the WHO's mortality database and the United Nations Office on Drugs and Crime's global studies on homicide.
Their estimates show a murder rate of 12.2 per 100,000 females in 2015 and 12.1 per 100,000 in 2016. This works out to a 0.8% decrease.
SA femicide rate per 100,000 female population
Year 2000 2010 2015 2016
Femicide rate 19.4 13.4 12.2 [7.1-20.2] 12.1 [6.8-20.1]
Source: World Health Organisation disease burden and mortality estimates 2000-2016
However, Burrows told Africa Check that she would not advise reporting on the percentage drop between the two years because the decrease was so small.
She added that the estimates were modelled and had wide confidence intervals (marked in square brackets in the table above). A confidence interval is the range of values in which a measure of population is likely to be found.
In the WHO's femicide estimates, the confidence intervals for 2015 and 2016 overlap. This means further analysis is needed to find out if the change between the two years is statistically significant.
Calculating femicide rate from police statistics
Data from the South African Police Service (SAPS) and Stats SA also disproves that there has been a 117% increase in femicide.
SAPS provides data on the number of women, 18 years and older, reported murdered in each financial year. Africa Check used this data and Stats SA population estimates for women in this group to calculate the femicide rate. (Note: We used population estimates from the end of September as this is the middle of the financial year, which runs from April to March.)
Number of women (18+) murdered in South Africa & the femicide rate
Year Woman 18+ murdered Femicide rate
2013/2014 2,354 13.1
2014/2015 2,234 12.2
2015/2016 2,416 13.0
2016/2017 2,639 14.0
Source: South African Police Service and Statistics South Africa
The data shows that the femicide rate increased from 13 murders per 100,000 women in 2015/16 to 14 murders per 100,000 women in 2016/17. This is an increase of 7.7%.
Conclusion: Femicide rise of 117% comes from incorrect comparison of figures, now withdrawn, in Stats SA report
Media reports and government statements claimed South Africa had experienced a 117% spike in femicide - the murder of women - between 2015 and 2016/17.
This figure was based on estimates of the female murder rate provided by Stats SA for 2016/2017. These estimates were incorrectly compared to World Health Organisation data from 2015.
The two sets of data were each the result of different research methods, and so could not be compared. Stats SA has since revised its report to exclude the estimates.
The most recent South African police data shows a 7.7% increase in the femicide rate between 2015/16 and 2016/17.
Based on this information, we rate the claim as incorrect.
Edited by Kate Wilkinson