Flawed reasoning and distorted statistics characterise attempts by some on South Africa's white right to prove black men are more violent than white.
In 2002 the Medical Research Council (MRC) undertook the first national study of female homicide in South Africa. Now, 12 years later, the study has been resurrected to justify a set of racially prejudiced claims that white woman are being killed in their numbers by “unknown black men” as part of a larger “genocide” against South Africa’s white minority.
There are two reasons for the flurry of excitement that the study has elicited from South Africa’s white right.
First, the study – which sampled national mortuary data from 1999 – found that in that year white women were killed more often by men other than their intimate partners. In statistical terms, 2.8 of every 100 000 white women were killed by their intimate partners while 5.8 out of every 100 000 white women were recorded as having been killed by “others”.
Second, 80.9% of killers in the “other” category were classed as “African”.
Drawing on these two findings, the more innumerate on the white right have reasoned as follows: white women in 1999 were more likely to be killed by “others” than their intimate partners; 80.9% of killers in the “other” category were African; therefore most white women are killed by African men.
Unfortunately, this reasoning is as horribly flawed as the small cottage industry in amateur statistics, facts and figures it has provoked.
Firstly, the term “others” does not mean “unknown” but refers to anyone “other” than an intimate partner. The killer could be a stranger, but could also be a brother, a father, an uncle, a colleague, a friend or an acquaintance.
Secondly, when it is calculated that 80.9% of “other” men are African, this does not mean that 80.9% of those who kill African women are African, that 80.9% of those who kill coloured women are African or that 80.9% of those who kill white women are African.
That sort of finding can only be made when separate analyses of each group of women is undertaken by perpetrator race. The MRC study did not undertake such analysis so we do not know the racial breakdown of perpetrators per category of women.
Given that 75.6% of the women killed by “others” were African, it is most likely that the overwhelming majority of African men who killed a woman, killed African women. But of course, given that we do not have racial breakdowns of perpetrators per category of women, it is also possible that some white men may have killed some African women, that a few African men may have killed a small number of Indian women and that some coloured men killed white women. Each of these permutations is possible when we are dealing only with percentages overall.
Importantly, arguing as though it’s 1999 assumes that somehow the violent events of that particular year have remained miraculously preserved in time, creating a kind of template for female homicide that replicates itself identically every year. This is not a plausible assumption.
Trends in homicide cannot be established from one year’s worth of data but must be discerned in patterns emerging over years. In other words, we do not know how representative 1999 was of patterns and trends in female homicide generally and we certainly cannot claim that what was the case in 1999 is still the case 15 years later in 2014.
To do so would be as logical as using crime statistics from 1999 for 2014.
At the risk of stating the obvious, things change, as the MRC’s follow-up study of female homicide in 2009 showed clearly. Just as trends in murder rates generally had declined over the decade, so too had female homicide decreased between 1999 and 2009.
But the decrease in the number of women killed by “others” was more pronounced than the decrease in women killed by intimate partners. As a consequence, where 50% of women murdered died at the hands of their partners in 1999, this percentage had increased to 57% by 2009.
But is the difference between the two years and the two types of murders important? This is a key question for researchers because in many instances such findings may be neither here nor there, perhaps even the result of chance.
To sort the important from the inconsequential, researchers run a series of statistical calculations to test for significance. Once a particular number crosses a certain threshold it is deemed “significant.”
In the case of homicides by “others”, the difference between 1999 and 2009 did indeed cross that threshold, while in the case of intimate partner homicide it did not.
This has bearing on how we interpret the murders of white women.
If the number of intimate partner killings remained more or less within the same range as the estimates for 1999 – but the number of “other” killings dropped significantly – then it implies that the gap between the killings of white women by their intimate partners and by “others” will have changed by 2009 – even if we cannot pinpoint the difference to a precise degree.
This brings us to a bigger question: Why does it matter if, say, only 49% – or 47% or 45% – of white women are killed by their intimate partners? Do those deaths only become relevant and cause for concern when they amount to 51% or more of women’s deaths? Are white, intimate male partners rendered less culpable and more acceptable because they might not constitute 50% or more of those who kill white women?
Indeed, the framing of this argument and the distorted statistics it has produced suggests that this debate is really about white men’s virtue versus black men’s vice, with white women’s deaths of interest only insofar as they support this claim. The remedy demanded by those who complained to the Human Rights Commission – a public apology to white men from Africa Check – implies as much.
That the deaths of all black women, whether African, coloured or Indian, do not even feature in the discussion serves only to underscore how marginal women are to the white right.
Lisa Vetten is a researcher specialising in issues of gender violence and is based at the Wits Institute for Social and Economic Research. She has previously been the director of the Tshwaranang Legal Advocacy Centre and head of the gender programme at the Centre for the Study of Violence and Reconciliation. As a specialist on violence against women, she has advised the Commission on Gender Equality and been called to testify as an expert witness in court cases involving intimate femicide.
- Report: Are SA whites really being killed “like flies”?
- Report: Police wrong to claim they were “vindicated” over official crime statistics
- Analysis: Where murder happens in SA
- Factsheet: The 2012/13 crime statistics
- Guide: Understanding crime statistics
- Comment: Racial scare-mongering in South Africa makes light of women’s murders
- Comment: Dodgy stats just a means to an end for Hofmeyr and Bridges