Let me mansplain statistics to you
- Ivo Vegter
- 16 Sep 2014 12:46 (South Africa)
With mandatory wall-to-wall coverage of the Oscar Pistorius trial, the media has turned it into a rote message: “Every eight hours a woman is killed by her intimate partner in South Africa”.
The Star says so. Women24 says so. The Guardian says so. A gun-control activist in this very publication says so. The only local source I could find that even tried to add some context was Eyewitness News, and its numbers did not add up.
It wrote: “The Medical Research Council of South Africa found that 2,363 women from the age 14 and older were murdered in South Africa in 2009 compared to 3,793 in 1999.
“Intimate partner violence is now the leading cause of death of female homicide victims, with 56 percent of female homicides being committed by an intimate partner.
“In 2009, one woman was killed by a partner every eight hours in South Africa compared to one women killed by a partner every six hours in 1999.”
But if you take 56% of 2,363, you get 1,323 and a third of a woman. Divide that by 365, and you get 3.63 women. This is exactly 21% higher than the 3.0 that the slogan “every eight hours” would indicate. It means one South African woman gets killed by her partner every six hours and 40 minutes. But this not only isn’t as marketable a slogan, it is wrong. The true rate is one every eight hours and 34 minutes.
So what happened here? The source of the phrase “every eight hours” is a Medical Research Council report published in 2012. It says that “intimate femicide”, as a woman’s murder by her partner or ex is apparently called, represents 1,024 of the 2,363 cases of femicide. That is 43%, not 56%. It represents 57% of the cases in which a perpetrator is identified, not 56%.
This last difference is important not because of its magnitude, but because it points to a misunderstanding. The intimate femicide rate, of 5.6 per 100,000, is the only number that has matching digits, but it represents 0.0056% of all women. And 0.0056% is neither slogan-friendly nor sensational.
Statistics such as “three a day” can be constructed from just about any data point. To demonstrate, I’ll do so for homicides involving male victims. Because the crime statistics released by the South African Police Services (SAPS) do not include such a breakdown, I’ll use the same data everyone else uses: the UN Office on Drugs and Crime (UNODC) statistics on homicides categorised by the gender of the victim.
Males represent 84.6% of all murder victims, or 13,755 out of a total of 16,259. That leaves us with 2,504 female victims, which is only slightly higher than the data from the MRC report, so it seems reasonable.
That gives us a male murder rate of one every 38 minutes.
But, the “every eight hours” activists say (or to be perfectly accurate, three of them told me on Twitter), that these men weren’t killed by their intimate partners just because of their gender.
This is true. Not all of them were. But let’s stipulate that we know the total murder rate for women is less than seven per day, while about 38 men are killed daily (this is no typo: 38 minutes goes into 24 hours roughly 38 times, bar a rounding error).
Now, the question is, how many of those men get murdered by their intimate partners? Well, one could speculate that since more than five times as many men get murdered in total, quite a few of those are likely killed by angry wives or girlfriends. In absolute terms, they may well match the “every eight hours” claim.
This has to remain speculation, because nobody actually measures that. The MRC report does not document “intimate manicide”. Neither does the SAPS. Nor does the UNODC. There isn’t even a word for it, although at least one reference site had the sense of humour to include “feminazi” among its suggestions for what I might have meant.
But let’s try our hand at informed speculation. We know the gender split for murder in South Africa is 84.6% male to 15.4% female. The UNODC does keep statistics about men and women killed by intimate partners or family members for countries that keep such data. Let us pick the UK as representative, because it is likely to have good records, there doesn’t seem to be any basis for choosing any particular country, and constructing weighted averages over all countries with good data would be overkill for our purposes.
In the UK about 67.5% of all women are killed by their partners or family members, while 13.7% of males are done in this way. (Recall that South Africa’s figure for intimate partner femicide was lower, at 57% of cases in which a perpetrator was identified, but excluded other family members, so it seems a reasonable comparison.)
So, if we suppose South Africa’s ratios are similar, would men die more or less often at the hands of their immediate family? (Arithmetically speaking, is 13.7% of 85% higher or lower than 67.5% of 15%? Bets?
As it turns out, if you’re going to count in numbers per day, more men get murdered by their intimate partners than women. Given these ratios, of all South African murders, 11.6% would be men killed by their partner or family, and 10.4% would be women killed that way. Or, to phrase it in the activists’ terms: a man dies at the hands of their immediate family once every four hours and 39 minutes, while a woman dies that way once every five hours and 11 minutes.
Now, remember that 0.0056% number? That is the proportion of women that are likely to be murdered each year by their intimate partners. Add the 4.2 per 100,000 that are killed by others, and you have 9.8 murder victims per 100,000, or roughly one in 10,000.
For men, the corresponding number is just short of 55 victims per 100,000, or a little more than one in 2,000. Allowing for rounding errors, this is consistent with 2009’s average murder rate of 34 per 100,000, as well as with the ratio we got earlier, that more than five men are killed for every female murder victim.
So, in truth, the disproportionately high component of South Africa’s murder rate is not women being murdered by their partners. It is men being murdered by people outside their family.
Having come this far, we should also note something else that isn’t captured in the phrase “every eight hours”. As only the EWN piece noted, it represents a significant improvement over the same number a decade earlier, which was “every six hours”.
You may also have spotted earlier that the total number of female murder victims has declined by almost 40% from 3,793 in 1999 to 2,363 in 2009, which is exactly consistent with the overall murder rate decline between those years, from 55 per 100,000 to 34 per 100,000. That rate has since declined further, to 31.1 per 100,000, but it falls to a foreign publication, The Economist, to point out that South Africa “is (a bit) less violent” than we are led to believe.
There is a fertile cross-pollination between the statistical tomfoolery about women’s murders and the claims of the white right about their supposed victimisation at the hands of black men. For example, the Afrikaner rights lobby group Afriforum is on record as claiming that one farmer is murdered every five days. Some people claim that this is indicative of a “white genocide” in progress.
The specific number may well be roughly accurate, as it is with the number of intimate-partner femicides, but it lacks broader context. Nechama Brodie researched the claims for AfricaCheck, and concluded that in fact, whites are victims of murder at a disproportionately low rate: “Whites accounted for 1.8% of the cases (although whites make up 8.85% of the population).”
Murder is a very emotional subject, of course. Even the MRC report that is the sole source for the femicide factoids has an exclamation mark in the subtitle.
These statistics are also polarising. For example, I now feel obliged to state what should go without saying: I don’t want to diminish the gravity of violence against women, or of the murder of farmers, white or otherwise. They are horrible crimes that society needs to fight with every legal means at its disposal.
But the same ought to be true of the high incidence of township murders. The same outrage should meet violence against men. It isn't like it happens any less often.
Is violence against men a sufficient cause for Men’s Rights Activism? I don’t feel particularly strongly about it. As with any form of activism, I am uncomfortable with its more belligerent elements, and the shrillness of its emotive appeal. (See? Men can be shrill too.)
That said, divorced or battered men surely deserve the same sympathy society reserves for women when roles are reversed. The movement has attracted both pro- and anti-feminist support, and not only from men.
I suspect this kind of movement is in large part a reaction to perpetual generalisations that cast one set of people as victims, and another group as aggressors.
Slogans like “every eight hours” or “every five days” are not only accusatory and shaming, but they are patently unfair. Trying so overtly and dishonestly to elicit sympathy for one group merely alienates the other. It is at best well-intended but self-centred stereotyping. Too often, it is demonstrably bigoted.
Murders are different only in terms of their preventative or law enforcement tactics. One is no less abhorrent than another. DM