More men killed by someone close than women?

Is this BBC article grossly misrepresenting the truth? OR (more likely) am I failing at basic maths?

It says

But if (all figures from the article) 80% of killings are of men and 36% of those are by someone close then 28.8% (0.8 * 0.36) of deaths are due to someone close. And if 20% of killings are of women and 64% of those are by someone close then 12.8% (0.2 * 0.64) of deaths are due to someone close. Which means that it is men who are actually more likely to be killed by someone close.

Or is my maths wrong?

Women who are killed are more likely to be be killed by someone close to them then men that are killed. You don’t need the 80%/20% part.

But that’s the thing: is it actually correct? If my maths is correct then because vastly more men are killed then more men than women in absolute numbers are killed by someone close.

The point is that statistically women are in more danger from people close to them than they are from strangers; men are in more danger from strangers than they are from people close to them.

The point is that the headline represents the truth in a way that is subtly but effectively misleading. Among women I know, it is common for them to believe that women are more likely to be killed, that it is more dangerous being a women, that members of their family are more likely to kill them than to kill other male members of their family. And conversely, that it is safer to be a man, that men are in less danger, and that when death or violence happens to men, they have provoked it.

Exactly. Women are at a far higher risk of being killed in a domestic violence incident, or an honor abuse incident, than a man would be. The victims are generally girlfriends, wives, in-laws, or relatives of the perpetrators, who generally fall into the “close” category.

Every once in a while, there’s a gang war, which are generally perpetrated by male gangsters, against male gangsters (female gangsters being very rare), and if any women are killed, it’s frequently a “civilian” who happened to be in the area where a drive-by took place, or they were a witness, or they happened to be home when someone came to kill their boyfriend. (Of course, male “civilians” are at equal risk.) The male gangsters might be “acquaintances” of the male gangster victim but they might only know the victim by face and street name. I figure if they were “close” it was probably an intra-gang killing.

…really? Even when war is taken into account? I’m curious if you ask that question how many folks, male or female, have that belief.

What’s wrong is the assumption that the two axis are independent and therefore their combination is obtained by multiplying both probabilities: the axis are not independent.

“Most pears are green” and “most fruits are below 4cm in size at the largest axis” does not lead to “since pears are fruits, most pears are green and less than 4cm in size at the largest axis”.

To answer OP, yes, the BBC are grossly misrepresenting the truth. The conflation of “women” and “female murder victims” is no more outrageous than the very existence of an article focusing on murders of women, the small minority of victims. What’s next, “The White people killed on one day around the world”? With the possible exception of intimate partner violence, where British figures have men as about 40% of murder victims, men and boys are more likely to be killed in every situation, from murder by the police to murder in the workplace to murder of children under five.

The data is poorly presented but the numbers are that 50,000 of 87,000 women are murdered by relatives or partners. Similarly,** 28,125** of 348,000 men murder victims are murdered by relatives or partners. Thus the 64/36% split in “Killings by partner or family member” in the second chart.

So to answer the OP, no and yes, but you aren’t on your own there.

The article is about the fact that women are disproportionately higher percentage of victims in just one category of killing. How is that misrepresenting anything? That disparity is the story.

Why “possible” exception? You acknowledge the numbers yourself.

I don’t see that as a gross misrepresentation, it is more that it is clumsy and imprecise.
It is looking at the issue from a particular point of view. The article is part of a series that relates specifically to women’s issues and particularly in this case their disproportionate murder by partners.

Right there in the article there is a big % chart that shows gender breakdown for overall homicides and it states.

So yes, in the modern world in general it is still a far more dangerous to be man than a woman but there are some limited circumstances where the reverse is true.
The focus of this article is to look at such circumstances.
Personally I’m not sure why homicides by partners or people you know are worthy of more interest than those carried out by strangers but there you are, that’s the area of investigation that the article has chosen.

One criticism I would have is that it gives three different figures for slightly different definitions. in one case it is 58% of all women killed are by family members or partners.
In another it is 8 out of 10 of all murders committed by intimate partners are female. In another it is 64% of killings by a family member are female victims. Some consistency and clarity is needed.

As for absolute numbers I can’t locate the raw data. I’m thinking that what they trying to get across is something like the following.

imagine 100 homicides, in total 80% male 20% female so 80/20 in actual numbers.

Of those 100 homicides, 10 were by intimate partner and the gender proportions are reversed: 20% male and 80% female so 2/8

Now their charts don’t make this clear and by using percentages as their scale in many cases you lose a grasp of absolute numbers involved.

I see no error in the way the BBC presents the story. You could present the data either way, but I think they make it clear that they are discussing the relative probability of who killed someone given that they have been murdered.

It’s not a question of independence, it’s conditional probability, i.e. the probability of who killed you conditioned on the fact that you were murdered.

The classic fallacy of failure to condition probabilities for known data was illustrated in the OJ Simpson murder trial, as it happens relating to an almost identical issue. I forget the exact numbers involved but it went like this:

Defence attorney: “The probability that a husband will kill his wife is only 1 in 10,000, so it is 99.99% probable that my client is innocent!”

But he ignored the known data. The relevant probability for O.J.Simpson was the likelihood that a husband has killed his wife given that we know she was murdered by someone. Which is more like 50%.

Yes, your maths is wrong. The article (and the raw data in the pdf) does not say that 36% of all murdered men were killed by someone close, that figure (not stated explicitly) would be lower.

It says that:
of all murders, 20% of victims were women;
of all people murdered by someone close, 64% of victims were women (and 36% men).

from the UN pdf:

The inferred proportions of all murders would be:

73% stranger kills man
7% stranger kills woman
7% partner or family member kills man
13% partner or family member kills woman

Just how are they describing “homocides”? I find the 1/5 killed by person known number quoted to be low.
Are they adding in deaths due to war as homocides or deaths due to political violence? The former is not homocide and the latter may or may not be.
The articles reasoning is already doubtful, they have called a judicial execution (for the killing of her husband!) a homocide.

No, they didn’t. If you’re going to be critical of the accuracy of an article, then you’d better do a better job of being accurate yourself.

First of all, this reported incident has nothing to do with the cited statistics from the U.N. study.

And the BBC said:

The article did not claim that this judicial execution (in Iran) was a homicide. It did say:

It seems to me quite appropriate to report this incident in the context of an article about the appalling treatment of women around the world. And although it would not be reported by Iran in homicide statistics, if Amnesty’s claims are true, arguably it is a homicide by any civilized ethical standard.

Ok. Do these civilised statistics also record imprisonments as kidnappings and fines as thefts?

To supplement the UN report, the BBC picked a recent day at random and reported on 47 “women killed… for probable gender-related reasons” around the world that day. These incidents were not inputs to the UN statistics, or any statistics, nor did the BBC claim they were all homicides, although most of them obviously were; the date is much too recent for trials and convictions to have taken place. One of these was a woman who was executed in Iran after possible torture for killing an abusive husband at age 17 after police failed to act to stop him abusing her.

Would you care to explain why you think these anecdotal BBC reports undermine the entirely separate U.N. statistics?

Since the U.N. report relies on consolidating statistics provided by each nation, and this incident would not have been reported as a homicide by Iran, it seems to me that it’s evidence that there may be a greater problem with homicidal violence toward women than the U.N. statistics imply.

Yes, but if you work the numbers as I did, it appears that more men in absolute numbers are murdered by someone close than women.