I’m posting this here on “Alas” so that, two or three years from now, when I run into this statistic again and think “oh, yeah, I wrote a thing on that, didn’t I?” I can search for here it and find it.
“An estimated 63 percent of young men between the ages of 11 and 20 who are imprisoned for homicide have killed their mothers’ batterers.”
— Kimberle Crenshaw, in her article Intersectionality and Identity Politics: Learning from Violence Against Women of Color.
Someone posted this on a Tumblr blog I read and asked if there were any sources for that statistic.
Someone else posted a link to the 1997 book in which Crenshaw said that; Crenshaw did not include a citation.
I found two older sources, neither of which I have any confidence in.
In this 1994 academic article, the 63% statistic is cited to a 1991 resolution voted for by the Texas Senate.1
And in another academic article, also from 1994, the 63% figure is cited to the 1991 testimony of Susan Kelly-Dreiss, who is a feminist activist (now retired) but was never, as far as I know, someone who did original research.2
I really doubt that this statistic is reliable.
1) At best, it comes from data gathered in the 1980s, so it’s of questionable relevance nowadays.
2) But also, although some congressional reports are very conscientious when they use statistics, many just grab any statistics that support whatever point they want to make, without any attention paid to quality.
3) I also think it’s unlikely to be true because the 1980s was a peak for drug gang violence in the USA (it leveled off in the 1990s and has dropped enormously since). In a decade famous for gangs hiring young men to shoot each other, it seems implausible that the majority of murders committed by young men and boys were about abuse in their family. [Deleted for being wrong wrong wrong, see comments.]
4) Also, the Bureau of Justice Statistics divides its age groups, when it releases statistics, into “under 14,” “14-17,” and “18-24.” Most researchers follow suit. So the age range given in this statistic – “11-20” – is extremely oddball, and would require either an original data source, or doing a lot of extra work.
5) It just doesn’t seem likely, and it’s reasonable to treat unlikely claims with skepticism until they are proven.
Maybe this is a true statistic, but I won’t believe it until I can read the original source and see how the numbers were gathered and calculated.
- The citation says “Tex. S. Con. Res. 26, 72d Leg., 1st Sess. 1 (1991)” [↩]
- The citation: “See Women and Violence: Hearings Before the Senate Comm. on the Judiciary on Legislation to Reduce the Growing Problem of Violence Crime Against Women, 101st Cong. 131 (1991) (statement of Susan Kelly-Dreiss).” [↩]
As someone who has done research on homicides in the form of secondary data analysis on FBI UCR data, the numbers of male individuals who committed homicide against a parent while under age 20 is relatively small.
I looked up the data from 2004 and here are the numbers I got when I ran my analysis:
First column is # of homicides where the perp was age 20 or under. The second column is 21-30, third is over 30, last is the total. And I myself did use age 20 as my cut off in this analysis in order to have a working ordinal variable.
Male Perp: 43, 59, 100, 202
Female Perp: 8, 7, 19, 34
Total: 51, 66, 119, 236
Male Victim: 28, 38, 41, 107
Female Victim: 22, 28, 59, 109
Total: 40, 66, 106, 212
As it is, in 2004, it appears there was a maximum of 28 male perp, male victim offspring perpetrated homicides where the perp was under age 20. At least that we know of.
And in case anyone was wondering why I had this analysis all handy, I did it because I was looking at the role dependency plays in intra-family violence.
Edit: I just noticed that the gender of the victim goes missing in 11 cases, which brings the potential total up to 39.
Jeremy, does this include numbers for young men who killed their mother’s romantic partner who was not their own parent or step-parent? If not, I don’t think it can be compared to Amp’s statistic, which merely says “their mother’s batterer”, which could as easily be a boyfriend. (And if we’re talking homicide, I would suspect it’s easier psychologically to kill a parent’s nonparent romantic partner rather than a parent.)
It includes parent and step-parent. No idea about boyfriend, but 20 and under was generally a low in terms of committing intra-family violence, even if you include “other family” (which I suppose police might have coded mom’s “boyfriend” under).
But then I’m not sure how it would have been coded as it were, it might have been placed under “Undetermined” and that relationship makes up a pretty large number of homicides overall.
Other family (includes in-laws and everything else), Same columns as before:
Male perp: 42, 78, 102, 222
Female Perp: 9, 7, 23, 39
Total: 51, 85, 125, 261
Male Victim: 35, 66, 83, 184
Female Victim: 16, 19, 41, 76
Total: 51, 85, 124, 260
Furthermore, I’ve seen no evidence from the data that I’ve seen that shared genetics/heritage helps deter homicides. The UCR data I looked at had mom/dad being the primary killers of individuals under the age of 5, although that included step-parents, they made up a significantly smaller portion of the total than those listed as primary parents.
Hey Barry:
* I suspect you are correct that this statistic is wrong.
* One of your points is so terrible it makes me think that you’re wrong, despite all evidence.
I’m talking about your point about gangs. Generally speaking, media panics don’t relate to actual crime statistics at all. I can’t find statistics on youth drug related homicide but almost all statistics about the youth homicide rate in the US show it peaking around 1994 (which probably accounts for all the studies), well after the drug gang scare, and relatively low during the 80s. If, as you say, drug gang activity leveled off in the 90s, it is clearly not a primary contribution to the youth murder rate.
[this next part is posted later, after I did some more digging]: I still can’t find great statistics on youth gang-related violence but here’s a link to some BJS stats and Figure 40 shows that gang-related homicides (not all kids, and I couldn’t find separate stats for “drug gangs” and “non drug gangs”) have a peak in 1994 just like every other kind of homicide, and no particular spike in the 80s aside from a growth like every other kind of homicide. http://www.bjs.gov/content/pub/pdf/htus8008.pdf
This is extremely nit-picky — and as I said I’m pretty much certain that you are correct and this statistic is wrong — but the argument “this statistic must be false because it doesn’t correlate to a media panic” bothers me so much I ended up posting about it anyway.
Do you know how many individuals under the age of 5 had only bio parents as in-the-home caregivers, vs those who had stepparents? (Especially if you weight the distribution by socioeconomic status the way the homicides are distributed.) Without that it’s hard to draw conclusions from those numbers.
You may be right, though, that my gut reaction “it’s harder to kill a parent than a parent’s SO” may be inapplicable to the situations where homicide happens.
Point well taken!
Edwards 1992 Reducing family violence: The role of the family violence council cites the source as Ackerman 1985 The War Against Women: Overcoming Female Abuse.
This is where it gets complicated. In general, when doing research, you can use variables within your survey/data set to control, or more specifically, you can assume your data set is your population and control for certain variables that way. When doing research on homicide using something like the UCR, your population is the population of the US. Unless the data was recorded in the UCR, or there is an accurate source about the US population one can draw from, the best you can do is… guess.
In some cases I was able to use census data to fill in a few blanks but something I found frustrating doing my research was precisely this problem.
Presumably there are fewer kids/adolescents raised by step-parents, then who are raised primarily by bio parents. There are fewer kids/teens listed as being killed by step-parents/baby-sitter, etc. then by bio-parents. Furthermore, the ratio (as I recall) was not outside what one would expect the ratio of kids being raised by mom/dad, to those being raised by step-parents. Without being able to control for socio-economic status (which is not recorded in the UCR) there wasn’t much that can be assumed beyond that.
And technically, when it comes to parental perpetrated homicides, the relationship being listed as mom and dad would not automatically mean that they were killed by their bio-parents, as they could have been adopted and there’s no reason to assume that those recording the data in the UCR would record such cases differently.
To go back to the original question, there are a few more points I want to make. There is a problem where women are told that they need a man to protect them for the dangers of the world (“Don’t walk alone” etc.) In criminology, we have what is routine activities theory, which claims that crime occurs when you have a motivated offender, a vulnerable victim, and the absence of a moral guardian. The point here, is routine activities theory would support the idea that women “shouldn’t walk alone”.
My problem here is that this claim (which appears to have no basis in actual data, men are more likely to kill people they know outside their families, than within their families, and the younger they are, the more likely they are to be killing, outside their own families) is that it simply substitutes one moral guardian for women with another.
That is, while this claim appears to be trying to raise awareness of intra-family violence by implying that it has such a large scope, that most male teens are in jail because they were trying to defend their moms from an abusive partner; it in fact minimizes it by only focusing on violence from one particular relationship. This means it is actually ignoring the full scope of intra-family violence. As it is, intra-family violence can occur in any family relationship and women would be a lot safer if they simply didn’t have any male relatives who were over the age of 30.
Okay the exact question was bugging and since I had the data available, I decided to run it on the 2004 UCR in SPSS.
In 2004 there were 2093 male perpetrated homicide where the perp was 20 or under.
Of those relationships that might include “the person abusing perps mom” we have:
Father: 21
StepFather: 6
Other Family: 56
Other Known to victim: 188
As well as: Relationship not determined: 414
Even if all the Relationship not Determined cases are included, you’re still at less than half of the total.
Thanks for the explanation. I wasn’t sure what data was in the UCR, so that’s helpful to know. This part:
was mostly what I was getting at (that what matters is the relationship between “percentage of kids murdered by stepparent” to “percentage of kids living with stepparent”, not the raw number for “percentage of kids murdered by stepparent”–and ideally you’d weight that to the characteristics of the households, but I understand that that would require a separate study). So if the ratios are equal, that’s interesting, thanks.
(And you’re right about the bio-parent label too: I noticed that when I was writing it, thought “I should change that”, and then forgot before posting. My apologies for the unthinking exclusion of adopted folks.)
That’s very cool detective work!
Wow, that’s great work! Thanks very much for working that out.
EDITED ON 7/20/2015 to add: See a new comment in an open thread, here.