Couple ngram associations. See them here.
Whenever I get a good indicator broken down by state, I head over to Google Correlate to see how it connects to America’s search behavior. Often what I find is a gun connection. This is very big in searches related to the election, so I’ll start with that before giving a couple other examples.
Odds of Romney winning
Taking yesterday’s New York Times 538 forecast chance of Mitt Romney winning each state, I entered the numbers into the search correlation machine. As you can see from the map on the left, these are very polarized numbers, with 42 of the states being above 90% or below 10%. Of the 100 Google searches whose relative frequency is most correlated with this pattern across states, 31 are about guns. Here is the search most correlated (.82) with Romney’s odds of winning: “marlin 30-30,” which is a classic rifle (available at Walmart):
News the other day was about the lives lost to unintentional injuries for people under age 20 — the most common causes of death in that age range. About half of this is from motor vehicle accidents, with most of the rest distributed between drowning, suffocating, fires, falls, and poisoning. The CDC put out a report that included a state breakdown, reported in terms of “years of potential life lost” per 100,000 population. That is just the number of deaths times the number of years between the age at which the death occurred and age 75 (so a death at age 1 is 74 years lost, a death at age 19 is 56).
The big inequalities here are in gender and geography. Males are about 1.8-times more likely to die from this stuff. And the most dangerous state (Mississippi) has more than 4-times the losses of the safest (Massachusetts). There are race differences as well — with American Indians having high rates — but the Black/White difference is not that large (Latino ethnicity wasn’t identified).
How are these rates of lost life correlated with search behavior? Guns. Among the 100 searches that most closely follow the pattern of deaths, 62 were about guns, starting with number 1: “shotgun for sale,” with a correlation of .93.
There were also 14 searches about cars and trucks on the list (mostly Ford F150s and Chevy trucks), four about wedding dresses and rings (“discount wedding dresses”) and three about Fox News.
I did this twice with divorce rates. Using the 2008-2009 divorce rates per 1,000 married women, I found a good gun correlation with gun searches, with “colt .45 automatic” scoring a .86:
There were 27 more gun-related searches on that 08-09 divorce-correlation list. I updated that for the new 2011 rates, and again came up with a list of gun-related searches (and other military or survivalist stuff). Here is the Norinco SKS and 2011 divorce rates, correlation .84:
Someone who knows more than me could probably read more into the searches for different kinds of guns and gun-related stuff — for example, the difference between sniper accessories, shotguns and handguns. These different gun results show variations in their geographic patterns.
Anyway, I can’t think of what else besides search data tells us so much about so many people’s behavior — not their stated interests, their reported behavior, their tax forms, or their consumption patterns. And yet I can’t really put my finger on what it does tell us. It’s a million miles wide and not that deep, but it’s endlessly fascinating. If someone can figure out how to explain the value of what this all shows, I’m all ears.
Are women becoming more violent as they move towards equality with men? Hanna Rosin says yes. But no.
I showed previously that Rosin, in her book The End of Men, grossly exaggerates the decline in sexual assault against women. What about the reverse — the increase in women’s violence?
In the radio show we were on together, the host asked, “Are [women] superior?”, and Rosin answered:
No. In fact, one thing I explicitly avoid in the book is this idea from Steven Pinker and others, that when women take over the world the world becomes a wonderful place, which is why I explicitly put in a chapter about violence, just to sort of make people understand that it’s not that women are wonderful and better … power has an effect on women like it has an effect on men.
That’s trouble, because when Rosin has the answer before she starts the research the outcome is a pretty sure thing. So, let’s see how much more violence women are perpetrating now.
Rosin starts her section on violence trends with this:
At the start of the aughts, criminologists began to notice something curious about the crime trends. The great crime wave of the mid-nineties was finally coming to an end. Rates of all violent crimes were plummeting — that is, violent crime committed by men.
She doesn’t actually say violence by women has increased, just emphasizes that it decreased for men. Her actual evidence for an upward trend in violence by women turns out to be from arrest rates. Using a report on the trend from 1992 to 2003, for example, Rosin describes the increase in juvenile assault arrests for girls, which was real.
Here is the trend she’s talking about: juvenile arrest rates for violent crimes, from the Bureau of Justice Statistics:
Even with the early-2000s decline, violent crime arrests for female juveniles were higher in 2004 than they were in 1980 — which was not true for men. She writes, “Women were by no means catching up to men, but they were fast closing the gap.” And that’s true, too — the male rate fell from roughly 8-times to roughly 4-times the female rate.
But arrest rates are tricky, since they reflect both (alleged) violence and police responses. Consider that the rate of homicide charges against female juveniles fell from around 1.0 per 100,000 in the early 1980s to about 0.5 in the late 2000s.
Setting aside juveniles and arrest rates, we can look at violence by women through reports of victims in the National Crime Victimization Survey (NCVS), and police accounts of homicide offenders.
From victim reports, there has been a slight increase in women’s representation among violent offenders in the last decade — edging up to about 20% of offenders. But the rate of violence by women is decidedly trending down.
The NCVS numbers show a drop in the number of women as violent offenders from the late 1990s. When you factor in population growth in that time, the drop is almost half — from more than 10 crimes per 100,000 women to less than 6:
Murder is probably the best-measured crime. According to FBI reports of homicide offenders, the number of male offenders was about the same in 2005 as it was 30 years earlier — but the number of women committing murder fell by more than 40%. As a result, the percentage of murders committed by women fell from more than 15% to less than 10%:
The FBI also releases rates of homicide for intimate partners. Here the evidence is dramatic and clear: From 1975 to 2005, the number of men murdered by intimates dropped by 75%, compared with a 25% drop in the number of women murdered:
From this evidence, it seems clear that women’s violence is on the decline — not rising, as Rosin says and implies. Even if the juvenile arrest rates reflect violence trends instead of just policing, those are also falling since the early 1990s.
When the evidence is the opposite
So far this goes in the category of exaggeration, ignoring existing evidence, and reading too much into too little evidence. But it gets much worse. A good tip from Ally Fogg (who elaborates here) pointed me toward a case where Rosin doesn’t just misread the evidence — she reverses the evidence to fit her argument.
After quoting a criminologist about how “unhinged” some people get when their narrative of female victimhood is disrupted, Rosin goes back to women’s arrest rates:
A recent British study showed the women were three times more likely to be arrested for domestic violence, and far more likely to use a weapon.
Fortunately, she provides a link to the study in question. The first half of that sentence refers to this passage. But read it:
As might be expected from the nature and severity of the domestic violence incidents, there were more arrests overall of men than of women. All cases [of couples with a violence charge] with seven or more incidents, most of which involved men, led to arrest at some time. This echoes US findings that male domestic violence perpetrators have more extensive criminal histories than female perpetrators. None the less, women were arrested to a disproportionate degree given the fewer incidents where they were perpetrators. Women were three times more likely to be arrested. (bold added)
This is not a case of just leaving out the context. The study’s fact is the opposite of what she implied, which was that women commit more domestic violence (remember, this is shortly after she used arrest rates to represent the prevalence of violence among juveniles).
On the second half of the sentence, that women were “far more likely to use a weapon,” it is true that the study found 24% of women accused of domestic violence used a weapon, compared with 11% of men. And the author wrote, “women were much more likely to use a weapon.” But in the table where that number appears, the difference is marked as not statistically significant (because this study was a small community-level study of just 126 perpetrators of domestic violence).
For the U.S., incidentally, Rosin should have cited evidence that men are in fact more likely to face a weapon in nonfatal intimate-partner violence — 81% versus 69%. Of course, men only experience 17% of nonfatal intimate-partner attacks. Overall, however, this BJS study from the 1990s reported that men were twice as likely as women to use a weapon in the commission of a violent crime. None of this suggests women are becoming more violent at their power grows.
Lest we forget
This 2005 NCVS study reported that women committed:
Men commit most of the violence. The growing employment rates, education levels and earnings of women don’t seem to have changed that much.
I am not coming unhinged. I am concerned about the damage done by Rosin’s corruption of the evidence to support the claim that women’s domination is nigh.
NPR science correspondent Shankar Vedantam the other day did a very nice job in a long story on how the recession may be affecting relationships and divorce. He interviewed me, and in this post I provide the sources for my comments.
The story began with a reference to this paper by Judith Hellerstein and Melinda Sandler Morrill, who find that divorce rates in the last few decades are lower in states with higher unemployment rates. They apply a rigorous set of tests and alternative explanations, and it looks pretty solid. This is tricky because research on couples shows that unemployment increases marital stress and the risk of violence. That’s not necessarily a contradiction, of course, since the research on states doesn’t show the people who are actually unemployed are the ones getting divorced — it could be other people in those states who are not getting divorced (people who fear unemployment, for example).
Vedantam’s story focused on two women: one who postponed a divorce because she couldn’t afford the legal fees and costs of starting over, and another who felt she was at risk of violence because economic scarcity was keeping her and her husband together after they intended to separate permanently. The examples made a good case for how unemployment could worsen marriages even as it prolonged them. NPR also has a new survey out, which shows marriages suffering when one partner is unemployed:
From my own explorations on divorce in the current economic crisis, it looks like there might be a delayed uptick in divorce, and higher rates in states where the crisis was worse — but that remains to be confirmed.
The NPR transcript reads:
Philip Cohen, a sociologist at the University of North Carolina at Chapel Hill, says that multiple studies have found that the marital distress that comes from money problems and feeling trapped is strongly associated with an increased risk of domestic violence. One study, for example, looked at women who showed up in hospital emergency rooms for injuries that were both intentional and non-intentional.
“When you compare the women who were injured intentionally and women who were treated for other conditions in the emergency departments, they found that those who were injured intentionally were more likely to have experienced recent unemployment in their families,” Cohen says….
At the same time, however, Cohen says the overall rates of domestic violence have generally been on the decline. But what’s clear, he says, is that unemployment increases the risk of domestic violence.
“I’m quite confident from the research on couples — and what drives violence within couples — that among the people who are experiencing economic shock or dislocation or unemployment, there is an increased risk of violence,” he says. “And I would not expect that to be any different during this recession.”
These were my principal sources:
In addition, my searching around over the last several years has not turned up any evidence that would lead me to doubt this association between unemployment and violence. Still, if you think I’m off on this, I’d be glad to hear about it.
New evidence against my earlier views.
Last year I cited a study of abusive head trauma in children which showed marked increases in four cities after the start of the recession (that was a conference paper that now has been published in Pediatrics). That followed a list of evidence supporting the idea that economic recession was increasing violence within families, from local intimate homicide rates and domestic violence court backlogs to social service reports and many news media anecdotes, along with general evidence that unemployment increases violence.
Since then, a 2011 review published in Sociology of Crime Law and Deviance reports that evidence for recession effects on intimate partner violence are weak at best, so there is no reason to expect an increase to show up when the data on this recession is put together. And now the latest data from the National Crime Victimization Survey shows a sharp drop in relationship violence against women in 2010:
But this is hard to evaluate because — as with divorce – it’s part of a long-term declining trend. To put that declining trend in perspective, I lined it up with the trends for violent crime, and the subset of serious violent crime. Showing these as percent changes, they are all following the same general trend. If anything, the decline in intimate partner violence against women has slowed relative to the others since 2007, which could be a sign of recession-related violence:
Finally, there is the problem of dating the “recession” to identify its effects. How to do that partly depends on what aspect of the the economic crisis you think might be affecting family violence — unemployment, home foreclosures, social service cuts, and so on. This has been dragging on long enough that we could assemble evidence in different directions several times. Here’s a final example: domestic homicide rates in New York state, which came out last month.
The total number of New York domestic homicides (which include intimate-partner homicides even if they don’t live together) increased about 10% in 2010, but the intimate partner portion of the total fell. You might say 2008 and 2009 constituted a recession-related spike in intimate-partner homicides, but 2010 doesn’t fit.
In short (not short enough), I’m not as convinced as I was that the recession has increased the level of violence over what it would have been otherwise. So, although there are sure to be many acts of violence triggered by the crisis, I recommend skepticism about broad conclusions on overall trends.
According to the latest research, unsupervised hanging out is nothing but trouble. But trouble is, what trouble? Or, compared to what?
Other research has shown that “unstructured socializing” by teens can lead to general delinquency, but this study is the first to suggest that it may also be associated with violent behavior.
Nothing but trouble.
As usual, I’m not an expert on this research. But I know enough to take issue with the conclusions. One co-author, Christopher Browning at Ohio State University, is quoted as saying, ”Parents are better off assuming that more structure is better for their teens.” As the press release further explains:
Browning said the study took into account a wide variety of characteristics that are also associated with violence, such as prior levels of violence of each adolescent, their levels of impulsivity, and the violence levels of each child’s peers.
But that is the difference between a statistical study – not that there’s anything wrong with that – and the advice to parents it generates. In real life, for real parents, those other things — the prior levels of violence, the level of impulsivity, and the level of violence among his or her peers — may be exactly what matters. For an individual, you should take them “into account,” but you can’t “hold them constant.” If a given child scores low on all the other risk factors, even an increase in the proportional risk of bad behavior is no big deal — compared with the possible benefits of unstructured socializing.
The article includes in the measure of “violent” behavior things that are very rare or unlikely in the case of many teens (like “gang fights”). If there is little or no risk of gang fights, doubling that risk doesn’t much matter.
Of course, if a given child has many risk factors, adding to them can make a big difference and be a bad idea. And I’m not diminishing the importance of preventing teen violence. But there is a difference between a population pattern and the advice it compels.
One additional beef — compared to what? Is increasing the risk of violence the only thing that matters? What about having fun, learning to socialize competently and make independent decisions? The study did not have a measure of the harm caused by the violence (the most common item was ”hitting” someone outside the family). Again, not to promote violence, but a little violence now might be better than a lot of some other problem later.
Everyone likes research with important direct implications, but pushing it to the personal level can be misleading.
OK, is this horse dead yet?
More bad news from the economic crisis.
A study of abusive head trauma in Pittsburgh, Seattle, Columbus (OH) and Cincinnati finds strong evidence of an increase in unequivocal cases since the start of the recession. Using data from the Child Protection Teams at four major hospitals from January 2004 through June 2009, the researchers found a rate of 4.8 cases of abusive head trauma in children before December 2007 — which they identified as the start of the recession — and 9.3 per month after that.
Abusive head trauma is the leading cause of death from child abuse. Poverty and stress are risk factors for abuse. During an economic recession, these risks are amplified while social service supports are often decreased.
The paper was presented at the annual meeting of the Pediatric Academic Societies, reported on here.
New data from the Bureau of Justice Statistics shows continuing decline in the rates at which both men and women report being the victim of intimate partner violence. Victimization rates for both groups are down by more than half in the last 15 years.
Source: My chart from BJS data.
The report is based on the National Crime Victimization Survey, and therefore is not subject to the same kind of under-reporting bias you get from reports to police.
That doesn’t mean it’s error free, of course. Another measure that tends to be accurate is homicides, since just about every death is counted, and these show declines too. Both trends show the greatest declines in the 1990s, when there were overall declines in crime and violence.
Women remain much more likely to report intimate partner violence, while men are more likely to suffer violence at the hands of friends, acquaintances and strangers.
If you are skeptical that things are getting better, I can think of two possible complications. First is changing age structure, so the aging baby boomers may be getting less violent (though by now their kids are old enough to start making things worse). The other thing is more young people being single, so there are fewer “intimates” to get hurt or killed. Maybe someone has already figured this out, but the BJS numbers don’t seem to account for it.