Poverty, Crime and Causality
I was bouncing around my twitter feed and landed on this tweet which in turn took me to a paper entiteld Childhood family income, adolescent violent criminality and substance misuse: quasi-experimental total population study. The paper appeared in the British Journal of Psychiatry in 2014. Here’s the basic summary:
Low socioeconomic status in childhood is a well-known
predictor of subsequent criminal and substance misuse
behaviours but the causal mechanisms are questioned.
To investigate whether childhood family income predicts subsequent violent criminality and substance misuse and whether the associations are in turn explained by unobserved familial risk factors.
Nationwide Swedish quasi-experimental, family-based study following cohorts born 1989–1993 (ntotal = 526 167, ncousins = 262 267, nsiblings = 216 424) between the ages of 15 and 21 years.
Children of parents in the lowest income quintile experienced a seven-fold increased hazard rate (HR) of being convicted of violent criminality compared with peers in the highest quintile (HR = 6.78, 95% CI 6.23–7.38). This association was entirely accounted for by unobserved familial risk factors (HR = 0.95, 95% CI 0.44–2.03). Similar pattern of effects was found for substance misuse.
There were no associations between childhood family income and subsequent violent criminality and substance misuse once we had adjusted for unobserved familial risk factors.
Declaration of interest
Because the British (let alone the Swedes) seem incapable of doing proper American, it might be worth translating the paper into something we English speakers can follow. Here goes. The study looked at 526,167 Swedish kids, or about 89% of all kids born in Sweden from 1989 to 1993. (Kids were excluded from the sample if they died or emigrated before their 15th birthday, if they were born with birth defects, if they couldn’t be linked to their birth parents, or if the authors were unable to determine the parents’ level of income.)
The authors found (no surprise to anyone) that kids born into the lowest income twentieth percentile of the population are far more likely to get convicted of violent criminal activity or become substance abusers. But, by accounting for changes in a family’s income over time and how that affected (or didn’t) criminality and substance abuse outcomes of siblings and cousins, the authors were able to conclude that a family’s income was not associated with violent criminal activity or substance abuse except insofar as income was being driven by some other unobserved factor(s) that itself was associated with negative outcomes. That unobserved factor (or factors) runs in families.
The authors are not as clear as I’d like in describing the data adjustment, and the process they use is not one I have employed myself at any point. But if I understand the limited description of the process correctly, they are basically noting that a kid in a 60th percentile income family is no less likely to become a criminal than his younger brother will be several years later when the family has dropped to below the 20th percentile of income. Furthermore, within each income level, crime tends to run in families.
To take the paper’s findings a bit further, there is a serious implication here: it isn’t so much that poverty drives people into crime, but that families whose members have a tendency toward criminal behavior have an increased likelihood of ending up poor. Perhaps those who lack empathy are both more likely to commit crimes and less willing or able to behave in ways that allow them to get and retain good jobs. Of course, some of the smarter criminals can fake empathy enough to do quite well for themselves. It is also important to note that most poor people are not criminal. Nevertheless, the reason crime correlates with poverty is not that poverty leads to crime, but rather that for a not insignificant piece of the population, criminal tendencies are associated with traits that increase a person’s likelihood of being in poverty.
update… grammatical error in the last line of the post corrected, July 7, 5:52 AM PST
Taking a line through Trump, maybe you are more likely to be convicted when you become poor.
I have no doubt that the effect you describe exists. To use two cases that were prominent during my lifetime, OJ and Robert Blake were each famously acquitted of killing their second wives (plus OJ was also acquitted of killing his ex-wife’s girlfriend) but it seems were widely believed to have done it and I believe both lost civil suits related to the deaths. Obviously nobody poor could have put up the same sort defense in a criminal trial.
However, I would imagine that effect is relatively small in the grand scheme of violent crime. For instance, in just about every major city there will be a par of the city known for having higher levels of violent crime. I am not familiar with any example where the neighborhoods that are best avoided also do not appear to be neighborhoods where the income levels tend to be quite low relative to the national average.
If the effect to which you point was high, we would expect to see the opposite – the most dangerous parts of the average city would be those where the wealthiest people lived.
The more interesting and very different question:
Is the kid in a family with below the 20th percentile income, more likely to become a criminal than his younger brother will be several years later when the family has risen to the 60th percentile of income.
Unfortunately, as I noted, the authors do not make every bit of the process explicit. To the best of my ability to figure out what they did, it seems that their calculations go both ways. So I noted in the post that:
Perhaps for completeness I should have added this:
If there was a difference between the two scenarios identified in the paper, I missed it. Again… this is my understanding and I do wish the paper had been more explicit.
What you say is no doubt true, but it may also be true that social pathology is expressed differently in the rich than in the poor. That does not mean it does not exist.
I note that the authors of the referenced study , if anybody bothered to even read it, make particular mention of the types of unmeasured “familial risk factors”
Ref: http://bjp.rcpsych.org/content/bjprcpsych/205/4/286.full.pdf :
“….familial risk factors (such as the quality of the parent–child relationship, family dissolution and parental criminality) are instead viewed as proximal risk factors because they tend to explain the majority of the variance in such outcomes.”
The majority of “such outcomes” referred to are:
” early socioeconomic exposures as distal risk factors because of their lack of direct associations with delinquency and antisocial behaviours ,…”.
Each of these familiar factors are also highly inversely correlated with family income… so its it cannot be at all concluded that family income s plays no role adolescent criminal behavior as readers and Mr. Kimel have tried to infer.
What the authors concluded is only that
” There were no associations between childhood family
income and subsequent violent criminality and substance
misuse once we had adjusted for unobserved familial risk
What they said is that when the unobserved family risk factors are removed then economics doesn’t account for the adolescent criminal behavior.
They did not conclude that family income is not a causal source of adolescent criminal behavior. They could not and did not draw that conclusion because they did not account for the “unobserved familial influences” on adolescent criminal behavior which types of factors are known by other studies and by the authors (in cited references in their report) to be highly correlated with economics.
In fact their recommendations based on their study is that solutions have to look beyond simple economics in so far as familial factors that are directly related to economics must be taken into consideration.
“We found strong inverse correlations that were explained fully
by unobserved familial risk factors shared by children growing
up in low SES households. Future research is needed to validate
these results in other contexts and elucidate the nature of the
mechanisms, including the relative contributions of genes and
“nature of the mechanisms” that are included in the “unobserved familial risk factors” need to be identified. The authors made no attempt to identify them. Nearly all studies I’ve read over the years have identified many primarily economic causes for those things that are correlated with youth criminal activities… parental educational attainment, (also found highly correlated in this study as well.. .if you bothered to read the report), single parent, divorced parents, parental criminal activity, nutrition, health care, family density per household room, adequate heat, cooling, etc. not to mention neighborhoods, all of which are included in “unobserved familial factors”.
By the way the report is in plain ordinary English words, sentence structure, and phraseology.
I decided to look a little further into Mr. Kimel’s source of finding the 2014 report he refers to.
Mr. Kimel stated:
“I was bouncing around my twitter feed and landed on this tweet which in turn took me to a paper …”
The tweet is from Brian Boutwell.
Who is Brian Boutwell?
He is a co-author of a genetics based determinate of criminal behavior in this paper which supports “differential k theory”:
What’s differential k theory?
Differential k theory: “… this theory explains race differences in fertility, IQ, criminality, and sexual anatomy and behavior The theory also hypothesizes that a single factor, the “k factor”, affects multiple population statistics Rushton referred to as “life-history traits”
“critics have argued that Rushton’s predictions based on the theory “are supported by selective citation and misrepresentation of the research literature and by the use of unreliable sources and that Rushton’s methodology “indicates a lack of familiarity with ecological thinking and scientific method in general.” Additional criticism of the theory has come from Edward M. Miller, who has argued that contrary to the theory, unpredictable environments select for k, not r, characteristics.”
Here’s another of Boutwell’s articles:
“Is Crime Genetic? Scientists Don’t Know Because they’re Afriaid to Ask”
“Do Parents Really Matter?”
Which not just btw, was pubiished by Spectator Life.
What is Specator Life?
“The Spectator is a weekly British conservative magazine on politics, culture, and current affairs. It was first published on 6 July 1828. It is currently owned by David and Frederick Barclay who also own The Daily Telegraph newspaper, via Press Holdings. Its principal subject areas are politics and culture.
Its editorial outlook is generally supportive of the Conservative Party, although regular contributors include some outside that fold, such as Frank Field, Rod Liddle and Martin Bright. The magazine also contains arts pages on books, music, opera, and film and TV reviews.
Editorship of The Spectator has often been a step on the ladder to high office in the Conservative Party in the UK – past editors include Boris Johnson (1999–2005), the incumbent Foreign Secretary, and former cabinet members Iain Macleod, Ian Gilmour, and Nigel Lawson.”
I’ll let readers of AB google more on Brian Boutwell and make up their own mind about the mind-set Mr. Kimel finds interesting.
My own opinion is no different than it has been from the first post I read by Mr. Kimel… and his forwarding of a 2014 paper from a tweet by a known “crime is caused by racial genetics” criminologist fits perfectly.
For some reason I keep giving you the benefit of the doubt. It took me a while to accept that you were seriously defending the rape of little boys and FGM as cultural. And for some reason, I keep assuming you are not going to misrepresent studies.
But to everyone else… here’s an interview with the lead author in Swedish (https://kit.se/2017/06/16/88686/del-3-tank-om-det-sitter-i-generna/). Google translate does a pretty good job of translating it. He sounds kind of irritated about people misrepresenting his work. For example, here’s a bit from google translate:
The interview goes on:
So what is the difference between his research and previous work? Well, he got himself a massive data set. (If you check his homepage you’ll find he’s been working that data set quite a bit.) Second, he didn’t start out with the assumption that poverty is the driver. He let the data go where it went.
But misrepresenting the work that falls out of his massive data set seems par for the course. Here he was on twitter just a few days ago noting another example of people doing what you are doing. As he notes:
It occurs to me that Longtooth might even misrepresent the latest set of quotes I provided. So let’s cut to the chase. Here’s the principal investigator telling people, on twitter, about the interview I cited earlier
Now, you might argue his analysis is wrong, or the data is wrong, or whatever, but it is pretty clear what he believes the data is showing.
Gee Mr. Kimel,
I was just providing what the paper said and concluded in contrast to what you told readers on AB it said:
“:But, by accounting for changes in a family’s income over time and how that affected (or didn’t) criminality and substance abuse outcomes of siblings and cousins, the authors were able to conclude that a family’s income was not associated with violent criminal activity or substance abuse except insofar as income was being driven by some other unobserved factor(s) that itself was associated with negative outcomes. That unobserved factor (or factors) runs in families.”
“To take the paper’s findings a bit further, there is a serious implication here: it isn’t so much that poverty drives people into crime, but that families whose members have a tendency toward criminal behavior have an increased likelihood of ending up poor.”
Those two statements by you, taken together and in context are in no way what the paper showed or stated in its body, much less concluded.
You misrepresented the paper’s findings to readers on AB: ” that families whose members have a tendency toward criminal behavior have an increased likelihood of ending up poor.”
That’s is patently pure unadulterated bulllshit…. the paper drew no such conclusion’s inference, or interpretations
I was just informing AB’s readers of the papers conclusions, not in teast bit misrepresenting them.
let me say it this way.
The paper used one economic factor .. family income… as the only and single economic factor it evaluated. There are literally dozens of known major economic factors in behavior and outcomes, family income being only one of them. The authors lumped all other unknown and all other unknown non-economic factors into one bucket; Unobserved familial risk factors.
In any experimental analysis one can regress only on the known and quantified variables. The authors had no data on anything but parents, siblings, cousins, and family income… each by age over a 15 max year time span. They had no data on any other major economic factors known to being highly correlated with childhood and adolescent behavior or any genetic information, or any parental lifestyles. Therefore all of these other known economically dependent variables ended up in the unobserved (unknown) Familial Risk factors.
The composite of family income, nuclear family siblings, and all other unknown factors = adolescent criminal behavior is highly negatively correlated with income quintiles as the authors show clearly.
Income quintiles have more than just the family income within those quintiles… single parents, divorced and remarried parents, employment types, unemployment, etc. not to mention neighborhood characteristics, lifestyles, habits, educational attainment, etc. All are well known to be highly correlated with income. But income is only one of the variables..
When sibling behavior is regressed on family income by Cox Hazard Ratio regression analysis, adolescent behavior is indistinguishable by family income. .. e.g. the hazard ratio’s were statistically identical with no hazards… actually with about a 10% lower hazard of adolescent crime than the reference basis group —which was the upper quintile … = 1.0..
This study does not conclude, infer, or conjecture that income does not have a major effect on adolescent behavior.. It says that if they account for income only, (within sibling nuclear families) the adolescent behavior differentiation by family income disappears.
Therefore all the other income related “unobserved familial risk factors” account for the entire difference among siblings in terms of adolescent crimes.
That is the only conclusion the paper drew. Your description of what the paper concluded in your description on AB was not at all what the authors concluded or tried to conclude, infer, or conjecture..
With regard to your source of finding the 2014 Swedish article, through your “twitter feed” which included Brian Boutwell’s twitter account, I’ll let readers decide for themselves what this infers about your agenda. I have already provded enough information on Boutwell (and further information I haven’t published on AB, on Boutwell and his groups and other publications) to enable the intelligent reader to be able to draw their own conclusions related to your agenda.
I bounced around on Twitter from one user to the other. I found Boutwell interesting enough that I am following him for the moment. He follows feeds in criminology. I follow people because they pull up interesting information, not because I agree with them or disagree them. In fact, I am more likely to follow someone with whom I disagree because they are likely to be bringing up things which don’t fit my preconceptions. ( And a paper looking at virtually the entire sample of children born in Sweden over a several year period is something interesting.) That forces me to think.
Now, this is exactly the problem that I keep describing, that we are ceding the field to the people we do not want make policy. This is the sort of study that people on the left should be following. After all, if it is right, and I suspect it is, it means that we are wasting a lot of of money on social policies that will not work and we should be considering different social policies that will benefit more people.
As to your repeated misrepresentations, to quote the lead investigator again: