Tuesday 23 June 2015

Dull minds and criminal acts

I have always had a warm spot for cold Finland. The people are much friendlier than their prices, which tend to be either high or higher. The Finns have a habit of singing about their landscape whilst drinking sahti, but no one is perfect. I spent some happy midsummer days in Vaasa, near the Artic circle, the guest of Per Fortelius and family, meeting his friends, photographing the local architecture and doing some artic temperature wind-surfing.

Finland is the sort of place where they do things thoroughly, things like testing the intelligence of a total population cohort of Finnish males born in 1987 and following up the results. Gold dust.

Joseph A. Schwartz, Jukka Savolainen, Mikko Aaltonen, Marko Merikukka, Reija Paananen, Mika Gisslerd. Intelligence and criminal behavior in a total birth cohort: An examination of functional form, dimensions of intelligence, and the nature of offending. Intelligence, Vol 51, July–August 2015, Pages 109–118.

https://drive.google.com/file/d/0B3c4TxciNeJZUnVoVy1YRHlNTjA/view?usp=sharing

They found that lower levels of intelligence are associated with greater levels of offending, that the IQ-offending association is mostly linear, with some curvilinear aspects at highest and lowest levels, and that the pattern is consistent across multiple measures of intelligence and offending. In some ways this is exactly as predicted and already observed, since the available literature shows that individuals with lower IQ are more likely to engage in criminal behaviour.  Criminal offending was measured with nine different indicators from official records and intelligence was measured using three subscales (verbal, mathematical, and spatial reasoning) as well as a composite measure. The results show consistent evidence of mostly linear patterns, with some indication of curvilinear associations at the very lowest and the very highest ranges of intellectual ability.

However, the advantage of these data is that they deal with an entire birth cohort, so there are no distorting effects caused by the loss of a few miscreants who might account for lots of crimes. The population is restricted to males n = 21,513 because only males in Finland do military service and sit the intelligence tests. Offending is judged from real documentary data, not from fallible self report, even more fallible when painful memories are involved. Lastly, they have verbal, mathematic and spatial IQ measures, so can investigate whether verbal intelligence has a particular effect, as some have argued.

Here are the results, for general intelligence, and all crime:

image

Note that violent crime is an order of magnitude higher in the bottom 20% of the population by ability than the top 20% of population by ability. The pattern is generally a linear one. The subscales of intelligence show the same pattern, though perhaps the spatial scores show a slightly less pronounced differential effect.

 

image

So, why do dull minds carry out criminal acts? The main effect is driven by general intelligence, so that raises a number of possibilities, in that highly g-loaded factors such as deficits in executive functions, including inhibition, processing speed, and attention are potentially linked to criminal behaviour. People with higher levels of intelligence are more dependable ( Deary et al., 2008b) and conscientious ( Luciano, Wainwright, Wright, & Martin, 2006), suggesting that they are more likely to think about the moral consequences of their actions compared to individuals with lower levels of intelligence. People with lower intelligence have been found to act more impulsively ( de Wit et al., 2007 and Funder and Block, 1989). People with lower levels of impulse control and related constructs, such as low self-control, have also been found to be significantly more likely to engage in various forms of criminal and antisocial behavior ( Gottfredson and Hirschi, 1990,Moffitt et al., 2011 and Pratt and Cullen, 2000). While only preliminary, current research suggests that lower levels of intelligence reduces the ability to weigh the costs and benefits of individual action, resulting in a greater propensity to make impulsive decisions, which in some cases involve illegal behaviour.

It is a minor finding, but the dullest are not quite the most criminal, an honour reserved for those in the 2nd decile of ability. It may be that those in the 1st decile are slightly restricted in their behaviours by their very low ability, and may be under supervision from care givers.

Equally minor, there is a slight uptick for criminality in the most intelligent, though hardly the torrent of criminal master-minds beloved of popular entertainments.

The authors say: “low intelligence is a strong and consistent correlate of criminal offending. For example, the risk of acquiring a felony conviction by age 21 is nearly four times (3.6) higher among those in the three lowest categories (1–3) of total intelligence as compared to those scoring in the top three categories (7–9). We observed differences of similar magnitude across each indicator of criminal offending and regardless of the measure of intelligence. We found no evidence for the hypothesis that deficits in verbal intelligence are more salient to criminal offending than deficits in other dimensions of cognitive ability.”

The authors mildly point out that, strictly speaking, these results may be confined to Finland. However, an easy test comes to mind: have a look at crime statistics in your country, and work out, for example, whether the crime rate for those below the 30th percentile rank is higher than those above the 70th percentile rank, and how much higher. Or, look at the crime rate in your country for those whose ability is equivalent to Finnish 30th percentile or below, which would be Greenwich IQ of 92 or below.

For group differences within nations use Emil’s calculator on tail effects for group distributions:

http://emilkirkegaard.dk/understanding_statistics/?app=tail_effects

For example, if IQ 92 is the point below which criminality increases considerably, then 30% of the blue group are at risk, and 68% of the red group. In this way one can model what levels of crime would be expected if IQ were the main cause. A hypothesis worth testing.

23 comments:

  1. Vaasa is about 500km south of the Arctic circle.

    ReplyDelete
    Replies
    1. Change "near" to "nearest I have ever been"

      Delete
  2. I wonder how much of this is driven by Sami.

    ReplyDelete
  3. Sounds like a nice article. Unfortunately the Google Drive link to it above doesn't seem to be working for me. It opens up to an .htm file which only seems to have the first section in it.

    ReplyDelete
    Replies
    1. Same here. Can't see the article, just lots of html source.

      Delete
    2. http://www.researchgate.net/profile/Joseph_Schwartz4/publication/279060987_Intelligence_and_criminal_behavior_in_a_total_birth_cohort_An_examination_of_functional_form_dimensions_of_intelligence_and_the_nature_of_offending/links/558979e308ae273b2876c451.pdf https://www.dropbox.com/s/tdibqiclqg4th0i/2015-schwartz.pdf

      Delete
  4. I wonder how much of this is driven by Sami.

    Approximately none. About 0.1% of Finnish citizens are Sami.

    ReplyDelete
  5. "Offending is judged from real documentary data, not from fallible self report": well that by itself makes it far superior to much "social science".

    Did they, by any chance, also record anything demographic e.g. the number of children these males had spawned?

    ReplyDelete
  6. http://www.ncbi.nlm.nih.gov/pmc/articles/PMC3404054/

    ReplyDelete
    Replies
    1. Thanks, Jayman. A very detailed supportive finding.

      Delete
  7. Finland has criminals?

    "It is a minor finding, but the dullest are not quite the most criminal, an honour reserved for those in the 2nd decile of ability"

    Smart enough to figure out that the system is rigged against them?

    ReplyDelete
    Replies
    1. Least able more likely to be supervised in sheltered settings.

      Delete
  8. All of the hyperlinks embedded in the references you have cited (i.e. Deary et al. etc.) actually link to the Schwartz et al. article and not the actual references themselves.

    ReplyDelete
    Replies
    1. I usually cut them out, but wondered if they would work. Seems not.

      Delete
  9. Can anyone think of a way for correcting for the possibility that the brighter are more adept at avoiding prosecution and conviction? Or even possibly less adept at it?

    ReplyDelete
    Replies
    1. Depends on the criminal justice procedures. In America, a person's best chance of getting off are to convince the arresting cop that they're not an underclass type of person. So divide rate of criminality by rate of police interaction times rate of arrest per interaction times rate of arraignment per arrest (correct?) times rate of conviction per arraignment times felony per conviction.

      Or something like that. Intuitively, these rates would also be linear with IQ and therefore the shape of the original correlation will be unaffected, although the slope might change a little. Honestly, I doubt the effect would be significant, excepting white collar crimes in the financial industry.

      Delete
    2. @dearieme, haven't their been experiments attempting to determine how honest people are, such as whether they'd cheat at cards when they think they can get away with it?

      I agree, the thought had occurred to me too. The brighter a young man is, the more likely he is to get away with a crime.

      I find it fascinating that it doesn't go to 0, even for the brightest. In all categories, though, the majority commit no recorded crimes at all.

      What if one excluded traffic crime and DUI? DUI, because the decision to commit DUI is strongly influenced by alcohol, which is known to impede good judgement. Traffic crime, because, well, as I look at the highways around me, at certain times of day, _everyone_ is speeding. They're all committing a traffic crime, but they're conforming to expectations.

      Delete
    3. "The brighter a young man is, the more likely he is to get away with a crime."

      What level of crime? I mean, I guess yeah, at the point of arrest, or in some elaborate scheme, maybe, but regular caught red-handed crime? How would that work?

      Delete
    4. I am a high IQ, high socioeconomic status. Police have caught me committing crimes a few times. Mostly I talked my way out of it. On two occasions, my expensive lawyer talked my way out of it. On one occasion, I made a deal with the person I offended against. No convictions.

      Delete
  10. A Steve Sailer commenter quotes:

    “All three subscales were normed and coded on a nine-point scale with scores ranging between 1 and 9 and higher values indicating higher levels of intelligence.”

    “Importantly, this coding strategy provides a distinct advantage for the purposes of the current study, in that it eliminates the need to categorize a continuously measured IQ measure, which is common in previous studies.”

    I confess that I don't understand that.

    ReplyDelete
  11. I was not convinced by that argument, but decided not to make anything of it. I prefer raw data always, so that one can see the true distribution. However good the norming on a 9 point scale there will always be some difficulties, and some loss of fine detail. It may indeed account for the curvilinear functions at the higher and lower points of the scale.

    ReplyDelete
  12. I'd like to add couple of quick points. The group studied here is NOT the full cohort, it is the part of young males who have been conscripted to military service AND not called it quits during two first weeks.

    The test battery is administered during boot camp to assess trainability of conscripts. So guys who have opted for civil service are not included, guys who washed out quickly are not there and guys who have been exempted for various reasons are not there either. You can get of the hook for psychological (mostly fake) or physical reasons - in other words, the Finnish military doesnt want lazy weirdos and has no use for retards. I'd say you wont find people with sub-80 IQ in the barracks these days.

    So the sample does not include the true bottom of the barrel in mental capabilities - if you are institutionalized or need help getting through the normal daily life, you dont take these tests.

    ReplyDelete