-
Features
Clear evidence for lead levels causing crime
- Thread starter Loren Pechtel
- Start date
zorq
Veteran Member
If I wanted headlines and links I would browse a news site. Can you offer a quick paraphrase and maybe your opinion of the article please? This is a discussion forum, not a link dump.
bilby
Fair dinkum thinkum
- Joined
- Mar 6, 2007
- Messages
- 40,281
- Gender
- He/Him
- Basic Beliefs
- Strong Atheist
I am not sure that a trend identified at least four years ago really counts as 'news', so a news site might not help much.
I presume that Loren's link (which I haven't clicked yet) refers to the same data as we discussed here almost exactly two years ago:
Some good news:
The article posits the lead exposure in childhood leads to violence hypothesis to explain the trend:
It was also touched on here, a year before that; By none other than Loren Pechtel.
It's old news.
bilby
Fair dinkum thinkum
- Joined
- Mar 6, 2007
- Messages
- 40,281
- Gender
- He/Him
- Basic Beliefs
- Strong Atheist
OK, I clicked on the link; And I immediately found a big issue.
That's far from 'obvious'; the non-intervention group weren't tested again, so it's perfectly possible that the 'intervention' was completely ineffective, and that both groups had a decline of similar magnitude due either to some external factor, or to the effects of 'intervention' on the supposedly 'non-intervention' group.
As the article goes on to say
Without blood test results for the 'non-intervention' group, these results are far from compelling. There may well be an effect, but the design of the study (as reported) appears to be such that it's not certain at all.
Even if the 'obvious' assumption is true, and the lead levels were markedly different between the two groups, it is possible that the intervention itself, and not the lead levels, was responsible for the result - perhaps the 'intervention' took the form of moving children to cleaner neighborhoods, away from main roads, with less lead and better schools.
Correlation is not causation, even if it seems 'obvious' that it is.
My bold.
That's far from 'obvious'; the non-intervention group weren't tested again, so it's perfectly possible that the 'intervention' was completely ineffective, and that both groups had a decline of similar magnitude due either to some external factor, or to the effects of 'intervention' on the supposedly 'non-intervention' group.
As the article goes on to say
and the similarity in lifestyle, demographics and environment that is lauded as making this study so clearly relevant has a sting in its tail - the controls could easily have been benefiting in some ways from the interventions applied to the test group. For example, if the intervention involved removing leaded paint from an apartment building, all the children in that building, both 'test' and 'control', would be expected to see similar drops in blood concentration - and this would go undetected, under the 'obvious' (but in this scenario, false) assumption that the non-intervention group did not see any decline.
Without blood test results for the 'non-intervention' group, these results are far from compelling. There may well be an effect, but the design of the study (as reported) appears to be such that it's not certain at all.
Even if the 'obvious' assumption is true, and the lead levels were markedly different between the two groups, it is possible that the intervention itself, and not the lead levels, was responsible for the result - perhaps the 'intervention' took the form of moving children to cleaner neighborhoods, away from main roads, with less lead and better schools.
Correlation is not causation, even if it seems 'obvious' that it is.
No, this is very different. They are looking at the violent crime levels of those who were just below the point of being treated for excess lead vs those that were higher and thus treated. Big difference.
ronburgundy
Contributor
This isn't a simple correlation. Its's It is a naturalistic quasi-experiment with an event-based time series design and direct comparison group. One group was assigned to an intervention based on an arbitrary cuttoff of BLL on the second measure, making an imperfect but still approximation of random assignment to control vs. experimental group. They also statistically controlled for numerous other factors, despite no reason to expect them to have varied between the groups.
There is no indication that any of the kids were moved or their daycare/schools changed. Rather the report says the lead was removed from their existing environments via renovations. The intervention is the most plausible cause of the difference between the groups. However, there was other aspects of the intervention that could have mattered, including having more thorough medical and psychological exams done with referral for treatment and changes in diet.
You are correct that they do not know that the control group did not also drop in BLL levels after the 2nd blood test. In fact, the actual research report shows that of the 500 kids who tested above the threshold on the first test, about 70% dropped below that threshold by the 2nd test without the need for any intervention. Thus, it is probable that the control group did also drop to some extent for reason unrelated to the intervention. However, if we assume that they spontaneously dropped just as much as a group with a targeted $8,000 per child intervention to removed lead from a child's environment, then it suggest that efforts to remove lead are a waste of public resources, because kids above acceptable levels somehow drop on their own just as much as an intense intervention.