My point is that the lead levels decreased back to where they were in the 1960s (-20) but crime rates only back to 1970s rates.
This is an incredibly strange graph
yeah sounds like there might be multiple things going on and that lead is just one of them.
1 Like
The fact that different countries phased out lead beginning at different times and essentially all saw a drop in violent crimes 20 years after they started is the most compelling evidence for the impact of lead. It seems incredible that such a simple factor could be behind relatively complex human behavior, but the evidence is super strong.
1 Like
Although this is reasonable, it is also the kind of thinking that could lead one astray. For example, what if countries that implement lead regulations are the same ones that implement a whole portfolio of socially oriented regulations? Broadly liberal reforms would tend to reduce crime and they’d also tend to correlate to environmental regulation.
I have no idea if any of this is the case and I have no reason to believe that lead regulations are bad. I’m just saying that comfortable narratives make me uncomfortable.
It’s not like the lead theory is unfalsifiable, but if there is a factor that had a stronger effect, we would expect to be able to isolate it in a similar fashion to lead and observe a tighter correlation. Up to now, nothing like that has been identified, and it’s not like no one has bothered looking. We also would expect to see that countries that banned lead but didn’t do X don’t see nearly as much effect or see it at a different time after they did X, and I don’t think that’s been observed, either. It’s true that correlation analysis can lead people astray (Freakenomics famously blamed this drop in crime on more abortions, but other countries that expanded abortion access at different times didn’t see the same temporal correlation with the drop in crime).
Social science experiments are really, really hard, so it’s not like this is epistemologically certain, but it’s about as strong an effect as you can observe in a study like this.
2 Likes
???
Lead seems like an interesting hypothesis, but it’s not uncontroversial.
https://twitter.com/sudonhim/status/1386488373234135041?s=20
Chris Hayes had a sociology prof/criminologist on his most recent podcast and lead didn’t come up, but the criminologist later tweeted that he didn’t think it was a major factor.
https://twitter.com/patrick_sharkey/status/1394746069561528335?s=20
So, the lead hypothesis may be less promising than I thought. I do suspect the recent uptick in murder is just more and more guns doing their work. I’ve heard guns are cheaper than they were in 1990 (though I have no idea if the uptick is “street crime” or more people killing spouses or baby mamas (see Miller, Jason)–seems an answerable question.))
This meta-analysis is predicated on the idea that there is fairly severe publication bias in studies involving lead and crime. The idea is that as the sample size of studies decreases, more error should be introduced into the results such that they scatter randomly on either side of the mean effect size. If they are instead skewed towards showing an effect, that might indicate publication bias. A confounding factor is heterogeneity of studies, e.g. smaller studies might just be looking at more severe lead exposure and thus be likelier to show an effect. They do address this in the paper but at that point it becomes above my pay grade to understand. But their results are that a study that is negative and significant would be 200 times less likely to be published than a study that is positive and significant, which seems unlikely to me because of how many studies on lead would be part of a larger study with more variables. That is, if you’re studying the effect of lead but also variables X, Y and Z on crime, and you come up with a result that lead doesn’t appear to be causative but variables X and Y are, you’re obviously going to publish that. So that smells off to me and that’s what their entire result is based around. Even with that, their result is that between 0% and 36% of the drop in homicide in the US is due to lead. While 0% is not very much, 36% is still a lot.
1 Like
Stephanie Grisham is like that fourth Marx Brother that nobody remembers. I don’t think she ever gave a single presser.
1 Like
Meaning, if he had simply apologized for being a racist asshole he would not have been fired? Lololol fuck off CNN
1 Like
Morning Joe was extra special during my morning routine today. A nice long segment with Frank Luntz and some other guy giving us the PULSE of the Trump voter. Because that’s what the MSNBC viewer tunes in to see!
My god this framing. “Biden gets pandemic under control, state tax revenues soar.” Nahh, this:
https://twitter.com/deborah_solomon/status/1397235306852651018?s=21
Oof.
