Hacker News Reader

>> If it's lead, did crime rates increase 20 years after leaded fuel became a thing?

Yes. The rise and fall of violent crime rates correlate very tightly to the rise and fall of leaded fuel, with a ~20 year gap. You see the same correlations in different countries that had different phase in / phase out times, and different U.S. states that had different timelines. (E.g. if a state phased out leaded gas 5 years earlier than average, their crime wave also started abating 5 years earlier than average)

Source? Crime in general decreased quite a lot after the depression, but the use of leaded gasoline was quite prevalent regardless.

Leaded gasoline was introduced in the 1920s but there weren't that many cars until starting in the late 1950s. (Construction of the interstates started with the Federal Aid Highway Act of 1956.) Then we saw the huge crime wave starting 20 years later, which peaked and started tapering off in 1993, 20 years after the phase out of leaded gasoline began in 1973.

There is also the point that the people who did have cars in the 1920s were the aristocrats, and twenty years after that...

> Then we saw the huge crime wave starting 20 years later, which peaked and started tapering off in 1993...

This seems to correlate fairly well with the government's war on (certain) drugs as well.

It also correlates with population density, public housing, migration, and crack cocaine use.

I highly doubt there's one contributing factor to crime, and I highly doubt it is solely from leaded gasoline.

There are many things that influence crime, the question is how much of the total variance is due to gasoline?

And the answer from this research is quite a lot, because the timing of the rise and fall in crime match the timing of leaded gasoline (with a 20-year lag) across cities, states, and countries.

To say that 'other factors are responsible' would be to posit that there is some other factor that holds both across and within different cultures and countries that perfect accounts for the pattern in crime rates that we see. That seems much harder to believe than simply 'leaded gasoline causes crime', which we both have robust evidence for now, and for which we have a very plausible mechanism of action.

Ok, so I'm not disputing that peope born in the 60s had more lead in their blood than people born in the 80s/90s. Additonally, I'm not disputing that lead causes behavioral and physical problems. It definitely does.

However, I would like to see the specific data that says the following (because I'm not seeing it when I read the article):

"Regardless of the crack epidemic in the 1980s, regardless of the decline in public housing in the 1980s, regardless of the rise of police surveillance technology, we have indisputable evidence that lead in the blood causes an increase in crime locally and nationwide by XMeaningfulPercent".

I'd be surprised if one could ever make such a statement.

The data we currently have shows that someone born in 1960 and committing a crime in 1985 has been exposed to lead, but they also might have been exposed to crack. Additionally, declining government project housing provided strong protection for criminals to hide from police, and the police surveillance technology in 1985 was pitiful compared to the massive levels of surveillance now.

None of these factors exist anymore in 2018. Crime is much lower. How can we say which variable contributed in which amount?

> "Regardless of the crack epidemic in the 1980s, regardless of the decline in public housing in the 1980s, regardless of the rise of police surveillance technology, we have indisputable evidence that lead in the blood causes an increase in crime locally and nationwide by XMeaningfulPercent".

> I'd be surprised if one could ever make such a statement.

"We found some cities where crack got big at different times but lead came at the same time"

"We found some cities where public housing declined at different times but lead came at the same time"

"We found some places where police surveillance..."

This is the aforementioned rabbit hole. People will have studied such evidence to various degrees.

Maybe they liked crack so much because they'd been exposed to so much lead.

But seriously, the point of this article is that studies still find a correlation between lead exposure and crime, after accounting for confounding.

And yet, that's not the same arguing that lead exposure was the only cause of the crime boom.

Your standards are bullshit and you know it. We can't say such statements of indisputable evidence about anything that happens in the world. Should we throw our hands up and just say, well it could be invisible llamas, or tree gnomes, or the boogey man, or should we use the best available evidence to guide future decisions?

From one of the papers cited in the parent article:

"A growing body of evidence in the social and medical sciences traces high crime rates to high rates of lead exposure. Scholars have shown that lead exposure and crime are positively correlated using data on individuals, cities, counties, states, and nations. Reyes (2007) exploits state-specific reductions in lead exposure due to the Clean Air Act to estimate the effect of lead emissions from gasoline on violent crime. She reports that reductions in childhood lead exposure in the 1970s and 1980s accounted for more than half of the violent crime decline of the 1990s.2 Stretesky and Lynch (2001) estimate that, from 1989 to 1991, counties with air lead levels equivalent to .17 μg/m3 had homicide rates four times as high as counties with air lead levels equivalent to 0 μg/m3. Mielke and Zahran (2012) show that air lead and aggravated assault rates were strongly associated in a panel of U.S. cities. Longitudinal studies of individuals document a positive relationship between pre- and post-natal lead exposure and delinquency (Dietrich et al., 2001) and arrests for violent offenses (Wright et al., 2008). Cross-sectional research on individuals (Denno, 1990; Needleman et al., 1996, 2002) and counties (Stretesky and Lynch, 2004), studies using cross-national panel data (Nevin, 2007), and analyses of national time-series (Nevin, 2000) have yielded similar results."

Are each of these going to be able to factor for every possible confounder you can think of? No. But it's not like they aren't controlling for anything. That's the whole gist of the parent article - three studies using three new ways of controlling for other variables. Have you seen multitudes of studies that point to any of those other confounder you mention having better evidence, across time and geography, while controlling for lead and all the other possible confounders? Again, nobody is arguing that there are no other factors. And nobody has a precise answer on just how much, because it's impossible. But all studies point to: very large effects, still happening in some cases, and largely preventable: https://www.dropbox.com/s/07z65p9a5wdppfl/LifeAfterLead_AEJ_...

> Your standards are bullshit and you know it.

Aggressiveness like this will get you banned on HN. Please read https://news.ycombinator.com/newsguidelines.html and follow the rules when posting here.

I mean I'd say that in civil conversation, face to face. Is swearing the problem? I was providing substantive contributions to the conversation, whereas conversant was, in my opinion, doing neither of these:

> Please respond to the strongest plausible interpretation of what someone says, not a weaker one that's easier to criticize. Assume good faith.

> Please don't post shallow dismissals

But perhaps doing neither of them more "civilly"? I'm genuinely looking for guidance on handling of this thread, not trying to argue. Thanks.

There's no need to say that I have bullshit standards. I literally agreed with you that lead is a contributing factor.

You seem to agree with me that there are other contributing factors.

Now that we're both happy, I will thank you for the work you put into investigating the sources and bid you a very happy, unleaded farewell.

I had too much lead in my coffee this morning

Nobody is stating there is a single factor. But it appears to be the largest that we can plausibly put a finger on. When you see the same correlations across different countries, states, and local jurisdictions that did not have identical drug policies, immigration/migration patterns, policing strategies etc. etc. etc, it becomes convincing. The three studies in this article provide even more evidence, using clever experimental designs to control for as many of these other variables as they can. Did you read the article? Do you have specific comments on how those studies are or are not contributing to our knowledge?

I did indeed read the article and replied to a different commenter below. I don't dispute that lead is a contributing factor.

I think it's impossible to say how much any factor contributes in a quantitative manner. "36% of crimes were committed in high lead areas, whereas 23% of crimes were committed in high crack consumption neighborhoods"?

I think it's all a mish mash of variables and it's best not to pick apart too many strands of data.

One thing we can say for certain is that murders are down 50% since 1980 despite population in the USA doubling. That is an amazing achievement and I highly, highly doubt it was chiefly due to our generation not having lead in their blood.

That’s an extremely strong statement to make without extensive examination of the actual data and studies by people using the actual data. Really, the first thing you try and find is the magnitude of impact which give exactly what you say is impossible to find.

Someone can go do that literature review then.

On an anecdotal basis only, which sounds more plausible, then:

1) Murder rates are 4x lower chielfy/solely because of unleaded gas

or

2) Murder rates are 4x lower because of quite a few qualitative observations that we think data would confirm each played a role.

It's like saying, "I think murder rates are higher in Norway compared to Sweden not because of some big societal difference between the two countries, but probably because I remember a news story where a depraved lunatic shot 90 kids on an island."

I can’t let this go. It’s nothing like that. Yes, you are just positing opinions as if they’re facts, but that’s not everyone. There has been reviews of the literature. E.g I can’t recall the book off the top of my head but something like out 120 studies on effectiveness of different policing strategies, only ~20% of studies showed statistical significant effects for any given strategy and a fair amount of those failed to replicate. Meanwhile we’ve continually had studies pointing to lead as a significant driver with significant effect sizes. You keep strawmanning this “solely” argument that no one is making while also ignoring the plethora of data pointing to it being a huge driver, and by the far the largest driver with any body of evidence. It’s not like this stuff isn’t being studied, yet I challenge you to find studies that can show anything near the evidence for these other factors.

No one said that lead is the only driver of crime. So lead and economic factors along with other causes can all jointly affect crime ratesand therefore correlate with it.

You see the same things with abortion. The states that phased in legalized abortions had their crime rates dropping before the ones that did. This too is in the freakanomics argument.

Industrialization/deindustrialization or something similar is not a confounding variable?

I like this argument which only looks at very long term statistics. Violent crime has been decreasing since colonial times if not earlier and globally. In this sense, a particular decrease in violent crime doesn't explanation as such (though the long term trend does. Perhaps lead exposure has been decreasing for a long time, perhaps not). Increases in crime, long term, are correlated roughly with veterans returning from war.

See this link for the argument.

http://blogs.berkeley.edu/2010/06/16/a-crime-puzzle-violent-...

Edit: also the original lead argument pretty hinges on a single moment of crime increase. I don't see how more detailed studies of the correlation could be more revealing since reality only provides one instance of a potential to leaded gasoline and reality is going to be noisy.

The author ends his post by stating that the rise of crime from the 60s through the 80s is a major puzzle (to him, in 2010 when he wrote it). I don't see this as an argument against the lead hypothesis in any way.

Re: your second point, one of the three studies in the parent article looks at city level crime in the 1920s/30s using city piping infrastructure (lead vs. iron) for test/control groups. The leaded fuel was a substantial source of increased lead levels but not the only one.

FWIW, it's not just lead in fuel, it's lead in paint (banned in the USA in 1978) that is a source.

My issue was that they stated very adamantly that there was no other explanations, when clearly they are.

Without stating which explanation is correct (which is impossible to prove). We can easily disprove the statement that no other explanations exist though.

Why would you expect a step effect from abortion legalization? It depends upon current criminals dying off or aging out, which happens gradually.

I'm no expert but I would think that if the mechanism is abortion, it's such an important decision that everyone who wants to get one will do so. So then ~20 years later all the aborted criminals vanish from the statistics and stay gone, leaving the other criminals to draw their own stats.

I'm not saying abortion is the ONLY factor, just that if it is, you'd see a step (or sigmoid) in the crime data and then that's it. As in there's a constant density of abortion sensitive criminals until they get aborted.

Similarly if abortion became illegal again, you'd see a step the other way. I'm not familiar with it, but I've heard it's gotten very hard in some places in the US to get an abortion.

Yeah there can be multiple explanations and even "anti-explanations". Things which should lead to increased crime but whose effects were swamped by other things.

Further data supporting your point: http://www.tylervigen.com/spurious-correlations