Lead and Crime: It Happened in Australia Too

Let's turn now to a less contentious topic: gasoline lead and violent crime. As you know by now (yes, there will be a quiz), there are three fundamental types of evidence in favor of the lead-crime hypothesis:

  • Retrospective statistical studies. These are called "ecological studies," and they rely on comparing past rates of lead poisoning in children to rates of violent crime two decades later. The more of these studies the better, but by themselves they're rarely convincing because it's very hard to disentangle all the possible causes of crime in an entire population based only on past statistics.
  • Prospective studies. These are studies that pick a sample of babies and then follow them in real-time for 20 or 30 years. The researchers can measure blood lead levels every few months and then compare that to behavior later in life.
  • Brain scan studies. This is just what it sounds like. You put people who have suffered from lead poisoning into an MRI machine and take a look at how lead affects specific parts of the brain.

All three of these suggest that lead poisoning in kids leads to more violent crime later in life. Today brings yet another bit of evidence. It's an ecological study done at Macquarie University in Australia, which makes it especially interesting. Here's the problem: As long as you do studies only of American crime, you can never know for sure if there's some hidden variable that affected all American kids and just happened to spread at about the same rate as lead poisoning. Maybe eating lots of McDonalds hamburgers causes crime rates to soar for some reason. But you'll never know because kids everywhere in the country ate more and more Mickey D's during the 50s and 60s and there's no data to tell you precisely who ate the most.

But if you study other countries, a lot of these confounding variables go away. Not all of them, perhaps, but most. If you think the problem is poverty or social welfare programs or policing styles or the unique American history of racism, then take a look at Canada. Or Italy. Or Australia. They have very different histories of all these things. If violent crime rates still match up with lead poisoning, that's good evidence that lead poisoning is truly the causal agent.

Rick Nevin has already done some of this work, comparing crime rates to lead poisoning in a number of other countries. But it's always better to get a detailed study from a local research team with access to more data. Interestingly, the Macquarie team didn't look at blood lead levels in children. They had direct access to atmospheric levels of lead, which Australia has tracked for decades, and this allowed them to look specifically at the effect of just gasoline lead. They were also able to make comparisons at the neighborhood, state, and national levels. Here are their conclusions:

Direct effects between air lead and assault rates across all suburbs were examined....For every additional μg/m3 of lead in air, assault rates 21 years later increased by 163 per 100,000 population. Lead in air was the strongest predictor in the model, accounting for 29.8 % of the variance in assault rates 21 years later.

....At the state level, strong positive correlations between petrol lead emissions and death by assault rates were found only for the states with the largest populations, highest population densities and greatest petrol lead emissions, namely, [New South Wales] and Victoria....Lead emissions in NSW accounted for 34.6 % of the variance in death by assault rates 18 years later....In Victoria...lead emissions accounted for 32.6 % of the variance in death by assault rates 18 years later.

....At a national level, the data also demonstrated a positive correlation between lead emissions and death by assault rates, but the association was weak. National lead emissions accounted for only 7 % of the variance in national death by assault rates 18 years later, as the health and behavioral effects of lead emissions are dissipated at larger geographic scales.

The chart on the right is typical of the suburban neighborhood-level data. Boolaroo is about an hour north of Sydney and has a history of high lead levels thanks to a local smelter that operated for decades. As you can see, the correlation between atmospheric lead and violent crime (lagged by 21 years) is very strong. In other suburbs, where lead was produced by gasoline and overall levels were lower, the correlation isn't quite as visually convincing, but it's still quite strong.

Australia banned leaded gasoline in 1985, and crime rates began dropping in the mid-2000s. That's quite different from the US, where leaded gasoline was phased out starting in the mid-70s and crime rates started dropping in the early 90s. This suggests that it's not just something about the specific time period from 1991-2010 that's responsible for our crime decline. Australia also has a very different racial history, a very different policing culture, a very different drug/gang culture, and a very different social welfare state. None of these are likely to be hidden variables that are coincidentally identical in both countries.

So we have a different country; a different time frame; and a different criminal justice culture. Yet crime still follows the trajectory of lead emissions, which appears to explain about a third of the change in assault rates. That's a lot. It's one more bit of data to add to the already persuasive pile of evidence in favor of the lead-crime theory.

I've gotten enough tweets like this that I suppose I should probably respond:

The topic is Gerald Friedman's paper suggesting that Sanders' domestic spending program will supercharge the economy in wildly unlikely ways. And it's true that Friedman isn't officially part of the Sanders campaign team. But they've previously relied on his analysis of their universal health care plan, and the campaign's policy director has repeatedly praised Friedman's paper:

CNN: "Sanders' policy director, Warren Gunnels, also defended the estimates, noting the candidate is thinking big. 'We haven't had such an ambitious agenda to rebuild the middle class since Presidents Roosevelt, Truman and Johnson,' he said."

Pittsburgh Post-Gazette: Warren Gunnels, policy director for the Sanders campaign, hailed the report’s finding that the proposals are feasible...'It’s gotten a little bit of attention, but not nearly as much as we would like,' Mr. Gunnels said....'It’s a very bold plan, and we want to get this out there.'"

NPR: "As for whether he was worried about these sorts of criticisms hurting the campaign in the future, he said no. 'That does not bother us at all,' he said. 'What bothers us is the fact that the U.S. has more kids living in poverty than nearly any major country on Earth.'"

Come on, folks. If your policy director is out in the media promoting Friedman's paper, then it means the campaign is standing behind it. There's no two ways about this.

The second line of defense for Sanders' supporters is that no one has proven that Friedman is wrong. In fact, the critics are "the establishment of the establishment": just a bunch of Wall Street shills on Hillary's payroll who have it in for Bernie. I'm at a loss about how to respond to this. Obviously you can't prove that a forecast of the future is wrong. But you can say that Friedman is forecasting a sustained level of economic growth that's literally never happened before in history. Not here, not in Denmark, not anywhere. Mature economies simply don't grow 5 percent a year for a decade. Labor productivity doesn't double just because you create a bunch of social welfare programs. The number of people in the labor force doesn't skyrocket to new records even in the face of increasing rates of boomer retirement.

The discouraging thing here is that Friedman's critics aren't saying that Sanders' proposals are bad. You can support every single element of his plan with a clear conscience. Their criticism is solely about forecasting how his plan will affect economic growth. And on that score, it's not even remotely realistic. It's about like saying his Medicare-for-all plan will increase life expectancy ten years. It's beyond belief. No matter who you support, you shouldn't do it based on fantasies like this.

A new poll says Americans are evenly divided about whether the vacant Supreme Court seat should be filled this year. Can you guess why they're so evenly divided? Huh? Can you?

The survey found voters were split deeply along party lines, with 71% of the Democrats favoring Senate consideration of an Obama nominee and 73% of Republicans supporting no action until the next president assumes office.

Yeah, that's a shocker, all right. By an amazing coincidence, partisans on both sides have accepted the rigorous and principled arguments set forth by their fellow partisans. However, the fight for the independents continues. They're split 43-42 percent, just like the country as a whole.

The big campaign news of the day is a new NBC/Wall Street Journal poll showing that Ted Cruz leads Donald Trump nationally, 28-26 percent. But this seems unlikely: Four new national polls have been released since yesterday, and three of them continue to show Trump with about 38 percent support compared to 17 percent for Cruz.

So what's going on with the NBC poll? If it's an outlier, it's a hell of an outlier. I couldn't even find a table extensive enough to tell me how unlikely it is to be just a sample error. One in a million, maybe? So maybe it's a problem with NBC's likely-voter filter? Could be. Or maybe there's been an enormous negative response to Trump's debate performance last Saturday? The NBC poll is the only telephone poll done entirely after the debate, so if that were the case it would show up most strongly there.

Very odd. I guess we wait and see.

I made a mistake a couple of days ago, and now all of you have to pay the price by listening to my explanation. Here's what happened.

My intentions were honorable. The Robert Wood Johnson Foundation put up a chart showing the growth of health care spending over the past two years, but it wasn't adjusted for inflation. You all know how I feel about that, don't you? Spending over time should always be adjusted for inflation. So I did the adjustment and posted a revised chart.

But then I got persuaded that I had used the wrong inflation measure. Instead of overall CPI, I should use just the medical component of CPI. So I did that. Unfortunately, I somehow slipped a digit and did it all wrong, which produced a rather odd looking chart. Long story, short, I've spent bits and pieces of the past two days trying to decide how to do this right. The answer is: CPI rather than PCE; medical rather than overall inflation; and year-over-year rather than compounded inflation. The original RWJF chart and the newly-corrected inflation-adjusted chart are on the right.

Ironically, what this ends up showing is...nothing. The numbers in the second chart are all lower, but the general trend is the same in both: spending growth peaks at the beginning of 2015 as Obamacare draws more people into the system, and then steadily declines as the flow of new consumers ebbs. The inflation adjustment didn't change anything.

This is because inflation has been low and pretty flat for the past two years. However, that's not always guaranteed. Sometimes inflation changes substantially over multi-year periods, and that can make trends in nominal dollars badly misleading. So even though it doesn't matter much sometimes, always adjust for inflation! The statistical gods will look favorably upon you.

The political fight over the current vacancy on the Supreme Court is very simple: Conservatives want a conservative justice to replace Antonin Scalia and liberals want a liberal justice. The end.

The PR fight, however, is far more interesting. As Greg Sargent says, it's a fight for the political center of America:

Democrats are betting that the American people will see their reading of what the Constitution obliges the Senate to do as the more reasonable one, i.e., that voters will agree that Senate Republicans should consider Obama’s nominee, and that their refusal to do so reflects the broader GOP strategy of scorched earth anti-Obama obstruction that has produced so much gridlock and chaos.

Republicans came out of the gate saying they'd refuse to even consider an Obama nominee. This was plainly tough talk aimed at the base, but in the underlying PR battle it was a blunder. Republicans have pulled this stunt before—on the DC Circuit Court, on the NLRB, on the CFPB—but people who aren't political junkies weren't paying much attention back then. That changes with a high-profile position like Supreme Court justice.

So now Republicans are backing off a bit. President Obama announced in mild tones that of course he'd nominate someone—that's what the Constitution tells him to do—and Republicans are kinda sorta saying that they'll hold hearings after all. If they do, they've probably dodged a bullet since most of America isn't really paying attention yet. The next stage in this PR battle is up to Obama: will he nominate someone who's scrupulously centrist and well qualified? That would rack up some points for Team Liberal in the battle to seem most reasonable. Will Republicans then run hearings that are at least tolerably efficient and fair-minded? That would rack up some points on their side.

Roughly speaking, every statement or action by anyone in the Supreme Court fight should be interpreted as a shot being fired in the underlying PR war. Most people won't care about this—they're already firmly on a team—but there's a small sliver of voters in the middle who do care, and they could make the difference in November. For that reason, it's worth it for each side to try to rein in its extremists and put up a show of being the most reasonable. Democrats have the early lead right now, but they won't necessarily keep it. After all, they have a base to keep happy too.

I'm not quite sure what I was doing last week when this first appeared, but I missed it. Here's a summary from UMass Amherst professor Gerald Friedman about the impact on the economy if we adopt all of Bernie Sanders' domestic spending proposals:

WTF? Per-capita GDP will grow 4.5 percent? And not just in a single year: Friedman is projecting it will grow by an average of 4.5 percent every year for the next decade. Productivity growth will double compared with Congressional Budget Office projections—and in case you're curious, there has never been a 10-year period since World War II in which productivity grew 3.18 percent. Not one. And miraculously, the employment-population ratio, which has been declining since 2000 and has never reached 65 percent ever in history, will rise to 65 percent in a mere 10 years.

The Sanders campaign hasn't officially endorsed this analysis, but we do have this:

Warren Gunnels, policy director for the Sanders campaign, hailed the report's finding that the proposals are feasible and expressed hope that more people will look into them. "It's gotten a little bit of attention, but not nearly as much as we would like," Mr. Gunnels said. "Senator Sanders has been fighting establishment politics, the establishment economics and the establishment media. And this is the last thing they want to take a look at.

"It shows that over a 10-year period, we would create 26 million new jobs, the poverty rate would plummet, that incomes would go up dramatically, and we would have strong economic growth…It's a very bold plan, and we want to get this out there."

I've generally tried to go easy on Sanders. I like his vision, and I like his general attitude toward Wall Street. But this is insane. If anything, it's worse than the endless magic asterisks that Republicans use to pretend their tax plans will supercharge the economy and pay for themselves. It's not even remotely in the realm of reality. If it were, France and Germany and Denmark would by now be consumer paradises to make Croesus blush.

A group of stuffy establishment economists says "no credible economic research" supports Friedman's analysis, which "undermines our reputation as the party of responsible arithmetic." Or, in Austan Goolsbee's more colorful language, Sanders' plans have "evolved into magic flying puppies with winning Lotto tickets tied to their collars."

Enough is enough. Everyone needs to get back to reality. This ain't it.

Corporations typically use data mining of personal information in order to sell more stuff to their customers. However, corporate wellness programs are mostly used in an effort to sell less stuff to their employees. For example:

Based on data such as an individual’s history, the firms can identify a person who might be considering costly procedures like spinal surgery, and can send that person recommendations for a second opinion or physical therapy.

Spinal surgery, which can cost $20,000 or more, is another area where data experts are digging in. After finding that 30% of employees who got second opinions from top-rated medical centers ended up forgoing spinal surgery, Wal-Mart tapped Castlight to identify and communicate with workers suffering from back pain.

To find them, Castlight scans insurance claims related to back pain, back imaging or physical therapy, plus pharmaceutical claims for pain medications or spinal injections. Once identified, the workers get information about measures that could delay or head off surgery, such as physical therapy or second-opinion providers.

So what do you think? Programs designed to lower health care costs are a good idea. Providing useful health information to employees is a good idea. But how about providing information specifically designed to influence a course of treatment? Is this an attempt to steer employees away from fly-by-night doctors who recommend back surgery for everyone? Or just another green-eyeshade attempt to persuade employees to forego expensive procedures?

Hey, those are good questions! Answers will be forthcoming some day.

A federal judge wants Apple to build a "back door" that allows it to access encrypted data on the iPhone belonging to the San Bernardino attackers. Apple is resisting:

The order, signed Tuesday by a magistrate judge in Riverside, Calif., does not ask Apple to break the phone’s encryption but rather to disable the feature that wipes the data on the phone after 10 incorrect tries at entering a password. That way, the government can try to crack the password using “brute force” — attempting tens of millions of combinations without risking the deletion of the data....Federal prosecutors stated in a memo accompanying the order that the software would affect only the seized phone.

In theory, this should be little more than a macabre joke. If Apple is truly using strong encryption, it wouldn't take ten million tries to crack the password, it would take more tries than there are atoms in the universe.

Unless, of course, the attackers are really stupid and used "123456" or "Jihad Forever" as their password. Which they very well might have. Folks like this aren't always especially bright.

In any case, I find it hard to side with Apple here. It's one thing for Apple to implement strong encryption that even Apple itself can't break. It's another to deny law enforcement the ability to even try to break the encryption. My initial reaction—which I admit might change if I think about this further—is that liberals have never opposed the right of the government to execute a search. We just want them to get a warrant first, and we want it particularized to a specific case. So we object to warrantless searches and we object to mass collection of surveillance data. A court order that applies to a specific case shouldn't be a problem.

Apple, of course, is arguing that if they create a special FBI version of iOS, it can be used anytime and anywhere, with or without a warrant. So that's the question for the court. If they compel Apple to create a version of iOS that can be hacked, are there legally enforceable restrictions on its use? Or does it become a permanent plaything for anyone who can issue a national security letter—which appears to include practically the entire FBI? This will be an interesting case going forward.

MSNBC Embarrasses Itself Yet Again

As John McEnroe might say, you have got to be kidding me:

This is just what the world needs: assisting Donald Trump yet again in his eternal quest for more television air time. This particular lovefest-cum-fuck-you is being hosted by one of his biggest fans, and is scheduled directly opposite the CNN town hall with Carson, Cruz and Rubio.

Look, I get it: ratings are everything, and Trump pulls ratings. But surely TV executives still have a sliver of self-respect left? Were they afraid they hadn't mentioned Trump enough in February, and he might get mad at them?