If Books Could Kill - Freakonomics

00:00:00.000 Peter. Michael. What do you remember about a book called Afreakonomics?

00:00:04.260 If I recall correctly, the thesis of the book is, had my mother aborted me,

00:00:09.180 I would not have committed so many crimes.

00:00:26.200 All right, welcome to If Books Could Kill.

00:00:29.660 premiere episode are we taglining this mike i don't think so i think that was the tagline i 0.98

00:00:33.760 don't know yeah i've done that for two shows already need a new thing yeah the the dumb books

00:00:39.160 that captured our collective imagination yes the books that did to our brains what jigsaw did to 0.74

00:00:46.280 robin hood i'm michael hobbs i'm a journalist and the co-host of maintenance phase i am peter

00:00:52.140 shamsheree i am a lawyer and the co-host of the five to four podcast both of us are fascinated

00:00:59.500 by dumb ideas and how they spread through the population. And so a couple months ago,

00:01:04.480 we started talking about how to do a podcast about the dumbest ideas of the last 50 years.

00:01:10.120 And the more we thought about it, the more we realized that like a good way to do it

00:01:13.740 would be by going through airport books, which are kind of like the super spreader events 0.99

00:01:19.980 at this point of American stupidity. Yeah, they're the natural vessel for 0.99

00:01:24.920 pseudoscience and fake history and just sort of quintessentially American, you know, all of this

00:01:32.460 complex knowledge and information boiled down into a mush and packaged and sold for $24.95 to

00:01:41.100 people who forgot to charge their Kindle for the flight. This is the seventh episode that we've

00:01:46.100 recorded, but the first episode that we're releasing because I started reading Freakonomics

00:01:51.960 And then I realized that this book is the perfect overture, the quintessential airport book, the quintessential, like wrong and bad airport book.

00:02:00.700 It's like really shocking how bad this book is.

00:02:04.380 And it's also like the badness of the book is also matched by like how influential it was.

00:02:10.280 Right. What do you actually remember about the book itself and like the era?

00:02:13.960 It comes out in 2005. Is that right?

00:02:15.640 2005. Yeah.

00:02:16.140 OK. So in the 80s, you have sort of like the Chicago school economists who start to like posit all of these theories that are that basically boil down to like we can solve most social problems with like economics principles.

00:02:31.080 And then you get something like Freakonomics, which feels like the mainstreamification of that concept, right?

00:02:38.140 Just taking that sort of nihilistic, neoliberal sort of viewpoint and bringing it to the masses.

00:02:46.740 Exactly. I mean, you cannot underestimate how popular this book was.

00:02:50.720 So it sold 4 million copies. It was on the bestseller list for 39 weeks.

00:02:56.000 I think an underrated aspect of this book's influence is the fact that they had a New York

00:03:01.620 Times blog and a podcast like five years before Serial. I mean, I listened to that podcast for

00:03:07.640 years. Like there wasn't that much else to listen to. It was like this or fucking Ricky Gervais.

00:03:14.040 And it's like, that's what you listened to when you were washing dishes. I also think the subtitle 0.99

00:03:18.500 of the book is important because it's a rogue economist explores the hidden side of everything.

00:03:24.380 like this guy's outside the mainstream and he's saying things they don't want you to hear

00:03:29.180 is also like one of the dominant paradigms of the ways that americans are liable to believe 0.85

00:03:34.880 bullshit we're coming off a decade of disaster movies each of which has one scientist that 1.00

00:03:40.560 nobody believes that's trying to get the truth to the president yeah people love this shit they're 1.00

00:03:45.560 primed for it everybody reading this thinks that they're pierce brosnan and dante's peak but 0.98

00:03:49.380 they're actually randy quaid and independence day so i'm gonna send you some paragraphs

00:03:54.080 The book began with a New York Times magazine article.

00:03:59.940 Stephen Dubner, who's one of the co-authors of Freakonomics, was at the time a story editor at the New York Times.

00:04:05.620 And he was working on a story about the psychology of money.

00:04:09.560 And that's how he met Stephen Levitt, who's his co-author and this University of Chicago economist.

00:04:14.640 So the book is co-written by both of them, but it's not kind of clear who wrote what.

00:04:18.660 And there's lots of stuff in the book that is based on Levitt studies, but there's also lots in the book that isn't based on Levitt studies.

00:04:23.780 It's just kind of random.

00:04:24.740 Got it.

00:04:25.340 Got it.

00:04:25.700 So I'm sending you the first five paragraphs of the New York Times story.

00:04:30.940 Oh, boy.

00:04:31.260 This is America's first introduction to the Freakonomics guys and like the Freakonomics

00:04:36.700 way of thinking.

00:04:37.640 The most brilliant young economist in America, the one so deemed at least by a jury of his

00:04:42.720 elders, breaks to a stop at a traffic light on Chicago's South Side.

00:04:47.540 It is a sunny day in mid-June.

00:04:49.460 An elderly homeless man approaches.

00:04:50.920 He wears a torn jacket, too heavy for the warm day, and a grimy red baseball cap.

00:04:56.620 The economist doesn't lock his doors or inch the car forward, nor does he go scrounging

00:05:01.200 for spare change.

00:05:02.340 He just watches, as if through one-way glass.

00:05:05.460 After a while, the homeless man moves along.

00:05:08.780 He had nice headphones, says the economist, still watching in the rearview mirror.

00:05:14.540 Well, nicer than the ones I have.

00:05:16.880 Otherwise, it doesn't look like he has many assets.

00:05:20.240 Stephen Levitt tends to see things differently than the average person.

00:05:24.460 Differently, too, than the average economist.

00:05:27.260 What do you think?

00:05:28.020 Is that seeing things differently?

00:05:29.980 I'm pretty sure that staring down a homeless person asking for money

00:05:33.940 and then making a snarky comment about the quality of their accoutrement

00:05:39.200 is a classic American tradition.

00:05:42.520 Is it not?

00:05:43.380 If he's so poor, why does he have stuff?

00:05:45.780 If he had taken that headphone money, Michael, and invested it in an ETF starting in 2002, he could have eight pairs of headphones by now.

00:05:56.820 I also love it because we're seeing this like he's different from the other economists in a story that's not remotely different from other like pretty well off people seeing a homeless man in public.

00:06:06.580 And also, it's basically trying to establish him as like a rogue economist and outside of the strictures of the field while acknowledging the fact that he's a tenured professor at the University of Chicago.

00:06:20.400 He has degrees from Harvard and MIT, and he won the John Bates Clark Medal.

00:06:26.460 So the article wants to use that as like, this isn't just some crank saying stuff, right?

00:06:30.140 Like, look how awarded he is within economics.

00:06:32.880 But then also we'll switch and be like, oh, she's different.

00:06:36.100 I almost feel like like remember when like the Trump campaign had like establishment Washington folks talking about like the swamp and you're and you're just like you feel like you're through the looking glass a little bit.

00:06:46.960 You're like, what are you? That's you. Exactly. Not just an establishment economist, but like the most establishment economist being like, I'm a sort of a bad boy in the economics industry who everybody really likes.

00:06:59.120 Yes. He also, there's another piece of foreshadowing that says one paper he wrote as a graduate student is still regularly cited. His question was disarmingly simple. Do more police translate into less crime? The answer would seem obvious. Yes, but had never been proved. Since the number of police officers tends to rise along with the number of crimes, the effectiveness of the police was tricky to measure.

00:07:20.800 Leavitt needed a mechanism that would unlink the crime rate from police hiring.

00:07:24.480 He found it within politics.

00:07:26.280 He noticed that mayors and governors running for re-election often hire more police officers.

00:07:30.480 By measuring those police increases against crime rates,

00:07:33.220 he was able to determine that additional officers do indeed bring down violent crime.

00:07:38.380 That paper was later disputed.

00:07:40.100 Another graduate student found a serious mathematical mistake in it.

00:07:42.860 But Leavitt's ingenuity was obvious.

00:07:45.100 so uh it's not real it's he found something that was fake and but let's all just be in awe of his 0.98

00:07:56.320 creativity this runner tripped and absolutely ate shit on his first lap but his speed was obvious

00:08:02.680 it's like what are we doing here yeah yeah and so the article ends with tax evasion money laundering 0.96

00:08:10.780 I'd like to put together a set of tools that lets us catch terrorists.

00:08:14.540 I don't necessarily know yet how I'd go about it, but given the right data,

00:08:18.040 I have little doubt that I could figure out the answer.

00:08:22.540 Small problems, just ending terrorism.

00:08:24.880 Yeah, what am I working on now?

00:08:26.200 Look, I've been thinking about stopping terrorism, you know, if someone can get me some numbers.

00:08:30.440 Stephen Levitt may not fully believe in himself, but he does believe in this.

00:08:34.480 Teachers and criminals and real estate agents may lie and politicians and even CIA analysts, but numbers don't.

00:08:43.440 I don't think that Stephen Levitt seems like he lacks confidence in himself, but perhaps I'm misreading. 0.97

00:08:50.380 But then one thing I genuinely really appreciate about Freakonomics is that everything that people will later accuse Freakonomics of, they just fucking say.

00:08:59.820 So, like, in this passage, he's saying, like, everybody else lies, but the numbers don't lie.

00:09:05.560 And it's like, right, the whole problem with this is this overconfidence in quantitative data that is completely stripped of all of its societal context.

00:09:14.220 Like, that's why we object to this.

00:09:16.000 This is a classic economics guy thing where they act like the narratives that they map onto the data are themselves just as, like, infallible as the data.

00:09:27.080 Exactly. That's a good way to put it.

00:09:28.300 The idea that there's like always something hidden, right, seems to be lurking here where there must be an explanation that is counterintuitive and fascinating and people are trying to hide it from you, right? 0.96

00:09:42.200 God, this is so fucking annoying.

00:09:45.700 You're having the same experience that I had in slightly distilled because like you can't believe they're just saying it. 0.97

00:09:51.000 That's what I cannot get over.

00:09:52.140 God.

00:09:52.540 So after the article got published, it was a huge sensation because of all these like bold ideas.

00:09:57.380 then like book publishers got in touch and then they're like quite open about the fact that this

00:10:02.580 was a rush job oh my god and they just like grabbed a bunch of random anecdotes i have a

00:10:08.120 lot of respect for being like look we're cashing in yeah exactly and they're like other people have

00:10:13.200 said there's no overarching theme to this book that's correct what we're doing is we're talking

00:10:17.760 about numbers and it's like okay you just have a bunch of cute anecdotes and you're gonna string 0.54

00:10:21.400 together the cute anecdotes with like the most fucking try hard transitions i've ever seen

00:10:26.080 So I've kind of broken apart this book and put it back together because even though the book is only 207 pages, they only spend a paragraph or two on each one of these anecdotes.

00:10:39.020 It's like a collection of basically like a hundred cute little stories.

00:10:43.060 Oh, hell yeah.

00:10:43.520 I'm trying to take a representative sample of the way that they present information.

00:10:47.900 But it's like we're only going to touch on like 10% of the book.

00:10:50.860 Right.

00:10:51.180 Because to debunk these like ridiculous paragraphs takes you four times longer than it took to 0.96

00:10:56.360 write them. 0.97

00:10:56.840 Right.

00:10:57.100 So we're not going to go through the book chapter by chapter basically because like it's too

00:11:00.360 much of a mess to do that.

00:11:01.620 But I do want to talk about the core ideology of this book and like why it exerted such a negative influence on the culture.

00:11:11.000 The reason I wanted to read that concluding paragraph from the original New York Times story is that the entire Freakonomics approach is setting up this binary between intuitive thinking and data-driven thinking, right?

00:11:26.900 So economics is uniquely positioned to allow us to see the world without all the human nonsense that comes along with asking people about it.

00:11:36.960 Right. You're cutting through the bullshit. 1.00

00:11:38.480 Right. And they're extremely explicit about this in the book. 0.99

00:11:42.100 So in the introduction, they say morality, it could be argued, represents the way that people would like the world to work, whereas economics represents how it does work.

00:11:52.120 No, no, no, no.

00:11:54.260 so like this is just a totally false binary and the best evidence that this is a false binary

00:12:01.920 is their own book yeah so in this episode we are going to talk about all of the ways that they

00:12:07.420 misuse data the first thing we're going to talk about is the way they use true data to reach false

00:12:14.920 conclusions so let me send you one more paragraph all right what about the election truism that the

00:12:22.480 amount of money spent on campaign finance is obscenely huge. In a typical election year,

00:12:28.260 campaigns for the presidency, the Senate, and the House of Representatives spend about one billion

00:12:32.780 dollars. That sounds like a lot of money, unless you care to measure it against something seemingly

00:12:37.800 less important than democratic elections. It is the same amount, for instance, that Americans

00:12:42.620 spend every year on chewing gum. This is like classic Freakonomics, where it's like phrased

00:12:47.720 as some sort of debunking you thought political spending was bad but wait till you hear about

00:12:54.160 chewing gum but like those things have nothing to do with each other you're just juxtaposing

00:12:58.700 an important thing with a frivolous thing to make them both seem frivolous right and also like i

00:13:04.180 mean the the objection to money in politics is a moral one right it's not like we are spending

00:13:11.260 too much money in a vacuum and that's it it's it's because it's like the literal manipulation of

00:13:17.500 society. That's the point of campaign finance. So yes, people have objections to that in and of

00:13:23.980 itself. Whatever. And they also, another example is they have this whole thing. They have a section

00:13:28.780 about cheating and how there's this infamous thing in 1987 where the IRS started asking people to

00:13:35.420 list the social security numbers of their children if they wanted the child tax deduction. So you get

00:13:41.060 like, it was like 2000 bucks at the time and you could just say, okay, I have little Timmy. And

00:13:45.580 then all of a sudden your tax bill would go down by $2,000. And then all of a sudden you had to

00:13:49.780 start providing the social security number for Timmy and 7 million children disappeared from

00:13:55.940 the tax rolls. They explicitly link this to cheating, that all of these people were lying

00:14:00.880 about their kids and then they had to prove that they had kids. And all of a sudden all these kids

00:14:04.740 disappeared, right? What they leave out is the fact that before 1989, children were not assigned

00:14:10.740 social security numbers automatically. So when the IRS announced this thing, you're going to

00:14:15.360 have to start putting social security numbers. Every single parent in America had to fill out

00:14:19.280 a form, send it to the IRS, wait two weeks, and get their kid's social security number back.

00:14:24.660 So of those 7 million people that didn't include their kids on their taxes that year,

00:14:29.080 a huge percentage of them were people who were like, oh shit, I forgot to do this. 0.89

00:14:33.520 I just can't include my kids this year. Most of the kids that disappeared were divorced parents 0.91

00:14:37.620 and both parents were putting the kid for the deduction on their taxes. And so some of that's

00:14:43.540 cheating but also it could also just be something of like they had never really talked about it

00:14:47.440 before or thought about it and didn't know that they couldn't both claim the kid right again the

00:14:51.280 number seven million appears to be true although i've seen somebody say that it was actually more

00:14:54.500 like two million but the interpretation of it like what they are using that number to say

00:15:00.060 is mostly wrong like we don't we don't know how much of that was cheating yeah as look as someone 0.58

00:15:06.460 who used to do his own taxes and then gave it to an accountant this year i understand fucking it up 0.76

00:15:10.960 completely my accountant was like what are you what have you been doing i'm like i don't know 0.80

00:15:14.560 man i just i kind of wing it and then i submit it and i haven't been arrested that's i thought i was

00:15:20.000 doing it right but then god this is not the worst one in the whole book but this is like

00:15:24.360 peak smug this this will give you flashbacks to the kind of dude who read this book

00:15:29.100 so this is in a section about parenting it's talking about risks and how people can be

00:15:33.520 irrational when they consider risks it says consider the parents of an eight-year-old girl

00:15:38.360 named Molly. Her two best friends, Amy and Imani, each live nearby. Molly's parents know that Amy's

00:15:44.060 parents keep a gun in their house, so they've forbidden Molly to play there. Instead, Molly

00:15:48.420 spends a lot of time at Imani's house, which has a swimming pool in the backyard. Molly's parents

00:15:53.680 feel good about having made such a smart choice to protect their daughter. But according to the

00:15:58.360 data, their choice isn't smart at all. In a given year, there's one drowning of a child for every

00:16:03.880 11,000 residential pools in the United States. In a country with 6 million pools, that means that

00:16:09.560 roughly 550 children under the age of 10 drown each year. Meanwhile, there's one child killed

00:16:15.300 by a gun for every 1 million plus guns. That means that roughly 175 children under 10 die each year

00:16:22.540 from guns. Molly is roughly 100 times more likely to die in a swimming accident at Imani's house 0.52

00:16:28.740 than in gunplay at amy's okay they're making a mistake here that i can't quite articulate but

00:16:35.900 it might have to do with um the amount of guns per household yeah um but i i want to take a step

00:16:42.020 back and say one of the least interesting things on earth is when people do this sort of like people

00:16:48.640 are assessing risks incorrectly oh my fucking god i know kind of analysis and it's like are people

00:16:54.060 supposed to know statistics like this before they like make everyday decisions? No. This is, 0.98

00:16:59.680 again, weirdly conservative where they're sort of being like guns aren't as dangerous as people

00:17:03.360 think. Exactly. I would love for them to be like actually undocumented immigrants aren't as

00:17:09.520 dangerous, you know, things like crime, for example, street crime are areas where people

00:17:16.720 are way out of whack. And yet these books don't seem to focus on it. So the obvious statistical

00:17:22.800 thing to say here is that it's absurd to say deaths per gun versus deaths per swimming pool.

00:17:29.020 Most people who own guns own more than one gun. And most people who have swimming pools have

00:17:34.040 exactly one swimming pool. So what you'd want to do is households with guns versus households with 0.66

00:17:39.820 pools. Right. But that's not that's like the sort of 101 bullshit thing. The much bigger thing is 0.98

00:17:46.100 you should not be using average mortality statistics to lecture other people on how to

00:17:51.560 parent their kids right most of the kids under 10 who drown this is like really awful is like

00:17:57.080 it's mostly very young kids in bathtubs uh-huh another very large portion is in like lakes and

00:18:02.620 rivers it's mostly poor kids a lot of it is like kids with disabilities like physical disabilities

00:18:08.720 who can't swim backyard pools actually are really dangerous compared to like municipal pools

00:18:13.360 yeah but the reason is not that kids drown when like they're playing at a friend's house and like

00:18:18.680 you know, usually mom is watching when kids are playing in the pool because they know it's

00:18:22.780 dangerous, right? Usually how kids die in backyard pools is the back door is unlocked

00:18:27.600 and they wander outside and they fall into the pool like at night when nobody's around and they

00:18:32.200 can't get out of the pool. If your kid can swim, they're probably fine. Like the dynamics of

00:18:36.820 drownings, you shouldn't just be looking at the average number of drownings across the entire

00:18:42.600 country. Like there are specific dynamics to this. And of course there's specific dynamics to

00:18:47.100 firearm deaths too but this whole thing is just like you might think that you have a good intuition

00:18:53.880 about this but what if i presented you with the worst oversimplification of the data that you've 0.90

00:18:58.860 ever heard in your fucking life what do you think now molly's parents it's also seemed totally 0.91

00:19:04.860 rational for me to be just kind of in general worried and uncomfortable around an object that 0.99

00:19:11.140 essentially only exists to cause harm like there's no reason for my child to be anywhere near a gun

00:19:17.040 whereas swimming pools like swimming is good for kids it's social it's exercise

00:19:23.420 you as a parent might say like oh that's actually really worth the risk for me what they seem to be

00:19:28.540 driving at in part is that maybe we're a little too uptight about gun restrictions and gun safety

00:19:34.880 right what they're missing is that maybe part of the reason that children are getting killed by

00:19:39.820 guns at relatively low rates is because people are cautious around them right and trying to like

00:19:46.760 drop that social stigma is just going to drive those numbers up right oh no i'm i'm about to

00:19:53.340 quote ruth bader ginsburg and i don't want to do that do it notorious rbg i know you love it

00:19:57.700 you have the mug next to you it's throwing away the umbrella in a rainstorm because you're not

00:20:02.920 getting wet right that's what they seem to be advocating for here what what drives me nuts

00:20:06.660 about this section and like this kind of way of doing statistics is like it doesn't give you any

00:20:12.240 understanding of drownings, of firearm deaths. All you have is like a little factoid that you

00:20:19.160 can drop at like a barbecue with the other dads and be obnoxious.

00:20:22.900 Molly's parents don't let her go over to Emily's house because Emily's parents

00:20:26.560 own a bear and they let it roam free. But did you know that bear attacks kill under 10 children per

00:20:35.220 year so that's the misuse of data part one part two is most of the content of the book it's the

00:20:44.580 over generalization from extremely specific data so there's some dude who's like an office drone

00:20:53.700 in dc and he starts selling bagels at work he brings in bagels and he puts like a you know 20

00:20:59.860 bagels in the kitchen and a bowl and it's like a trust system people are supposed to take a bagel

00:21:04.580 leave a book. Okay. And so he starts making so much money from the bagels that he decides to

00:21:09.240 do this full time. So now he delivers like, I don't know, 10,000 bagels a day to various offices

00:21:13.520 around DC. And he does the same thing. He leaves a big bowl of bagels and he leaves a bowl for

00:21:17.480 money. And allegedly this guy has kept meticulous records for years. And so he has basically the,

00:21:23.840 the honesty of various customers, right? Cause they don't have to put in a dollar. They can

00:21:27.900 just take a bagel. So according to this guy's data, it's like 90% of people pay for the bagel.

00:21:32.980 And there's some, like, borderline interesting stuff around, like, around the holidays, people are less likely to pay for the bagels.

00:21:40.140 Certain kinds of companies, like people in big companies, are more likely to pay for the bagels than at small companies.

00:21:46.360 And people in the executive suite, like on those floors, are less likely to pay for the bagels.

00:21:51.000 So, like, rich people are stingy or whatever.

00:21:53.080 I believe that.

00:21:54.040 It's kind of interesting.

00:21:54.920 Like, it's a cute story, this guy.

00:21:57.260 But, like, is this generalizable?

00:21:58.900 I don't really know.

00:22:00.800 Freakonomics.

00:22:02.980 thinking like a freak will this guy give me a dollar for the bagel uh on the honor system

00:22:09.780 maybe not freakonomics it's very funny to me that like the only good parts of this book are

00:22:16.100 just the descriptive parts yeah where it's like they just like these are the phenomenon you're

00:22:19.580 like oh interesting but then as soon as they try to turn them into like pat little lessons

00:22:23.560 they're like and that's why you're like i don't know that we can really learn anything from this

00:22:30.120 They have a whole thing with, like, incentives and, like, it turns out people cheat way less than you think they do.

00:22:35.740 Right.

00:22:35.900 Well, maybe, but it's like these are bagels at work.

00:22:38.600 And they cost a dollar?

00:22:39.540 Yeah, they're cheap.

00:22:41.600 Yeah.

00:22:41.860 It just seems like a very unique situation that I'm not sure you can really say anything about, like, humans' propensity to cheat based on this.

00:22:49.200 They then have a whole section about sumo wrestlers.

00:22:52.600 This is another study of Levitz.

00:22:54.680 There's a weird thing in sumo wrestling tournaments where you do the best of 15.

00:23:00.560 So you have to win eight matches out of 15, right?

00:23:03.280 Okay.

00:23:03.620 But the problem with sumo tournaments is that oftentimes people reach their eight wins and

00:23:09.200 then they still have like three more matches to go.

00:23:11.680 So basically, you have all these matches between like people who it just doesn't matter if

00:23:16.360 they win or not because they've already gotten their eight matches.

00:23:18.920 And sometimes they are matched with people who are like seven and seven and like really

00:23:22.800 need to win.

00:23:23.460 really needed so sure levitt runs the numbers and he finds it like you would expect a sort of 50 50

00:23:28.560 split you know winning and losing percentage on these matches but it turns out it's 80 20 for

00:23:33.720 people who need to win end up winning these things and the the only explanation for this is like

00:23:40.020 widespread criminal conspiracy it must be fixed but then what's wild so i i had the same reaction

00:23:46.340 as you i was like this seems like a really big leap but then in 2011 there was an actual like

00:23:51.720 huge scandal in sumo wrestling that confirmed that like was a massive criminal conspiracy

00:23:56.540 among these dudes oh hell yeah and like wow it's interesting in that there wasn't actually that

00:24:01.300 much selling of matches but like these guys would just kind of meet in the dressing rooms and be

00:24:06.580 like dude you don't need to win this i need to win this do you mind just letting me win and they'd

00:24:09.700 be like yeah yeah you're fine uh-huh this is one of the only examples in freakonomics where like 0.51

00:24:13.960 he was fucking right like good for him kudos i will give you this one steve economics could 0.69

00:24:18.800 have never predicted this only for economics could have predicted this well this is this is 0.97

00:24:23.520 the other thing that i learned when i was researching the match fixing scandal is that

00:24:26.620 like sumo fans have been complaining about this for literally decades like they they changed the

00:24:31.680 rules in the 1970s to try to prevent this obviously not effectively enough but like everyone knows

00:24:38.180 that like these matches are kind of fake right i do think that like providing numbers to these

00:24:42.720 things and giving evidence to something that feels true is like a very important role for academia

00:24:47.440 But I don't know that there's like human human behavior there other than the super banal finding that like, yeah, when it matters to one person and not the other, they're probably going to trade.

00:24:58.120 Right. You know, this reminds me of like, you know, in like econ 101, when you learn about moral hazard.

00:25:05.260 Yeah. And you feel like really smart for a day.

00:25:08.420 Yeah. This this is like some of the most basic human behavior stuff that you could ever conceive of.

00:25:14.460 But they present it like they're blowing your mind.

00:25:16.640 Well, this is something I learned from reading a bunch of extremely scathing reviews of this

00:25:20.740 book by economists. One thing this book does that I think became very prominent in the early 2000s

00:25:26.240 was this idea that like incentives explain everything, right? And if you want to understand

00:25:30.060 a situation, you sort of look at the incentives of all the actors involved. And of course,

00:25:33.980 this book does that, right? It like presents the bagel anecdote and like 50 other anecdotes. And

00:25:38.140 it's like, oh, the incentives. Economists can understand things better than other types of

00:25:42.780 scientists, because they look at the incentives. But then when you look at the bagel example,

00:25:47.860 it's like, well, everybody has the incentive to steal a bagel. Yeah. But only 10 percent of people

00:25:53.920 do. And they spend almost an entire chapter on this example of Chicago school teachers and how

00:25:59.740 Stephen Levitt designed an algorithm to detect teachers who were like erasing bubbles on

00:26:05.500 standardized tests and filling in their own answers to make sure that they didn't get fired.

00:26:10.140 and it's like oh the incentives of the teachers yeah but then they mentioned sort of offhand that

00:26:15.580 it's only five percent of the teachers who cheat so it's like 100 of the teachers have the incentive

00:26:22.360 like very strong incentives to cheat but very few of them do so like what does it actually mean to

00:26:28.920 say incentives matter right right like you could just as easily say that people pay for a bagel

00:26:34.500 because of their upbringing you could say it's because their psychology you could say it's

00:26:38.400 because they want to be moral people and they don't want to be the kind of person who steals

00:26:42.240 a bagel. Like those are incomplete explanations too, but it's not clear to me that those are less

00:26:45.920 scientific than just saying like incentives over and over again. I just, this book has a very

00:26:52.300 complicated relationship with like morality and it sort of sees morality as like this weird

00:26:57.680 irrational thing that people do. That's a classic conservative economist tick, right? Where the goal

00:27:04.940 of a lot of the work is to critique liberal sentimentality right right in their view so

00:27:11.240 the third way that this book misuses data is leaping to conclusions on some things while

00:27:18.800 refusing to reach conclusions on others this is where we get into the black names stuff

00:27:28.000 throughout the book there's like various examples of kind of weird race stuff stephen levitt did

00:27:34.340 a study on do you remember the weakest link the game show yeah it's like a cross between

00:27:39.480 who wants to be a millionaire and survivor you like vote people off for getting questions wrong

00:27:43.720 yeah it's with the mean british lady yes the mean british lady was the host yeah so he did a study 1.00

00:27:50.020 of everyone who's ever been kicked off of that show and it's like you'd expect the black contestants

00:27:55.300 to be kicked off right but actually it wasn't and the female contestants weren't kicked off either

00:28:00.460 it turns out the hispanic and the elderly contestants were the ones who faced discrimination

00:28:05.260 okay other people have questioned this because there were only 22 hispanic contestants on the

00:28:10.480 show out of a thousand contestants so like you can't you can't really like make claims about

00:28:14.840 that but anyway it's like okay whatever then we get to the final two chapters of the book

00:28:20.120 which are all about like cultural explanations for poverty yep here we go so here's a couple

00:28:26.460 of paragraphs that I don't want to read, but I'm going to make you read. They're talking about a

00:28:32.040 researcher named Roland Fryer. In addition to economic and social disparity between blacks

00:28:36.760 and whites, Fryer had become intrigued by the virtual segregation of culture. Blacks and whites

00:28:41.820 watch different television shows. Monday Night Football is the only show that typically appears

00:28:46.180 on each group's top 10 list. Seinfeld, one of the most popular sitcoms in history, never ranked in

00:28:52.340 the top 50 among blacks they smoke different cigarettes and black parents give their fucking 1.00

00:28:58.580 shit it's happening and black parents give their children names that are starkly different from 1.00

00:29:03.640 white children's friar came to wonder is distinctive black culture a cause of the 0.99

00:29:09.980 economic disparity between blacks and whites or merely a reflection of it time to ask are poor

00:29:15.740 people poor because it's their fault yeah i mean are you poor because of socioeconomic structures

00:29:24.100 or are you not watching seinfeld you know is that the problem the the black names thing like this is 0.82

00:29:30.820 tale as old as time right people being like well if you have a black name you're less likely to get 0.99

00:29:36.740 uh job offers and then like their conclusion is it's stupid to give your kid a black name 0.96

00:29:42.780 instead of like wow must be some serious racism at play which is the obvious conclusion it's it's 0.98

00:29:49.080 worse peter it's okay okay this is in a chapter called would a roshanda by any other name smell 0.94

00:29:55.480 as sweet oh shit long silence long silence freakonomics we meet this roland fryer guy who 0.56

00:30:06.120 has a database of every single person born in california since 1961 and he starts combing 0.97

00:30:11.240 through like the demographic data and like cross-checking it with like the names so it says

00:30:16.220 the data show that the black white gap is a recent phenomenon until the early 1970s there was a great

00:30:22.060 overlap between black and white names the typical baby girl born in a black neighborhood in 1970

00:30:27.180 was given a name that was twice as common among blacks than whites by 1980 she received a name

00:30:32.580 that was 20 times more common among blacks boys names moved in the same direction but less

00:30:37.100 aggressively, probably because parents of all races are less adventurous with boys' names than

00:30:41.840 girls. A great many black names today are unique to blacks. More than 40% of the black girls born 0.56

00:30:47.680 in California in a given year receive a name that not one of the roughly 100,000 baby white girls

00:30:53.240 received that year. The California study also shows that white parents send a strong signal

00:30:57.740 in the opposite direction. More than 40% of white babies are given names that are at least four

00:31:02.620 times more common among whites consider connor and cody emily and abigail this is interesting

00:31:08.800 yeah that is interesting descriptive statistics it's like wow social trends yeah although this

00:31:13.820 is also like a few years before the great i don't know how to what to call it uh dipshitification

00:31:19.760 of white names the gwyneth effect yeah yeah we're at a time where um the desire among

00:31:28.120 white parents to throw a y where an i used to be is that just it's at peak we then get a long

00:31:34.940 section that's basically just like riffing on black names so they talk to a judge in family

00:31:43.180 court in new york who is like presumably a friend of one of theirs who just like tells them the

00:31:48.820 funniest black names that he's ever seen they start out with the story of a girl named temptress

00:31:55.560 who's arrested for prostitution at age 15 you're literally make making cracks about a 15 year old 0.99

00:32:02.740 who's probably being sexually trafficked yeah just that's the joke good stuff and then we get

00:32:07.700 the story of someone named amcher who had been named for the first thing his parents saw upon

00:32:14.980 reaching the hospital the sign for albany medical center hospital emergency room okay and then we

00:32:21.680 have this paragraph. Roland Fryer, while discussing his name's research on a radio show, took a call

00:32:28.020 from a black woman who was upset with the name just given to her baby niece. It was pronounced

00:32:33.280 shiteed, but was in fact spelled shithead. Or consider the twin boys Orange Jell-O and Lemon 0.97

00:32:40.440 Jell-O, also black, whose parents further dignified their choice by instituting the pronunciations

00:32:46.260 orangelo and limongelo now that first one was like 80 a prank call but yeah go on this thing 0.81

00:32:54.300 of like black people giving their kids weird names is like a very well-known urban legend

00:32:59.080 there are numerous snopes articles about this this is something that was like huge in email

00:33:05.040 forwards in the 1990s there are stories of this going back to 1917 these were like vaudeville

00:33:11.960 jokes this thing of naming your kid after like the emergency room where you were born this is a 1.00

00:33:17.420 really old joke it's like black people are so stupid that they name their kids like no smoking 1.00

00:33:21.920 right because that's like the sign above the you know place where they're filling out the birth 1.00

00:33:26.360 certificate right there's another one where they they name their kid female because that's like

00:33:30.660 the word in the box and like they don't understand how to fill out the form but it's pronounced

00:33:35.220 famale do you have you heard this there's like this urban legend that there's someone named 0.75

00:33:40.980 ladasha yeah how's it spelled l-a-dash-a right like just obvious bullshit i was amazed that

00:33:48.520 this wasn't in the freakonomics book because like every other urban legend about this is in 0.77

00:33:53.340 the freakonomics book and then the shateed shateed thing i've seen this in a kevin hart routine i

00:33:59.260 think that he did years ago was the first place i came across it i like the idea that kevin hart

00:34:03.840 is pulling jokes from freakonomics he's like good one guys but so this like in a book that is meant

00:34:12.680 to be like data driven you know and like exploring the world through quantitative data to fall for

00:34:18.280 this just rank bullshit that i don't know what the google situation was in 2005 but like two 0.98

00:34:23.840 minutes on google yeah it's like do black people name their kids shithead ah yeah this has been 0.99

00:34:27.800 bouncing around for decades right we then get into you know these studies about they send in 0.99

00:34:31.900 resumes. And if you have a black name, you're less likely to get a call back than if you have

00:34:35.560 a white name, right? This says, according to one such study, if Deshaun Williams and Jake Williams

00:34:41.020 sent identical resumes to the same employer, Jake Williams would be more likely to get a call back.

00:34:46.100 The implication is that black sounding names carry an economic penalty. Such studies are

00:34:50.820 tantalizing but severely limited for they can't explain why Deshaun didn't get the call. Was he

00:34:57.200 rejected because the employer is a racist and is convinced that deshawn williams is black

00:35:01.480 or did he reject him because deshawn sounds like someone from a low-income low-education family

00:35:07.680 a resume is a fairly undependable set of clues a recent study found that more than 50 percent of

00:35:13.260 them contain lies so deshawn may simply signal a disadvantaged background to an employer who

00:35:19.800 I believe that workers from such backgrounds are undependable. 0.99

00:35:25.220 You might think this is racism, but what if I told you that they're simply associating the name with a set of undesirable qualities that they attach to black people? 1.00

00:35:36.800 Are you fucking kidding me? 0.98

00:35:38.960 Did they not hire somebody named Mohammed due to Islamophobia? 1.00

00:35:42.940 Or did they simply believe that he was going to strap a bomb to himself and blow up the building?

00:35:47.200 that's a longer way of saying islamophobia also how come this is the one time in the book that 0.98

00:35:53.400 they're like demanding more data yeah no shit everything else there they'll hang on to two 0.95

00:35:57.900 data points and be like we've proven that uh people are irrational about guns vis-a-vis swimming 0.98

00:36:04.000 pool yeah but with this one they're like let's not get crazy before we start calling people racist 0.56

00:36:09.160 also dude this is why i mentioned the fucking weakest link study because like three chapters 0.75

00:36:13.240 ago you're like whoops turns out racism doesn't exist we looked at evidence from a game show 0.90

00:36:17.760 and now they're looking at like real world examples these studies are extremely consistent

00:36:23.820 there's been like a million of these by now this is like some of the strongest data for racism in

00:36:30.740 hiring because exactly you can eliminate so many variables that otherwise might complicate the

00:36:36.560 process right it's just resumes and they're identical and also how is this for economics

00:36:42.500 Where's the Freakonomics here?

00:36:44.400 Well, I mean, Stephen Levitt did a study on this where he's basically asking the question of like, should Deshaun change his name?

00:36:53.600 And so he concludes, so does a name matter?

00:36:57.300 The data show that on average, a person with a distinctively black name, whether it's a woman named Imani or a man named Deshaun, does have a worse life outcome than a woman named Molly or a man named Jake.

00:37:07.320 But it isn't the fault of their names.

00:37:09.740 If two black boys, Jake Williams and Deshaun Williams, are born in the same neighborhood, into the same family and economic circumstances, they would likely have similar life outcomes.

00:37:20.060 The kind of parents who name their son Jake don't tend to live in the neighborhoods or share economic circumstances with the kind of parents who name their kid Deshaun.

00:37:27.940 But Deshaun is more likely to have been handicapped by a low-income, low-education, single-parent background.

00:37:33.400 His name is an indicator, not a cause, of his outcome. 0.90

00:37:36.840 Good Lord. I mean, first, he's just making that up, right? The whole point of the resume studies

00:37:42.300 is that they show that that's not true. Exactly. That actually, there are disadvantages to the

00:37:47.060 name in and of itself because people are racist. So he's just saying, no, let's ignore those studies

00:37:52.740 and just get the causation exactly backwards, or at least eliminate some complexity.

00:37:58.980 But I feel like this is another feature of these books, is that oftentimes they'll set up

00:38:03.820 this like straw man to debunk so in this he's like you thought the only reason dashawn can't 0.95

00:38:11.000 get a job is his name but it turns out most poor black people can't get jobs it's like right that's

00:38:18.620 what i thought in the first place i didn't think it was only the name this is this is like a little

00:38:23.980 darker than like what gladwell does which is always like often a little cuter like gladwell

00:38:30.820 be like how one soccer team used jelly donuts to win a championship and you're like what's going

00:38:36.440 on there and levitt's is like it feels just a little more racist every single time yeah there's 0.96

00:38:42.260 a huge amount of conservative ayn rand bullshit presented in this book as like harsh truths 0.57

00:38:49.540 yeah yeah i and again i would like to circle back to the whoever thought of freakonomics 0.98

00:38:55.280 as a title because Black People, a Critique is a much worse title.

00:39:03.000 God, I didn't realize he was a Chicago guy, but it's starting to all click together in my brain.

00:39:08.740 Chicago and bio. We could have stopped at that. So speaking of which, the final way that this book

00:39:16.160 misuses data is waltzing into huge pre-existing debates and pretending to solve them.

00:39:25.280 Chapter five of Freakonomics is dedicated to the question, what explains the crime drop of the 1990s?

00:39:32.880 Oh, yeah.

00:39:33.520 We both know where this one is going, but we're going to let them work up to it.

00:39:36.400 So first of all, the massive crime drop of the 1990s is probably one of the biggest social shifts to happen in our lifetimes.

00:39:43.960 Yes.

00:39:44.320 The murder, the national murder rate went down by 50 percent.

00:39:47.560 Murders in New York City went from 2300 a year to 600 a year.

00:39:52.020 This is like an actual huge deal.

00:39:54.400 And there's a whole field of criminology dedicated to explaining this.

00:39:59.300 I interviewed three different criminologists for this.

00:40:02.200 So Stephen Lovett says there's actually three things that explain the massive crime drop

00:40:08.620 in the 1990s, right?

00:40:09.900 So the first is imprisonment, mass incarceration.

00:40:14.000 You will love this because this is Supreme Court.

00:40:16.380 First of all, to explain the crime drop, we have to explain why crime rose so much in

00:40:20.500 the 1960s, like massive increase in crime.

00:40:23.340 He says,

00:40:24.180 In retrospect, it is clear that one of the major factors pushing this trend was a more lenient justice system.

00:40:30.200 Conviction rates declined during the 1960s and criminals who were convicted served shorter sentences.

00:40:35.160 This trend was driven in part by an expansion in the rights of people accused of crimes.

00:40:40.520 A long overdue expansion, some would argue, others would argue that the expansion went too far.

00:40:45.880 At the same time, politicians were growing increasingly softer on crime.

00:40:49.900 For fear of sounding racist, the economist Gary Becker has written, since African-Americans and Hispanics commit a disproportionate share of felonies.

00:40:59.480 So the reason we got more crime is because America was famously not racist in the 1960s. 0.71

00:41:06.160 Yeah, we hadn't hit that Goldilocks just right amount of racism that we need to drive crime down to historic lows.

00:41:13.860 So the only citations in like this section of like mass incarceration, reduced crime are three articles that Gary Becker wrote in Business Week. All three criminologists told me that like this is not remotely an accepted explanation. Like we gave too many rights to people and then we got more crime.

00:41:33.000 I want to point out some big picture social science shit before we advance just to get it off my chest. 0.91

00:41:38.300 First of all, taking away the like inherent moral concerns with mass incarceration. 0.97

00:41:44.920 A lot of what it's actually doing is just containing crime, right?

00:41:49.540 Placing crime into prisons where it's generally not recorded and doesn't add to the crime rate.

00:41:54.960 Another thing is that when we talk about the decrease in crime, what we're talking about is something very specific.

00:42:01.200 and it's really the decrease in certain crimes, generally violent crimes, right? We're talking

00:42:06.520 about murders, assaults, et cetera. The sort of phenomenon of corporate level crime and

00:42:14.240 government level crime, right? Crime by massive institutions is completely excluded from these

00:42:20.000 calculations. And so I'm not saying that crime didn't go down, but our understanding of crime

00:42:25.300 is tunneled through street-level violent crime.

00:42:30.520 But this is one of the reasons why I think this book has been such a negative force in

00:42:34.460 American life, is that a lot of policymakers read this and I think adopted the conflation

00:42:41.560 that the book is making, where it's toggling back and forth between what is the most effective

00:42:47.640 policy for reducing crime and what is the right policy.

00:42:52.060 I remember when I was in grad school, I took a class on crime and punishment.

00:42:54.640 And on the first day, the professor told us that, like, if you imprisoned every teenage boy on their 16th birthday and released them on their 25th birthday, you would prevent 80 percent of crime.

00:43:05.660 Right.

00:43:05.840 That's a horrifying policy for many reasons, but it would be very effective to deter crime.

00:43:11.720 Right.

00:43:12.080 If every speeding ticket came with the death penalty, just immediate shot to the back of the head, we would have no speeding in America.

00:43:18.040 Right.

00:43:18.400 Many, many, many policies would be effective at reducing crime, but that does not mean that they are the right policy.

00:43:26.360 You know, other countries which did not have mass incarceration also had huge crime drops in the 1990s.

00:43:32.960 This is a worldwide phenomenon. 0.99

00:43:35.120 So I actually think that it's probably the case that mass incarceration reduced a lot of crimes, mostly because you're just imprisoning a bunch of fucking teenagers. 0.96

00:43:43.760 But that doesn't mean that it was the right policy. 0.92

00:43:46.100 And that also doesn't mean that there weren't other policies that would have had the same

00:43:49.280 outcome with a lot less like human misery.

00:43:52.640 And also it's not an assurance of long term declines in the crime rate.

00:43:58.580 Right.

00:43:58.860 Right.

00:43:59.160 The benefit, quote unquote, is extremely immediate.

00:44:01.920 Right.

00:44:02.360 This person is off the streets, you know, so to speak, and not committing crimes.

00:44:07.140 What happens in 10 years when they're out and unemployable?

00:44:10.640 Yeah.

00:44:10.820 I don't know what else to say about these sorts of analyses. 0.98

00:44:12.540 But the idea that someone is saying, like, mass incarceration is uniquely effective, it's like, what fucking country are you looking at? 0.98

00:44:18.860 That's why America has so few murders. 0.98

00:44:20.600 Yeah, right.

00:44:22.700 Right.

00:44:23.280 So that was reason one for the crime drop.

00:44:25.720 He says that explains 40 percent of the reduction in crime, mass incarceration.

00:44:30.160 Reason number two is that we had more cops on the street.

00:44:34.000 Of course.

00:44:34.500 This is the one time in the book they talk about, like, actual methodologies and how difficult it is to measure many things.

00:44:40.300 So they basically say you can't just do a correlation between like this city has more cops and less crime and divine any kind of causal analysis.

00:44:48.960 But then Stephen Levitt comes up with this unique model that does allow you to do causation where he says after mayoral and gubernatorial campaigns, they often hire more cops.

00:45:02.520 It's like a campaign promise.

00:45:03.700 Like I'm going to hire 50 more cops, put them on the street, whatever.

00:45:07.040 And then you get less crime.

00:45:08.600 The main thing to know about Levitt's work on this is that it's it doesn't hold up. People have found coding errors in it. There's like mistakes in the data.

00:45:16.520 Shocker.

00:45:16.920 Other people have pointed out that like mayors and governors don't hire enough cops after elections to make much of a difference. Like you can only really get to this through like weird statistical mumbo jumbo. And also there's other reasons why crime might go down after a giant election that don't have to do with cops.

00:45:34.160 And then the number one problem with this, which I feel like should be much more front and center in like every debate about crime and policing, is that crime statistics are not statistics about crime.

00:45:47.300 They are statistics about reports of crime.

00:45:50.200 Right.

00:45:50.480 Which run through police departments.

00:45:53.060 Exactly.

00:45:53.660 And there are many, many things that would increase reports of crime but not increase crime.

00:46:01.060 Yes.

00:46:01.300 A lot of countries that have done like large scale public information campaigns on like sexual assault have like sky high rates of rape and sexual assault, not because they have more of those crimes, but because people are willing to come forward about them.

00:46:14.060 You also find after like large corruption scandals or police brutality incidents, people become more reluctant to report crimes to the police.

00:46:22.740 And that results in what appears to be a drop in crime because you have fewer reports of crime and the police list that as a triumph.

00:46:29.860 Like, look, the police are working.

00:46:31.160 We're reducing crime.

00:46:32.180 But it actually means they're not working because people don't trust them.

00:46:34.800 Right. 0.93

00:46:34.900 So every single, like, study of crime and, like, discussion of crime has to start with the fact that, like, for most crime statistics, we don't know what the fuck we're talking about. 0.57

00:46:44.700 Most crimes are not reported to the police.

00:46:47.280 And there's all kinds of weird stuff about, like, what counts as an aggravated assault versus not an aggravated assault.

00:46:52.820 Even violent crime statistics include robbery, which is stealing from somebody through violence or the threat of violence.

00:47:01.060 That's also something that depends on a judgment call.

00:47:03.440 Right.

00:47:03.880 So with the exceptions of homicides where like basically a body is a body and like those tend to get counted.

00:47:11.240 And one of the criminologists I talked to said that like motor vehicle theft is also weirdly reliable because people have to report it to their insurance.

00:47:16.320 Yeah. I also want to point out this is 2005, another big cultural phenomenon, 2005, The Wire.

00:47:24.320 Oh, yeah.

00:47:24.600 And if you watch The Wire, you would know that when there's a new mayor, they put pressure

00:47:28.660 on the police departments to keep the crime stats down, artificially deflated, OK?

00:47:35.100 And I want to be a guy that learns lessons about city government from The Wire, but I

00:47:39.780 do trust it more than Stephen Levitt at this point.

00:47:42.420 There are two separate teams of researchers that tried rerunning his data and weren't

00:47:46.760 able to replicate it.

00:47:47.960 So the first one just found this coding error.

00:47:49.620 And then the second team of researchers said that when you run it, you actually find more

00:47:54.480 reports of crime but that's probably because when there's more cops on the street they just see more

00:47:59.740 stuff yeah right and like it it gets included but it's this weird thing where a reduction in crime

00:48:05.320 reports means the cops are working and an increase in crime reports also means the cops are working

00:48:09.620 yeah yeah so we just really don't like know very much about what's going on and so again do cops

00:48:16.340 reduce crime is like a huge debate in like three different fields and they're just like waltzing

00:48:23.700 into this and being like obviously police reduce crime because like i did a single study freakonomics

00:48:29.340 so that explains so mass incarceration and policing explain 50 of the crime reduction

00:48:38.300 according to this argument and the third reason is smush motion do you remember like this theory

00:48:44.940 yeah uh i mean in in broad strokes the theory is that i'm i'm gonna try to be somewhat polite

00:48:53.300 about it, that crime disproportionately emanates from poor communities and is within poor

00:49:00.600 communities.

00:49:01.820 Abortion is something that is utilized disproportionately by poor communities.

00:49:07.100 Therefore, the decline in crime can be explained by Roe v. Wade and the spread of abortion.

00:49:15.640 Yes.

00:49:16.300 So this is from the chapter of the book.

00:49:17.620 It says,

00:49:18.760 Leavitt and his co-author, John Donahue of Stanford Law School,

00:49:21.720 argued that as much as 50% of the huge drop in crime since the early 1990s

00:49:26.680 can be traced to Roe v. Wade.

00:49:29.020 Their thinking goes like this. 0.96

00:49:30.680 The women most likely to seek an abortion, 0.99

00:49:32.720 poor, single, black, or teenage mothers, 0.99

00:49:34.700 were the very women whose children, if born, 1.00

00:49:37.640 would have been most likely to become criminals.

00:49:40.260 But since those children weren't born,

00:49:42.420 crime began to decrease during the years

00:49:44.640 they would have entered their criminal prime.

00:49:46.660 In conversation, Levitt reduces the theory to a tiny syllogism.

00:49:51.480 Unwantedness leads to high crime.

00:49:53.340 Abortion leads to less unwantedness.

00:49:55.460 Abortion leads to less crime.

00:49:57.100 Look, I don't even want to say that this is like that there is like no causal connection here.

00:50:02.940 I have no idea.

00:50:04.380 But the amount of information you would need to draw that conclusion is enormous, right?

00:50:11.140 The fact that this comes after them casting doubt on a much clearer correlation with the 1.00

00:50:18.000 resume, like the black names and resumes, it's fucking hilarious. 0.98

00:50:21.600 But why, Peter? 0.99

00:50:22.840 But why?

00:50:23.560 Yeah, it's a real mystery.

00:50:25.080 They begin with a cute anecdote about Romania, where abortion was a really common form of

00:50:33.640 birth control.

00:50:34.680 For every four live births, there was one abortion.

00:50:38.380 Okay.

00:50:38.600 And then in 1966, Ceausescu was doing some like national Romanian greatness thing and

00:50:43.100 he banned abortion.

00:50:44.140 So like overnight, abortion went from like extremely common to non-existent, essentially.

00:50:50.480 And so they say, compared to Romanian children born just a year earlier, the cohort of children

00:50:56.500 born after the abortion ban would do worse in every measurable way.

00:51:00.760 They would test lower in schools, they would have less success in the labor market, and 0.86

00:51:04.260 they would prove much more likely to become criminals.

00:51:07.040 Okay.

00:51:07.260 And then they sort of lay out the argument for the way that abortion reduces crime, which basically it reduces the percentage of unwanted kids in the population.

00:51:16.020 And so when you have abortion being legalized, so the opposite of what Romania did, it's like clockwork.

00:51:22.360 18 years later, you start to see these really significant reductions in crime. 0.69

00:51:26.760 To make this argument, they have four pieces of evidence.

00:51:31.140 The first is that five states legalized abortion two years before Roe, and all five of those states had earlier crime drops.

00:51:42.940 The second piece of evidence is states with higher abortion rates at that time also had bigger crime drops.

00:51:50.160 The third piece of evidence is people born after Roe v. Wade have lower crime rates.

00:51:56.340 If you look at people born after 1973, it's like, oh, they commit less crime.

00:51:59.640 And number four is data from other countries confirms the result.

00:52:04.540 So they say studies of Australia and Canada have since established a similar link between legalized abortion and crime.

00:52:11.360 Of these four pieces of evidence, two are dubious and two are straight up lies. 0.90

00:52:18.420 So the main thing to know about all this stuff about like states that are legalizing abortions had less crime and states with more abortions had the biggest crime drops is like mostly the data is just garbage.

00:52:31.500 So all of the states that legalized abortion earlier had like huge rates of abortion because people were traveling to those states to get abortions.

00:52:39.760 There's no guarantee that those people are living in those states 18 years later.

00:52:43.260 Right, right.

00:52:43.720 And I really couldn't believe this.

00:52:46.040 They don't actually track like a rise in abortion.

00:52:51.320 For this, I interviewed a guy named Ted Joyce, who's written a bunch of articles about this

00:52:55.240 because he's an economist who specializes in abortion policy.

00:52:59.760 He points out that Stephen Levitt is just assuming that there were zero abortions in

00:53:05.180 all of these states before Roe v. Wade. 0.92

00:53:07.700 Right.

00:53:08.200 You're proposing that abortion explains 50% of the crime drop.

00:53:13.460 This is a huge effect, right? 0.62

00:53:15.460 Right. 0.98

00:53:16.040 To demonstrate that you would have to have like doubling, tripling, quadrupling of abortions, 0.51

00:53:21.620 right? But a lot of the states that legalized abortion early were fairly liberal states that

00:53:28.200 had a lot of abortions going on, even when it was technically illegal. It's not big enough

00:53:32.680 to explain this huge effect. And then Ted Joyce also points out that like abortion didn't actually

00:53:38.980 change the birth rates all that much. Like it was like the same number of people being born.

00:53:43.460 One of the weird things about this is that abortion in and of itself is just another way of talking about birth rates, right?

00:53:52.240 You're saying there's this thing that decreased birth rates among poor people.

00:53:56.340 Then why not just look at birth rates among poor people, right?

00:53:59.500 Why do this whole rigmarole?

00:54:01.180 And also the biggest thing to me, if it was Roe v. Wade, right, you have this cohort of kids that are born after Roe v. Wade is legalized, right?

00:54:10.300 So if there were 20% unwanted kids in the population, now there's like 10% because there's

00:54:15.640 higher access to abortion, right?

00:54:17.820 But then when you look at the specifics of how crime dropped, it wasn't young people

00:54:22.140 who drove the reduction. 1.00

00:54:24.060 It was old people. 0.77

00:54:25.180 I read this fascinating article on the reduction in adult homicides.

00:54:30.420 This was also a time when the demographics were shifting, where there were, you know,

00:54:33.080 the baby boomers were aging into later adulthood.

00:54:36.440 So basically, you just had the entire population getting older.

00:54:39.320 there's just fewer teenagers in the population. So like that drove a lot of reduction in crime

00:54:43.740 rates to begin with. And at the same time, you had reductions, like fewer adults killing each

00:54:49.560 other. And like the biggest decrease was among wives killing their husbands.

00:54:54.640 Dudes rock. Let's go, boys. 0.86

00:54:57.820 That's also something that draws upon all these other social shifts at the time, right? There's

00:55:01.320 like there's no fault divorce, people waiting longer to get married, people moving in together

00:55:08.560 before they get married it's easier to leave somebody rather than have this sense of desperation

00:55:12.260 guys just getting nicer you know give us some credit michael no it's it's definitely the

00:55:17.460 decreasing murder ability of straight men they stopped wearing such provocative clothing

00:55:22.000 i i don't want to swap like one cute counterintuitive explanation for another one

00:55:28.440 yeah there's not that many wives killing their husbands in the country no it's just this is

00:55:32.680 insanely noisy exactly it's just so noisy the amount of variables we have going on here is

00:55:37.540 just crazy. And so you would expect for something like this, for like way fewer teenagers to be

00:55:43.460 killing each other. The cascade through the population is exactly the opposite. It starts

00:55:48.360 with like 40 year olds and teenagers were actually killing each other more because this was right

00:55:53.180 during the crack epidemic. Right. So it's this extremely weird thing where Stephen Levitt says, 0.52

00:55:58.620 oh, abortion explains the crime drop among teenagers. And then people are like, oh,

00:56:02.820 well, teenagers were actually killing each other more. Oops. But then he says, like,

00:56:06.680 if you control for the crack epidemic then we have the effect well then what's this based on 0.82

00:56:13.120 sorry but control for the crack epidemic is killing me this is too good

00:56:17.800 control for the causes of crime please but then to me this is where it gets really cynical so

00:56:23.600 in the book they say studies of australia and canada have established a similar link 0.99

00:56:29.820 between legalized abortion and crime this is a lie this is just a straightforward fucking lie 0.97

00:56:34.600 in Canada abortion was legalized in two waves one in 1969 and one in 1988 and the crime drops don't 0.97

00:56:41.000 line up with either one of them and then the paper that they're citing about Australia finds mixed

00:56:47.260 results it's like in Australia it's also some states did it before other states the researchers

00:56:52.540 who clearly agree with Levitt and are like trying to make the data show that they're like uh we

00:56:58.000 found it earlier in some states but not in others and also we don't have data on like the age

00:57:04.440 of perpetrators of homicide so we can't actually say whether it was teenagers so it's like oh so

00:57:10.040 you just can't say anything right there's other studies where they found that in the uk the crime

00:57:14.760 rate fell almost equally in england and northern ireland even though northern ireland didn't

00:57:19.860 legalize abortion there's a review of 20 countries which found no link between abortion and crime

00:57:27.100 Yeah. This also brings us back to the Romania stuff. So in the book, the way that they describe this is like this perfect mirror image of what happened in America. So in America, we legalized abortion and then we got less crime. And in Romania, they banned abortion and then they got more crime.

00:57:49.100 Right. So again, the way that Freakonomics frames this is these children would turn out to have miserable lives. Compared to Romanian children born just a year earlier, the cohort of children born after the abortion ban would do worse in every measurable way. They would test lower in school, they would have less success in the labor market, and they would also prove much more likely to become criminals.

00:58:11.060 So like slam dunk, right? One of the first things you find when you start Googling around for this is you find a series of studies by an actual Romanian who looked into the data on the abortion ban and then this huge explosion of birth rates right after the abortion ban. 0.76

00:58:26.840 And what he finds is exactly the opposite. The kids born in the wake of the abortion ban committed less crime. So the reason for this is all about who was getting abortions in Romania. So the people who were getting abortions were mostly middle class, more educated women, partly because you would get abortions from like the official medical system. And you had to be able to afford to go see a doctor.

00:58:56.140 You had to be educated enough to know that abortions were available to you, right?

00:59:00.280 And like living in a city where you could access them.

00:59:02.460 And so when they banned abortion, the birth rate spiked. 0.99

00:59:06.380 But the women who were having babies were mostly educated, middle class, relatively well-off women who could afford to give their kids the resources to make sure they sort of ended up okay in life.

00:59:19.660 And then when you look at America, it turns out to kind of be the same thing.

00:59:24.440 One of the things they mention in the Freakonomics book is that after abortion is legalized in the United States, the cost of an abortion goes from roughly $500 to around $100, which is like a huge difference, right?

00:59:37.220 Yeah.

00:59:37.520 But also $100 in the early 1970s is still a decent amount of money.

00:59:41.200 Right.

00:59:41.480 And a lot of people do not have access to abortion clinics, right?

00:59:45.660 If you live in a rural area, if you're not educated enough to know, like, what the signs of pregnancy are, you don't you don't realize what's happening.

00:59:52.960 You might have super religious parents that don't allow you to go get an abortion.

00:59:57.000 It would be very odd to act like Roe v. Wade didn't increase abortion access for poor women because, like, obviously it did. 0.95

01:00:04.620 Right. But there's not enough of a shift in who was getting abortions to explain all that much.

01:00:11.660 Right. Right.

01:00:12.040 The poorest women, the most marginalized women in America were having babies when abortion was illegal and they're having babies when abortion is legal. 0.82

01:00:21.540 Right. He's he's implying that legality was the real barrier to abortion access, but it's only one barrier. 0.90

01:00:28.420 Right. And I really could not believe this.

01:00:30.880 So when when I read the Freakonomics book, Googled around, found this Romanian study and I was thinking of like how to explain this to you.

01:00:37.760 i was like okay so they wrote freakonomics and then later this romanian guy looks into the data

01:00:42.600 and he finds that it doesn't hold up and like fair enough we've all written stuff in popular

01:00:47.280 media that like eventually it turns out not to be true when we get better data whatever right

01:00:51.300 yeah so i went back and double checked this romanian study the study came out in 2002

01:00:56.480 three years before freakonomic oh my god so this is not a debunking of freakonomics this is this

01:01:04.120 their source this is the source they're using in the footnotes oh my god so it's like this to me 0.80

01:01:10.360 is like a new level of cynicism for this fucking book there's something so wild about this because 0.93

01:01:17.360 what you expect from these like little pop science books is oversimplification and what you so often 0.98

01:01:26.280 get is just incorrect information right well this is what is so fascinating about this book and like 0.91

01:01:33.240 the discourse around this book is that whenever you see criticism of these dumb airport bestsellers,

01:01:39.440 the defense of them is usually like, well, you got to sand down some rough edges, right? You're 0.95

01:01:44.400 trying to convey complicated issues to the lay public. And like, you know what? Fine. I have

01:01:50.040 spent a lot of my career doing this. It is hard to simplify complex ideas and like entire fields

01:01:56.660 of academic studies. I get that, right? But it's very odd to use that defense when what we're

01:02:02.620 talking about is a study says crime went down and you are saying crime went up no one would 0.95

01:02:10.540 understand that as like simplification of a complex idea that is a fucking lie there's something 0.99

01:02:15.980 completely insane about it was 40 this 10 this 50 this and it's like what the fuck like yeah yeah 0.97

01:02:22.940 let's just take a big step back and this is speculative but also almost certainly true 0.88

01:02:27.360 part of the crime drop i don't know how much maybe even a very tiny amount part of it is probably

01:02:32.220 just cultural norms shifting part of it is probably education going up you know like all

01:02:37.840 these like little things and when you're saying three things account for 100 of it you're allowing

01:02:44.900 for no flexibility but like i i looked into this i there's a really good post by john roman

01:02:50.720 where he goes through like 25 explanations for the crime drop all of which are backed up by like

01:02:57.220 pretty good data right but like you can't combine all of them right like everybody has their kind

01:03:02.540 of pet theory and you can use statistics however you want to like bolster your pet theory yeah as

01:03:08.100 far as i could tell the consensus seems to be that it's like some combination of various like

01:03:15.280 justice system things so like incarceration policing i mean this is also a period where

01:03:20.380 there's like huge technological changes you know people are less likely to carry around cash now

01:03:25.860 there's one theory that it's air conditioning where people were just indoors more in comfort

01:03:31.160 and not like outside hanging out where they're like interacting with people. People talk about

01:03:35.680 the VCR. Yeah. Because crime goes up in the summer, right? That's like a classic sort of

01:03:40.200 social psych 101 thing. Yeah. And the reason for that, of course, that people tend to be

01:03:44.780 outside more. And the idea that things are keeping people indoors more seems to make some sense.

01:03:50.360 Yeah. There's also a lot more people are on medications for various mental health things.

01:03:55.200 There's also I think a really underrated factor is better medical procedures.

01:03:59.540 So if somebody gets shot, they're much more likely to live now than they were 40 years ago.

01:04:03.100 So a lot of those murders just became assaults.

01:04:05.660 There's all kinds of other social changes which like you can't really disentangle from the crime stuff.

01:04:10.280 So like teen pregnancy, teen fertility is way down.

01:04:14.080 That's another like long, slow shift that has happened in their lifetimes.

01:04:18.120 Teens are less likely to use drugs and alcohol now.

01:04:20.320 Adults drink less alcohol.

01:04:21.800 There's less, you know, whatever.

01:04:23.340 Inflation and unemployment is higher.

01:04:25.200 And just like living standards are higher.

01:04:27.040 One theory is that it's immigration, you know, because immigrants commit less crime.

01:04:30.680 Oh, Michael, I think you have that backwards.

01:04:33.120 Surely we are under siege, sir.

01:04:36.400 There's the lead stuff.

01:04:37.940 People are kind of like very attached to this one.

01:04:40.600 And like, I think it's in there. 0.94

01:04:42.380 I think it might explain why crime went up so much in the 1960s, basically because baby boomers were becoming teenagers. 0.79

01:04:48.640 And then crime didn't go up as much when like Gen X and millennials became teenagers. 0.85

01:04:53.100 Like maybe the lead thing is in there. 0.68

01:04:55.140 Although it totally breaks down internationally and doesn't really work.

01:04:57.860 I'm not invested in like it's fake, but I'm also not accepting that like that's what really

01:05:02.980 explains it.

01:05:03.660 I don't think there's any like quote unquote real explanation.

01:05:06.040 Yeah, but what they're doing is the equivalent of just being like, you're just looking at

01:05:10.180 one trend and being like, wow, it looks like ice cream sales went up and crime went down.

01:05:17.140 There it is.

01:05:17.920 Ice cream and crime. 0.99

01:05:19.940 That's all they're fucking doing here, right? 0.99

01:05:21.700 Just like throw a dart at a board, hit a data point, throw another at another board, hit another. 0.99

01:05:27.060 And you're like, those two, bang.

01:05:28.600 Ultimately, you're looking at correlations.

01:05:31.400 You're looking at very noisy data on crime, very noisy data on abortions.

01:05:37.480 And like I'm not even covering all of the statistical debate that has gone on around this.

01:05:43.280 Like there's been a huge number of papers about like there are basic coding errors in Levitt's study.

01:05:49.680 he uses arrests as a proxy for crime oh he uses the raw number of arrests rather than the arrest

01:05:57.300 rates so it's like new york has more crime than wyoming it's like well yeah because it has more

01:06:01.520 people one of the criminologists i interviewed said that like no responsible sociologist would

01:06:08.000 ever say that it's one thing or even like three things it's going to be like 12 and they're all

01:06:13.200 going to be interlinked right we may just never be able to untangle this that's a sound right but

01:06:18.800 But then what's so weird to me is like, you know, the international comparisons don't hold up.

01:06:23.060 The timeline doesn't hold up.

01:06:25.100 The statistics don't hold up.

01:06:27.040 And yet Levitt continues to double down.

01:06:31.020 Yeah.

01:06:31.600 I remember a couple years ago, right?

01:06:34.240 Yeah, they're still publishing papers on this.

01:06:35.940 Yeah. 1.00

01:06:36.460 I mean, what the fuck, man? 0.99

01:06:37.560 So he just released an updated version of this. 0.99

01:06:40.320 I think it was 2018.

01:06:41.460 Maybe it was 2019.

01:06:42.160 And they did an updated Freakonomics episode about this, which I listened to.

01:06:47.220 and again they find oh it explains 50 of the crime drop but then in the freakonomics episode

01:06:53.080 levitt just sort of drops in he's like well it might even explain as much as 80 or 90 hell yeah

01:06:57.940 like well that's not in your paper but okay i love the idea of like a serious academic and

01:07:03.800 this actually happens i think to some like more than you might expect but an academic puts a

01:07:09.060 thesis out there and they're just in need of therapy and so when people attack the thesis

01:07:15.280 Rather than being like, hmm, you know, that's interesting.

01:07:18.700 Maybe if we reconceptualize it like this or maybe I'm wrong, they just get defensive for 20 years.

01:07:24.600 Yeah, yeah, yeah.

01:07:24.980 And by the end of it, they're just a complete crank.

01:07:27.760 What's so weird to me is like I don't even feel all that strongly that like this effect doesn't exist.

01:07:34.100 Yeah.

01:07:34.460 If it's really important to you to say that like abortion is one of the things that's in there.

01:07:39.320 Yeah.

01:07:39.500 I can't really disprove that given the data that we have.

01:07:42.640 but to me it's it's on the level of like vhs or air conditioning you know it's vaguely plausible

01:07:48.860 you can kind of use statistics to say almost anything you want if you're trying to explain

01:07:54.000 10 of this massive social shift you can say like the decline of vinyl records explains 10 of the

01:08:00.420 crime drop if you want to using like modern statistical techniques so i'm not i'm not going 0.94

01:08:04.640 to say that like he's full of shit the data does not allow me to say that i mean look yeah i mean 0.87

01:08:09.880 this is an incredibly complex set of phenomena. And if you wanted to say that abortion is like 1.00

01:08:16.860 part of this tapestry. Sure. When you're once you're saying it's 50 percent. Right. Then you

01:08:23.060 need to show an unbelievable amount of data. Yeah. And instead he gives you fucking nothing.

01:08:30.140 Right. Some incorrect data from Romania and some hypothesizing. To wrap up. I mean, I read a bunch 0.99

01:08:36.140 of reviews of this book and there was only one that pointed out what an ideological project this

01:08:43.080 is right once you get to the actual things that you learn from reading this book it's like okay

01:08:47.640 campaign finance doesn't matter that much and like discrimination isn't that big of a deal

01:08:53.320 they have a whole section that i fucking skipped about how like protecting the forests to save the 0.77

01:08:58.840 spotted owl is like not worth the money and like throughout the book they're setting up this binary 0.95

01:09:04.920 between acting on intuition and acting on data. Right. But what we see throughout this book is

01:09:12.000 that all of their quote unquote data driven presentation is riddled with ideology. Right.

01:09:19.340 Yeah. They're leaving out important information. They're using data that doesn't indicate what

01:09:24.680 they say it indicates. They're miss citing existing research. Right. I don't want to set

01:09:30.900 up like a weird QAnon thing where it's like it's bullshit to look at the data or like we

01:09:34.820 we shouldn't look at research. Like that's obviously just as shallow as saying that like 0.82

01:09:38.960 research will solve everything. But I just want to stress that this is a false binary. There is

01:09:44.740 no such thing as using data to remove all human judgment, value, ideology from the way that we

01:09:53.580 make decisions. We should make decisions based on values. Right. The book has a confidence to it

01:10:00.520 That carries forward into its readers almost, right?

01:10:05.060 Where people had like this belief that this was sort of groundbreaking in a sense.

01:10:11.400 And, you know, I have to say as much as they present that dichotomy between like intuition

01:10:15.780 and science, the appeal of most of their ideas is really intuitive.

01:10:21.080 Right.

01:10:21.260 You know, they're making these claims that might not be the consensus, but are not really

01:10:27.060 counterintuitive, right?

01:10:28.720 Oh, well, swimming pools are dangerous too.

01:10:30.340 I mean, that's something we all sort of know, right, that kids around swimming pools is a dangerous combination.

01:10:36.700 I actually think what they're presenting is basically something that is designed to appeal to your intuition.

01:10:43.280 And that's why it's effective.

01:10:45.000 And I think very importantly, they're repackaging it and selling it back to you while telling you that it's science, right, while telling you that it's objective, which I think is like somehow worse.

01:10:54.760 Because if you know that you're you're going on your gut, you can compare that to other people's guts and have a little bit of humility about like, oh, I see it this way.

01:11:02.800 Somebody else sees it the other way. 1.00

01:11:03.840 But it's like, no, no, I'm doing science and everyone else is a fucking rube. 0.99

01:11:07.740 That's actually like really dangerous. 0.99

01:11:10.080 Right. And there's an there's an implication that institutions like major academic institutions are either ignoring or covering up this information in some way.

01:11:21.220 Right. And that shit is wildly dangerous. 0.99

01:11:24.760 Right. And there's a sense to which sometimes I hear about a book like Freakonomics and it conjures up like very specific discussions like the abortion component. And then you hear all the things that the book contains. And rather than getting clarity, it just ends up being a jumble of bullshit in my mind where I'm like, damn, that book was a lot dumber than I thought. 1.00

01:11:48.980 It really is shocking how dumb this book is. I was expecting like Gladwell where you have to kind of look for it. This is also why I wanted to leapfrog this to the first episode that we release because we've talked about a lot of other books. We've already recorded episodes on like one of the Gladwell books and The End of History and Clash of Civilizations. But I think Freakonomics is really one of the worst entries in this both because of how stupid it is and also how like no one seemed to comment on that at the time. 0.99

01:12:17.880 Right. It's kind of incredible. And, you know, Michael, you once described this phenomenon to me that the TED talkification of American discourse as a common consumer. I think you would assume that to the degree that this sort of like pop science exists above it somewhere is a level of serious science that the serious people are discussing. 0.95

01:12:41.440 but in reality there are no serious people and they're all reading this bullshit yeah that is

01:12:46.940 like a very true and haunting phenomenon but where like you know the most powerful people 0.99

01:12:51.540 in the world are absorbing the same dumb ideas that the rest of us are 0.71

01:12:57.400 Freakonomics. 0.99

01:12:58.780 Freakonomics.

If Books Could Kill - November 02, 2022

Freakonomics

Episode Stats

Transcript