Self-Help for Data Nerds

My guest today, Seth Stephens-Davidowitz, has a very special gift. He’s a data whisperer. You give him a pile of data and he will find an interesting story.

Seth STEPHENS-DAVIDOWITZ: We’re living in an incredible era — a data explosion that’s giving us answers to all these kinds of age-old, very important questions. What makes you rich? What makes a successful dater? What makes people happy? What makes you look better?

Welcome to People I (Mostly) Admire, with Steve Levitt.

Seth parlayed his Harvard Ph.D. thesis, which analyzed data on Google searches, into a data scientist job at Google and a bestselling book entitled Everybody Lies. Now he’s back with a brand new book. It’s called Don’t Trust Your Gut. And it’s more or less a self-help book for data nerds.

Steve LEVITT: A great example of the kind of thing you do so well is the part of the book where you talk about how you have a lifelong frustration with the way you look, and then you finally decide to do something about it. Can you explain how a data scientist approaches a makeover?

STEPHENS-DAVIDOWITZ: Yeah. So there’s all this research on how you look influences your life outcomes. So, Alexander Todorov — I think he’s with you at Chicago — has done these studies that you can predict 70 percent of the winners of gubernatorial elections, just based on showing people a face of the candidate. It’s really, actually, depressing. It’s like, geez, are we really that superficial? Yes. You can predict how long someone’s sentence is going to be based on what they look like or how far they’re going to rise in a military career based on what they look like. I have, lifelong, been insecure about how I looked. The way I usually responded to it was just ignoring it, and dressing poorly, and making some self-deprecating jokes. And I’m like, “Wait a minute, maybe I can influence this a little bit.” And there also is research that how you look can vary a lot. Like subtle differences can change dramatically how you’re perceived. So there’s this app FaceApp where it uses A.I., and you can just totally change what your face looks like. What do I look like with a beard, a shaved head, long hair, smile, no smile, pink hair, blonde hair, anything you can imagine. So I created like a hundred different versions of myself. And then I asked people — I did a survey — how do I look on each of these different versions of my face? And then I did a regression analysis on those answers.

LEVITT: Where do you find people who want to take the time to tell Seth whether he looked better or worse with pink hair?

STEPHENS-DAVIDOWITZ: My friend Spencer Greenberg created this platform, Positly, which allows researchers to ask any question to a large sample of people.

LEVITT: So you’re paying these people small amounts of money —

STEPHENS-DAVIDOWITZ: Yeah, I was paying people. Yeah.

LEVITT: — to look at hundreds of pictures of you. But what question are they trying to answer?

STEPHENS-DAVIDOWITZ: So I focused a lot on competence. I would have done attractiveness if I was still single, but now I’m happily in a relationship. So I don’t feel like I need to be more attractive.

LEVITT: So that’s an interesting statement about your relationship. You don’t think it would be a good idea to remain or become more attractive for your significant other?

STEPHENS-DAVIDOWITZ: Uh —

LEVITT: That’s a mistake. Let me just tell you, I think that’s a mistake. I think you want to rethink what you’re optimizing.

STEPHENS-DAVIDOWITZ: That’s a good point that I didn’t totally think through. She wouldn’t have been with me if she wasn’t attracted to me.

LEVITT: I would just tell you as someone who’s been in relationships longer than you, the more goodwill you can bank early on the better. ‘Cause it only gets harder over time.

STEPHENS-DAVIDOWITZ: Competence and attractiveness are highly correlated, and I was getting pretty similar responses with each. And if I really just wanted to impress my girlfriend, I could do a lot simpler study and just show her the 100 pictures and ask her which one looks best.

LEVITT: But you did not do that, interestingly.

STEPHENS-DAVIDOWITZ: I did not do that. And I instead focused on competence because competent is very, very important in your career. I’m trying to impress people as a writer. Competence was the number one predictor in Todorov studies of politician success. So I’m, like, I really do want to look competent. And there were two variables that dramatically improve how competent I look. One, having a beard, and two, having glasses. And everything else was kind of irrelevant. I thought smiling would be huge. That didn’t have an effect. Small negative effect from pink hair, which shouldn’t be too surprising. But like gray, mature hair, no real effect. Except there were huge effects from beard and glasses. Now, I’m bearded and with glasses. And I hope I look competent to you, like someone you can trust my opinions on.

LEVITT: So Seth, I don’t know if it was data-based, but LeBron James beat you to this conclusion many years ago. He began wearing glasses to his press conferences, even though they just had blank frames. He didn’t need glasses. He decided he looked more competent when he was wearing glasses. So interesting that the data backup what LeBron instinctively thought was true.

STEPHENS-DAVIDOWITZ: The next question, which I haven’t done, is to see how much it varies by individual. I think beard, it’s definitely particular to my face. Like a lot of the things I didn’t like about my face, I have a very full mouth, and I think the beard kind of covers it up and hides it. And glasses, it’s possible that’s a universal thing where everyone looks a little more competent in glasses. It costs a pretty trivial amount of money to test all these photos because survey research is getting so cheap. So it would be really interesting to do a follow-up study, where I tested 100 faces on 1,000 people and saw what the patterns were of what makes people look competent and how much individual variation there is. Do glasses look better when you have a beard? All these subtle things.

LEVITT: So it’s funny you go that direction because I had a very different take away from your study, which is, why don’t you launch a makeover business? With huge markups, because no one else needs to know how cheap it is. And you put your own personal data science touch on it. I would think there would be big demand for that.

STEPHENS-DAVIDOWITZ: I hadn’t thought of it. Maybe if this thing takes off and I get a lot of requests from people. Can you do a makeover for me? I sent the book out to a bunch of people. And a couple of the writers have been like, “I want to write about your book, but could you do a nerdy makeover for me? And as part of the review, we’ll include it.” So if there does turn out to be a lot of demand, maybe I will turn this into a business. It’s a good idea.

LEVITT: So you say that sounds like a good business idea, but you actually have the data to know whether things are good business ideas. Because there’s a part of your book that actually looks at data on entrepreneurs and projects.

STEPHENS-DAVIDOWITZ: So for this book, Don’t Trust Your Gut, I probably read about a thousand academic studies to see what data’s out there on big life decisions and what could people learn from the best research these days on the most important questions of life. And occasionally you’re reading a paper and something, like a sentence there, just blows your mind. And there was a paper by two friends of mine in my Ph.D. program, Eric Zwick and Danny Yagan and two other researchers, where they studied American taxpayers. And they did the most comprehensive study of rich people in the United States. That’s not entirely how they advertise the paper. But there was one sentence where they say the typical rich American is the owner of a regional business, such as an auto dealership or beverage distribution company. And that sentence just blew my brains away. I’m like first of all, auto dealerships, I’d never really thought of as a path to great wealth. And, I had never heard of what a beverage distribution company even was, let alone that’s who I should think about as the typical rich American. It’s not usually who we think about.  

LEVITT: What does it mean to be rich? What’s your definition of rich?

STEPHENS-DAVIDOWITZ: So their definition of rich is the top 0.1 percent, which is about $1.5 million a year. So it’s pretty high. Doctors and lawyers are going to dominate making a mid-six-figure salary. But there definitely is a difference between the top 1 percent, top 0.1 percent. One of the things is that wealth rather than wages is the big path.

LEVITT: Ownership. Yeah. You’ve got to own something.

STEPHENS-DAVIDOWITZ: You’ve got to own something. There are these fun stories, the richest N.F.L. player in history is not Jerry Rice or Peyton Manning or one of these great players. It’s Jerry Richardson, who played for two seasons in the N.F.L., and then bought a whole bunch of Hardee’s franchises and became a billionaire out of it. That kind of shows the value of ownership versus wages towards getting rich. But then within ownership, there are huge variations. There’s an appendix from this paper, where they say how many people in different businesses are in the top 1 percent and the top 0.1 percent. And it turns out there are tons of people in that business and very few people are entering the top 1 percent, let alone the top 0.1 percent.

LEVITT: What’s an example? So, like, terrible business to start, I would think would be the ones that people perceive as fun. So starting a bookstore or, restaurants, obviously, go out of business constantly, so they’ve got to be terrible.

STEPHENS-DAVIDOWITZ: You should actually trust your gut on that because you’re right. The quickest business to go out of business is a record store. The average record store lasts about 2.5 years, which is the shortest of any business, in any field. And clothing stores, toy stores — really, really low. There’s kind of this idea you should be in a boring business, which I think is generally right, and may go against your makeover business because that sounds kind of sexy and exciting. But then there are a whole bunch of businesses that are super boring. And lots of peoples are in them, but there are actually very few people in the top 1 percent. So building equipment contractors — terrible. Residential building construction businesses, auto repair and maintenance, gasoline stations — when you look at the census data, there are lots and lots of people starting businesses in these. And basically, just very few entering the top 1 percent, let alone the top 0.1 percent. I tried to do an analysis combining the census data and this data from this paper, and there were only a select few where more than 10 percent of people who start a business in them were getting really rich. And auto dealership was one of the best ones.

LEVITT: Is anybody actually starting an auto dealership? Or are these just auto dealerships that have survived for 30 years?

STEPHENS-DAVIDOWITZ: So you can’t take from this analysis that you should start an auto dealership because the people who own auto dealerships, if they’re making $2-3 million a year, maybe more, they’re not going to sell it to you. But, when you think about why are such a high percent of people who own an auto dealership getting rich? It’s basically a legalized local monopoly, where there is legal protection against people starting a competing auto dealership. And obviously that’s going to give you a huge advantage in business, where you can be the only auto dealership serving a particular car company in a particular market. Beverage distribution has some legal protection as well. There are other fields that maybe are a little more promising if you want to actually start a business. So market research seems to do really well. And that also kind of surprised me. What’s so great about market research? I think it has a similar flavor to an auto dealership. It’s not a legally protected local monopoly, but if you really become an expert in some small area —

LEVITT: Like makeovers?

STEPHENS-DAVIDOWITZ: Yeah, like makeovers. I don’t know if I’d call it a market research business.

LEVITT: Oh, it’s exactly a market research —

STEPHENS-STEPHENS-DAVIDOWITZ: Yeah, maybe.

LEVITT: Literally, it is market research. It couldn’t be more market research.

STEPHENS-DAVIDOWITZ: I hadn’t thought of that. Yeah. So if you’re an expert in a niche field, you can basically write up reports and just sell them to all the companies in the field. And it’s very hard if you have this very specialized expertise, built up over many years, for someone to come in and knock you out of business. It’s not impossible, but it’s difficult.

LEVITT: How does writing nonfiction books play into this analysis?

STEPHENS-DAVIDOWITZ: I think writing nonfiction books, certainly not $1.5 million a year, in my case, but it does have a local monopoly flavor. I was surprised in the paper that I looked at there were 10,000 independent creatives in the top 1 percent, which is not a huge number considering how many people want to be independent creatives.

LEVITT: So independent creative would be an actor, actress, musician, writer, artist.

STEPHENS-DAVIDOWITZ: Yeah, unfortunately, they were not able to break it down by field because I’m sure there’s huge variation in what fields allow this. And I do think some of these fields do have a local monopoly flavor. So Freakonomics — I love all your books, but to be fair, just about anything you write at this point, is going to get a good advance and pretty good sales because you have this fan base, and if someone unknown, wrote another book with similar materials, similar ideas, it’s not going to do nearly as well as a Dubner-Levitt creation would do. And of course, as a creative, you are an owner, so you do own all your content. You’re not a wage earner. It’s maybe not as crazy as it sometimes sounds. I quit my job at Google to become a writer. And maybe I shouldn’t admit this, but my salary actually did go up, which seems a little weird, but I think it does show the value of owning versus being a wage earner.

LEVITT: There’s going to be a mass exodus from Google upon the release of this podcast episode. That clashes a little bit with another analysis in your book, which I found fascinating, about what factors predict which entrepreneurs will succeed or fail. What people find in the data goes completely against all of the narratives that exist in society.

STEPHENS-DAVIDOWITZ: So the research on entrepreneurship plays into this idea of the counter-counterintuitive idea, which I think is an idea that only nerds like me and you and maybe your listeners will appreciate. But you look at the data on what makes a successful entrepreneur, and there are a whole bunch of studies. Successful entrepreneurs tend to be middle age. And the odds of success in entrepreneurship — this shocked me — go up, up until the age of 60. You don’t really think of 60-year-old entrepreneurs. And if you ask people, “What’s the age of successful entrepreneur?” People say, “20, 25, 30.” People think of Mark Zuckerberg, Steve Jobs, Bill Gates. They founded their companies when they were either 19 or 20 years old. They’re the exception.

LEVITT: And not only do people think you can’t be 60 and start a new company, you couldn’t possibly even get hired at a startup if you weren’t in your twenties. There’s enormous age discrimination. Like you wouldn’t think of hiring a 40-year-old to work in your startup when you’re in your twenties. And yet what you’re saying in the data is that, actually, the startups that are really working by and large are the ones owned, started, run by 40-year-olds and 50-year-olds.

STEPHENS-DAVIDOWITZ: Yeah. And it’s true even in tech. You think, “No way is that true in tech. You need to know all the newest tools.” Even in tech, the median age of a successful entrepreneur is in their forties.

LEVITT: Okay, but this is where the counter-bcounterintuitive comes in, because you’re saying once you actually stop to think about it, well, why would you think a 24-year-old would be any good at running a business? They’ve never even been in a business. Whereas the 50-year-old has spent his life learning about stuff and finally makes the break when he or she sees the exact moment where, oh my God, I actually see a great opportunity. And I know it’s great because I’ve been in this industry for 20 years.

STEPHENS-DAVIDOWITZ: Exactly true. So Mark Zuckerberg starts Facebook when he’s 19. And everyone goes, “Oh my God, that’s amazing. A 19-year-old started a global media empire.” There’s actually data — right after that business started, there was a huge rise in teenage entrepreneurs. Everyone was like, “I want to be the next Mark Zuckerberg.” And the fact that it’s so surprising initially, a 19-year-old running a company, makes it talked about more. And there are stories about it. There are movies about it. And pretty soon, everybody starts thinking, “This is really, really common.” The fact that it’s counterintuitive makes everyone talk about it so much that eventually it becomes conventional wisdom and everyone’s like, “Oh, yeah, of course, a 20-year-old starting a business. That makes plenty of sense.” And then you look at the data and you’re like, “No, no, no, no, no, wait, that’s the exception.” There are Mark Zuckerbergs out there, yeah. There are Bill Gates out there. But they are swamped in the data by people like Tony Fadell, who started Nest in middle-age, and many other examples. And there are anecdotes of people who created businesses with no experience. My favorite is Suzy Batiz, who created this company, Poo-Pourri. She’s one of the richest self-made women in the United States. She had this idea at a dinner party that we should get rid of the smell of poop. And she mixed essential oils, and she created a product that, she claims, gets rid of the smell of poop and is worth hundreds of millions of dollars. She had zero relevant training to do this. No training in chemistry. No training in plumbing. We hear these stories and, again, they’re so counterintuitive that eventually we’re like, “Wow, maybe the reason is that she actually had an advantage from being outside the field.” And then you look at the actual data, and it’s not true at all.

LEVITT: Well, I’ll give you an example with my sister Linda. So back when the internet was just starting, she liked to make soaps at home. And she liked to put fragrances in her soaps. And she was really knee-deep into the online soaping world. And the thing that frustrated her was that when she would want to make a soap, she needed very little of the fragrance oils, but you could really only buy them wholesale in a big jug that was maybe 50 times more than she needed. So she started a business — this is literally at the beginning of the internet — where she would buy big jugs of fragrance oils, and she would pour them into little, tiny containers. And she would relabel them. And she would sell them at a markup of 30 or 40 times. And, this is a business that still exists, and for 20 years made enormous amounts of money. She only could do that because she, number one, understood the needs of the people in this industry, because she was part of it. And number two, because she had a reputation and was in the right place to market it. This was at a time when Amazon was losing money, and nobody was making money on the internet, except my sister who was raking in money pouring big bottles of fragrance oil into little bottles.

STEPHENS-DAVIDOWITZ: That’s a great example. And I suspect your sister also has some Levitt talent. Wasn’t she the one who came up with the title Freakonomics?

LEVITT: Yeah, not some Levitt talent. She passed away, but she had the most amazing creative talent of anyone on the planet. So it wasn’t just that she poured them in the bottle. She made really great labels, and she got people excited to buy it. But it is an example of where, when she suggested the idea, as an outsider, I thought, “This is really dumb. How could anyone possibly make money doing something so simple?” But she understood better what the circumstances were.

STEPHENS-DAVIDOWITZ: Yeah. Since I have zero knowledge on how people should look good, except for my one study, maybe I shouldn’t start this business. Do you think I’m an insider or an outsider?

LEVITT: I think you’re an insider in the sense that you understand exactly how to do data science and how to use the platform to make it happen. I think the outsider would be the person who’s good at giving makeovers but doesn’t know anything about data platforms. This is totally a technology play, not really a fashion play, I would say. Another part of your new book that I really liked was your analysis of the relative importance of genetics across different sports. Can you explain how you get at an answer to that question?

STEPHENS-DAVIDOWITZ: Yeah. There’s this book, Sports Gene, by David Epstein, which I loved.

LEVITT: So David Epstein was a guest. He was an excellent guest. We had a fantastic conversation on this podcast.

STEPHENS-DAVIDOWITZ: Yeah, he’s amazing, and he researches the hell out of these books. It’s really impressive. And he talks about how important genetics are in sports. And there’ve been studies that they found particular genes that may increase your muscle mass in various ways. I think one of the things that’s interesting in that book is it didn’t distinguish that much between different sports. Like what sports are more genetic, and what sports are less genetic? The way that generally people test how genetic something is, you use twin studies. Fraternal twins share only 50 percent of their genetics, and identical twins share 100 percent of their genetics. And if something’s very genetic, then identical twins should be much more similar than fraternal twins.

LEVITT: The reason you’re focusing on twins is because in this nature versus nurture argument, you want to hold the nurture part constant. So you say, “Well, twins experience the same lives, essentially, whether they’re fraternal or identical.” And so if the identical twins end up being more likely to both end up in the N.B.A. than fraternal twins, then that tells you basketball is really genetic.

STEPHENS-DAVIDOWITZ: Yeah, there are about twice as many same-sex fraternal twins as identical twins. But the N.B.A. is a great example because, in the N.B.A., there have been 10 pairs of twins. And eight of them, they’re confirmed identical. Eight pairs of the 10. And two of them are a little uncertain. One of them, Charles and Carl Thomas — I reached out to Charles Thomas on LinkedIn because I couldn’t find any evidence of whether they were identical or fraternal. And he told me that they’re identical. And Joey and Stephen Graham, there’s a little back and forth. So it’s either nine or 10 out of 10 are identical twins in the N.B.A.. Which is way above the ratio you’d expect if genetics aren’t a factor.

LEVITT: So if genetics are not important at all, then what you’d expect to see — that if you go and look in the N.B.A., there should be twice as many fraternal twins as identical twins. And your contention is that either nine or 10 of those sets of twins are identical twins. And you’d only expect it to be three or four. And so that difference tells you that genetics is really important to basketball.

STEPHENS-DAVIDOWITZ: Yeah, it’s enormous in basketball. It’s 75 percent or a little bit higher, maybe even 90 percent of basketball talent, would have to be genetic to explain those enormous number of identical twins in the N.B.A..

LEVITT: What I love about this is that it’s a case where the method is as interesting as the results, because until you explain it, I would’ve said, “How do you get at nature versus nurture?” But once you hear the tool, it’s super intuitive. What other sports also seem to have a big genetic component?

STEPHENS-DAVIDOWITZ: Track and field seems to be incredibly genetic. In track and field, there’ve been 20-something or 30-something identical twins.

LEVITT: And do they tend to do similar events?

STEPHENS-DAVIDOWITZ: Yeah, it’s not like there’s a runner and a —

LEVITT: Shot putter.

STEPHENS-STEPHENS-DAVIDOWITZ: — a javelin thrower.

LEVITT: Yeah.

STEPHENS-DAVIDOWITZ: And there’s obviously complications with this approach. So one of the sports that has had an enormous number of identical twins is synchronized swimming. But that’s because they’re meant to be synchronized. So identical twins have a clear advantage in being synchronized in that they’re probably going to be about the same height and look about the same. But there have also been sports where there have been huge numbers of athletes who have competed in them over the years in the Olympics. And there have been very few, or even in some cases zero, identical twins. So there’s never been identical twins who’ve competed in diving, equestrianism, or weightlifting despite, almost as many athletes in this field as in track and field where you have 20-something pairs of identical twins. And if you look at the numbers, do the math, do the standard errors, it’s very clear that genetics has to be lower than 10 to 20 percent in these sports. If you’re not genetically blessed with any great skill, you’d probably want to focus on one of these sports that shows up very low in the list and that there haven’t been a lot of identical twins in that sport.

You’re listening to People I (Mostly) Admire with Steve Levitt and his conversation with Seth Stephens-Davidowitz. After this short break, they’ll return to talk about how to handle rejection.

*          *          *

Morgan LEVEY: Hey, Steve.

LEVITT: Hey, Morgan, how are you today?

LEVEY: Good, thanks. How are you?

LEVITT: I’m doing great.

LEVEY: So we’ve received a few questions recently about the leaked draft of the Supreme Court’s opinion, which would overturn Roe v. Wade, which of course gives women the right to seek abortions in the U.S. And we’ve gotten so many questions about it recently because of your research with John Donohue.

LEVITT: So John Donohue and I wrote a paper over 20 years ago, where we posed the hypothesis that potentially crime had fallen in the U.S. in large part because of the legalization of abortion. And the theory behind it’s really, really simple. The idea is that unwanted children are at risk for crime. And after abortion was legalized, there were fewer unwanted children. And so you put those two together and it would seem like the legalization of abortion should lead to a reduction in crime 20 years later. And then it really just becomes an empirical question. How important did it turn out to be? And our research suggests that it was really important — that maybe half of the incredible decline in crime that we’ve seen in this country over the last 20 or 30 years would be due to the legalization of abortion.

LEVEY: And we should say that you and John Donohue did follow up your original predictions with a paper 20 years later. And everything you predicted turned out to be true, correct?

LEVITT: Right. We wrote one paper that was based on the first 10 or 15 years post Roe v. Wade. And then we also made predictions about what would happen over the next 20 years. And indeed, when we went back and we checked those predictions, the results were remarkably in line with what we had expected. And I would have thought that was kind of the last word, the last thing that one could say using U.S. data on abortion and crime. But if indeed Roe versus Wade is overturned, and if indeed that makes abortion access very difficult in some states, which would lead, in our theory, to more unwanted births, then if somebody had the patience to wait 20 years until that cohort has grown up and gotten to peak-crime ages, our theory suggests that you would see in states that make abortion harder to get that you would expect an increase in crime 20 years from now.

LEVEY: Steve, do you think either side of the abortion debate can use this body of research to bolster their argument?

LEVITT: No, I really don’t. Because we asked a very narrow question and really, it was a question about crime, not about abortion. We were interested in understanding what would happen to crime and abortion law just happened to be a vehicle, a natural experiment, by which we could learn something about crime. And it’s interesting, ‘cause almost never in the U.S. do people talk about abortion when they don’t have some strong vested interest in pushing either a pro-choice or a pro-life agenda. And John Donohue and I were really unique in that regard, because we weren’t saying abortion is good or bad. We were just using it because it affected something that we thought could lead to a change in crime, which was unwantedness. And I think many people have had a hard time looking at our research objectively because they think, “Okay, what’s their agenda? They’re talking about abortion. And since everyone who talks about abortion has an agenda, what’s their hidden agenda?” But the thing is we didn’t have one and that was confusing to people.

LEVEY: So Steve, if Roe v. Wade is overturned, are you going to put on your research hat and look at what happens?

LEVITT: I tell you, I’m not going to be doing academics in 20 years. So someone else is going to have to do the research, but I have no doubt that there’s going to be a lineup of economists ready to answer that question 20 years from now. And I hope I live long enough to see what they find.

LEVEY: Thank you everyone who wrote in. If you’re interested in learning more about Steve’s abortion and crime research, check out Freakonomics Radio’s episode 384 called “Abortion and Crime, Revisited.” If you have a question for us, we can be reached by email, that’s pima@freakonomics.com. P-I-M-A@freakonomics.com. Steve and I read every email that’s sent and we look forward to reading yours.

*          *          *

In the second half of my conversation with Seth Stephens Davidowitz. I want to talk about mental health. So it’s been very light hearted so far. But Seth is also someone who battled serious depression for years. I wonder whether he’s got some data-driven ideas for beating depression. Also the analysis of Google searches that he did in graduate school is just so incredibly clever. But it also led to his academic demise. Let’s find out why.

LEVITT: You’re open about your struggles with depression in your book. Do the data tell us anything that can help someone who’s fighting depression?

STEPHENS-DAVIDOWITZ: So, I’m actually thinking of writing an entire book, “A Data Scientist Versus His Depression,” where it’s just on this topic because I’m doing much better now, but definitely in my twenties, I was suffering through a severe depression. And it did dominate my life. I did have a section on happiness. And I was reading all these studies on happiness. There’s this research by George MacKerron and Susana Mourato where they ping people at different times on their iPhone and say: What are you doing? Who are you with? How happy you are? And they built a data set of 3 million happiness points, which is just incredible. It’s called the Mappiness project. And I tell my friends, like, “Did you know that people are much happier — the same person doing the same activity at the same time — is much happier if he’s near a lake or he’s in a beautiful environment?” Or: “Do you know people are really unhappy when they’re working, but they’re much happier when they’re doing these different activities?” Or: “People are really happy with friends and romantic partners, but not with other people or acquaintances or colleagues?” And I’d get really excited because I thought the methodology was so cool. And I’d tell my friends. And they’d all be like, “Duh, did we need scientists to tell us this? Like, those are the most obvious things in the world.” And I kind of lost a little momentum. Everybody wants something surprising, counterintuitive, the key to happiness. And then I stepped back and thought about it. You know, all my friends who are telling me that these studies, this research, it’s so obvious — when I look at their lives, many of them are very unhappy. And second of all, if you look at how they’re spending their days, they are spending very few of the hours doing the things that tend to make people happy. They tend to live in cities. Don’t spend much time in nature. They’re career-oriented, and spend all their time working, chasing after money. So I’m like, there’s something profound in the obviousness of the happiness research, where we’re all told that there’s some great answer to happiness out there if we just make enough money, if we buy the right good, if we move to the right place. There’s something that’s going to give us that jolt of happiness we’re looking for. What I took from the research is really the things that make people happy are blindingly obvious. And it’s just up to us to avoid some of the noise of modern life and do more of those obvious things. So I conclude the book, “What’s the data-driven answer to life’s biggest question: How we can be happy?” I say, “The data-driven answer to life, only uncovered in this modern research: Be with your romantic love on an 80 degree and sunny day overlooking a beautiful body of water having sex.” If you do that, you’re probably going to be happy. And if you’re not happy, you do have to ask yourself: How far is how you spend your day from that type of life? Because those are the types of things that tend to make people happy.

LEVITT: Your book is more or less a self-help book for data nerds. And I love how you use data to figure out that your second book should indeed be a self-help book. Could you talk about that?

STEPHENS-DAVIDOWITZ: So my first book, Everybody Lies, it was mostly what you can learn from people’s Google searches. But also just what we can learn about people from data. And after the book came out, I did my own market research. And I asked people, “What do you like about this book?” And people told me, “I really found it interesting when you talked about child abuse,” or “I really found it interesting when you talked about inequality and how we could solve these problems.” And I’m like, “Oh, okay, that’s what really registers with people — how to improve the world in various ways.” And then I looked at Amazon Kindle, they now show you how frequently sentences are underlined. And I had this section where I talked about what you should say on a first date if you want a second date. And a lot of people were underlining that. Like, “Oh, I want to know what to say on a first date to get a second date.” And I’m like, wait a second. People seem really interested in how to improve themselves in various ways. And there was a recent study where they studied a whole bunch of Kindle data on 50 bestselling books. And they found that the word that most improved the odds that a sentence would be underlined was the word “you.” So, like, anything with “you, you, you,” people are really like, “Yeah, you’re speaking to me now. I really am interested in what you have to say.” And then I looked at the bestselling books. By far, the most popular category is self-help books. So I’m like, “Okay, if I really want to give people what they want, I’ve just got to tell them how to improve their lives. Tell them whom they should marry, how they should date, who is secretly rich so they can get rich themselves, how to start a business, how to be happy — all these questions.” So I spent a few years just researching, reading papers, and learning the best data in the world on these self-help topics.

LEVITT: I’ve been intrigued also by the Kindle underlining. And I’ve wanted to do a different research project exploiting that underline feature, just for fun. ‘Cause I’ve long believed that many of the most frequently talked about and purchased books, the ones that people most prominently display on their bookshelves and their coffee tables — they’re not actually reading those books. And I think one good proxy for how far someone gets in the book is the point at which the underlining stops. And I think that many authors, probably myself, maybe even you, if we were able to see that data, we might be discouraged — horrified — to see the results of that study.

STEPHENS-DAVIDOWITZ: So Jordan Ellenberg actually did the study you suggested. He analyzed the underlined sentences in some of the bestselling books, Thomas Piketty’s book on capital and Daniel Kahneman’s book Thinking Fast and Slow. And he found that the rate at which sentences were underlined dramatically decreased towards the end of books. I think there were 3 percent as many underlined lines at the end of Thinking Fast and Slow as at the beginning of Thinking Fast and Slow. I actually once had lunch with Danny Kahneman, where he said he knew that without looking at the data, because he said that the best sentence in his book by far, was, “Nothing’s as important as it seems the time you’re thinking it.” And nobody talks about that sentence. Like none of the reviews mentioned it, nothing. And it was way down near the end of the book. And actually Everybody Lies, which is about people lying, the final chapter of my book was titled “How Many People Finish Books?” And the lesson I took from that study — maybe it should have been how to keep people engaged — but the lesson I took instead was, I could just phone in this conclusion. The last sentence of Everybody Lies was: “Too few of you big data tells me are still reading.” I’m just like, “F*ck it, I’m done. I’m finished. I’m going and having a beer with friends.”

LEVITT: So I do want to talk a bit about your first book, which is called Everybody Lies, because it really is a testament to your creativity. Let me paint the backdrop. So Google released a tool and it was called Google Trends, which allowed anyone to see how frequently a particular search term was used over time and across space. And lots of people, myself included, played around with Google Trends. And I thought it would be fascinating to use it in academic research. But when I tried to think of a specific way to use it, I came up empty, and I abandoned it. And it was the fact that I had tried to get some mileage out of Google Trends and failed that made me stand in awe of the fact that you took that same tool and somehow you delivered an entire book full of interesting, surprising insights. Can you just regale us with a couple of the interesting insights you pulled out of Google Trends?

STEPHENS-DAVIDOWITZ: I quickly concluded that the big advantage of Google Trends — well, there are a couple of big advantages. One of them is just the size of the data set. The sample size is enormous. But another big advantage is that people are honest on Google. So there’s an incentive to tell the truth and say what’s really on your mind. I kind of focused just about all my energy on topics that are socially sensitive, where there may be a disconnect between what people admit in a survey and what people really think and really do and would tell Google. So the first study I did was on the effects of racism in the United States. So if you asked people in the United States of America in 2012, when I did the study, or 2022: “Are you racist?” Everyone’s going to say, “No. Of course not.” So I went to Google Trends, and I was shocked by how frequently people make explicitly racist searches on Google, like “N-word jokes.” It’s really dark, shocking, disturbing material. And in the time period I was looking at, people were making these searches at the same frequency as searches like “Lakers,” or “Daily Show,” or “migraine,” “economist.” So it was millions of people making these searches. I had this map based on Google of where in the United States racism was highest. It was just kind of a surprising map. If you had asked me, “Where is racism highest?” Based on everything, I knew about the history of the United States, I would have said that racism is predominantly concentrated in the South, the deep South. And certainly some of the areas that scored very high on this kind of secret racism measure on Google included Southern Louisiana, Southern Mississippi, parts of South Carolina. But then the number one state was West Virginia. Other areas that were really high: Upstate New York, Western Pennsylvania, Eastern Ohio. If you look at the map, it’s much more an East-West divide these days where racism is much higher in the Eastern part of the United States and much lower in the Western part of the United States. Nevada, Idaho, California — Hawaii’s the lowest state. And then there are also, like, patterns that are disturbing. The highest single week of racism was the week when Barack Obama was first elected president. So you have on T.V., all these videos of people saying how great it is we have an African-American president. Meanwhile, on Google, there are more searches for things like “N-word jokes,” than we’ve ever seen before. And, they rise on Martin Luther King Jr. Day, on average. Actually, during the Black Lives Matter protest, I did some work with John Donohue, a good friend of yours, and racist searches rose during the Black Lives Matter protests. Kind of anytime there’s progress for African Americans, there’s also a rise in racism. The first study I did — okay, I had this map of racism. Well, what do you want to see the effects of? Obama, when he was running for president, how many votes did he actually lose from racism? Because if you asked people in surveys, very few people said that they cared that Obama was Black. So when Obama was running for president, this is way back in 2008, you don’t want to just compare the racism measure to Obama’s vote share because, there are many things that influenced that. Obama was Democrat. Democrats would be more likely to support Obama than his opponent. So you compare Obama to previous democratic candidates, such as the Caucasian candidate, John Kerry, and you see just a stark relationship that in areas that had made historically more racist searches, places like West Virginia, Western Pennsylvania, Eastern Ohio, Upstate New York, Obama just did way worse than the previous democratic candidates.

LEVITT: Do you remember in percentage terms how much less of the vote Obama got in West Virginia than you might have predicted based on John Kerry?

STEPHENS-DAVIDOWITZ: Overall, it was about four-percentage points, which actually means when you do the math, that about 10 percent of white Democrats didn’t vote for Obama just because he was Black. And in West Virginia, he would’ve lost something like 15 percentage points and something like 30 percent of the white Democrats didn’t vote for him just because he’s Black. Actually in the 2012 primary, a white, former-convict ran against Obama. And I think the convict got like 45 percent of the vote or something. It was insane. So they really did not like that Obama was Black. And it’s the type of thing that you wouldn’t see in surveys. You would see it a little higher — there are more white people in West Virginia who will just say in a survey, “I didn’t like Obama because he’s Black.” There literally are people who still say to surveys they oppose interracial marriage. But those people are becoming much, much rarer. Whereas there still are plenty of people who aren’t going to tell a survey that they care that Obama’s Black, but are doing things like searching “N-word jokes,” or even there was a common search during Obama’s run, “N-word Obama,” and not voting for Obama. And since I created that dataset, a lot of other scholars have used this and found that areas that make more racist searches, Blacks had worse health outcomes. They were more likely to be stopped by police. They’re more likely to die early. They had lower wages. So it does seem to play out in many different ways, this secret, explicit racism in the United States, that isn’t seen in surveys, but does show up in Google searches.

LEVITT: So I would think a lot of people listening to this would say, “Wow, that is awesome research. That’s exactly what social science should be doing.” And you were an economics Ph.D. student at Harvard at the time you were doing this Google Trends research. And I suspect like most Ph.D. students, you were hoping to land a top academic job. But I also imagine, knowing my own profession, knowing economics, that this is not the kind of research that sells very well in economics. And I imagine along the way, you had advisors at Harvard telling you that your Google Trends work might not appeal to economics departments. Is that true?

STEPHENS-DAVIDOWITZ: Yes, there’s an interesting backstory in this. So not to be too mushy, but I should be interviewing you for the People I Admire podcast, because, you basically inspired my career. I was a philosophy major at Stanford. I was reading Hume, Nietzsche, Heidegger — lost in the weeds of thinking about the meaning of life with really no plan for what I was going to do with my life. There aren’t really jobs in philosophy. And I have this distinct memory. I was with my girlfriend and her family and the father of my girlfriend was reading The New York Times Magazine. And he’s just like, “Oh my God, you got to hear this. This professor at University of Chicago does all these crazy studies with data. He says that the legalization of abortion caused the drop of crime in the 1990s, and Sumo wrestlers cheat based on their record before the match and how much they need to win.” And it was like a mind-blowing moment — 10 seconds later, I’m like, that’s what I’ll do. That’s exactly how I think about the world. I had zero training, no qualifications, and I applied to be a researcher for John Donohue, who co-wrote the abortion crime paper. I have no idea why he took me on as a research assistant.

LEVITT: I think he didn’t get many applications, my guess would be.

STEPHENS-DAVIDOWITZ: Oh, maybe. That could be, but I was blown away that he took me. And I applied for Ph.D. programs a short time later. Harvard took me. And I was in the Ph.D. program. I was already excited. I’m like, “I’ll be the next Steve Levitt.” And then I’d go to my professors, and I’d have 10, 20, 30 ideas for papers I wanted to do. And this is probably 2009, 2010. And they’d be like, “That’s so Freakonomics.” And I’d go, “Yeah, yeah, yeah. That’s so Freakonomics, exactly.” And they’d go, “No, no, no. That’s so Freakonomics.” Like, I didn’t realize that when I read that New York Times Magazine article in 2003, by 2009, you had gotten so famous and so much attention that there was a lot of professional envy. There was a little bit of a backlash. So now I was in my Ph.D. program, a little bit screwed, where all I wanted to do is do these clever studies of these mischievous explorations of life. and people are like, “No, no, no. You want to go back to inflation and interest rates and inequality and all these topics that are super important, but I know nothing about and have zero interest in.” So when I wrote this paper on the effects of racism on Obama, there was a little bit of an idea that this was cute. This would get a lot of attention. Newspapers would love it, but this wasn’t serious economics. And I applied for a whole bunch of academic jobs and didn’t get a single interview for an academic job. Yeah. Which I think ended up maybe being a blessing in disguise because I’ve done a lot of cool things that I probably wouldn’t have done if I had gotten an academic job. But, obviously, anytime you’re rejected universally for something you’re applying to, it’s a pretty depressing, dark experience.

LEVITT: Yeah. I’m frustrated by the profession of economics because I think we have the wrong priorities. But for a long time, I’ve given a particular piece of advice to my grad students. I tell them, “You should do the research you love, that’s fun for you. And if it turns out that the economics profession loves it too, that’s fantastic. Then you’ll have an amazing academic career. And if it turns out the economics profession doesn’t share your taste in research questions or approaches, that’s okay too. It just tells you that you should be doing something different.” Because I think the very worst outcome is when young people try to pretend that they’re interested in inflation, when in fact they aren’t. And then, they get a job. And it bores them. And you just can’t succeed in a profession that you’re bored by. You’re an interesting case because you are really the poster child for that piece of advice I give because you did what you loved. And it didn’t work out in academics, but it’s obviously worked out incredibly well in life. By being authentic to what made you you, you’ve managed to do fun stuff and had an impact.

STEPHENS-DAVIDOWITZ: And when I become a billionaire with my nerd makeover business, then it’ll really be true, right?

LEVITT: Do you have any advice for people who want to be luckier?

STEPHENS-DAVIDOWITZ: There’s one aspect of luck that is really important, which is just taking a lot of shots on goal. There’ve been these studies by like Dean Simonton and others, that if you look at how much work artists produce, it correlates very highly with various measures of quality, various awards. Just putting more work out there really increases your odds of maybe getting a lucky break. Dating’s a great example of this. There’s data from online dating sites. They’ve actually looked at, what happens when someone of different desirability ratings ask people of other desirability ratings. And there are many different ways they measure desirability. Sometimes it’s just simple physical attractiveness. Sometimes it’s complicated network analysis using the page-rank formula of Google. But they all kind of converge on this idea that even if you’re very low on conventional desirability, if you asked someone very high on desirability — if you messaged them, the response rate for the least desirable men ask the most desirable women, it’s something like 15 percent. And for the least desirable women asking the most desirable men, it’s only like 30 percent, which is a lot higher than I would have guessed. And of course, there’s a lot of steps between getting a response and then getting a date and then getting a relationship and getting a marriage. But I think these numbers make it very clear that everybody should just be asking out more people, but people are so scared of rejection that they hold themselves back and don’t allow themselves to get that lucky break.

LEVITT: In some sense, the secret is how do you minimize how bad it feels to get rejected.

STEPHENS-DAVIDOWITZ: There’s something called rejection therapy. People invented this game where they give you cards, and they ask you for an absurd request. So you’d be like, “The next person you see, ask them like — if I’m in Manhattan right now giving this talk, like, I need a ride to Long Island. Can you give me a ride?” And the person presumably is going to say, “No,” or think you’re ridiculous and weird and crazy. The idea of rejection therapy is that if you do that enough times, you get desensitized to rejection. The more I’ve gotten rejected for things, the easier it is to handle the rejection.

LEVITT: There was a golfer named Gary Player. And he said something like, “People say I’m lucky on the golf course. But the funny thing is, the more I practice, the luckier I get.” I think what he was saying is: there’s actually an interaction between luck and being able to take advantage of it. I look at my own career, every interesting research paper I wrote, there was a lot of luck. On the abortion and crime paper with John Donohue, I happened to flip to a page in a random book that listed the number of abortions that were taking place in the U.S. each year. Had I not done that, I never would’ve thought about it. If I had never run into Sudhir Venkatesh, I never would’ve studied gangs. It goes on and on. But in each instance, I can point to a certain kind of preparation and hard work I had done in advance that when I stumbled onto something, I already had this backlog of ideas and interests that this stumble fit right into. And I was able to take advantage of it. So I really think that’s powerful advice for people.

STEPHENS-DAVIDOWITZ: I totally agree. But one thing you also think about is that you should expect a certain amount of luck in life, right? If you got zero lucky breaks in your entire life, you’d be the least lucky person on the planet. Like, over the course of a lifetime, you’re going to run into people who can help you. You’re going to read a sentence that is relevant to your work. But doing things like running into a lot of people, reading more books, applying for more jobs, asking out more people, just allow you more opportunities to get lucky. And then you’re ready to take advantage of it.

Seth says you should take more shots. My guest a few weeks back, psychologist Dan Gilbert, he said the key to his happiness was doing less better. Can both those pieces of advice be true? Well, I’ve been pondering that, and maybe there is an answer. I think take more shots is good advice when you’re getting started, like when you’re starting your adult life trying to figure out what you’re good at and what you enjoy, and also in specific settings that involve new starts, like dating. Do less better, that seems to me to be powerful advice once you’re more settled, once you’ve figured out the arc of your life. I think it’s no coincidence that the 30-something Stephens-Davidowitz says take more shots and the 60-something Dan Gilbert says, do less better. It would not surprise me one bit to discover that a young Dan Gilbert endorsed taking more shots or that 30 years from now Seth is singing the praises of doing less. What’s that mean for you? I’m not sure. But I can tell you how I run my own life. I don’t take that many shots, but I still do too much, and not very well. In other words, I’m the worst of all worlds. I guess that’s why I’m in the business of asking for advice and not giving it. One last thing, we asked for your feedback two weeks ago at the end of the episode and many, many hundreds of you wrote and you shared the kindest, most touching messages, along with some great ideas for improving the show. I would like to write back personally to every one of you, but I know I won’t manage to. But I did read every one, and it meant absolutely the world to me that so many of you devoted so much time and wrote such amusing and personal emails. Thank you so much. And we’ll see you next week.

People I (Mostly) Admire is part of the Freakonomics Radio Network, which also includes Freakonomics Radio, No Stupid Questions, and Freakonomics M.D. All our shows are produced by Stitcher and Renbud Radio. Morgan Levey is our producer and Jasmin Klinger is our engineer. We had help on this episode from Alina Kulman. Our staff also includes Alison Craiglow, Greg Rippin, Gabriel Roth, Rebecca Lee Douglas, Zack Lapinski, Julie Kanfer, Eleanor Osborne, Ryan Kelley, Emma Tyrrell, Lyric Bowditch, Jacob Clemente, and Stephen Dubner. Our theme music was composed by Luis Guerra. To listen ad-free, subscribe to Stitcher Premium. We can be reached at pima@freakonomics.com. That’s P-I-M-A@freakonomics.com. Thanks for listening.

*          *          *

LEVITT: So I said, “Now, if you’re the worst player on the worst team, what does that say about you and soccer?” And I raised him well, because he said, “It probably means I should quit.”