Seth Roberts, Guest Blogger (Part III)

Here’s a third post from our guest blogger, Seth Roberts, a psychology professor at Berkeley and, apparently, the next American diet guru. If you need to get up to speed on Seth’s unorthodox research with weight-loss, mood, and sleep, click here (our N.Y. Times article about him), here (research extras and pix), here (the first round of reader comments), here (for Seth’s first guest-blog, including comments and questions), and here (his second guest-blog).

“Freakonomics and Me”
Wed., Sept. 14
When Dubner contacted me about writing about my self-experimentation, I thought I finally knew what freakonomics meant: freakishly lucky.

My ideas had been written about before. A columnist for the San Francisco Chronicle, call her Person X, is a friend of friends of mine. She interviewed me and wrote about my idea that faces on TV can improve mood. I had no complaints but a year later a friend said, “She made fun of you, Seth” — gentle fun, to be sure — and he was probably right. My work also attracted the attention of Andrew Solomon, who interviewed me for The Noonday Demon: An Atlas of Depression, a great book. I like and admire Andrew. I was quite happy with what he wrote but even I could see that it wasn’t flattering in a normal way. It was flattering to have been mentioned at all.

Person X and Solomon were perfectly good writers. They told the truth. My claims about mood were so non-standard that they were indeed humorous (Person X) and highly alternative (Solomon) — no doubt about it. But Levitt and Dubner would be the most sympathetic audience I could hope for, I suspected, because Freakonomics and my research were similar in certain ways:

1. Popular beliefs questioned. At one point, an alternative title for Freakonomics was Ain’t Necessarily So. I said that breakfast is bad for you and that drinking sugar water causes weight loss.

2. Delayed causality claimed. Freakonomics said that Roe v. Wade caused a decline in crime that began many years later. I claimed that, under the right conditions, seeing faces in the morning causes big changes in mood that begin about 12 hours later.

It was freakishly (and freakin’) lucky for me that these similarities existed, I thought, that someone who had done such research had been empowered, from my point of view, at just the right time.

I met Dubner and Levitt at the Commonwealth Club in San Francisco, where they spoke. Levitt told the audience that when he was a graduate student, personal computers had recently become available. You no longer had to be wealthy to study a big data set. He took advantage of this to study data sets that economists had not previously studied. Last week I had lunch with Stefano DellaVigna, a professor in the Berkeley economics department. I asked him why economics had become much more eclectic — studying a much wider range of questions — over the last 15 years. He said the same thing as Levitt: the availability of PC’s.

Finally I understood. With cheap, small computers it became much easier to study many things, including the large data sets that Levitt uses and the home data that I collect (e.g., my sleep). When I got a computer at home, self-experimentation became much easier. It was much easier to move data 20 feet (bedroom to PC) than 1 mile (bedroom to campus). Moreover, the statistical software on my PC was better than what I had on the computer in my office. With a PC at home, I started to keep much better records of my sleep. Within a year, using statistical graphics that the PC made possible, I found a surprise in my data: A sharp drop in how much I slept at about the same time that I had lost twelve pounds. Telling my students about this led circuitously to a bigger surprise: Breakfast was causing me to wake up too early. With that discovery I began to realize self-experimentation was more powerful than I had thought.

It is another case of delayed causality. About 15 years ago, Levitt started taking advantage of the expanded opportunity that PCs offered: the ability to study large data sets cheaply. He crunched many numbers not previously crunched. After 15 years, there was enough of this new-fangled research, which dealt with everyday issues, to make a book: Freakonomics.

Likewise, about 15 years ago my self-experimentation started making real progress. Via self-experimentation, I could, like Levitt, study many everyday questions that had previously been much harder to study. After 15 years I had enough data to write a (long) paper. I titled it “Self-experimentation as a source of new ideas” – not so different than Ain’t Necessarily So.

PCs are like small, cheap, high-powered telescopes, which Levitt and I pointed not at outer space but at everyday life. One thing they made visible was delayed causality, the everyday equivalent of very faint stars. As in all areas of science, when it became possible to study something for the first time, many popular beliefs turned out to be wrong. Attacks by real-estate agents on Levitt and Dubner are essentially the same as attacks by creationists on evolution. It was not freakish luck that Levitt’s research resembled mine; there was a reason for it.

Note: I wrote the preceding post a week ago, before seeing the article. I failed to realize a third reason to expect sympathetic treatment: Levitt himself had been written about.

Leave A Comment

Comments are moderated and generally will be posted if they are on-topic and not abusive.



View All Comments »
  1. Anonymous says:

    For an outstanding visual display self-collected behavioral data see Ben MacNeill’s Trixie Update.

    For instance, here’s his daughter’s Sleep Record for the last 14 days.

    Thumb up 0 Thumb down 0
  2. Anonymous says:

    I have been getting a kick out of the people who have been asking Seth to post the specifics of his weightloss plan.

    Come on people, the point of the article was on self-experimentation. Seth has pointed the way, let’s grab the self-experimentation ball and run. Try some sugar water solution, track your weight and see what works for you. Maybe we can get a webpage going of posted results.

    Thumb up 1 Thumb down 0
  3. notDebFrisch says:

    Yes anon above: we have everything we need: time, food, inspiration, & more body weight than we need. We needn’t understand “set point” or even believe drinking x amout of sugar water every day will lead to loss of appetite. It works or it doesn’t.
    I can disbelieve in the force of gravity all I want, it (the force) doesn’t care a whit, it’ll hold me to this freaky non conventional orb regardless.
    Thanks to Levitt & Dubner for bringing Seth Roberts into the fold. The premise of Freakonomics: unconventional wisdom works, is secondary to the caveat that we should question things, think for ourselves.
    [Whatever became of the combative Frisch?]

    Thumb up 1 Thumb down 0
  4. Bill Quick says:

    A few comments for Seth:

    1. The flavor concept explains a phenomenon I have long noticed, but not understood: on my rare vacations driving cross country, I always lose about a pound a day, even though I spend hours sitting on my rear, and eat nothing but highway junk food. Turns out, that is the only time I eat highway junk food. Hmmm…

    2. How long does the body “remember” flavor/calorie combinations? Is the initial imprint, once complete, permanent? If so, it might help to explain the problems of long-time dieters who have tried all sorts of diets: they have memorized so many flavor/calorie combinations that just about everything they eat is fattening for them.

    3. Do you think your work has any relevance to Dr. Ray Walford’s (UCLA) work and his Calorie Restriction Optimal Nutrition (CRON) diet, which is based on the much documented research showing that rats (and other test subjects) on extreme calorie restriction live much longer lives? Your picture indicates that you appear to be much younger than your chronological age, and the calorie intake you say is your current norm – 1200 cal/day – would put you squarely in the calorie restriction range necessary to demonstrate anti-aging or life extension phenomena. Have you considered looking into Walford’s research and checking to see if your vital signs are consistent with what he predicts for life-extension calorie restriction regimens?

    Just some thoughts. I’m following this discussion from my perch here in San Francisco, and from my blog, The Daily Pundit. I’m trying the mechanism you’ve identified: after digesting all your papers, it becomes clear that there are several possible approaches, and that neatness doesn’t necessarily count. My method is simple: Fill a two-liter pitcher with purified, distilled, or filtered water. Dump in one cup of white table sugar. Stir till dissolved. Drink half the pitcher at intervals during the first day, making sure not to drink within an hour of eating food. Repeat on day two, and then refill the pitcher.

    I’ve been on the diet for three days. I’ve lost seven pounds. I’ve not had a moment of hunger, nor have I experienced any odd cravings. I’ll keep at it for at least a month, to make sure there isn’t some expectation mechanism at work. I’ll monitor weight change versus sugar water/calorie intake to try to reach an optimum personal weight loss trajectory, and intend to publish the results on my blog, along with before and after photos (assuming the program is a success).

    Thumb up 0 Thumb down 0
  5. Anonymous says:

    Seven pounds in three days? You must be huge!

    Thumb up 0 Thumb down 0
  6. Glen says:

    This post has been removed by the author.

    Thumb up 0 Thumb down 0
  7. Glen says:

    Come on people, the point of the article was on self-experimentation.

    Yeah, but we need some specifics as a starting point. Case in point: me. I drank fructose water in an amount I thought was consistent with his experiments…and gained about 5 pounds over a month, which caused me to give up on the diet. I still think Seth is fooling himself as to the mechanism. Saying it has to do with adjusting a set point or modifying a pavlovian flavor/calorie connection makes it publishable in his field, but there are simpler explanations.

    There’s also the traditional selection bias issue common to psychology studies – the people for whom it doesn’t work (like me)stop doing it and drop out, so you’re left with a population for whom it “worked”. In a random sample of people whose weights are taking a random walk, if all the people who gain (or don’t lose)weight drop out, you’ll have “evidence” at the end of the study that whatever you were doing causes people to lose weight.

    Thumb up 0 Thumb down 0
  8. Anonymous says:

    Here is an article that relates to nutrition studies of just one individual:
    “Diet and Genes:
    It Isn’t Just What You Eat That Can Kill You, and It Isn’t Just Your DNA That Can Save You–It’s How They Interact.”

    By Anne Underwood and Jerry Adler

    The article points out how the developing field of nutritional genomics shows vast differentiation in the interaction of genes, food consumed, and a variety of reactions of the human body.

    Quote: “This [research] supports what we know from observation, that some individuals are better adapted than others to survive a morning commute past a dozen doughnut shops. Pima Indians in the Southwest get type 2 diabetes at eight times the rate of white Americans. Individuals have widely varying responses to high- or low-fat diets, wine, salt, even exercise.”

    So how do we know that sugar water as a diet aid will work for everyone? Maybe it’s only effective for a person with a particular genetic combo — and whether that combo is frequent or rare.

    Thumb up 0 Thumb down 0