Harvard got conned.

Oh no! The Harvard Data Science Institute did a podcast with “Drs. Julia and John Gottman from The Gottman Institute about the secrets of long-lasting love.”

Spoiler: Julia and John’s secret is that they train their evaluation on the same data they use to measure their method’s success. Unfortunately it does not seem that the Harvard Data Science Institute revealed this secret in their podcast. It’s a disappointment that Harvard got conned in this way. Not as embarrassing as that Jesus’s wife thing, though!

I suggest that, for a followup, Harvard can interview Laurie Abraham, the journalist who skeptically looked into some of the Gottmans’ claims a few years back.

Similarly, Alexey Guzey informs us that the author of Why We Sleep has a new podcast. Dude should interview Guzey for his next episode. Also Ted should get on the horn to Guzey and get him to speak too. Could be fun, no? It’s so frustrating when people duck legitimate criticism.

7 thoughts on “Harvard got conned.

  1. Hey! Let’s start a new trend!

    To heck with the science podcast. It’s time for the Anti-Science Podcast!

    Alex Guzey and Walker can debate that traffic data linked to a few days back, taking note of the strong relationship of TOD of fatalities to the diurnal sleep patterns practiced by humans!!! We can all look at the data (which shows fatalities peak at 4-12pm), then Guzey can present his N=1 study of himself proving sleep doesn’t matter and Walker can present a chart with that’s missing the 4p-12p bins, showing fatalities are most common on Monday morning!

    Which anti-science explanation is most incorrect? Listeners can chime in right here on the blog!

    • Anon:

      It’s wrong to lump Guzey with Walker. Guzey makes speculations based on his readings and life experiences, and these speculations are entirely transparent. Walker makes speculations based on nonexisting data, misreports studies, etc. No comparison at all. Guzey’s speculations are not anti-science at all. Theorizing, including theorizing with weak data, is part of science. Misrepresenting the literature, that’s anti-science.

      • (Another anonymous).

        I’m not convinced!

        It’s a bit like what you said about super-vague priors earlier on this blog: it’s sort of information to assume that some treatment could have an enormous effect. But here the issue’s the other way round!

        That Guzey (if I remember his position correctly) throws away/ignores so much data and anecdotes about the very real effects of sleep deprivation is kind of like that… I mean, really, do we have to pretend that we don’t know anything about how sleep (or lack of it) affects people? His N=1 study should be overwhelmed by prior KNOWLEDGE about the subject matter, but somehow that prior doesn’t seem to be there, or it’s unrealistically vague.

        I don’t think that’s really “scientific”, but is it ANTI-science? I dunno, I don’t think that question really matters. It’s all too muddy.

        One could argue – like I did above – that Guzey is indeed misrepresenting or willfully ignoring lot of known things about sleep deprivation; on the other hand one could take the more optimistic view and say that he just has weak data which he is sincerely presenting and that’s part of science, sometimes the data is weak.

        But when you get into his evolutionary arguments, that – to me – starts slipping into at least pseudoscience. Maybe even that isn’t anti-science, but pseudoscience, yes, I could use that word for that.

        • Yet:

          Interesting discussion. There are two questions here. First, when Guzey offers theories based entirely on speculations, introspection, anecdotes, and a very partial reading of the literature, where does this stand on the spectrum of science vs. pseudoscience or antiscience? Second, is Guzey misrepresenting his experiences or willfully misinterpreting the literature?

          Let me consider these in turn.

          1. I do think that theorizing—even when based only on speculations, introspection, anecdotes, and a very partial reading of the literature—is still science, or at least it’s part of science.

          2. I have no reason to think that Guzey is intentionally misrepresenting anything. Nonetheless, it’s possible to fool oneself! I remember my friend Seth Roberts, who went off the deep end with his theories based on self-experimentation. And I agree that, at some point, theorizing unmoored by careful experimentation can become pseudoscience.

          So, yeah, it’s complicated. Walker still seems like in a different category, in that he’s misrepresenting his sources and presenting speculation as scientific fact. Guzey is openly presenting his speculation as speculation, which is a plus.

        • Andrew, I’ve been pondering this during the day, and a lot of what I’ve been thinking circles back to Guzey’s intentions and what he knows or can be assumed to know.

          For example, you say that you have no reason to think that Guzey is intentionally representing anything. That could be the case. But I find the negative effects of sleep deprivation to be so self-evident and at least anecdotally well-known that to me it seems that ignoring this prior information must be intentional.

          But of course, maybe somehow Guzey hasn’t been exposed to this massive amount of data?

          Or maybe someone is now saying that anecdotes don’t equal data. I disagree. Especially when it comes to psychology etc. anecdotes are invaluable, they might be the only reasonable source of information. Otherwise me might slip into scientism. (Curious side point: firefox wants to correct scientism into antiscience!).

          A while ago on your blog you discussed a study that purportedly “showed scientifically” that bisexuals exist by measuring (if I remember correctly) volume of erectile tissue in response to erotic stimuli . That was an example of scientism, trying to make something measurable just for measurements sake, whilst ignoring our everyday knowledge of the subject.

          These kinds of sleep studies, in which Guzey tries to measure effects of sleep deprivation by playing video games or doing math problems – in my opinion – fall into the same category.

  2. Really, Harvard got conned? All University faculty, staff and students were deceived into believing Gottman’s claims about factors that predict long lasting romantic relationships based on the fact that the data science institute aired a podcast on the topic? This is such an absurd click bait title that is ridiculous on its face. It’s really too bad to see you go down this path.

    • Sentinel:

      That’s right, all of Harvard didn’t get conned. The Harvard Data Science Review got conned. But, y’know, reputations go both ways. The Harvard Data Science Review (which is pretty good; I’ve published two papers there myself!) gets some of its reputation from the Harvard name. It’s only fair that when the HDSR screws up, Harvard gets some of the reputation hit.

Leave a Reply

Your email address will not be published. Required fields are marked *