Stanford prison experiment

Mark Palko points us to a review by Alison Abbott of a book by Susannah Cahalan telling a disturbing story of a psychology professor at a prestigious university who had stunning academic and popular success based on research that he seems to have incorrectly and misleadingly reported.

Disturbing—but not surprising, given we now have a template with many examples of “psychology professor at a prestigious university who had stunning academic and popular success based on research that he had incorrectly and misleadingly reported.”

#NotAllPsychologyProfessors

28 thoughts on “Stanford prison experiment

  1. I never understood the excitement that the pseudo patient study produced. It isn’t particularly hard to fake some symptoms that will lead to hospitalization. In that era, a good chest pain story would have landed you in the hospital for at least three days of observation. When I was a junior in medical school I remember a man who came to the local VA right after Christmas complaining of coughing blood. His chest X-ray had some areas of scarring, and we brought him in to exclude TB. The identification of TB bacillus included trying to grow it which took six weeks. His skin test was positive for TB exposure. I had worked in a TB lab and became the point man while the more senior house staff avoided possible exposure. One day he casually mentioned another VA hospital. I made some calls and found that he had been well treated previously, but he had developed a good story that with his X-ray and skin test had gotten him several six weeks stays in VA hospitals as an escape from Midwestern winters. Fake patients can fool real doctors. That’s not a big surprise, is it?

  2. I am unconvinced. A journalist fails to be able to find all of the participants in a study done about fifty years ago. That doesn’t mean there was anything wrong with the study. It could be just a case of poor record keeping.

  3. A search of this blog for “Zimbardo” turned up only one brief mention in 2015 and there are several (other) mentions of the Stanford Prison Experiment. Perhaps the following are too well-known to warrant citing:

    http://rationallyspeakingpodcast.org/show/rs-241-thibault-le-texier-on-debunking-the-stanford-prison-e.html

    is an extended interview with Thibault Le Texier, the author of a 2018 book (in French) on Philip Zimbardo’s Stanford Prison Experiment. The documentation on which the book is based comes from Zimbardo’s own records, which are now archived by Stanford. The author subsequently published

    Thibault Le Texier, “Debunking the Stanford prison experiment,” American Psychologist 74(7) (2019, August 8) 823–839 (with a rejoinder to a reply from Zimbardo at https://psyarxiv.com/9a2er/download?format=pdf).

    There’s also https://medium.com/s/trustissues/the-lifespan-of-a-lie-d869212b1f62 from 2018.

    [http://blogs.law.columbia.edu/critique1313/bernard-e-harcourt-the-dishonesty-of-thibault-le-texier/ criticizes Le Texier’s professional behavior concerning another matter and is not directly about Le Texier’s criticism of the Stanford experiment.]

    Google Scholar provides thirty citations to Le Texier’s American Psychologist paper. I’ve not yet read these. Among the citing works are:

    Tomasz Witkowski, Shaping Psychology Perspectives on Legacy, Controversy and the Future of the Field (Springer, 2020) https://link.springer.com/book/10.1007/978-3-030-50003-0#about

    which contains an interview with Carol Tavris (“Carol Tavris: Writing About Psychological Science and Skepticism: Tavris discusses the Stanford Prison Experiment, positive psychology, and the so-called crisis in psychology resulting in part from the shift from studies based on observing behavior to those based on self-reports.”)

    Augustine Brannigan, The Use and Misuse of the Experimental Method in Social Psychology A Critical Examination of Classical Research (Taylor & Francis, 2021)

    also discusses the Stanford Prison Experiment.

    [Tavris and Elliot Aronson, Mistkes Were Made (but not by me) 1st ed (Harcourt, 2007)

    accepts the inflated correlations of Gottman’s research on enduring relationships. I’ve not seen the two later editions.]

  4. To write a review of the review, I felt it just kind of stopped once it was getting to the good part. “Had Rosenhan invented them, she found herself asking?” Well, did he? The review gives no hint about how the book author went about pursuing this after her initial attempts, nor is it even clear from the review whether additional investigation was done. Instead, the reviewer tries to tempt us somehow, “She writes that she cannot be completely certain that Rosenhan cheated. But she is confident enough to call her engrossing, dismaying book The Great Pretender.”

    Why can’t she be sure? Why is she confident? Were there other things this guy clearly made up? What were the consequences for herself, for the field of psychiatry, etc.?

    I guess on the one hand, the review encourages me to read the book so I can learn the answers to these questions. But on the other hand, the review is so vague about what those answers might be that I’m ambivalent whether it will be worth reading the book.

    It’s like when scientific papers have question titles, “Does X influence Y in the context of Z?” If the title doesn’t know, why should I think the article has the answer?

  5. Now, this was a weak post I have to say. There are many problems in science but the study of the fake patients was so fundamental to transforming psychiatry and psychology (and showing how it is certainly not a science, they cannot measure anything, they lack reproducibility and love to blindly go around naming concepts made out of measurement error (call it grit, call it intelligence, call it personality, all error error error that cannot even be quantified)

    At least people are now waking up and the DSM V is no longer even useful for diagnosis! see for a start here https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4060802/

    THANK YOU David Rosenhan!

  6. Somewhat related: I just started reading Think Again: The Power of Knowing What You Don’t Know, by Adam Morgan. A “Number 1 New York Times bestseller”. Morgan describes himself in the book as the (I quote from memory) “top ranking professor in Wharton seven years in a row”, stuff like that. I just started reading the book, but it already begins by citing some decisive study (reference probably hidden somewhere in an end-note, not easy to get to in Kindle without messing up the reading order). The main message of the book seems pretty reasonable, and something I can get on board with: one should always re-examine one’s beliefs and assumptions, and be ready to change one’s mind. Don’t get committed to one idea. All great stuff, very Bayesian (Cromwell’s rule and all). But the way it’s all packaged (decisive psych studies all pointing in one direction; yeah right) makes me very skeptical of the book.

    Morgan also has a fancy podcast, with custom-made music and lots of assistants. I heard one of the episodes, on Burnout, and it has a not-so-clever product placement (a plug for SAP) embedded into the content of the espisode. A case study of how nicely SAP treats its employees, so please go work for them.

    It’s all so depressing to see these professors hyping up stuff, even when what they’re hyping up is a call to remain skeptical of one’s own position. Zimbardo also briefly appeared in the news some time ago, some new book about how young boys/men were not real men anymore because porn, something like that. I think Stanford desperately back-pedalled away from his claimed association with them once Zimbardo started doing the rounds on TV.

  7. The review is entertaining more because it highlights a story, and it’s definitely telling that some of the details from the Science paper don’t line up with what Cahalan finds in Rosenhan’s files. Yet this doesn’t really tell us much. Yeah, she found stuff that doesn’t line up, just as any reviewer might find if you carefully review the code of another paper but there’s not a lot convincing evidence to say that this happened on purpose or that the experiment is completely invalidated. It might be interesting to see where this ends up from Cahalan’s point of view.

  8. I think the title of this post and several commenters mix up the Stanford prison experiments by Philip Zimbardo with David Rosenhan’s fake mental hospital patients, which is the topic of the linked review. To be clear, the Zimbardo’s prison experiments also had issues, see https://gen.medium.com/the-lifespan-of-a-lie-d869212b1f62. And a write-up at Vox: https://www.vox.com/2018/6/13/17449118/stanford-prison-experiment-fraud-psychology-replication.

    Googling it now, I found that Zimbardo has written an extensive reply: https://www.prisonexp.org/response

    • Oystein:

      It wasn’t confusion on my part! I read the linked article and I know it doesn’t refer to the Stanford prison experiment. It refers to a different example of a psychology professor at that same prestigious university who had stunning academic and popular success based on research that he had incorrectly and misleadingly reported. That was kind of the point, that these things are so common that there were two such examples at Stanford alone. If you read my post carefully, you’ll see that I never claim that the linked study had anything to do with the prison experiment.

  9. So I guess my question is:

    have the social sciences – ex economics – accomplished anything, ever? Can we look back at the steady accumulation of knowledge in the social sciences the way we look back at the accumulation of knowledge about physics? No, not even close. Today’s social scientists are little more than alchemists, proposing various types of bloodletting to cure various purported ailments that frequently aren’t even real.

    • Jim:

      I have mixed feelings about your question.

      On one hand, yeah, I don’t think social science has accomplished anything along the lines of what has been done in physics (bombs, microchips, etc.), chemistry (more bombs, also lots of new materials), biology (the coronavirus vaccine), and other natural sciences. All I can think of that social science has given us in terms of real practical utility is Keynesian economics, and that remains controversial!

      On the other hand, I’m a social scientist for a reason. I love social science, and I think it’s important—even thought it has not given us any real practical advances, except for sports analytics, I guess. And opinion polls. I agree with George Gallup that it’s a good thing when politicians and the public have some sense of public opinion beyond what one might gather simply by listening to the loudest voices. That said, baseball analytics and public opinion, if not nothing, are a lot less than what the natural sciences have given us. So what about social science? I’d like to think that social science gives us understanding, and understanding is something we’re always looking for. In the absence of (somewhat) rigorous social science, we’ll search for understanding using messier tools, and at that point it helps to have some sense of issues of measurement and modeling. So maybe I’d say that social science, for all its flaws, is unavoidable.

      It’s worth pushing this discussion further, I think.

      • Well seems like bombs and microchips are an outcome of a more fundamental accomplishment, which is a rigorous theory of how the physical world works.

        In economics there are fundamental truths about pricing and supply and demand etc which Keynes built upon. I guess this is more what I’m looking for: the emergence of a structure that knowledge can be – and is being – built on. I’ve never really thought of it that way before but I think it’s the case that social sciences have basically no scaffold.

        Your comment on polling is interesting because it seems like that’s one of the few endeavors in social sciences where some rigorous framework has developed and adjustments are made for the varying representation of various groups. It does seem to be accepted in polling that these adjustments are required. I suppose that’s because of the discipline brought about by the reality of an election.
        And I guess as I think about it more this principle really is applied elsewhere, which is why the smoking gun for racial bias in policing is such a stinker to find.

        So I guess that’s a real accomplishment: there is a fundamental set of rules that are evolving.

        But I’d love to hear what other people have to say about the question.

    • So on its face, this question “have the social sciences – ex economics – accomplished anything, ever?” is ridiculous to me. All I have to do to answer it is provide one example of someone who is not an economist collecting a set of population statistics that led directly to some practical application. You don’t think anyone has ever done a paper that established a useful bus stop or something? I’m pretty sure I applied some social science at work yesterday. Did some K-means on human population centers for marketing targeting, stopped some pointless emails from going out. They paid me for it so I guess it’s a non-zero accomplishment. In 1994, research on the impact of same sex parents motivated the supreme court to rule that same-sex couples could adopt children.

      In no particular order:

      1. Many statistical methods were developed for the study of the social sciences. Almost all of what a data scientist does at a company today is using statistical methods to study a human population

      2. Political science is very effective at generating sophisticated techniques to subvert the spirit of Democracy. For example, you could consider the entire enterprise of gerrymandering to be an applied political science, and it only gets more efficient every year. Political science has also produced methods for detecting gerrymandering or generating fair districts, but I don’t know how much traction they’ve gotten, at least stateside.

      3. Applied social choice theory has created pretty good ranked choice voting systems with well-understood stability properties that are used in both elections outside the US and in non-state polling systems.

      4. If epidemiology and public health can be considered social sciences, then some triumphs are pretty obvious. Most loudly, robust contact tracing systems and well-organized public health authorities have allowed East Asian countries like Taiwan to actually have GDP growth in 2020.

      5. Psychology is often considered a social science. Many people pay for psychotherapy and swear by it, and psychophysics, though nowadays an unexciting field, has found very successful application in engineering, industrial design, and art.

      6. It is my understanding that you like standardized tests. Their existence is largely due to the social sciences.

      7. This is a real stretch, but if you consider political philosophy to be social science, as it once was, then the modern Democratic nation-state is a triumph of social science.

      > Can we look back at the steady accumulation of knowledge in the social sciences the way we look back at the accumulation of knowledge about physics? No, not even close.

      If you’re looking for universal rules that apply across a universal human context, then study something else. If you rule out specific study of specific populations and specific solutions to specific problems, then you’ve ruled out all useful applications of the social sciences.

      • “So on its face, this question “have the social sciences – ex economics – accomplished anything, ever?” is ridiculous to me.”

        I’m sure that successfully placing a bus stop can be done well enough without a social scientist. Own goal.

        I pointed to evolution of knowledge in physics as an example of what I’m looking for and, for the most part, I don’t see that in social sciences, or not nearly as long of an evolution. I don’t think it’s a ridiculous question at all. The average person wouldn’t have a clue how to answer it, while many people could mutter off at least a few historical leaps in physics.

        Some of your points are pretty good: demographics is obviously a social science and is doing pretty well. I overlooked that. Others are not so good. Lots of people swear by psychology and therapy. Lots of people also swear by homeopathic medicine. But there doesn’t appear to be a successful theory about psychology as a whole that is leading to breakthroughs, as there is about genetics. I’m vaguely aware that cognitive behavioral therapy is quite successful, but I don’t know if there is some fundamental foundation behind it.

        “If you’re looking for universal rules that apply across a universal human context, then study something else.”

        Why should I not expect that of social sciences? They have a different standard? As I wrote about this, though, I realized there are some breakthroughs. For example we now understand that economic growth leads to slowing population growth, so we don’t have to worry about The Population Bomb.

        Actually the more I think about it the more I see economics as becoming social sciences.

        • > I’m sure that successfully placing a bus stop can be done well enough without a social scientist. Own goal.

          If you were placing a bus stop, you really wouldn’t plot people’s transportation patterns on a map or collect some survey data to make sure it’s where it’s the most useful?

          > I pointed to evolution of knowledge in physics as an example of what I’m looking for and, for the most part, I don’t see that in social sciences, or not nearly as long of an evolution. I don’t think it’s a ridiculous question at all.

          That’s not the question I’m talking about. You asked, and I quoted,

          > have the social sciences – ex economics – accomplished anything, ever

          It’s a ridiculous question because the answer is obviously yes.

          > Actually the more I think about it the more I see economics as becoming social sciences.

          Please god no.

          > “If you’re looking for universal rules that apply across a universal human context, then study something else.”

          > Why should I not expect that of social sciences? They have a different standard?

          You want a universal causal model that predicts human behavior in every context, that can be transplanted across time and cultures? Then you necessarily have to restrict yourself to studying a small subset of human activity. Much of human behavior changes over time and differs across contexts—that’s a fact. Those things are worth studying anyways. Science is not defined by invariance properties of its theories or particular subjects, it’s a method.

        • Somebody:

          This is related to my point above that: (1) we use social science all the time, and when we’re not using formal “social science,” we’re using informal social science based on hunches, stories, unprocessed data, etc., and (2) social science is important but very difficult, and I agree with Jim that the things have have been learned in social science are much less impressive than what’s been learned in the naturals sciences.

        • I think much of what’s impressive about the natural sciences is when they are:

          1. Counterintuitive; things like radio and airfoils defy our expectations to the point of seeming magical
          2. Universal; Newton’s formulation in the 17th century still works today
          3. Precise; even where error is large the magnitude of error is quantifiable

          I do think the first part is a big part of why the natural sciences are so cool—South Korean test and trace is an enormous achievement, but doesn’t seem so magical to me really. I particularly think counterintuitive phenomena is hard to come by in human affairs since we evolve with an intuition for human behavior and continuously revise that intuition as we experience life. The only example of a persistently counterintuitive social phenomenon I can think of is Ricardo’s comparative advantage.

        • This is very good post. I’m gonna steal those three bullet points, and cite somebody on the Gelman blog.

        • “It’s a ridiculous question because the answer is obviously yes.”

          OK. I guess you interpreted it literally. In that sense it’s obviously a ridiculous question but I think given the context of my reference to the history of accumulation of physical knowledge you might have seen beyond the literal words.

          “If you were placing a bus stop, you really wouldn’t plot people’s transportation patterns on a map or collect some survey data to make sure it’s where it’s the most useful?”

          Plotting transportation patterns is “social science”?!!! :) Shocking, but somehow that seems to fit the pattern of social science giving itself credit for common knowledge.

          “You want a universal causal model that predicts human behavior in every context”

          I want a method that leads to general principles. Yes I want one that extends across cultures. Physics extends across materials and planetary bodies, social science should seek general principles that extend across cultures. Yes, creating increasingly broad generalizations from specific observations is an important goal of science.

        • > Plotting transportation patterns is “social science”?!!! :) Shocking, but somehow that seems to fit the pattern of social science giving itself credit for common knowledge.

          > I want a method that leads to general principles. Yes I want one that extends across cultures. Physics extends across materials and planetary bodies, social science should seek general principles that extend across cultures.

          Science as it is typically discussed here refers to a method of inquiry, not a set of facts and laws with attractive invariance properties. It’s something you do, not something you read and know or a group of people with particular titles and degrees. Plotting transportation patterns and analyzing likert scaled survey data is “social science,” even when it’s done by somebody like me who has a degree in physical science and has never read a social science textbook.

          If you prefer another definition, so be it, but the methodological one is typically what we use.

  10. Usually, reviewers of non-fiction give a pretty good idea of what shockers are to be found in the book. This review doesn’t just bury the lede but leaves it out completely. Or maybe there is no lede. Either that, or it’s like the way every kid in my fourth grade class finished their oral book report: “If you want to find out what happens, read the book?”

    I am curious though as to what Rosenhan got wrong about mental hospitals – I mean besides whether one of his pseudopatients was in for eight days rather than seven.

Leave a Reply to jim Cancel reply

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