5 ms·

How Priming Research Went off the Rails (2017)

34 points by scott_s 6y ago

camjohnson26 6y ago
What’s frustrating about this situation is how obviously wrong the experiments were as described in the book, and yet how unshakably the findings were defended by Kahneman and others. I read the book before the crisis and stopped reading after the priming chapter, it was clearly fantasy, especially the experiment about reading words related to old age and then walking more slowly down a hallway. Now that the crisis is public knowledge it’s hard to believe these findings were ever accepted by the mainstream, but they were. You just have to think for yourself and weigh the evidence, be humble and open to other perspectives. Mindlessly following perceived experts won’t save you.
- 0xddd 6y ago
  It seems to me that people put too much faith in statistics despite not having a fundamental understanding of the methods. I'm sure in plenty of cases these frivolous results get published due to academic fraud and the pressure of publish or perish. But in other cases, it seems like researchers really think "p<.05" translates to "your alternative hypothesis is irrefutably true" and will believe their flimsy conclusions because, as far as they're concerned, numbers don't lie. (If I remember correctly, when Brian Wansink was exposed for p-hacking, it was partly because he wrote a blog post openly describing his sketchy research methods, where he seemed totally unaware that he was churning out illegitimate results.)
  - anitil 6y ago
    I've been trying to find this blog post, any chance you can remember more of it to help my googling?
    - 0xddd 6y ago
      I managed to dig it up: https://web.archive.org/web/20170312041524/http:/www.brianwansink.com/phd-advice/the-grad-student-who-never-said-no https://web.archive.org/web/20170312041524/http:/www.brianwa.... It really is incredible to read Wansink's clueless non-rebuttals to the comments that point out his academic fraud.
      - lowdose 6y ago
        Staggering replies from him on the negative comments on his methods and thinking, what a character. What a gem to read, absurt funny title "The Grad Student Who Never Said "No". And the fact that at least 54 times people took a snapshot of the webpage. We should print this on toilet paper and sell it to universities.
      - anitil 6y ago
        Thankyou!
- wyattpeak 6y ago
  On what, at the time, did you base the assumption that people walking slowly after reading words related to old age was clearly fantasy? It doesn't sound likely to me, but neither do a bunch of other cognitive results (the gorilla running through the basketball court comes to mind). I'm all for questioning and rerunning tests, but dismissing them because they seem clearly fanciful without data just brings us back to pre-science. Science which is only allowed to discover plausible-sounding things isn't science at all.
  - aabhay 6y ago
    The burden of proof is on science, not me. I’m not going to follow every conjecture and hypothesis floated by a scientist. If something sounds fanciful and violates my common sense, I need a robust experiment. Accepting results without that just brings us back to pre-science.
    - jbay808 6y ago
      Sounds like you're properly applying priors! There's an interesting chapter in Jaynes' Probability Theory: the Logic of Science where he talks about how much evidence it would take to convince him that ESP is real. At the time that book was written, there were experiments supporting it that failed to replicate. Basically you're totally allowed to read a scientific study, say "okay, that's an interesting result", and then still not believe the conclusion if you have enough reason to doubt the premise a-priori.
c3534l 6y ago
It's really weird when you hear people saying things like: > disbelief is not an option. The results are not made up, nor are they statistical flukes. You have no choice but to accept that the major conclusions of these studies are true. It's as if they know they're wrong, but are trying to force themselves to believe it any way.
- 0xddd 6y ago
  I think that is quite harsh on Kahneman, especially considering how reasonable he sounds in the open letter that is linked. But I am also completely perplexed as to how he was so confident about these priming studies given the research on cognitive biases that he's so well-known for (as mentioned in the article).
  - aabhay 6y ago
    A great deal of his fame is premised on this priming effect. It’s been taken almost as gospel for decades, to the point that a typical applied psychology paper will actually include a priming step even if unnecessary just to spruce up the methodology. E.g. “subjects were primed by having them read articles on spicy food” for a study on water fountains, etc.
    - 0xddd 6y ago
      Oh, wow. Is any of his research still taken seriously? For whatever reason, I was under the impression his work on researcher bias like "the law of small numbers" was quite influential and still cited here and there by trustworthy figures.
Vinnl 6y ago
The selection effect is also easily explained. 'Top' journals have as explicit policy not to publish the most robust research, but the research they think will have most impact in the field. (Hence 'impact factor'.) That's why with the project I volunteer for (Plaudit.pub) researchers recommending research have to explicitly indicate whether they think a work is robust and/or exciting, where neither block publication nor the recommendation.
thu2111 6y ago
Something I've been reflecting on over the past couple of months is how the university system loudly claims to promote critical thinking and peer review, whilst it seems to me that it actually does everything possible to discourage it. If you look at the unfolding replication disaster in epidemiology, re: Ferguson and his totally non-replicable Report 9, it's really turning over the rocks on an academic culture in which: • Academics flat-out refuse to question each other across disciplines, practically as a matter of honour. • Nobody else being able to understand your research is something boasted about on Twitter and in the press rather than being a source of embarrassment. • Scientists routinely refuse to reveal their data and code, making a mockery of both peer review and replication as a process. • Anyone who does engage in critical thinking about their work gets shut down, censored, attacked as an "armchair epidemiologist" etc - often it seems by people with academic backgrounds. After years of non-stop hiring, tech firms are now flooded with naive post-docs steeped in academic culture, who appear to have totally bought into the notion that critical thinking about the output of "experts" is not only morally wrong but outright dangerous and needs to be suppressed. This has led to the absurd sight of Google/Twitter trying to delete anyone who disagrees with the WHO at a time when the WHO routinely disagrees with itself as of a few weeks ago, or where the WHO disgrees with the CDC, or where experienced doctors disagree with both, and it ends up being the doctors that get shut down because they are mere doctors and not "experts"! It feels like this whole conflation of academia with "expertise" is leading society to disaster; why should any of us believe universities when they say they teach critical thinking? Someone trained in critical thinking wouldn't just accept such a claim at face value, they'd demand a weight of evidence. But no such evidence is ever presented.
- pjc50 6y ago
  This situation has been exacerbated by the "barbarians at the gates": stepping outside the zone of people who are trying to do inquiry honestly, however badly, leads rapidly into the swirling internet drain of quacks and cranks who feel under no such obligation to try. Hence all the "miracle mineral solution" nonsense. The landmark of bad epidemiology is of course Andrew Wakefield, who almost single-handedly destroyed the effectiveness of measles vaccination. Experts are in trouble. Some of the "non-experts" are right. But as often as not it's the confidently extremely wrong who get promoted by the media and social networks.
jimmySixDOF 6y ago
Big shout out to the "Reproducibility Project: Psychology" [1] by Brian Nosek and the Center for Open Science in 2015 that started this trend to shine light on "establishment" "conclusions" in the social sciences. People are an awfully tricky subject of study. [1] https://en.m.wikipedia.org/wiki/Reproducibility_Project https://en.m.wikipedia.org/wiki/Reproducibility_Project

34 points by scott_s 6y ago

What’s frustrating about this situation is how obviously wrong the experiments were as described in the book, and yet how unshakably the findings were defended by Kahneman and others. I read the book before the crisis and stopped reading after the priming chapter, it was clearly fantasy, especially the experiment about reading words related to old age and then walking more slowly down a hallway. Now that the crisis is public knowledge it’s hard to believe these findings were ever accepted by the mainstream, but they were. You just have to think for yourself and weigh the evidence, be humble and open to other perspectives. Mindlessly following perceived experts won’t save you.

0xddd 6y ago
It seems to me that people put too much faith in statistics despite not having a fundamental understanding of the methods. I'm sure in plenty of cases these frivolous results get published due to academic fraud and the pressure of publish or perish. But in other cases, it seems like researchers really think "p<.05" translates to "your alternative hypothesis is irrefutably true" and will believe their flimsy conclusions because, as far as they're concerned, numbers don't lie. (If I remember correctly, when Brian Wansink was exposed for p-hacking, it was partly because he wrote a blog post openly describing his sketchy research methods, where he seemed totally unaware that he was churning out illegitimate results.)
- anitil 6y ago
  I've been trying to find this blog post, any chance you can remember more of it to help my googling?
  - 0xddd 6y ago
    I managed to dig it up: https://web.archive.org/web/20170312041524/http:/www.brianwansink.com/phd-advice/the-grad-student-who-never-said-no https://web.archive.org/web/20170312041524/http:/www.brianwa.... It really is incredible to read Wansink's clueless non-rebuttals to the comments that point out his academic fraud.
    - lowdose 6y ago
      Staggering replies from him on the negative comments on his methods and thinking, what a character. What a gem to read, absurt funny title "The Grad Student Who Never Said "No". And the fact that at least 54 times people took a snapshot of the webpage. We should print this on toilet paper and sell it to universities.
    - anitil 6y ago
      Thankyou!
wyattpeak 6y ago
On what, at the time, did you base the assumption that people walking slowly after reading words related to old age was clearly fantasy? It doesn't sound likely to me, but neither do a bunch of other cognitive results (the gorilla running through the basketball court comes to mind). I'm all for questioning and rerunning tests, but dismissing them because they seem clearly fanciful without data just brings us back to pre-science. Science which is only allowed to discover plausible-sounding things isn't science at all.
- aabhay 6y ago
  The burden of proof is on science, not me. I’m not going to follow every conjecture and hypothesis floated by a scientist. If something sounds fanciful and violates my common sense, I need a robust experiment. Accepting results without that just brings us back to pre-science.
  - jbay808 6y ago
    Sounds like you're properly applying priors! There's an interesting chapter in Jaynes' Probability Theory: the Logic of Science where he talks about how much evidence it would take to convince him that ESP is real. At the time that book was written, there were experiments supporting it that failed to replicate. Basically you're totally allowed to read a scientific study, say "okay, that's an interesting result", and then still not believe the conclusion if you have enough reason to doubt the premise a-priori.

c3534l 6y ago

It's really weird when you hear people saying things like: > disbelief is not an option. The results are not made up, nor are they statistical flukes. You have no choice but to accept that the major conclusions of these studies are true. It's as if they know they're wrong, but are trying to force themselves to believe it any way.

0xddd 6y ago
I think that is quite harsh on Kahneman, especially considering how reasonable he sounds in the open letter that is linked. But I am also completely perplexed as to how he was so confident about these priming studies given the research on cognitive biases that he's so well-known for (as mentioned in the article).
- aabhay 6y ago
  A great deal of his fame is premised on this priming effect. It’s been taken almost as gospel for decades, to the point that a typical applied psychology paper will actually include a priming step even if unnecessary just to spruce up the methodology. E.g. “subjects were primed by having them read articles on spicy food” for a study on water fountains, etc.
  - 0xddd 6y ago
    Oh, wow. Is any of his research still taken seriously? For whatever reason, I was under the impression his work on researcher bias like "the law of small numbers" was quite influential and still cited here and there by trustworthy figures.

Vinnl 6y ago

The selection effect is also easily explained. 'Top' journals have as explicit policy not to publish the most robust research, but the research they think will have most impact in the field. (Hence 'impact factor'.) That's why with the project I volunteer for (Plaudit.pub) researchers recommending research have to explicitly indicate whether they think a work is robust and/or exciting, where neither block publication nor the recommendation.

thu2111 6y ago

Something I've been reflecting on over the past couple of months is how the university system loudly claims to promote critical thinking and peer review, whilst it seems to me that it actually does everything possible to discourage it. If you look at the unfolding replication disaster in epidemiology, re: Ferguson and his totally non-replicable Report 9, it's really turning over the rocks on an academic culture in which: • Academics flat-out refuse to question each other across disciplines, practically as a matter of honour. • Nobody else being able to understand your research is something boasted about on Twitter and in the press rather than being a source of embarrassment. • Scientists routinely refuse to reveal their data and code, making a mockery of both peer review and replication as a process. • Anyone who does engage in critical thinking about their work gets shut down, censored, attacked as an "armchair epidemiologist" etc - often it seems by people with academic backgrounds. After years of non-stop hiring, tech firms are now flooded with naive post-docs steeped in academic culture, who appear to have totally bought into the notion that critical thinking about the output of "experts" is not only morally wrong but outright dangerous and needs to be suppressed. This has led to the absurd sight of Google/Twitter trying to delete anyone who disagrees with the WHO at a time when the WHO routinely disagrees with itself as of a few weeks ago, or where the WHO disgrees with the CDC, or where experienced doctors disagree with both, and it ends up being the doctors that get shut down because they are mere doctors and not "experts"! It feels like this whole conflation of academia with "expertise" is leading society to disaster; why should any of us believe universities when they say they teach critical thinking? Someone trained in critical thinking wouldn't just accept such a claim at face value, they'd demand a weight of evidence. But no such evidence is ever presented.

pjc50 6y ago
This situation has been exacerbated by the "barbarians at the gates": stepping outside the zone of people who are trying to do inquiry honestly, however badly, leads rapidly into the swirling internet drain of quacks and cranks who feel under no such obligation to try. Hence all the "miracle mineral solution" nonsense. The landmark of bad epidemiology is of course Andrew Wakefield, who almost single-handedly destroyed the effectiveness of measles vaccination. Experts are in trouble. Some of the "non-experts" are right. But as often as not it's the confidently extremely wrong who get promoted by the media and social networks.

jimmySixDOF 6y ago

Big shout out to the "Reproducibility Project: Psychology" [1] by Brian Nosek and the Center for Open Science in 2015 that started this trend to shine light on "establishment" "conclusions" in the social sciences. People are an awfully tricky subject of study. [1] https://en.m.wikipedia.org/wiki/Reproducibility_Project https://en.m.wikipedia.org/wiki/Reproducibility_Project