5 ms·

Eugenics Shaped Statistics

22 points by NN88 4y ago

dekhn 4y ago
I've noticed a bunch of people have tried to update the basics of stats by using this argument. No.
- JHonaker 4y ago
  I mean the discussion of the reasons and motivations for the development of many classical techniques actually does lend credence to the argument that they are misleading (by design). That said, I find the technical arguments against NHST to be just as necessary (and I also find them just as convincing). I'm curious what your actual objections are to this?
  - dekhn 4y ago
    I don't know what the official term for it is, but because I do not believe the nature of abstract statistics is specifically bound (other than historically) with eugenics (IE, the founders of statistics were also eugenecists, but I don't think they embedded the principles of eugenics into statistics). Instead, the argument should be entirely about the abstract details of statistics. Other historical articles can bring up (and attempt to make the case for) the linkage between the two concepts. Eugenics is like a strange attractor for scientists: over and over again, smart quantitative people who learn about genetics, selective breeding, and human phenotypes all end up converging on the idea of using selective breeding humans to "improve the phenotype" in their biased and preconceived notion of what "improve" and "phenotype" mean. It's not at all surprising that Galton (one of my personal heros for his statistical work; I made a Galton board) and Fisher (not one of my favorites) both explored these ideas alongside their scientific efforts.
    - JHonaker 4y ago
      I don't think using frequentist methods makes anyone a eugenicist by transitivity, but I do think it's important to understand the motivations behind the development of notoriously confusing, misapplied methods. If you consider the context, I think the author's position that the confusing aspects of it are very likely because of the desire to (falsely) profess that we have removed all aspects of judgemental decisions from a particular procedure which was driven by the desire to convince people that it wasn't their opinion, but fact, that they were genetically superior beings.
JHonaker 4y ago
I'm a statistician, and I have to say, I agree with the fundamental premise of the author: The problems with null hypothesis significant testing (NHST) are a direct result of two things: 1. the false air of objectivity it provides 2. the lack of consideration of the relative likelihood of the data under any alternative I also agree that Pearson, Fisher, and Neyman knew what they were doing when they appealed to these methods' "objectivity" to support their terrible views. I also find it just absolutely hilarious that Fisher's adamant opposition to Bayesian inference was the (in his opinion) absurdity of the uniform prior distribution, and yet one of his most famous contributions (Maximum likelihood estimation) is essentially the same result you'd get from the Bayesian estimate of the most likely parameter with a uniform prior. If anyone is interested in the history of this stuff along with a discussion of the philosophical meaning of probability from the same author, his book Bernoulli's Fallacy is a pretty good read. My only qualm with it is the lack of much exposure on how to actually do Bayesian inference or a discussion on what it's like to do it in a modern setting. There's a brief few pages at the very end that mention things like Markov Chain Monte Carlo and the fact that most modern Bayesian inference is necessarily numerical approximations to the posterior, not the analytical solution (but that's not necessarily a bad thing nor is it an escape-able problem). If you want a book that makes many of the same philosophical arguments about probability but also backs it up with math and descriptions on how to do Bayesian inference, Edwin Jaynes's Probability Theory: The Logic of Science is for you. It's quite literally a modern cult classic. People that read this book tend to go out and evangelize for it. My final comment is that there is a place for frequentist methods, even if you're a modern Bayesian. The theory of Markov Chain Monte Carlo (one of the key methods in making modern Bayesian inference work) is exactly the domain of frequentist probability: Long runs of repeated things. Many of the diagnostic tools for assessing whether the inferences of MCMC results can be trusted are hinged on results from the frequentist interpretation and are steeped in their strange-ness. For instance, the lack of an issue is taken lack of evidence of an issue rather than evidence of no issue. The interconnected-ness of the two is unfortunately missing from the author's work (that I have seen).
- dekhn 4y ago
  I worked with a guy who is a legit science genius and he read Efron (not Jaynes) and it changed how he thought about statistics, specifically how bayesian bootstrap is a very simple and easy way to get more utility from your data set. https://www.ams.org/journals/bull/2013-50-01/S0273-0979-2012-01374-5/S0273-0979-2012-01374-5.pdf https://www.ams.org/journals/bull/2013-50-01/S0273-0979-2012... Here's something he publisehd on it https://threeplusone.com/pubs/Crooks2005a.pdf https://threeplusone.com/pubs/Crooks2005a.pdf https://threeplusone.com/pubs/Price2005.pdf https://threeplusone.com/pubs/Price2005.pdf
wrp 4y ago
It seems that his real target is significance testing and he is trying to cast it in a bad light by associating it with eugenics. I started skimming after about a third, so I may have missed a critical point, but I don't see a valid argument there. I'm not a fan of eugenics or significant testing, but they are orthogonal issues.
- JHonaker 4y ago
  I read this article when it originally came out, and I think it's mostly a historical discussion. But, I can summarize the "leap of logic" that significance testing makes for you. If T(X) is a statistic designed to measure a quantity A, then significance testing are statements that say P(T(X) is close to A | A) = P In other words, the statistic T is close to the true value of A (which is fixed) with probability P. So far, I think this is pretty straightforward, but the tricky part comes when you apply the idea that "closeness" is symmetric. In the physical world, if A is close to B then B is close to A, but that isn't necessarily a valid thing to infer based on a significance test. The thing ostensibly implied by a hypothesis test is that the these two are reasonable to exchange: P(T(X) is close to A | A) = P (A is close to T(X) | T(X)) The interpretation of the right side is what you really want. The probability that the thing I want to know about is close to my statistic given my statistic. Unfortunately, this is not true in general. It's this misunderstanding that's at the heart of "logical fallacy" of frequentist hypothesis testing. No matter how careful we statisticians are about p-values, empirical evidence has shown that this leap is what people make time and time again (including us). This is to say nothing of the absurdity of interpreting probability as a long run frequency for events that happen once or relying on results that formulate things that way when you're doing iterative analysis and development (as all actual inferential processes are).
  - wrp 4y ago
    I think we agree on NHST. I expend a lot of effort to remind students (and colleagues) that the popular interpretation of significance tests gets the direction of implication backwards.
    - 4y ago
zen_1 4y ago
foreigner 4y ago
Couldn't statisticians apply some sort of anonymization to the data before analyzing it, similar to the way medical experiments are carried out with "double-blind" studies?
- boxed 4y ago
  Yea. In fact they do. For example in double blind studies! And if they do look at the data and try to fit their preconceived idea then that is P-hacking, which is statistical fraud.

22 points by NN88 4y ago

dekhn 4y ago

I've noticed a bunch of people have tried to update the basics of stats by using this argument. No.

JHonaker 4y ago
I mean the discussion of the reasons and motivations for the development of many classical techniques actually does lend credence to the argument that they are misleading (by design). That said, I find the technical arguments against NHST to be just as necessary (and I also find them just as convincing). I'm curious what your actual objections are to this?
- dekhn 4y ago
  I don't know what the official term for it is, but because I do not believe the nature of abstract statistics is specifically bound (other than historically) with eugenics (IE, the founders of statistics were also eugenecists, but I don't think they embedded the principles of eugenics into statistics). Instead, the argument should be entirely about the abstract details of statistics. Other historical articles can bring up (and attempt to make the case for) the linkage between the two concepts. Eugenics is like a strange attractor for scientists: over and over again, smart quantitative people who learn about genetics, selective breeding, and human phenotypes all end up converging on the idea of using selective breeding humans to "improve the phenotype" in their biased and preconceived notion of what "improve" and "phenotype" mean. It's not at all surprising that Galton (one of my personal heros for his statistical work; I made a Galton board) and Fisher (not one of my favorites) both explored these ideas alongside their scientific efforts.
  - JHonaker 4y ago
    I don't think using frequentist methods makes anyone a eugenicist by transitivity, but I do think it's important to understand the motivations behind the development of notoriously confusing, misapplied methods. If you consider the context, I think the author's position that the confusing aspects of it are very likely because of the desire to (falsely) profess that we have removed all aspects of judgemental decisions from a particular procedure which was driven by the desire to convince people that it wasn't their opinion, but fact, that they were genetically superior beings.

JHonaker 4y ago

I'm a statistician, and I have to say, I agree with the fundamental premise of the author: The problems with null hypothesis significant testing (NHST) are a direct result of two things: 1. the false air of objectivity it provides 2. the lack of consideration of the relative likelihood of the data under any alternative I also agree that Pearson, Fisher, and Neyman knew what they were doing when they appealed to these methods' "objectivity" to support their terrible views. I also find it just absolutely hilarious that Fisher's adamant opposition to Bayesian inference was the (in his opinion) absurdity of the uniform prior distribution, and yet one of his most famous contributions (Maximum likelihood estimation) is essentially the same result you'd get from the Bayesian estimate of the most likely parameter with a uniform prior. If anyone is interested in the history of this stuff along with a discussion of the philosophical meaning of probability from the same author, his book Bernoulli's Fallacy is a pretty good read. My only qualm with it is the lack of much exposure on how to actually do Bayesian inference or a discussion on what it's like to do it in a modern setting. There's a brief few pages at the very end that mention things like Markov Chain Monte Carlo and the fact that most modern Bayesian inference is necessarily numerical approximations to the posterior, not the analytical solution (but that's not necessarily a bad thing nor is it an escape-able problem). If you want a book that makes many of the same philosophical arguments about probability but also backs it up with math and descriptions on how to do Bayesian inference, Edwin Jaynes's Probability Theory: The Logic of Science is for you. It's quite literally a modern cult classic. People that read this book tend to go out and evangelize for it. My final comment is that there is a place for frequentist methods, even if you're a modern Bayesian. The theory of Markov Chain Monte Carlo (one of the key methods in making modern Bayesian inference work) is exactly the domain of frequentist probability: Long runs of repeated things. Many of the diagnostic tools for assessing whether the inferences of MCMC results can be trusted are hinged on results from the frequentist interpretation and are steeped in their strange-ness. For instance, the lack of an issue is taken lack of evidence of an issue rather than evidence of no issue. The interconnected-ness of the two is unfortunately missing from the author's work (that I have seen).

dekhn 4y ago
I worked with a guy who is a legit science genius and he read Efron (not Jaynes) and it changed how he thought about statistics, specifically how bayesian bootstrap is a very simple and easy way to get more utility from your data set. https://www.ams.org/journals/bull/2013-50-01/S0273-0979-2012-01374-5/S0273-0979-2012-01374-5.pdf https://www.ams.org/journals/bull/2013-50-01/S0273-0979-2012... Here's something he publisehd on it https://threeplusone.com/pubs/Crooks2005a.pdf https://threeplusone.com/pubs/Crooks2005a.pdf https://threeplusone.com/pubs/Price2005.pdf https://threeplusone.com/pubs/Price2005.pdf

wrp 4y ago

It seems that his real target is significance testing and he is trying to cast it in a bad light by associating it with eugenics. I started skimming after about a third, so I may have missed a critical point, but I don't see a valid argument there. I'm not a fan of eugenics or significant testing, but they are orthogonal issues.

JHonaker 4y ago
I read this article when it originally came out, and I think it's mostly a historical discussion. But, I can summarize the "leap of logic" that significance testing makes for you. If T(X) is a statistic designed to measure a quantity A, then significance testing are statements that say P(T(X) is close to A | A) = P In other words, the statistic T is close to the true value of A (which is fixed) with probability P. So far, I think this is pretty straightforward, but the tricky part comes when you apply the idea that "closeness" is symmetric. In the physical world, if A is close to B then B is close to A, but that isn't necessarily a valid thing to infer based on a significance test. The thing ostensibly implied by a hypothesis test is that the these two are reasonable to exchange: P(T(X) is close to A | A) = P (A is close to T(X) | T(X)) The interpretation of the right side is what you really want. The probability that the thing I want to know about is close to my statistic given my statistic. Unfortunately, this is not true in general. It's this misunderstanding that's at the heart of "logical fallacy" of frequentist hypothesis testing. No matter how careful we statisticians are about p-values, empirical evidence has shown that this leap is what people make time and time again (including us). This is to say nothing of the absurdity of interpreting probability as a long run frequency for events that happen once or relying on results that formulate things that way when you're doing iterative analysis and development (as all actual inferential processes are).
- wrp 4y ago
  I think we agree on NHST. I expend a lot of effort to remind students (and colleagues) that the popular interpretation of significance tests gets the direction of implication backwards.
  - 4y ago

zen_1 4y ago

foreigner 4y ago

Couldn't statisticians apply some sort of anonymization to the data before analyzing it, similar to the way medical experiments are carried out with "double-blind" studies?

boxed 4y ago
Yea. In fact they do. For example in double blind studies! And if they do look at the data and try to fit their preconceived idea then that is P-hacking, which is statistical fraud.