SoylentNews is people
SoylentNews
Paul Meehl is responsible for what is probably the most apt explanation for why some areas of science have made more progress than others over the last 70 years or so. Amazingly, he pointed this out in 1967 and it had seemingly no effect on standard practices:
Because physical theories typically predict numerical values, an improvement in ex-perimental precision reduces the tolerance range and hence increases corroborability. In most psychological research, improved power of a statistical design leads to a prior probability approaching ½ of finding a significant difference in the theoretically predicted direction. Hence the corroboration yielded by "success" is very weak, and becomes weaker with increased precision. "Statistical significance" plays a logical role in psychology precisely the reverse of its role in physics. This problem is worsened by certain unhealthy tendencies prevalent among psychologists, such as a premium placed on experimental "cuteness" and a free reliance upon ad hoc explanations to avoid refuation.
Meehl, Paul E. (1967). "Theory-Testing in Psychology and Physics: A Methodological Paradox" (PDF). Philosophy of Science 34 (2): 103–115.
https://dx.doi.org/10.1086%2F288135 . Free here: http://cerco.ups-tlse.fr/pdf0609/Meehl_1967.pdf
There are many science articles posted to this site that fall foul of his critique probably because researchers are not aware of it. In short, this (putatively fatally flawed) research attempts to disprove a null hypothesis rather than a research hypothesis. Videos of some of his lectures are available online:
http://www.psych.umn.edu/meehlvideos.php
Session 7 starting at ~1hr is especially good.
(Score: 2, Insightful) by Anonymous Coward on Saturday January 23 2016, @12:20AM
If you know anything about the Bayes v. Fisher battle, which has raged since the 1930s. Fisher was a very smart, but very intellectually domineering personality who publicly and brutally railed against Bayes. He essentially bullied into dominance the field, out of which came his many frequentist-based tools that have been much abused ever since. Bayes Theorem came back out of the shadows in the 1950s, and in the 60s you had people like Tukey using it to call elections. Papers like this appeared all the time. They are different tools for different kinds of problems. There is no one approach to data analysis! It depends upon the problem you're facing [xkcd.com]! The problem is that Fisher tools are easy to apply and just "feel" right intellectually (provided you pretend you have an infinite ensemble of random outcomes to pull from), so they get applied everywhere. A physicist makes a measurement and he gets some numbers; just plug them into these formulas and you get an answer back on its significance! In psychology you get problems that don't make sense with frequentist tools, so you use Bayes because it "feels" right because it seems like common sense for these kind of problems.
(Score: 1, Interesting) by Anonymous Coward on Saturday January 23 2016, @12:41AM
No, that is not what this is about at all. Choice of equations to use does not address this problem in any way, only the choice of null hypothesis. Please read the paper, where this "objection" is dismissed at the top of the second page (one of the things that makes it such a great paper is that it preempts so many of these nonsensical objections):
Also, Fisher was not at all a frequentist. He railed against use of that philosophy for scientific means for the last half of his life, and I do mean railed against it. He thought it would be the downfall of western civilization.[1] The best term to describe Fisher's statistical philosophy is probably "inductivist".
[1] Fisher, R N (1958). "The Nature of Probability". Centennial Review 2: 261–274
http://www.york.ac.uk/depts/maths/histstat/fisher272.pdf [york.ac.uk]
(Score: 0) by Anonymous Coward on Saturday January 23 2016, @02:21AM
But his later views were modified by the acerbic exchanges he had with Jeffreys in the journals. After those changes, Fisher's arguments started picking up tones reminiscent of Bayes, but he was too proud and bull-headed to have conceded any of Jeffreys points that he ended up co-opting into his own work.
I'd say Fishers statistical philosophy was anti-inverse-probability, in whatever that form happened to take at the time.
(Score: 0) by Anonymous Coward on Saturday January 23 2016, @02:59AM
In that paper he disagrees:
Obviously check it yourself to see I'm not doing some selective citing. Of course, that is irrelevant to the issue brought up by Meehl which is much more important. However, if you can back some of your claims about Fisher with citations I would be interested.