4 ms·

P .05 cut-off for significant scientific result is due to copyright restrictions

19 points by samwillis 5y ago

HWR_14 5y ago
This title is pretty misleading and not supported by the tweet. The point it makes is the original simplification to 0.05 as a binary cutoff was for copyright reasons over a century ago. The current use of 0.05 is inertia from that decision. There are no current copyright issues.
- upsidesinclude 5y ago
  This should be the example under the definition of pedantic
  - HWR_14 5y ago
    I think that's very important distinction. "Copyright law needs to change because p < 0.05 are legally scientific results that are outside copy protection and can be freely copied" is a different case than "bad facts taught and here's a better solution; oh, and the origin was a historical copyright issue"
- samwillis 5y ago
  Quite right, I struggled to trim the tweet down in length to make it fit the HN title restrictions. Not ideal, however the trivia is interesting.
  - HWR_14 5y ago
    Not just the trivia of the origin. If we know why it was chosen, and have a better solution, we can start to advocate for it. I had an issue with the title, but I'm glad you posted it.
dekhn 5y ago
No.
derbOac 5y ago
So I was really suspicious of this, both because it seemed too tidy and the citation to "Kendall Pearson" was suspect. I recommend the cited paper by Hurlbert & Lombardi (2009), which makes a lot more sense and explains things more clearly. The actual discussion in that paper claims that Kendall was relating a story told to him by Fisher, about Pearson's defense of copyrighted tables. So Pearson publishes books of tables, which were expensive to print at that time, and is defensive of them, leading to challenges for Fisher, who relates this to Kendall. The claim that copyright played a role seems partially true (if you believe Hurlbert & Lombardi, 2009, which I do). Hurlbert & Lombardi make clear the dichotomous thinking about significance, use of p<.05 etc, was prevalent before and the copyright obstacles, but the copyright obstacles may have reinforced the trend further. It's pretty clear standard significance thresholds (.05, .01) were being thrown around in various forms for other reasons having nothing to do with copyright, but the copyright issues kinda nudged things further in that direction.

19 points by samwillis 5y ago

This title is pretty misleading and not supported by the tweet. The point it makes is the original simplification to 0.05 as a binary cutoff was for copyright reasons over a century ago. The current use of 0.05 is inertia from that decision. There are no current copyright issues.

upsidesinclude 5y ago
This should be the example under the definition of pedantic
- HWR_14 5y ago
  I think that's very important distinction. "Copyright law needs to change because p < 0.05 are legally scientific results that are outside copy protection and can be freely copied" is a different case than "bad facts taught and here's a better solution; oh, and the origin was a historical copyright issue"
samwillis 5y ago
Quite right, I struggled to trim the tweet down in length to make it fit the HN title restrictions. Not ideal, however the trivia is interesting.
- HWR_14 5y ago
  Not just the trivia of the origin. If we know why it was chosen, and have a better solution, we can start to advocate for it. I had an issue with the title, but I'm glad you posted it.

dekhn 5y ago

No.

derbOac 5y ago

So I was really suspicious of this, both because it seemed too tidy and the citation to "Kendall Pearson" was suspect. I recommend the cited paper by Hurlbert & Lombardi (2009), which makes a lot more sense and explains things more clearly. The actual discussion in that paper claims that Kendall was relating a story told to him by Fisher, about Pearson's defense of copyrighted tables. So Pearson publishes books of tables, which were expensive to print at that time, and is defensive of them, leading to challenges for Fisher, who relates this to Kendall. The claim that copyright played a role seems partially true (if you believe Hurlbert & Lombardi, 2009, which I do). Hurlbert & Lombardi make clear the dichotomous thinking about significance, use of p<.05 etc, was prevalent before and the copyright obstacles, but the copyright obstacles may have reinforced the trend further. It's pretty clear standard significance thresholds (.05, .01) were being thrown around in various forms for other reasons having nothing to do with copyright, but the copyright issues kinda nudged things further in that direction.