Great summary, thanks Dr. Briggs. I'm not a statistician, and will accept that's how statisticians proceed, but a *scientist* SHOULD proceed differently -- okay to gather relevant data to understand the problem (*The Science*), but one's primary duty is to try to slay the proposition. Inability to slay it after numerous tries isn't proof of the proposition. See Einstein's various comments. Obversely we have these days a commentariat that tries to amass clues and think they are constructing a resolute edifice. I don't have to tell you how this has poisoned our discourse in recent years.
Testing. The test is to see whether you're ready to suppress intuition and common sense, in favor of MATHY NOTATIONS. Forty years later I will now throw my Mendenhall in the trash, thanks to Briggs.
I always had this feeling that the NULL HYPOTHESIS was a kind of RITE OF INITIATION, where you had to LET IT TAKE OVER YOUR PSYCHE, in order to have any chance of professional success.
I was just reading chapter 10 of PT:TLOS over lunch and once again struck by the good humor and power of Jaynes' writing. He is wonderful on "games of chance" and "random." I confess that your quote on "random" is one that GG and I return to in almost every conversation.
Best wishes and hope you're staying warm. Dale
P.S. I'm going to email you an outline for your thoughts on a course for lawyers I'm working on that teases out the implications of PT in the non-quantitative realm - i.e. the courtroom - where we use words rather than numbers to quantify what we'll accept for "p" - "more likely than not", "clear and convincing evidence", "beyond a reasonable doubt," for example.
Nice summary. How does one work in that the Pr for a continuous variable being a specific point is always zero, a.k.a. infinitesimally small, e.g. real high Temp = 70 ? Or must we always account for a measuring device having non-infinite precision, as knowing is observation, which is dependent upon a measuring device? (Alternatives being knowing without observation, or use of a model). Perhaps not a summary level thing. Cheers.
And I thought the introduction of SPSS was bad. Now it’s AI. I don’t use AI much but I have noticed ChatGpt frequently will not admit its wrong when it’s made a mistake. It even becomes argumentative.
One of my many mathematical statistics textbooks says this about parameter estimation:
"The rationale behind all these criteria is to have an estimator which is reasonably close to the unknown parameter."
This statement is ridiculous. How can we judge “closeness” to an unknown quantity? If we knew how close we were, we’d already have information about the true parameter—which defeats the entire purpose of estimation. This kind of reasoning exemplifies the vague, circular logic that pervades so much of statistical theory.
Great summary, thanks Dr. Briggs. I'm not a statistician, and will accept that's how statisticians proceed, but a *scientist* SHOULD proceed differently -- okay to gather relevant data to understand the problem (*The Science*), but one's primary duty is to try to slay the proposition. Inability to slay it after numerous tries isn't proof of the proposition. See Einstein's various comments. Obversely we have these days a commentariat that tries to amass clues and think they are constructing a resolute edifice. I don't have to tell you how this has poisoned our discourse in recent years.
Testing. The test is to see whether you're ready to suppress intuition and common sense, in favor of MATHY NOTATIONS. Forty years later I will now throw my Mendenhall in the trash, thanks to Briggs.
I always had this feeling that the NULL HYPOTHESIS was a kind of RITE OF INITIATION, where you had to LET IT TAKE OVER YOUR PSYCHE, in order to have any chance of professional success.
... like just enough EXTRA LAYER OF ABSTRACT COMPLEXITY, in order to keep out the non-science majors ....
Listen whatever the nun says ok?
She’s only there to keep our asses alive and out of prison until we turn 18 anyway. Don’t argue.
These women civilized the Irish, a tribe so barbarous that even today are second only to the Chechens in Caucasian Conflict Chaos. I know I am one.
A summary E.T. Jaynes would be proud of, Matt.
I was just reading chapter 10 of PT:TLOS over lunch and once again struck by the good humor and power of Jaynes' writing. He is wonderful on "games of chance" and "random." I confess that your quote on "random" is one that GG and I return to in almost every conversation.
Best wishes and hope you're staying warm. Dale
P.S. I'm going to email you an outline for your thoughts on a course for lawyers I'm working on that teases out the implications of PT in the non-quantitative realm - i.e. the courtroom - where we use words rather than numbers to quantify what we'll accept for "p" - "more likely than not", "clear and convincing evidence", "beyond a reasonable doubt," for example.
Say! I didn’t even know you had a Substack. Good news. Look forward to hearing from you.
There's a series you might like on aviation crack-ups. First one I talk about might be one you're familiar with.
https://theabjectlesson.substack.com/p/aviation-mishap-story-time-introduction
Will look!
Nice summary. How does one work in that the Pr for a continuous variable being a specific point is always zero, a.k.a. infinitesimally small, e.g. real high Temp = 70 ? Or must we always account for a measuring device having non-infinite precision, as knowing is observation, which is dependent upon a measuring device? (Alternatives being knowing without observation, or use of a model). Perhaps not a summary level thing. Cheers.
We’ll come to those details very soon. It’s all the same in spirit, though.
If you're sweating it is hot.
If you're shivering it is cold.
In between it is temperate.
:-|
or, if you´re doing both at the same time, it´s COVID-25!
And I thought the introduction of SPSS was bad. Now it’s AI. I don’t use AI much but I have noticed ChatGpt frequently will not admit its wrong when it’s made a mistake. It even becomes argumentative.
I have a high probability the Penguin terrifies dangerous men
https://youtube.com/clip/UgkxSGHUJbSJGur4HFsK-8sI0DnBlkyfZVzQ?si=8hJ1fGvV55-kfuYG
One of my many mathematical statistics textbooks says this about parameter estimation:
"The rationale behind all these criteria is to have an estimator which is reasonably close to the unknown parameter."
This statement is ridiculous. How can we judge “closeness” to an unknown quantity? If we knew how close we were, we’d already have information about the true parameter—which defeats the entire purpose of estimation. This kind of reasoning exemplifies the vague, circular logic that pervades so much of statistical theory.
Amen.