(2017-10-21, 10:27 AM)Chris Wrote: [ -> ]It seems to me that there's a lot of confusion in what fls is writing between different concepts - in particular between hypotheses and tests of hypotheses.
Exactly this, Chris. I have been offline for a bit but had been planning to respond to Linda when I returned, and that was the main point I was going to make. Once again, you have beaten me to it.
Linda, in your first reply to me you seemed confused as to whether this "confirming"/"disconfirming" distinction you've been trying to make applied to hypotheses themselves, or to tests of hypotheses.
First you applied it to hypotheses:
(2017-10-17, 01:41 PM)fls Wrote: [ -> ]Falsification is about whether your hypothesis reflects what you would see if your idea is true or whether it reflects what you would see if your hypothesis is false.
But as I pointed out, this is purely a semantic distinction - any positive hypothesis can (I think) be reframed as a negative hypothesis, and vice versa, without any change to the way the experiment is conducted. For example, the positively-framed hypothesis "That when psi is present, a statistically significant effect will be observed" can be reworked into the negatively-framed hypothesis "That when psi is present, a statistically significant effect will
not be observed". Nothing about the experiment would change.
You seem to agree with me about this. So, let's move on to whether the distinction can apply to hypothesis
testing. Of this, in your initial response to me, you wrote, of trying to test whether "p implies q" as per the paper you referenced:
(2017-10-17, 01:41 PM)fls Wrote: [ -> ]You are attempting to confirm the idea when you choose to look at "p". Your result may be "q" (alternative hypothesis confirmed) or "not q" (null hypothesis confirmed).
You are attempting to falsify the idea when you choose to look at "not q". Your result may be "not p" (alternative hypothesis confirmed) or "p" (null hypothesis confirmed).
However, as both the paper and logic dictate, whilst, yes, starting by looking at the consequent "not q" can potentially falsify the hypothesis (in the case that the antecedent turns out to be "p"),
starting by looking at the antecedent "p" can also potentially falsify the hypothesis (in the case that the consequent turns out to be "not q").
Both tests can potentially falsify the hypothesis.
In favour, though, of your labelling the first test "confirmatory" is that only this test can "confirm" the idea. In other words, if you turn up "q" as the consequent, then you have confirmed that "p implies q".
Unfortunately, this means that you are mistaken in implying that the second test can "confirm the alternative hypothesis": you cannot conclude from "not p implies not q" that "p implies q" (the hypothesis in question).
[Edit: the above two paragraphs are mistaken, as Chris pointed out later in the thread. Re the first paragraph: if you turn up "q" on a card whose reverse is "p", all you can conclude is "p and q",
not "p implies q". This is, as the paper Linda referenced calls it, an "ambiguous verification" rather than a "confirmation". Re the second paragraph, I once again confused conjunction (logical and) with implication: you would not be starting from "not p implies not q" but rather from "not p and not q". Even though this
does entail "p implies q", it is again more accurate to refer to it as an "ambiguous verification" than as a "confirmation".]
So, what I would say overall is that, as I initially wrote to Steve, generally, scientists (at least in parapsychology and related fields) form a falsifiable hypothesis, and then design a test which either
does falsify that hypothesis,
or confirms it - and this is fundamentally tied to whether or not the null hypothesis is rejected.
You are, though, correct that it is sometimes possible to design tests that are
only capable of falsifying an hypothesis, and not of confirming it. We can (paraphrasing you and reframing in a way that makes most sense to me) take a
massive caricature of the Jaytee hypothesis as follows:
When Pam is returning home {R}, Jaytee is
always at the window {A}. Also, when Pam is
not returning home {¬R}, Jaytee is
never at the window {N}.
Now the hypothesis can be written as (where ∧ is the conjunction aka "and" operator, and → is the implication operator):
(R→A)∧(¬R→N)
So, yes, you
could in this scenario choose to test only
one of the conjuncts - and thus potentially falsify the entire hypothesis without being able to confirm it. Why might you do this? Maybe testing the other conjunct is extremely expensive, and if you can falsify the hypothesis via the other conjunct, then you save a bunch of money. Let's say though that instead of falsifying it, you
confirm the conjunct in question - then you would probably want to test the other conjunct - expensive as it is - to see whether you can confirm it too, and thus confirm the entire hypothesis.
If all of that doesn't clarify my position for you, Linda, then please just ask.
As for that which you wrote in an earlier post: "
I jumped in with more information in order to be helpful. I'm not sure why you are affronted by this". Maybe now you can see that I was irritated because you tried to make out that my simple and unobjectionable description of how hypothesis testing usually works was wrong, and you did this by sowing confusion in the process! It is a character flaw though that I expressed this irritation to the extent that I did, and I'm sorry for my antagonistic approach.
By the way, I read enough of the two papers that you referenced in your initial response to me to see if/how they applied, and came to the conclusion that they aren't particularly relevant. I could write at length on that, but this post is long and tedious enough as it is.
No hard feelings, Linda. Thanks for trying to be helpful.