Hi:

A controlled experiment is expressed as:

Ctrl group: no drug
Expr group: with drug

And we expect to see that:

Ctrl group: no effect
Expr group: has effect

The above looks like the logic relation: iff

A iff B
T T T <- Expr group and has effect
T F F
F F F
F T T <- Ctrl group and has no effect

Could we say that the degin of a controlled experiment is intended to show that there is an iff logical relation between the drug and the effect?

Are there some other implications for the above similarity?

--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)

On Mon, 8 Jun 2020 07:06:49 -0700 (PDT), Cosine <[email protected]>
wrote:

Hi:

A controlled experiment is expressed as:

Ctrl group: no drug
Expr group: with drug

And we expect to see that:

Ctrl group: no effect
Expr group: has effect

The above looks like the logic relation: iff

A iff B
T T T <- Expr group and has effect
T F F
F F F
F T T <- Ctrl group and has no effect

Could we say that the degin of a controlled experiment is intended to show that there is an iff logical relation between the drug and the effect?

I was hoping someone else would pitch in an answer,
because several re-readings still leave me drawing a blank
as to what is designated by your A and B, and what
questions are answered by your T and F. So, in general, here
is a description including IF and Only IF -


The RANDOMIZED, controlled experiment gives the result that
was hoped for
IF the drug shows effect, (and)
ONLY IF the no-drug group shows no effect (or, much smaller effect).

If neither or both groups show effect, the experiment
was an apparent failure to distinguish them, for whatever
reason.

If the no-drug group has the better effect, something is screwy
about the expectations or the carrying out of the experiment.

I think you know all that.

Are there some other implications for the above similarity?

Sorry, I don't imagine what you are driving at.

--
Rich Ulrich

--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)

Rich Ulrich <[email protected]> wrote:

On Mon, 8 Jun 2020 07:06:49 -0700 (PDT), Cosine <[email protected]>
wrote:

A controlled experiment is expressed as:
The above looks like the logic relation: iff
Are there some other implications for the above similarity?

Sorry, I don't imagine what you are driving at.

Cosine might like to read the books by Judea Pearl on causal inference,
who says "I call the levels 1. Association, 2. Intervention, and 3. Counterfactual...At the Intervention layer, we deal with sentences of
the type P(y|do(x), z) that denote ???The probability of event Y = y,
given that we intervene and set the value of X to x and subsequently
observe event Z = z. Such expressions can be estimated experimentally
from randomized trials or analytically using causal Bayesian
networks..." The Pearl type causal models can be reformulated as a "non-monotonic modal logic" eg https://www.aaai.org/ocs/index.php/AAAI/AAAI15/paper/viewFile/9686/9417

--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)