Class 71: Never Trust ‘Research Shows’ Headlines: We Bring Cause To Data

All uses of hypothesis testing are fallacies. The results could be true, or likely, but only accidentally. A proof based on the Resurrection and GPA.

Bonus Song

Anon sent in this catchy tune, which I understand is Grammy-worthy. Beware Correlations!

Video

https://youtu.be/54UOI8ImRfs

Links: YouTube * Twitter – X * Rumble * Bitchute * Class Page * Jaynes Book * Uncertainty

HOMEWORK: Given below; see end of lecture. Do the code!

Lecture

Research Shows headlines are generated from papers by academics, and these all have explicit or tacit claims of cause, all purporting to explain some set of observations (whether gathered in history, the world, or by experiment). To explain is to state or to tacitly point to a cause.

The problem is that the methods science has developed to affirm or claim cause are often wrong: they are not right; they are in error; they are incorrect; they are fallacious; they sometimes make the right decisions, but only accidentally. By which I mean, cause is arrived at not by the methods, but by other means, yet the methods are credited.

I hope it is clear when I say that these methods are not to be used.

But are.

The worst tool, and one whose use is always and every instance a fallacy, is the so-called hypothesis test. We have done (“wee Ps”) in Class so many times, we’re sick of it (or I am). But I want to prove to you “tests” are fallacious another way, using one familiar example and one common situation. And with no math!

The idea is simple: a researcher makes observations, runs a “test”, and makes a pronouncement the cause he thought of is the one correct explanation for the observations.

He might be right about this, but it will only be accidentally, and not because of the “test”. For that same “test” could be used in support of an infinite number of other possible causes. That is the proof against “tests”: that they can support anything.

This will always be the case. As in every time. As in it is inescapable.

Our familiar example comes from A Global Enlightenment: Western Progress and Chinese Science by Alexander Statman (his surname guaranteeing I would read the book). It is mostly a review of the 18th Century Jesuit mission to China, and those Jesuits’ interactions with major figures of the so-called Enlightenment.

An important observation Statman makes is man’s proclivity to look for wisdom in the past or the future. One believes those in the past had superior knowledge, knew more secrets and could communicate with God (or the gods) with greater ease and facility, yet somehow that knowledge was lost (wiped away in the flood, say; China, having the oldest extant civilization was thought to hold vast repositories). Or one believes those to come will be better than we, will know more, and will lead easier and happier lives, if only they are not held back by those who look to that past (China was also by others thought to have stagnated and could only copy their betters).

As an example of the former, we are introduced to Louis-Raphael-Lucrece de Fayolle, Comte de Mellet, a so-called orphan of the Enlightenment, men who thought there was more to the world than “facts and logic” (sound familiar?).

Mellet began his search for ancient wisdom in Egypt. In this sense, his project was not just generically esoteric, but out-and-out Hermetic. Although he did not often mention the discredited Corpus Hermeticum, he accepted its central idea that the account of the New Testament was foreshadowed by Egyptian mythology, a common theme for magi from the Renaissance on down. He developed it further through Court de Gebelin’s techniques of comparative etymology. For example, Osiris, both the son of the sun and the sun itself, was born on December 25, and his name was written with the hieroglyphics of “sun”; thus, he was Jesus. Isis, both mother and virgin, was written with the hieroglyph depicting the constellation Virgo; thus, she was Mary…How had Egyptians known these Christian truths thousands of years before the birth of Christ? [Mellet’s] answer was that the hieroglyphs had descended from a “sacred language” which emphasized divine knowledge.

Modern Reddit atheists, using those same observations, offer them as proof Jesus was not God. Others say that because of these observations, Jesus was himself only a “myth”, a fictional creation, an invention of some First Century Jews who were keen on storytelling. (The Romans, alas, did not enjoy the antics of the storytellers, and dismantled, with extreme prejudice, the storyteller’s infrastructure.)

These two (or three) positions on who Jesus was do not exhaust the theories offered to explain the same observations, as you will know. That we know this “controversy” so well is perhaps why we lose sight of its lesson. Which is (to repeat) that the same set of observations can be used to “prove” more than one cause.

This example is informal. No “test” is offered for it. But it wouldn’t make any difference if one was.

In Class, we did the example of first-year college GPA, cut by Whites and non-Whites. We saw that there were differences in the observations, with many more non-Whites scoring low and vice versa. A “test” gave a wee P-value.

Now something, or things, caused the differences. The “test” rejected the “null” that there were no causal differences between races. Thus, because we reject “no cause”, we must embrace its logical contrary, which is “there is a cause of the differences”. And the cause most researchers might pick, more academics being in universities, which are hives of DIE, is “racism”.

But “racism” is not the only possible cause. “Racism” was not in the observations. “Racism” was in the mind of the researcher. The “test” only said “no cause” was false. The “test” did not say which cause was the cause. It only said a cause existed.

Here’s a list of possibilities, necessarily incomplete:

• Whites crush the intellectual efforts of non-Whites using occult powers of “racism”
• Whites are superior intellectually
• Administrators cheated and gave Whites more GPA points
• Administrators cheated and stole non-White GPA points
• Whites took classes which gave easier As
• Non-Whites took classes which gave harder As
• The Whites were all older and paid to take their classes
• Non-Whites were recruited for being non-White and were not enthusiastic students
• Non-Whites were playing a joke
• White students formed a pool to see who could get the best GPA
• Non-Whites sabotaged their own grades to have people accuse Whites of “racism”
• Whites bribed professors to get better grades
• Whites bribed professors to give non-Whites worse grades
• Professors lowered non-White grades purposely to illustrate the plight of non-Whites
• Whites were mostly male, and males use “sexism” to get better grades
• Non-whites were mostly female, who are Victims of “sexism”
• Non-whites had more immigrants, who were cowering and fearful of deportation
• Whites cheated to get better grades
• Whites worked an average amount, but non-Whites put in little effort
• Just one White used “racism”
• Just two Whites used “racism”
• Etc.
• Just one White used “racism” half the time
• Etc.
• Lord Xenu, in league with rogue Mormons, sent Thetans to curse non-Whites

I could have gone on and on, ad infinitum.

If I were writing a paper in the International Journal of Dianetics, I would emphasize that last theory, declaring the evidence (the wee P) was clearly in support of it.

And I would have been right. The evidence is clearly in support of that theory. The data we saw is clearly not against the theory.

There is no escaping that conclusion. Not if you believe statistical testing confirms, or even gives evidence for, causal theories.

None of the causal theories in the list was in the data, and of course cannot be, or in the test themselves. Data is just measurements, and is always silent on cause. It is we who bring causal notions to data. That is inescapable, and must be obvious given it was we who said “Let’s measure this, and not measure that.” We picked those items because we thought they would be related to the causes and conditions of whatever theories we entertained.

The test cannot possible confirm, or deny, any theory. If it could, then a test would confirm every theory put forward as a causal explanation.

There is no escape for “controlling for” data that might be measured in any of the hypotheses in the list. For one, “control” is not control, unless you have manipulated the particular items in a real way, as a chemist does in his experiments (see this post). And anyway, even if you believe you have “controlled for”, say, sex, there are still an infinite list of causal hypotheses that have nothing to do with sex, or whatever you “controlled for”.

My use of the example is now hackneyed, but it is still apt. A chemist wants to measure the rate of heat change due to a reaction. He measures out his substances to great precision, he isolates his test apparatus from all those things he can.

That’s it. That’s the proof.

Which means you must investigate every claim of “Research Shows”, because if they used ordinary methods to “confirm” cause, their arguments will be fallacies. Even if accidentally correct.

So what’s the answer? How to find cause? Well, stick around.

HOMEWORK

Jessica Utts writes in a peer-reviewed paper “An Assessment of the Evidence for Psychic Functioning” this:

Discuss

Subscribe or donate to support this site and its wholly independent host using credit card click here. Or use PayPal. Or use the paid subscription at Substack. Cash App: \$WilliamMBriggs. For Zelle, use my email: matt@wmbriggs.com, and please include yours so I know who to thank. BUY ME A COFFEE.