And yet you point to these studies as reasons why some medical treatments are not efficacious.
The answer is simple, Loudmouth, but you don't get it because you know nothing about math.
To determine the probability that something is true after a test, you must know two things. First, you must know the
a priori chance that the condition is true, and then you must know the power of the test. You routinely spew nonsense here without paying attention to either of these two factors.
Let's suppose, for example, that we have a test that is 99 percent accurate and produces false results only 1 percent of the time. Let's also suppose that we run this test on a man, and it comes back positive. What does that mean? Well, if it's a
pregnancy test, that means that the test is
wrong because
men don't get pregnant. The
a priori chance that a man is pregnant is zero, and so it doesn't matter if the test is 99.99999999999999999% accurate the test is still wrong, and the chances that the man is secretly pregnant remain stubbornly stuck at 0.0%
So now that the concept has been explained (though I despair that you will ever understand it), let's move into the realm of treatment for diseases. Let's suppose that a person has disease X, and someone claims that eating fruit Y will cure you of disease X. What's the
a priori chance that this claim is true? Well, we cannot know, but suppose that we observe that disease X regularly occurs in a region where fruit Y is eaten, we might suppose that the chances that fruit Y cures the disease is low. Let's assume that we guess, calculate, or in some other way come up with the idea that fruit Y is only about 1 percent likely to cure the disease.
Then we conduct a scientific study. Lo and behold, fruit Y is shown to cure disease X with p = 0.05 on the button. What does that mean? Well, it means that we should update our estimation of the probability that fruit Y cures disease X. How do we do that? Well, we use
Bayes' Theorem. That theory states that:
P(H|E) = P(E|H) * P(H) / P(E)
Where P(H|E) means "the probability of the hypothesis in light of the evidence."
(number crunching on an Excel spreadsheet).
All right, that means that the new probability that fruit Y cures disease X has risen from 1 percent to slightly over 16 percent. Now that's a big jump in probability, but it still means that I am
far from convinced that fruit Y cures disease X. Now let's suppose that another test is done and this test determines that fruit Y is no better than a placebo.
At this point, I am obviously going to again re-evaluate my belief that fruit Y cures disease X down substantially. In fact, the most logical conclusion is that the result of the first test was due to
p-hacking, which is known to be prevalent in the sciences.
Now at this point, you will start whining that I am using science when all I am doing is using a calculator and a bit of common sense.
