Hypothesis Testing

A Flim-flam, par excellence

Hypothesis Testing


When it comes to testing a hypothesis, people outside the science disciplines believe the matter is fairly well cut-and-dried:  You construct a hypothesis and attempt to prove it scientifically valid.  Wrong.  According to most scientists (astronomers included) you attempt to disprove the hypothesis.  Such a superb scientific exercise is actually accomplished with a little sleight-of-hand and a certain small amount of doublespeak.


Even those outside science know that an assumption is a belief or a logical construct, or to some, an “unconfirmed fact.”  Some of your friendly neighborhood scientists go even further and will classify certain assumptions as hypotheses or theories for no apparent reason but to further their favorite preconceptions.  Their method for testing a hypothesis can be downright bewildering to someone not “in the know.”


Scientists will generally agree that a hypothesis is a tentative explanation for an observation, a phenomenon, or a scientific problem that can be tested by further investigation.  However, the procedures commonly used for “further investigation” appear to have been designed to ensure that a “tentative explanation” becomes an accepted fact of science.


In hypothesis testing, an alternative hypothesis is the proposition that is accepted if the null hypothesis is rejected.  [It is the “maintained hypothesis.”]  This is where doublespeak comes into play.  Usually, an alternative hypothesis is, in fact, the primary hypothesis (the favored hypothesis); this is the hypothesis that is scheduled to be accepted once the dummy null hypothesis has been duly and properly rejected.


Then, there is the null hypothesis.  This is the hypothesis to be tested and [accepted(?) or] rejected in favor of an alternative.  Specifically, a null hypothesis is the hypothesis that an observed phenomenon is due to chance alone and not due to a systematic cause.  Generally, the null hypothesis is the one that is disproven, not the alternative one.  The real purpose of the null hypothesis is to challenge the alternate hypothesis so that the null hypothesis can be rejected and thereby give more credence to the alternative hypothesis (the favored hypothesis).


As is obvious, the alternative hypothesis and the null hypothesis are the two rival hypotheses whose likelihoods are compared by means of a hypothesis test. Usually, the alternative hypothesis is the probability that an observed phenomenon is genuine and the null hypothesis is the rival possibility that it has not occurred.  Even though it is said that both hypotheses are subjected to testing, it is only the null hypothesis that is actually tested (with the premeditated intention of disproving it).


In hypothesis testing, an alternative hypothesis is accepted if the null hypothesis lacks sufficient evidence to require rejection of the alternative hypothesis. However, in most cases, the acceptance of the alternative hypothesis is the “pre-ordained” and expected conclusion (which is why the test was completed in the first place) and the null hypothesis is, quite simply put, the dummy hand (with no trump).


After ‘sufficient’ testing, an alternative hypothesis is often considered ready to be elevated to the status of a theory.  But note:  the hypothesis has not been proven; its rival null hypothesis has been disproven (the main purpose of the null hypothesis is to provide a basis for accepting the favored alternative hypothesis without actually having to prove the alternative hypothesis is factual).  Disproving a null hypothesis not only spells failure for it, but also for the alternative hypothesis, for the simple fact that the alternative hypothesis has not itself been subjected to testing.


NOTE:  “One must realize that it does not matter how well-constructed a hypothesis is, or how intelligent the originator is, or what his/her name is, or how many scientists believe in the hypothesis.  If a hypothesis fails [does not pass] the testing experience, it is wrong [and must be discarded].” --  Richard Feynman, Ph.D.   (edited)


Something else one must realize:  The rejection of the null hypothesis in a test does not automatically (or otherwise) mean that its corresponding alternative hypothesis has passed the testing experience.  That is the back door approach.  A hypothesis cannot pass a test without itself being tested.  It is rather foolish (and unscientific) to imagine there is such a thing as a surrogate hypothesis in testing.


Mainline scientists say that attempting to disprove a hypothesis is the practice of true science, whereas attempting to prove a hypothesis is the practice of pseudoscience.  Consider the following three hypotheses:  heliocentricity, evolution and relativity.  Neither of them (not one of them) can be proven to be scientifically valid.  So, what does one do?  Do as the scientists do.  Contruct a series of null hypotheses (including built-in fallacies)  for each one of them.  When an adequate number of the series of three null hypotheses have been rejected, declare that heliocentricity, evolution and relativity are ready to be considered theories since the null hypotheses were unable to find cause for rejection.  (Then explain why such a practice is not pseudoscience.)


NOTE:  A theory is a plausible or scientifically acceptable general principle or body of principles offered to explain phenomena (e.g., the wave theory of light).


Thomas E. Cobb, Se.M.

“If you take a highly intelligent person and give them the best possible, elite education, then you will most likely wind up with an academic who is completely impervious to reality.”                                    -- Halton C. Arp, Ph.D.