Sciency Words: Null Hypothesis

Hello, friends!  Welcome to Sciency Words, a special series here on Planet Pailly where we look at the meaning and origins of scientific terms.  Today on Sciency Words, we’re talking about:

THE NULL HYPOTHESIS

Whenever there’s a big scientific discovery in the news, my first question is always: should I take this seriously?  The answer is usually no.  The popular press may say one thing, but when you dig into the actual science, you often find the facts do not support the hype.

So when I started reading about a second possible planet in the Proxima Centauri system, I wanted to know: should I take this seriously?  In this article from Scientific American, the astronomers who discovered this possible planet are quoted as saying:

Since the very first time we saw this [potential planetary] signal, we tried to be its worst enemy.

The astronomers are then quoted saying:

We tried different tools to prove ourselves wrong, but failed.  However, we have to keep the doors open to all possible doubt and skepticism.

For me, this is the most reassuring thing any scientist could say.  Too often in popular culture, scientists are portrayed a certain way.

But no, good scientists are not out to prove to the world that they’re right.  They’re trying as hard as possible to prove to themselves that they’re wrong.  Which brings me to the null hypothesis.

According to the Oxford English Dictionary and other sources (like this one), the term “null hypothesis” can be traced back to British statistician Ronald Fisher.  Fisher first wrote about the null hypothesis in 1935, in a book titled The Design of Experiments.

As a way of introducing the concept, Fisher tells us the story of a woman who claimed to have an oddly specific talent.

A lady declares that by tasting a cup of tea made with milk she can discriminate whether the milk or the tea infusion was first added to the cup.

The Design of Experiments, by Ronald Fisher

Fisher then describes an experiment to test this woman’s claim.  She’s given eight cups of tea, four with the milk added first, and four with the milk added afterward.

In the context of this experiment, the null hypothesis predicts that the woman will not be able to tell which tea is which—she’s only guessing.  Or to put that in sciencier language, the null hypothesis asserts that there will be no statistically significant relationship between the way this woman’s tea was prepared and the way she believes her tea was prepared. As Fisher explains:

[…] it should be noted that the null hypothesis is never proved or established, but is possibly disproved, in the course of experimentation.  Every experiment may be said to exist only in order to give the facts a chance of disproving the null hypothesis.

The Design of Experiments, by Ronald Fisher

A null hypothesis is usually paired with an “alternative hypothesis,” which asserts that a statistically significant relationship does exist.  In Fisher’s tea tasting example, the alternative hypothesis would be that the woman really can tell which tea is which.  You can never really prove that either the null hypothesis or the alternative hypothesis is true, but a well designed experiment should be able to prove that one hypothesis or the other is false.

Going back to that possible planet in the Proxima Centauri system, the article from Scientific American does not explicitly mention the null hypothesis; however, the spirit of the null hypothesis is clearly in play.  Astronomers are trying their best to prove that that planet does not exist, and so far they can’t do it.  And that’s enough to convince me that I should take this new planet seriously (at least for now).

Next time on Planet Pailly, we’ll find out what this not-yet-disproven planet might look like.