10/07/2013 01:10 pm ET Updated Dec 06, 2017

What Is A Hypothesis? Or, Why Can't I Dismiss This Theory in Favor Of My Religion?


Last month we discussed what science is, namely a way of approaching the universe in terms of measurable empirical evidence. We started examining the scientific method and how to ask a question, refine and build upon it, and now we are going to learn what to do with that question.

Having researched a question into something we can study, it is now time to apply that research to the question and come up with a model proposing a possible answer. This model is called a hypothesis. It draws from our question and is the substructure of the experimentation that will ultimately test our hypothesis and serve to answer our question.

A hypothesis should not therefore shy away from making a prediction. In fact, a hypothesis that makes no prediction is no hypothesis at all, it is just a statement. This means the model you propose must be testable.

Examples of good hypotheses:

  • "The vinegar will not visibly react with the baking soda."

This hypothesis predicts no visible reaction between vinegar and baking soda. While false, this hypothesis provides for a test of something observable and when contradicted provides another avenue of experimentation.

  • "Gravity is a function of space-time geometry and will therefore warp light. During a solar eclipse, stars that are actually behind the sun will be visible at these precise angles."

This hypothesis also makes a prediction, but far more strict and precise predictions. This is a part of General Relativity which was observed in 1919, and can be clearly seen in the Hubble Extreme Deep Field. General Relativity has made many such precise predictions that have been observed, including observed "double" quasars that are in fact single quasars whose light is being deflected by gravitational lensing as predicted by General Relativity.

Examples of bad hypotheses:

  • "I predict that unicorns could be real, and since there is no evidence to the contrary I am right."

This is a bad hypothesis because it seeks to prove the existence of something that does not exist by demanding contradicting proof that does not exist either... namely because the thing itself does not exist. Thus this hypothesis offers nothing in terms of measurable predictions, since it is defined by observing nothing.

  • "Gravity is a function of space-time geometry because God willed it so, and therefore will warp light. During a solar eclipse stars that are actually behind the sun will be visible at these precise angles."

On the surface this hypothesis may sound reasonable, perhaps a God in the Gaps compromise. It does, after all, make a prediction that can be tested -- and has been tested positively in fact. So this is a good hypothesis, right?

No, it is not a good hypothesis because it is offering a supernatural subtext for otherwise mundane phenomena. In other words adding God to this hypothesis does not change the prediction. It can exist perfectly fine without God, and since it can exist perfectly fine without God there is no reason to have God in it, unless we are being forced to satisfy some theological demand. Galileo and Copernicus had to do this to protect themselves from the church, and they were by no means alone.

To recap, a hypothesis proposes an idea that makes testable predictions about a given question. We then set up an experiment to test this model by looking for those predictions. This is why predictions are very important. No prediction, no test, no science.

This is the fundamental difference between scientific questions and religious questions. Someone once tried to pass off, "Life carries on just as it does," as a Creationist prediction. By definition this may be a prediction, but it is not testable because it can be applied to any phenomenon.

When we prove a hypothesis it becomes a theory. This may be confusing. Why do we call proven hypotheses theories? And, more to the point, do we ever stop calling them theories? No, we don't. Going back to Carl Sagan, "Science invites us to let the facts in, even when they don't conform to our preconceptions." In this statement Carl is saying that in order for science to work we cannot ever be truly final in our conclusions. They must always be open ended and subject to revision as new evidence arises. The very best we can do is provide a theory that can be tested in various conditions, much like General Relativity will be tested repeatedly as venture forth into the Universe. It will always be: Einstein's Theory of General Relativity no matter how many times it is proven.

Now, go forth and hypothesize!