Formulating Hypotheses

In an effort not to write a Toltsoy-esqe blog post, I’m breaking some background information out into separate posts.  There’s a point, I promise, it’ll just take me a couple posts to get there.

The Scientific Method is a standard process which is used to properly frame an experiment and guide the experimenter (at least it should be, but there’s a lot of junk being done these days).  Various descriptions of the Scientific Method give it a variable number of steps–from 4 to 7.  Those with fewer steps combine a few things that others break out into separate steps–there’s no big divergence in process.  For the sake of discussion, we’ll use the description at

One step, in this case #3, is always stating your hypothesis.  Classically, one wouldn’t just make a single statement and test against it; one would make several statements, each predicting a different outcome of the experiment.  Usually there is a statement of no difference, called the Null Hypothesis, and abbreviated H0 (that’s H sub zero).  One would then offer Alternate Hypotheses predicting different effects, and abbrevizted Ha, Hb, etc.

Let’s look at a simple example: will fertilizer help us grow bigger tomato plants?  We’ve made our observation that Farmer Brown grows nice tomatoes, and he uses fertilizer.  Our hypotheses would be:

H0: Fertilizer will have no effect

Ha: Fertilizer will help our plants grow

Hb: Fertilizer will hinder our plants’ growth

Simple, and easily tested.

Not every hypothesis can be tested.  The human lifespan is too short, and the vastness of space and time and the complexity of our own bodies (let alone the ethical considerations) make some hypotheses virtually untestable.  In these circumstances, even with some solid evidence, the discussion is more academic and philosophical than scientific, especially if there are equally strong (or preposterous) interpretations of the same evidence.

Additionally, another important component of the Scientific Method is repeatability.  Performing an experiment numerous times, and getting the same (or wtihin statistical probablity) answer each time is essential.  In our tomato experiment, we can easily reproduce this hundreds or thousands of times under very controlled conditions.  Referring again to vastness and complexity, some experiments can’t feasibly be repeated.

Quod erat demonstratum (“that which is to be proven”) sometimes always remains as such.

 <background information>

For anyone who’s new to the conversation, I hold a BS in Environmental Science and an MS in Biological Sciences.  I earned my MS studying the genetics of pollution-degrading bacteria and the external proteases of a pathogenic bacterium.  While in graduate school, I was both a teaching and research assistant; I was awarded two teaching awards, and was rated “excellent” by my students when I taught MCAT prep classes for The Princeton Review.  After graduation, I worked as a molecular biologist in medical research for many years before switching to computer programming.

</background information>

Now playing: BoDeansTrue Devotion