A scientific hypothesis generates measurable predictions. The requirement of falsifiability means your hypothesis also has to survive rigorous attempts to find examples where the predictions are wrong. This means an established scientific theory is simply a theory that hasn’t been disproven – yet!

It may sound flimsy. Especially if you find the idea of absolute truth attractive. But put it this way: if scientific theories were boxers, who’d be more credible? The guy who welcomes a challenge, goes round after round over thousands of fights but has never, ever been beaten? Or the guy who boasts he’s the strongest, best fighter in the world, but won’t step in the ring?

The Encyclopedia Britannica puts it beautifully: “Scientific theories are instead incrementally corroborated through the absence of disconfirming evidence in a number of well-designed experiments.” Like a fighter, a scientific theory is a reigning champion until it’s overthrown by a theory that survives more and better challenges.

2. Replicability

Not only does a scientific theory have to be testable, it’s got to be a test anyone can repeat and get the same results. Makes sense. Theories are intended to tell us about how the world works–not just how it allegedly worked that one time in the lab. If you publish a recipe for a fluffy soufflé, but everyone who tries it gets matzo bread, you have not made an advance in culinary science. Scientific theories have to be replicable.

3. Correlation is not causation

As a child, did you think that going to bed at night caused the sun to come up in the morning? It’s a pretty sweet and innocent example of how correlation (the sun coming up after you’ve gone to bed) is not causation (going to bed doesn’t cause the sun to come up.) Confusing correlation with causation is a logical fallacy.

Examples of this fallacy are countless: A studio head takes credit for the new blockbuster movie, even though it was her predecessor who green-lit the film three years before. A newly elected politician takes credit for a recent economic upturn which resulted from long-term international events beyond his control.

In combination with confirmation bias, the fallacy has a huge potential to mislead.  So science has this one big important rule that just because something is correlated with something else doesn’t mean it caused that something else. Maybe it did… But let’s go seek some disconfirming evidence first.

For a crash course in scientific literacy check out Beyond this Vale of Testing: Post-Empirical Science from the podcast Stuff to Blow Your Mind.