It’s always interested me that inspite of what you might expect, PhD training doesn’t tend to come with any formal lessons on how-to-be-a-scientist. There’s a distinct lack of any proper training in philosophy of science, even though it’s something we’re supposed to use every day in order to be good at what we do. The most important pieces of scientific lore tend to be imparted to you as you go along, and perhaps the most important thing which you’ll almost never see any officially taught lessons for is how to construct a scientific hypothesis.
You see, as scientists, we revolve around hypotheses. Science is all about crash testing ideas for a living, but a hypothesis is more than just an idea. Firstly, it needs a point. Is there something which can’t be fully explained using existing theories? Or perhaps something which you could explain better? You need to take your idea and make sure it’s worth investigating. Analyse it, check it and check it again. It seemed like a good idea in the shower this morning, or while you were on the bus last night (good ideas seldom happen at opportune moments), but is it still a good idea when you hold it up to the light and check it for imperfections? If you had to defend your idea would you be able to do so? Would you be defending it because it’s actually any good, or simply because it’s yours? These are all things you need to consider if you’re trying to evolve your idea into a hypothesis.
In truth, at least 90% of ideas never make it as far as becoming hypotheses. For one reason or another, those ideas just weren’t good enough, and were discarded. And If you’re serious about making a hypothesis, you must be your own harshest critic. As a scientist you can’t afford to become attached to your ideas, no matter how much you might like them. If it doesn’t work, it doesn’t work. Move on. Think up some new ideas instead.
For an idea to make a good scientific hypothesis, it needs to satisfy five main points:
To be a proper scientific hypothesis, it needs to be testable. If there’s no way to test your hypothesis, then it can’t actually be a hypothesis. You need to be able to decide if your hypothesis is true or false based on real experience (in other words, you need to experiment). Also, your hypothesis must be falsifiable. There may be an experiment or observation which gives a result which is in conflict with it. Importantly, the results of your experiments must be reproducible. If someone elsewhere in the world tries the same experiment, they should get the same result.
Your hypothesis should be clear and simple. It shouldn’t require anything unnecessarily convoluted, and nor should it need any oddly specific circumstances. If two hypotheses can exist to explain the same set of phenomena, then the simplest one will always be favourable. In other words, you need to satisfy Occam’s razor.
Explaining one thing is good. Explaining lots of things is better. A good hypothesis can, in one broad sweep, explain a number of cases of phenomena. If you came up with a good hypothesis, you may well find that it explains other phenomena without even trying.
Does it stop at what it does now? Or is there a chance it could continue to be useful? A good hypothesis will continue to explain things in the future. Ideally, future discoveries will mesh with it, and be explainable using the same principles.
Radical ideas are good, but exactly how radical is your thinking? How well does it fit? Do you have to tinker with existing knowledge a little to make space for your hypothesis, or would it require you to re-write half of Physics? Too much conservatism is a bad thing, and I’ve heard people lamenting the number of times the phrase “our work agrees with the previous studies” is uttered in conclusion to things. But established fact is established for a good reason, and if your work says that Einstein and Newton were both wrong, then actually chances are good that it’s you who’s wrong and not them.
So… Is it looking like your idea will make a good hypothesis? Great. What now?
Well first off, you’re not going to do anyone any good if you just put it on a pedestal and admire it. No, your hypothesis is not an exhibit. Test it! Try dropping it. Did it break? Well then give it a kick. Still intact? Get your colleagues to throw it around a bit. Smash it into a wall. Still solid? Ok, then it has a chance. Let the whole world see it. Let a team of postdocs from Germany run over it with a car. Get a couple of PhD students in Thailand to drop it off a bridge. Have a blogger from Kenya stomp all over it. Let a professor from Brazil hit it with a hammer for 14 hours straight. As comical as this sounds, it’s pretty much how peer review works.
So… did it survive? It takes a lot to get this far, but if anyone in the world who’s paying attention can’t dismantle your hypothesis, then congratulations! Now you can take it one step further and evolve it into a theory. People may use the words “theory” and “hypothesis” interchangeably, but they’re not the same at all. Not in science anyway. To formulate a theory, you need to have a hypothesis which has survived at least a couple of attacks by manic hammer-wielding professors. Once it’s been found to be bulletproof, then and only then can you build it into a full framework of ideas and concepts that qualify it to be a scientific theory.
Image credits: Unknown
(These images were acquired from somewhere amongst the internets. If you know where they’re originally from, please let me know!)