Otherwise known as the scientific method, or How Science Works. You were probably taught the basics of this in secondary school/high school/your local equivalent, but the number of physics undergraduates who came through my university not having the first clue about how science works is large enough that I suspect it’s not quite being hammered into people’s brains the way it should be. Which is a shame1, because not knowing anything about the scientific method is what allows so much anti-science to flourish in the press and media at large. I guarantee you homeopathy – for example — wouldn’t last a second in a world where people were the least bit curious about what was going on under the bonnet, and neither would the hundreds of news “stories” about how chocolate is actually good for us.
So this is going to be a brief précis on how science works with a couple of classic examples thrown in that should hopefully prompt at least a couple of people who read it to not just accept a piece of research’s final conclusions, but instead to spend a second or two looking at their methodology to see if it’s solid and if it adheres to the basic structure of the scientific method. If it doesn’t then there’s a fairly high chance that there’s some commercial bias involved; I know that more than a few of the “chocolate is good for us” pieces are actually funded by chocolate manufacturers who want people to feel less guilty about buying and consuming their product, while the last thing the peddlers of homeopathic pills want is for someone to take a magnifying glass to their method because it makes no scientific sense whatsoever.
Science is an evidence-based discipline. This is the key feature of science, so I’m going to repeat it a couple of times. Science is an evidence-based discipline. Science is an evidence-based discipline. If you don’t have some kind of objective, verifiable evidence that proves what you are saying could be true, what you are doing is not science. Whatever scientific idea you’ve come up with needs to be built on previously existing evidence that provides it with a strong support, otherwise it’s going to collapse alarmingly quickly under the detailed scrutiny of other trained scientists.
Speaking of, science is also based heavily around the concept of peer-review. Scientific research has to be as open as possible, with each stage of the process exhaustively documented so that other people can exactly reproduce that research if they want to. Getting independent confirmation of a result is very important, and so is letting other scientists in the field pick over the details of your research for any flaws. You might have carried out your experiment with the best will in the world and merely reported what you saw at the end of it, but that doesn’t necessarily mean you have done a science. Your apparatus might have been broken in some subtle way. You might have set it up in the wrong place. There might be some other unknown factor at work that you just didn’t think of but which alters the experimental results significantly. Exposing your research method to the rest of scientific establishment means that it’s going to be looked at by dozens of experts in the field2 you’re exploring, and vastly decreases the chance of any flaws making it through into the final work.
Anyway, let’s say you do have some kind of coherent scientific idea based on previously-existing evidence. Contrary to what you might think, this is not a theory. What you have there is a hypothesis, which you might recognise as the thing most experimental research sets out to test. The reason we differentiate hypotheses from theories is because it’s rather easy to tweak a “theory” to fit pre-existing scientific evidence. If you have one cat-shaped jelly mould and somebody uses it to make some jelly, predicting that the jelly is going to come out of the mould in the shape of a cat isn’t exactly going to indicate astounding precognitive abilities on your part because you knew the mould was shaped like that in the first place. It isn’t enough for your nascent hypothesis to make sweeping statements about the universe we already know and think we understand; in order for it to be accepted as a new theory that is better than the currently existing theory, it has to make a unique, experimentally verifiable prediction about the universe that is at odds with – or at least more accurate than – the old theory. It’s easy to come up with a hypothesis that accounts for experimental outcomes that have occurred in the past; it’s rather more difficult to come up with a hypothesis that accounts for experimental outcomes that have yet to occur at all. If you manage it without access to a time machine there’s a very good chance there might be something in your hypothesis.
The most dramatic example of a successful theory overturning an old one in this fashion was when Einstein’s general relativity proved more accurate than Newton’s theory of universal gravitation at predicting the deflection of light as it passed close to a large gravitational source – in this case the Sun. General relativity had already accounted for aberrations in the precession of Mercury that Newtonian gravity couldn’t, but the existence of those aberrations had been known for centuries and it was entirely possible that Einstein might have written his theory with the sole purpose of explaining them away. In order to be accepted general relativity had to predict something new that had never been tested before, and in this case Einstein and Arthur Eddington decided to attempt the measurement of the deflection of light waves by the Sun in 1919. Cavendish had predicted this would happen in a Newtonian universe waaaay back in the late 1700s, but the key difference between Newtonian gravity and general relativity was that Newton’s formulation predicted a degree of deflection that was half that predicted by general relativity. Eddington and other collaborators around the world waited for a total solar eclipse (this being the only time that deflected beams of light wouldn’t be completely drowned out by the ambient light from the Sun) and then measured how much the positions of stars close to the edge of the Sun’s disc appeared to shift from their actual, known positions in space. After they’d cranked through the calculations the result was unarguably final: Einstein’s prediction fell within experimental error of the actual observed deflection of light, while Newton’s was a long way outside it. After this result was independently confirmed Newton’s theory wasn’t exactly thrown out, but it was seen as just an approximation of what was really going on with gravity which was more adequately explained by general relativity.
That’s a good example of the scientific method working exactly how it’s supposed to: a hypothesis is formulated, tested, found to be more accurate than the currently existing theory and subsequently supplants it in the scientific lexicon. It’s also possible for negative outcomes to result in this rewriting of the scientific handbook; just look at Michelson and Morley’s interferometer experiment back in 1887. Before we understood that space is a near-total vacuum it was thought that in order for light waves to make it to the Earth from the Sun (as well as other sources) it would need some medium through which it could propagate, like ocean waves propagate through water and sound waves propagate through the air. Scientific orthodoxy therefore held that the entire universe was permeated by a substance called the lumineferous aether through which light could travel. Such a medium would have produced fluctuations in the velocity of light depending on which way the Earth was travelling at a given time, and so Michelson and Morley built an experiment that would show these velocity shifts as an effect of the Earth’s motion against the “aether wind”3. They were rather surprised when they discovered that the speed of light was completely invariant no matter which way the Earth was going, which neatly disproved the whole concept of the lumineferous aether in the first place and directly led to Einstein’s formulation of special relativity two decades later.
Not being able to account for a subsequent experimental result can lead to the death of an old theory even if there’s nothing to replace it. This is the core of the scientific method; not only do theories have to make it through this arduous process of formulation, testing and gradual acceptance into the scientific mainstream, but they have to be on their metaphorical toes even after they’ve made it in life as one null result could end up invalidating the whole thing. Scientific theories are constantly – constantly – being tested in this way to increasingly insane degrees of accuracy, and so you can be sure that anything that still carries the “theory” nomenclature is about as correct as we can make it. Anyone who says “But it’s just a theory!”4 is betraying their basic ignorance of how science works and probably has trouble tying their shoelaces in the morning to boot. The Big Bang may have happened 13.7 billion years ago but Big Bang theory has made several crucial predictions about the universe which have subsequently been confirmed experimentally. Evolution is trickier to spot in action, but even there the weight of observed evidence supporting the theory’s case both before and after its formulation is so overwhelming that it is (so far) the only plausible explanation for that observed evidence. The scientific method isn’t without its flaws – there’s an old adage that new theories are only fully accepted once the supporters of the old theory physically die out – but it’s the best we’ve been able to come up with and it is astoundingly successful.
Next week: what happens when people ignore the scientific method, otherwise known as “making stuff up”, otherwise known as “newspapers will print anything because journalists don’t understand how science works,” Hopefully you do now, though.
- Actually what I want to say here is that it’s a tragic shortcoming in our societal attitudes that science is perceived as a “hard” subject and something that should be mostly left to experts, which allows anyone who can dress themselves up like an expert to peddle whatever crap they want under the guise of science.
- Peer-review isn’t perfect by any means, as the various scientific paper databases are set up so that they’ll prioritise research that is referred to a lot by other people. This is reasonable, as a paper that gets mentioned a hundred times in the reference sections of other papers is probably a very important piece of work, but it can also turn the submission process into a Kafkaesque version of social media where some of the people commenting on the work you’ve done have a vested interest in getting you to mention their research at some point in order to increase their pageviews. You would be amazed at how petty things can get, especially in niche areas of research.
- The lumineferous aether was supposed to be moving relative to the Sun’s motion around the Milky Way, meaning that the Earth’s motion around the Sun should have produced apparent shifts in the relative velocity in the aether wind – think running down a track on a still day and nevertheless feeling the movement of air around your body. By taking measurements six months apart when the Earth was travelling in completely different directions through space it was expected that this would produce a large shift in the relative velocity of the aether wind that would be reflected in the light waves that travelled through it and showed up on an interferometer. It didn’t.
- The notable exception here being string theory. I’ve gone over my objections to string theory before and I’m not entirely sure how it’s ended up with the “theory” moniker since it is entirely based on pre-existing scientific evidence and has never – and probably will never – be experimentally verified in any way that counts.
Interesting. A lot of what you say here can be cross-applied to other things, like political polling.
Political polling can almost seem like black magic at times. I remember during the 2010 election the BBC had a guy from (I think) Ipsos Mori on who had done exit polling that showed the Lib Dems had done nowhere near as well as their high of 30-odd points after the first debate had indicated. The BBC presenters were somewhat incredulous, but as the night wore on it became obvious that the Lib Dems had only made marginal gains despite their previous high poll position. Ipsos Mori guy had correctly predicted this, but every other poll up until that point hadn’t which seemed to me to point towards a flawed methodology. Of course I know nothing about political polling so I have no idea what really happened there.
Measuring people’s preferences is difficult because people often aren’t sure what they want themselves, or else will lie to pollsters (a typical one was where people said they’d vote Labour in 1992, but didn’t want to be taxed higher so voted Conservative instead). This is a problem in selling products, because you’re never sure of the best price point until you try and see whether people buy at that price. People may say “I’ll buy that!” until they have to actually spend money, at which point they’re like “eh”.
You’ve gracefully left out the bit when scientist turn human and ego and deliberately and on purpose try to destroy each others theory/work.It’s just not science and tea party…
I know of one anecdote where in Italy archeologists threw fake stone into a field at night so that by morning the other team would “hopefully” pick up their stones, and the the carbon dating would have given false dates on the find – thwarfing some theory about human migration. (I am still searching for the paper, it’s fairly old, supposed to have happened in ’50s or so)
Excellent article thanks!
As Josh says, it’s obviously applicable outside of Science (or Science is more applicable to things that aren’t necessarily science). It’s not a hard concept to understand, but lots of people don’t really take it on board.
As you say, it’s not just people who do humanities at university. They are just as capable of understanding and using the principles of the scientific method as scientists are of not or misusing it.
Also – not the best Holmes! Does Jeremy Brett have an MBE for it like Vasily Livanov does? No, didn’t think so.
Also, don’t trust the Scientific Method. The Scientific Method leads to Communism. Don’t want that.
Huh, why does scientific method lead to communism? Never thought about that before.
Wait, chemistry leads to scientific method, which leads to biology and physics. Brainhurt.