Thursday, 9 February 2012

Good science, bad science

Something I try to get in books and talks that an awful lot of people presenting science skim over is that science isn't about absolute fact. All science can ever be is our best guess given the current data. Tomorrow new data may emerge that totally overthrows our current thinking.

This has happened repeatedly in the past. Something like Newton's laws of motion can seem set in stone (hence the 'laws' label, which I don't think is a good idea). And then along comes Einstein and shows they are wrong. Not wrong enough to throw them away - they still work well in many circumstances - but wrong nonetheless.

This is why I get a bit irritated when popular science presenters and writers make sweeping statements like 'the universe began 13.7 billion years ago in the big bang.' What they should be saying is 'according to our best supported current theory, which fits the data well (though it's not entirely surprising as bits of it have been changed to do so), the universe began 13.7 billion years ago in the big bang.' Now, I admit that's rather clumsy - but I think every popular science book should have a proviso that what is being described as if it were fact is the current best theory, and this may change.

Note that this isn't an argument in support of the 'evolution is just a theory, so we ought to teach intelligent design' brigade. I didn't say a 'current theory'. I said the current best theory - the one best supported by the evidence and that works well with our other current best theories. This is not a recipe for taking any old hypothesis with the same confidence as the best theory. But it is bad science to suggest that our favourite theories (especially those like the cosmological ones which are based on very indirect data) are fact.

That's one kind of bad science. Another is cheating. We tend to think of scientists as emotionless seekers after the truth, but if you ever meet a scientist (treat them nicely - buy them a drink!) most are normalish human beings. With human tendencies. And there is a well established psychology of the way we fool ourselves in order to get the results we want. This inevitably happens in science. Some of the best known experiments in science have not really produced the results the scientist wanted. So they just went ahead and ignored the results. Newton, for instance, didn't get what he really wanted in his experiments with light using prisms. No matter. He knew the desired result and that's what he wrote up. Whenever anyone writes about general relativity, they talk of Eddington's 1919 Principe expedition which proved the expect effect of the Sun bending light. Only it didn't - certainly not within acceptable margins of error. But Eddington announced the result he wanted.

It would be silly to think that this doesn't happen all the time, and I'm delighted to be able to show the graphic below (technically an 'infographic' but I hate the term) from Tony Shin and his team (I've seen it elsewhere, but it's worth repeating) on the subject. I ought to say that we shouldn't take this as all bad. Many experiments are't fiddled. Just because a result is tweaked doesn't mean it's wrong. And over time science's system of repeating results ensures that problems are ironed out. But we can't pretend it doesn't happen:

Bad Science
Created by: ClinicalPsychology.net

1 comment:

  1. For another take on this graphic, and the larger issue of accuracy in infographics, you might want to take a look at http://researchnews.osu.edu/blog/?p=1279.

    ReplyDelete