Science: Why we can’t say yes…
No, I’m not trying to suggest that we are all grumpy, but rather, after reading more than a few dissertations over the last few days, I identified a common mistake: Virtually all of them tried to proof prove (ouch! Thanks! SD) that something worked. This made them rather “flat” pieces of work, which usually were pretty boring to read and – dare I say it – appear pretty “shallow”. The problem is, that science does not work that way – and that really the “scientific method” can never prove something works, but rather that it “probably isn’t wrong” (which really isn’t quite the same thing).
The typical rational I was reading went like this: We have a hypothesis (or educated guess based on previous research), we have observations – and based on some statistical magic (usually represented in the form of p) we claim our hypothesis is true. The fundamental error is, of course, that p does not say that the hypothesis is true – but that the likelihood of the null hypothesis being true is whatever the value of p is. So, in social sciences we generally go with 5%, if p is lower than 0.05 – then that does not mean that we can be 95% sure of the hypothesis being true – but we can be 95% sure of the null hypothesis being true. In other words we can show that the explanation we used to derive the null hypothesis is probably not right – but that does not make the hypothesis right.
Take a simple example: One could come up with the hypothesis that the current rainy weather is caused by the steam coming from all the kettles being boiled because of the cold weather (this is a rather clumsy adaptation of the butterfly effect, really). We can now try and prove that this is in fact the case – but will never succeed, as we simply don’t have the tools to do this (apart from arguing the case, of course). Thus, we need an alternative approach (in the form of the scientific method, really). In other words, we try and figure out all the ways in which we could prove that this wasn’t the case – and then try and disprove every possible alternative explanation (and we do have the tools to show that these are actually NOT true). Thus, once we have done that, then we are left with no other possible explanation other than suggesting that possibly the rainy weather is indeed caused by all the kettles. In other words, our hypothesis of the link between kettles and rainy weather has matured into a theory.
However, once we have that theory, it would potentially take only one observation to destroy the theory – even if we had hundreds of previous observations supporting the hypothesis. In other words, being scientific about it, we would never be entirely sure that kettles really cause the rain – but (unless we have found a convincing observation which disproves it) we can accept it as the best possible explanation.
Realising this fundamental point, really changes the dynamics of a good dissertation (or other academic work). In other words, we have a real case of a threshold concept here (my colleague Simon Manyiwa explained the importance of threshold concepts here). The structure of the discourse changes from “Yes, I’m right” to “Yes, I’m not wrong” – or in other words disproving every argument that could be brought up against the main argument, which, ultimately makes a really good piece of work which is interesting to read – rather than a flat piece of work that only aims to prove that the author is right in the first place.
And now, it’s time for a cup of tea… (and sorry for the rainy weekend ahead!)