You might say this is a small practical experiment. Following on from our discussion of science and humanism the other day, it occurred to me that while reason is of course a very good thing, and God can never be anything other than absolutely rational (if only we knew what the rational thing was), the scientific method is a different matter. Life is not a laboratory, and I don't mean that metaphorically. There are indeed circumstances in which we can determine what course of action to adopt by testing hypotheses, such as how much incense a particular congregation can stand wafting from a thurible. The thurible, the incense, and probably the people, remain the same from instance to instance: the variables are limited. But most of the time we inhabit immensely complex, nay chaotic, systems, and the accurate replication of experimental circumstances which is central to the scientific method becomes impossible. For instance, if you are a finance minister debating whether to adopt a particular fiscal strategy, you can't repeatedly test your guesses out on the level of an entire national economy and exclude all the variable circumstances which would make the resulting data meaningful. It is exactly those variable circumstances which make for so much dissension among economists (I wait for Dr Abacus to comment). There's only so much the scientific method can do: and beyond it, we're left with the arguable.
Absolutely. Some things are pretty easy to predict. If you increase the price of petrol, people will buy less of it. They will do this by driving less, and driving more slowly on motorways. People with two cars will use the more economic ones more often. And when people buy a new car, they will buy a more economic one if petrol prices are high. You can see this very clearly in the US - low gas prices lead to more SUV sales, high gas prices lead to more Toyota Prius sales. Economists new the sugar tax on soft drinks would work. And that cigarette taxes work.
ReplyDeleteYou can also test some things. HMRC have tested different letters, to see whether they can persuade some people to file their tax returns on time. Some letters work better than others.
Predicting the extent of a response is harder than predicting the direction. If we added 20p to the price of petrol I would be very confident in predicting a fall in petrol sales. But predicting the extent, in either the short term or the long term, would be much harder. As a rule of thumb, a 1% rise in price causes a 1% fall in sales, but the margins of error around this are large.
The hardest thing for social scientists to changes in tastes, and the effect of new products. The rise in vegetarianism is not caused by a rise in the price of meat relative to veg. It is exogenous to prices. And when Henry Ford built the first affordable car, he was confident he would sell a lot of them. But how many? No-one knew. And no-one knew what things people would stop buying. (The answer was pianos - piano sales drop dramatically, as people spend their money on cars instead. Both were commonly bought on credit).
The hardest thing to predict is foreign affairs. WW1 was nuts - there was every incentive to avoid it. But it still happened.