Umm ... speaking as someone who's done some psychology at A-Level, the best I can come up with is 'to an extent'. To phrase it another way ... huge numbers of people do tend to behave certain statistically-predicatble patterns of behaviour - an example, there are 60m people in Great Britain, most of whom wear clothes. However, you always get exceptions - example, those occasional streakers at big sports events. Essentially, the problem with forecasting human behaviour is that of demand characteristics; when you do the research upon which to base your maths, you have to do it in such a way that the research itself doesn't prompt a particular kind of behaviour - you want people acting as naturally as possible, so as to correspond to how they actually do behave. And finding ways to control for this experimental bias are one of the biggest nightmares in social science. It's why psychologists don't seem to bother much with error bars - you'd rarely be able to wrestle the b***ers down *as far* as 10%, let alone to the point where you actually have a reliable set of conclusions. Essentially, no model is ever better then the data that went in, and people can be a bloody pain when they put their minds to it.
![[personal profile]](https://www.dreamwidth.org/img/silk/identity/user.png){kind=small}
(no subject)
Date: 2003-10-17 05:43 am (UTC)To phrase it another way ... huge numbers of people do tend to behave certain statistically-predicatble patterns of behaviour - an example, there are 60m people in Great Britain, most of whom wear clothes. However, you always get exceptions - example, those occasional streakers at big sports events.
Essentially, the problem with forecasting human behaviour is that of demand characteristics; when you do the research upon which to base your maths, you have to do it in such a way that the research itself doesn't prompt a particular kind of behaviour - you want people acting as naturally as possible, so as to correspond to how they actually do behave. And finding ways to control for this experimental bias are one of the biggest nightmares in social science. It's why psychologists don't seem to bother much with error bars - you'd rarely be able to wrestle the b***ers down *as far* as 10%, let alone to the point where you actually have a reliable set of conclusions.
Essentially, no model is ever better then the data that went in, and people can be a bloody pain when they put their minds to it.