Tuesday, 29 June 2004

Regression to the Mean

The title might suggest a post about the English on holiday, but it's more about my holiday. Or, at least, the above phrase, which I used a lot on holiday.

It's a simple idea. If you do something really well (or badly) the next time you do that thing it probably won't be as good (or bad). Put like that it's quite obvious. All things people do tend to have a random factor built in, it's probably what makes them fun. That random factor means that something exceptional will be followed by something not so exceptional. It also means this will happen more often than not.

It's why the next time you go to that really great restaurant you just discovered, you'll probably be disappointed. And that really bad one must have hired new staff, or something.

One odd side-effect of this is that it often reinforces the wrong message. In that, when you notice something good (or bad) and praise it (or complain) that that thing will get worse (or better), it would have done that anyway, though. What this means is that it often seems that we get results when we tell people off, but it looks like they are getting complacent after we've complimented them.

Tests have shown, however, that positive reinforcement is better for people than negative. Short-term results mask long-term ones.

Anyway, I've found an article from the Skeptical Inquirer explains this much better than me and also shows how certain magazine curses are actually true... Sort of.

No comments: