I knew Tom and I would eventually agree on something, and his spotted dog example is right on target. We are programmed by evolution to see patterns around us, it's a survival tactic, and we are so sensitive to these patterns we often see them when none exist. Here's an experiment you can try for yourself. Next time a program on dance comes on TV, be it ballet, hip-hop, country line, ballroom, whatever; turn off the sound and turn on the radio and randomly tune it to a music station, any station. This was a game we used to play in college all the time (smoking a joint first helps, but its not an absolute requirement). After a few minutes, the dancers appear to start dancing in time to the music, because our mind imposes that order where none exists.

Science is full of this fool-yourself type of phenomenon, Lowell's Martian canals come to mind, and for years biologists thought there were 48 chromosomes when there are only 46 (or is it the other way round?). The first guy got it wrong and nobody bothered to verify it for years! Psychologists are notorious for this type of pattern error. They come up with a new model of the mind about every six months. You can even argue that this is what led the Bush team astray over WMDs, they only saw what they wanted to see. I've never blamed the Bush staff for making that mistake because I made it too, faced with the same evidence I saw what I wanted to see.

This is why we can't trust astrologers to evaluate their science, it's not that they're dishonest, they're human. They will impose an order on their data which does not exist, no matter how objective they try to be, because they want it to be true. And they are not trained to analyze their data statistically the way the hard scientists are. This pattern error plays a huge role in our political debates too...we remember what meets our expectations, and forget what doesn't. This is why left and right can argue interminably when neither is lying.

Here's another example. People are always telling you how the moon affects behavior, emergency room staff, bartenders, cops, people who see a lot of tragedy, will swear that things are always crazy on full moon nights. However, actual attempts to capture this effect statistically have repeatedly turned up nothing. I once got access to the city police traffic accident records for a whole year and binned them by age of the moon in days, from the Nautical almanac. As I expected, there was absolutely no corellation with age of the moon and auto accident frequency. I did find all sorts of other patterns, school year, tourist season, start of rainy season, weekends, Fridays (happy hour), Thursdays (pay day) but not the moon. The variation from new to full moon for ever day of the lunar month was within the tolerances you'd expect for a truly random phenomenon.

Even if there HAD been a "Lunar Effect", how would those cops, bartenders and ambulance drivers know it? I'm an astronomer and I don't keep track of the phase of the moon, and half the time it's overcast, so you can't see it anyway. The full moon does have the property that it rises at sunset and sets at sunrise, so if we notice the moon at all it is most likely to be the full moon, its up all night.