Sometimes I could swear there's a pattern in my data. Not in any one experiment, but across all of them. You know how, in an ideal world, every experiment is done double-blind to guard against experimenter bias creeping into the data? Well, it's like that, except the exact opposite: it feels like every experiment I run produces the result I least expected. That's not the reality, of course; if I step back and examine this one series of studies that's confounding me at the moment, I'm about 3 for 6. In part, it's just that the unexpected results hit me harder. Still, 3 for 6 feels kind of lousy. Fundamentally, (theoretically?) all data is good data, and when the data tells you to modify your theory, you modify your theory, and get closer to the truth. But if you go 3 for 6, that means you've modified your theory 3 times, and you start questioning whether you're making progress, or just stupidly meandering around in theoretical circles.
Much of the time, I love science.
Today, science has stamped "return to sender" on my metaphorical valentine.
Arrest This Man, He Talks In Maths 
Wednesday, February 15, 2006