Friday, April 03, 2009

Statistically Gtalk doesn't wear Capri pants


Hello everyone out there! I don't post much because I am usually busy telling people why they should keep paying us to keep posting for y'all. Quick bio, I am a physicist trained at University of Texas (hook 'em) who now works for APS and likes to knit, play video games watch reality TV.

Statistics are funny things. You read about statistics all the time in the media. "4 out of 5 dentists agree....." but what does that mean? How many dentists did they ask? Were they all buddies of the person asking? How were they asked? Are they real dentists or do they just play them on TV? These are all things you need to think about before believing a statistic. One of the biggest marketing ploys is to tell you that a correlation, meaning when one variable changes another seems to change at the same time, implies that one variable is causing the change in the other: causation. But why should it? Every time I mention Capri pants in gtalk, my gtalk crashes. Absolutly every time. So obviously there is bug in gtalk that hates Capri pants as much as I do. Now, really, does that make sense to you? Why would gtalk care about poor clothing choices? This is an absurd example, but hopefully one that gets you thinking. What are the stats you are seeing really telling you?

Statistics can lie to you, but they are also extremely valuable in science if used correctly. Stats are the basis for almost every scientific study, but only when used correctly. When physicists really investigate if correlation implies causation in the case they are studying they can learn about the world. But they have to work hard at learning what their statistical data means and throwing out the data that shouldn't be in there. This is one of the most important parts of the scientific method. What if you take some data and realize that for one or two data points you were standing on an important hose and blocking air flow (true story). Can you use that data in your statistics? This is what physicists have to ask themselves all the time.

So next time you look at a statistic, look critically and figure out what it is really saying. Oh, and if you want to stop global warming, become a pirate.

No comments:

Post a Comment