Thursday, December 09, 2010

A cautionary tale from a statistics instructor

Imagine that, some time in the future, you have the worst insomnia anyone in history has ever experienced. So you sit on your favorite chair and turn on the TV to pass the time, in hopes of finding Ferris Bueller's Day Off playing on TBS; sadly, it isn't. While you flip through channels, you notice a commercial for insomnia treatment drugs. In the commercial, which features images of beautiful people sleeping restfully in beds with the finest fabrics, a voiceover says that Brand A and Brand B were compared to see which is more effective. The voiceover continues and says that these comparisons indicated that one of the brands is more effective, so effective in fact that the FDA says that anyone who ever has trouble falling asleep should take this drug now!

But it doesn't tell you which one works better.


After you throw a slipper at the TV, the voiceover then continues and says that the test of Brand A and Brand B was carried out on both men and women, and that the brands worked differently for men and women.


But it doesn't tell for which group the brands work better or worse. Just that the effectiveness of the brands differs for the groups.


After you throw your other slipper at your pet, you realize that you should write a note to that stats teacher you had years ago and apologize for saying that main effects and interactions are significant without saying anything more. And then you cry a little, wishing that the commercial were more informative. And eventually, finally, slumber ensues, but your dreams are filled with a voice saying the words significant effect ... and nothing else.


Please help your statistics teacher keep his or her slippers on, and from abusing any pets.