Thursday, March 22, 2007

Lies, Damn Lies, and Statistics

Screen after screen of tables flashed across the screen. Here and there an asterisk appeared beside a bar graph or table. "You know what that means," exulted the speaker. "That means this difference is statistically significant. It means that every time you repeat this study, you'll get the same results."

Umm. Not exactly. At least not when I studied statistics. I know it's been a long time and what the hell, I took statistics for Psych majors, not statistics for nurses, but it can't be that different, can it? They accepted the credit when I transferred.

Oh, and the Chi Square was 0.1, so it really means there's only a 10% probability that the results were due to chance. Given the nature of the study, I'd say there's a hell of a lot more probability than that. They didn't appear to control for much more than the one element they were trying to prove was significant.

I very nearly raised my hand again to correct the speaker, but I'd already done that twice, so I settled for rolling my eyes and figuring she's either not as smart as she looks, or she's bought into the whole big pharma promotion of this particular product.

At least I got a good dinner out of the deal.

No comments: