Thursday, October 03, 2002

[ email from Ananda -- ]

Hey, how's it going? I was just checking in on your blog and I wanted to pick a nit -

> The drug that helped our little girl measured a 65% rate of
> success among the group who were administered the drug.
> But because the placebo result was a startling 50% percent,
> it meant that the drug was only 15% better than nothing.

As a journalist I have to deal with this like this occasionally -- it's not 15 percent better, it's 15 _percentage points_ better (65 percent minus 50 percent). It's actually 30 percent better (15 percent divided by 50 percent).

[ /email from Ananda ]

Crap, you're right. I seem to have fallen on the same statistical landmine that I believe the journalists who first reported on the landmark report that seriously challenged the effectiveness of hormone replacement therapy for menopausal women.

I have enrolled in three post-secondary scholarly programs in my lifetime. I have taken 'intro to stats' twice now and may have to take again if I want to complete a degree that I'm sort of doing in my spare time. Maybe it's time I take the refresher.

No comments: