Lies, damned lies, and statistics: Polling means nothing (yet)

This recent poll asking voters whom they preferred in the Toomey vs. Sestak race is more than mildly amusing, it’s ridiculous. First off, the headline, “Toomey leading Sestak in latest Poll” is technically accurate, but if you read on you’d find that 38 percent of those surveyed are undecided.

Let’s break this down: 577 people were asked their preference. Toomey (31%) is “winning” by 3 percent more than Sestak (28%).

That’s 17 people who say, unequivocally, that they support Toomey over Sestak. There’s another 219 people who are undecided, and apparently 19 who don’t count at all.

161 say they will vote for Sestak.

178 say they will vote for Toomey.

219 say they are unsure.

Of course, the best part about the survey is that it has a margin of error of 4%, which means that you could read it:

• 35% support Toomey, 24% support Sestak (Yay Toomey!), or …
• 27% support Toomey, 32% support Sestak (Yay Sestak!), or …

You could just ignore the whole thing, as a more appropriate headline would be:

“Race is wide open: 38% of voters for Senate seat still haven’t made up their mind.”