At what point does a poll’s margin of sampling error become ridiculously huge?

The National Journal has new swing state polls out. Minnesota is included, and here, Barack Obama sports a 10-point advantage, 50 to 40 percent.

But what caught my eye was the survey’s margin of sampling error — a hefty 4.9 percentage points, plus or minus. That puts Obama’s double-digit margin barely outside the margin of sampling error. (Obama’s actual support could be as low as as 45.1 percent, and McCain’s as high as 44.9 percent.)

Each of NJ’s state polls — which include Pennsylvania and Wisconsin, where Obama has similar double-digit leads — sampled a paltry 400 voters. Why even bother at that level?

Oh yeah, to get guys like me to link. And by the way, hat tip to MPR’s Tom Scheck at Polinaut.

No comments yet

Leave a Reply