Margins of Error

The usually indispensable, which I check eagerly every morning, today contains the following howler:

Remember that Zogby saying PA is a (47-47) tie means the pollster is predicting that there is a 95% chance that the true score for each candidate falls in the range 43% to 51%, no more and no less.

Of course it means much more; for example it means that the candidates are substantially more likely to be about one point apart than to be about four points apart. But compounds the error:

All the battleground states are statistical ties. A couple of percent lead means nothing

Acutally, as no reader of will need to be told, the candidate with a 2% lead among those surveyed, if that lead is within the reported “margin of error” has less than a 95% chance, but still greater than a 50% chance, of having the lead among the public at large (ignoring, of course, all of the potential problems with surveys other than sampling error.)

Since everyone knows this, it’s hardly worth pointing out, except for the fact that these silly statements about margins of error have become ubiquitous in the press, and it’s worth asking why. Surely it must be obvious to all reporters and most readers that it’s better to have a two point lead in the polls than a two point deficit. So why does everyone keep saying otherwise?


Comments for this post are closed