Perhaps the most important thing to keep in mind is that, although national polls are extremely useful in giving us a good sense of how the race is going, we choose our President through the electoral college, not by direct popular vote. It's an obvious point, I know, but one that is often ignored by the chattering classes on network and cable television - and even by those in the print media. The state-by-state polls are what really matter.
Secondly, the margins of error (MOEs) in any poll - national or state - are very important. These MOEs are typically in the range of plus or minus 3 to 5 percentage points. Thus, real voter sentiment can vary as much as six to ten points from what the raw poll numbers indicate. So far, we've seen only few national polls in the presidential race which give one candidate or the other a lead outside the margin of error. The Washington Post/ABC News poll, released today and discussed in detail in the blog below, is a notable exception. Obama's nine point lead is well beyond that survey's three point (plus or minus) margin of error.
Finally, it is helpful to remember how many real voters these national poll numbers represent. Assuming that the turn-out in this year's election is at least as high as four years ago, each percentage point equates to roughly 1.3 million voters. So, a nine point lead in a national survey, such as we see in today's Washington Post/ABC News poll, equals over eleven million voters. No small potatoes, that.
1 comment:
Another thing to warn about in polls is not to pay too much attention on swings between a subgroup in one particular poll.
If say, we're talking about the Catholic vote which usually accounts for 25-30% of the vote, the sample of Catholics will typically only be 25-30% of the poll. So if there's a big swing among Catholics in that poll, keep in mind the MoE will be extraordinarily high.
Post a Comment