Polls Apart

After reading and listening, over the past five days, about the growing inevitability of Barack Obama winning the New Hampshire primary, I was somewhat surprised with the actual voting results.

After all, the polls had predicted a large Obama win, based on a succession of tracking polls taken since the Iowa caucus.

What happened? Well, polls are based on statistic sampling of "populations". Opinion polls are inherently risky because opinions can change suddenly. In the case of the New Hampshire primary, it was hard to define the "populations" to sample; since the pool of voters includes Democrats, Republicans and Independents who can wait until the last minute to decide which, if any, primary in which to vote.

Many independent voters may have decided to vote in the Republican primary because of the predictions of an Obama win (taking them out of the pool of possible Democratic primary voters).

The TV media, which did an abysmal pre-vote job in overly concentrating on the poll predictions and touting an expected Obama win and the demise of the Clinton candidacy, was quick on election night trying to explain "what went wrong" without exploring the inherent limitations of political polling.

In any case, Iowa and New Hampshire are small parts of the American electorate and may be poor predictors of which two candidates will ultimately win their party's nomination.