Poll, baby, poll!
But will polling accurately predict the outcome of November’s election?
IN LATE August, when most polls showed Barack Obama losing his lead over John McCain, Mr Obama’s campaign manager, David Plouffe, pooh-poohed the reports. “We don’t pay attention to national polls,” he said. Today, the question on many Americans’ minds is whether they should either.
The volatility of polls give good cause to wonder. Each day, a slew of new ones hits the American press, but they very seldom agree. Polls this week, for instance, showed Mr Obama with a lead as great as 14 percentage points or as small as zero.
One way that polls can be wrong, some say, is because of the high percentage of young people without landlines. Polling organisations usually call landlines, because federal regulations targeting telemarketers makes it illegal to dial mobile numbers automatically. But after a recent study by the Pew Research Centre, a non-partisan opinion research group, found that the exclusion of “mobile-onlys” (who are mostly young and pro-Obama) could introduce a bias into survey data, many polling organisations now feel pressure to invest the money and time to have humans call more mobile phones. Still, only some of them do so, and to differing extents, which could help explain the wide variation in polls on any given day.
To read the full article: The Economist
