Some People are Very Confused

The 2012 election may be remembered as the election of the polls. Contemporary media coverage requires constant, 24-hour stories to entertain the public. However, often there are no entertaining campaign stories on a given day. Oh boy, Obama and Romney both held extremely conventional rallies in Ohio. Thrilling. Polls, on the other hand, are released constantly, giving the news media a stream of news stories. The campaign is treated as a horse race, with constant updates on which candidate is ahead.

This has made poll analysis remarkably popular. Nate Silver of the New York Times blog fivethirtyeight provides daily analysis and commentary of the polls. Major TV networks have their own poll analysts who chip in their own commentary. The problem with this influx of polling analysis is that sometimes polls do not create an interesting story. For example, if current polls reveal the state of the race accurately, Barack Obama is clearly winning the election. That is not a particularly interesting story. As a result, some pundits have lashed out against polling, often in ways that demonstrate that they do not understand what they are critiquing.

TV pundits have frustratingly pulled predictions out of nowhere. Jim Cramer of CNBC predicts Obama will win 440 electoral votes, and carry states such as Texas. Meanwhile, conservative pundits have consistently predicted a Romney victory. Polls appear to indicate such a victory is unlikely, so these pundits are ignoring polls altogether. Unfortunately, this makes their predictions somewhat absurd, such as Dick Morris’s prediction that Romney will win by more than 100 electoral votes or Michael Barone’s prediction that Romney will win by nearly that much. Barone doesn’t even pretend to examine polls when he makes his predictions. Morris says that the polls are skewed because they oversample democrats (and by skewed, he means about 10 percentage points off). Historically, polls have not oversampled either party, and it is perplexing that every poll would make the same oversampling mistake. Morris’s argument fails basic scrutiny.

Obama could certainly lose, because of low enthusiasm or other factors. It’s just not very likely, based on the actual data out there. The disdain for polling data and actual facts in favor of gut calls and baseless assertions demonstrates the extent to which many pundits are entertainers rather than analysts. I hope and expect that Obama wins by the margin data-driven analysts project, so that hopefully when 2016 rolls around pundits who pull predictions from the ether will be laughed off the TV. A future where analysts actually analyze would be a tantalizing reward after a long, bitter campaign.

Share your thoughts