Iowa Polls and Statistical Foes

I haven’t yet confirmed my intention to secure the Democratic nomination, but that won’t stop journalists from forecasting my demise to a Republican who has yet to officially throw his name into the ring for the third, consecutive, presidential election.

A new Des Moines Register/Bloomberg Politics poll shows that, were Mitt Romney and I to compete for the presidency in 2016, I would lose Iowa to a Republican who lost there in 2012. Or so journalists with either no understanding of margins of error or thirsty for page views (maybe both?) insinuate.

From a sample of 1,017 likely Iowa voters, the poll found that 44 percent would cast a ballot for Mitt Romney and 43 percent would cast a ballot for me. In other words, among the poll’s sample, I would lose to Mitt Romney by one percentage point.

The problem with deriving such a steadfast conclusion from the poll results, however, is that the poll has a margin of error of 2.9 percentage points. In other words, the poll forecasts that, were an election between Romney and I to happen today, I would receive somewhere between 40.1 and 45.9 percent of the vote in Iowa and he would receive somewhere between 41.1 and 46.9 percent.

Despite what the headlines and stories written about this poll say, then, it is impossible to say who is leading who in Iowa. If anything, the poll only definitively forecasts a dead heat between Romney and I. To derive any other conclusion is simply irresponsible.

Maybe instead of writing articles predicting my losses in an election 764 days away, journalists could spend that time brushing up on their abilities to accurately analyze and report on polls.