The results of the Republican presidential primary in Michigan are another blow to the use of pollsters by mainstream news operations, which had declared John McCain likely to to win the Michigan election. Here’s a graphic used by MSNBC to report a poll on the morning of the Michigan primary.
John McCain at 27 percent and Mitt Romney at 24 percent is not what the actual results looked like. Mitt Romney got about 39 percent of the vote, and John McCain got about 30 percent.
Pollsters acknowledge that their work isn’t reliable as a prediction of an election, yet news companies keep on using the polls as if they do predict elections well. As a result, the narrative of the elections get distorted.
In the case of the Michigan primary, Mitt Romney‘s victory is now being told as a come-from-behind stunner, when in fact it was entirely predictable that Romney, a Michigan native with a lot of his father’s political contacts in the state, should come in first on the Republican side. John McCain’s loss is being described as a dissapointing disaster for his campaign, when it was never really clear why McCain should have won the state in the first place… except for what the polls said.
Will TV news stop using pollsters’ work as a predictor of the results of elections that haven’t been held yet? Only when they get some other shorthand way for predicting which candidates are ahead and which are behind. It’s just too boring for the TV audience, they’ve concluded, to discuss the ideas, strategies, and dynamics that could contribute to a more substantial analysis of a primary race. The numbers of a poll can be described quickly, and with pretty graphs that don’t require a lot of words to describe.
