« »

Monday, October 29, 2012

Politico provides excellent example of shitty polling reporting


Sen. Scott Brown is locked in a tie with his Democratic challenger, Elizabeth Warren, according to a poll released Monday, with Warren’s popularity slipping over the past month.

Brown and Warren are neck and neck with 47 percent each, according to the Boston Globe poll. Without including voters who are leaning toward one candidate, Brown has a slight edge, 45 percent to 43 percent.

Most of the polls conducted in the past month have shown Warren gaining an edge on the incumbent, who is trying to hold on to the seat he won in a January 2010 special election. A WBUR poll released last week showed Warren with a 6-percentage-point edge, 50 percent to 44 percent.

The piece goes on to report in extremely definite terms how Warren’s favorables have plummeted and her campaign is suddenly shrouded in doom. Remember how I warned to to watch out for headlines like “LONE POLL SHOWS SURPRISING RESULT; RACE TURNED ON ITS HEAD!”? This would be one of those headlines.

If we head over to FiveThirtyEight, we see that Warren has led in the last five polls by between 2 points and 9 points. The polling average for the race is Warren +4.5. It’s rated not as lean democratic or likely democratic, but as safe democratic. Warren is given a 95% of victory.

In short, judging from the history of the race and the polling, this is almost certainly an outlier. Politico is hyping a poll result that turns the race on its head — mostly to get you to click on the link. The odds of this race going the way this single poll shows are extremely slim.

This isn’t reporting, this is a traffic strategy. I provided the link to the story to be thorough and because of best practices — there’s no reason you should reward this sort of behavior by following it.

Search Archive:

Custom Search