Posted: 8:46 pm Wednesday, November 5th, 2014
By Jamie Dupree
After seeing polls in 2012 that often produced a result which underestimated support for President Obama and overestimated the backing for Republicans, the quick review of polling in the 2014 elections raises even more questions about the accuracy of those surveys.
Clearly a rough night for us and much of the polling industry and we'll own that and try to figure out what happened in the coming days
— PublicPolicyPolling (@ppppolls) November 5, 2014
“This was a difficult year for polling,” wrote Doug Rivers, reviewing his work on polls by YouGov,
in which he acknowledged they ended up overstating support for Democrats.
An initial study by the website fivethirtyeight.com shows that 2014 had a bias of 4.0 points in favor of Democrats in races for Governor, and a significant amount in certain Senate races:
+ In Iowa, the polls overestimated Democratic support by 7.0 percent
+ The bias in favor of Democrats was almost 9 percent in Kentucky and over 12 percent in Arkansas
+ In Georgia, the poll average tilted 7 percent in favor of Democrats
+ In New Hampshire, it was 1.2% in favor of the GOP
Two years ago, the controversy was that the polls were supposedly “skewed” against the Republicans – but when the election was over, we found out that the polls had gave better numbers to Republicans than they deserved.
In 2014, it was the opposite – the polls were skewed more in favor of Democrats, making it look like they were doing better than reality.
“The polls may be biased again in 2016,” wrote Nate Silver of the fivethirtyeight.com website, “we just won’t know much about the direction of it until votes have been cast and counted.”
Here is a graphic produced by Silver on how the polls have missed, and which side they have favored in recent elections:
Covering politics with and without the polls
It’s sort of hard to cover an election for Congress without getting involved with polling data; what I tried to do in 2014 was not get overly focused on the polls, simply because the 2012 misses were so fresh in my mind.
Instead of taking poll results which showed one candidate ahead or behind, I tried to look for data on where swing voters or independent voters were going, and to pay more attention to the direction of the polls – were they moving to one candidate or another.
But the allure of polling data is difficult to resist in the news business – and in the election business.
Man so did those polls suck or what? Hey, any new polls out this morning?
— Kyle Kondik (@kkondik) November 6, 2014
Rightly or wrongly, polls can drive news coverage – if we think a race is a toss up, it tends to get more attention, rather than a race where one person is ahead by double digits.
As we move towards 2016, the polls will again become very important in the race for the White House.
But no matter how hard people work to fix them, I’m still left (as a reporter) wondering if the data I’m seeing is right, or is it skewed to one party or the other.