The results are in — Bernie Sanders won Indiana, his 18th primary victory. But the polls were wrong. Again. Why?
Online research does not reveal a primary in which Hillary Clinton was predicted to do worse than Sanders, and then things majorly flip-flopped. Not once has the surprise been that Clinton won big where she was predicted to lose.
But on numerous occasions, the media has reported “surprise” with a Bernie Sanders’ primary win, because he won despite polls that indicated otherwise.
With his latest win in Indiana, the AP published the headline, “Bernie Sanders pulls out surprise win in Indiana.”
Politico used the headline, “Sanders surprises Clinton in Indiana.”
RealClearPolitics recently reported that Clinton would likely beat Sanders in Indiana, with Clinton winning 50 percent, and Sanders 46 percent. Polls typically have a margin of error of about plus or minus 4 percent. But this still doesn’t account for the margin by which Sanders won.
Sanders actually won Indiana by 5 percent. How did he go from down-in-the-polls by 4 percent to up-in-the-actual-primary by 5 percent? In other words, how were this and virtually all other polls wrong by nine percentage points?
If the answer is that polls are just a prediction and by their nature will often be wrong, why are the polls in this election contest usually wrong by predicting Clinton will do better than she often does and Sanders will do worse?
It is not just the polls that end up wrong. Several media stories relying on the polls publish information in favor of Clinton when Sanders ends up winning.
On the same day of the Indiana primary, the Orange County Register published a story about sources that claimed Clinton would win in that evening’s Indiana primary.
Their headline ran, “Who will win Indiana? Five views point to Trump, Clinton victories.”
All were wrong about Clinton.
In some of the primaries, Sanders has won by so much more than the polls predicted that his outperformance of the polls became a main part of the media reports, such as in the New York Times story about the Michigan primary.
“Jaws dropped Tuesday night as Bernie Sanders defied the embarrassingly incorrect polls and shocked election watchers with a narrow but important upset of Hillary Clinton in Michigan.”
In Michigan, Sanders won when polls had predicted he would lose by 20 points.
Despite losing the New York primary — amid voter accessibility issues — Sanders won the majority of counties in that state, which was not predicted by the polls.
In Wisconsin, Sanders lost by a lot less than predicted, prompting stories to write that, “Bernie Sanders shocks with 41 percent of the vote.”
And when Sanders has won in landslides, no polls predicted such big wins. ABC reported that Bernie won Alaska, Hawaii, and Washington with 82, 70, and 73 percent, respectively — shattering all predictions.
Sanders has now won 8 of the last 9 primaries:— Ben Norton (@BenjaminNorton) April 9, 2016
There is not a current, publicized investigation underway to figure out why the polls are almost always off, in a way that predicts that Sanders will not do as well as he ultimately does.
But since the polls have consistently predicted that Sanders will earn fewer votes than he goes on to win, the following questions arise.
If the latest RealClearPolitics poll says that Sanders beats Trump nationally by 14 points, does that mean that Sanders would actually beat Trump by more than 14 points, given that this polling source is one of the ones that has been wrong by underestimating Sanders?
If, as the Washington Examiner is reporting, Sanders is beating Clinton in national polls by two percent, and in the Indiana contest the poll was off by 9 percent, is Sanders really beating Clinton by 11 percent nationally?
No one has officially answered these questions. But Bernie Sanders has won another primary — Indiana — a “surprise” to those who continue to rely on polls that consistently under-predict Sanders’ performance.
[Photo by Joe Raedle/Getty Images]