Commentary
With Trump Win, Polls And News Orgs Face Crisis Of Credibility
- by Erik Sass , Staff Writer @eriksass1, November 9, 2016
Armed with such data, the media made confident predictions of a Clinton victory.
Indeed, it’s remarkable just how wrong so many of the polls were in retrospect.
Casting an eye back over the surveys on the last day before the election, Monday, November 7, as compiled by RealClearPolitics, in the four-way election (including Libertarian candidate Gary Johnson and Green Party candidate Jill Stein) Bloomberg had Clinton ahead of Trump by three points, CBS News by four, Fox News by four, Reuters/Ipsos by three, ABC/Washington Post by four, Monmouth University by six, Economist/YouGov by four, Rasmussen by two, and NBC News/SM by six.
In a two-way comparison of Trump and Clinton, the same polls generally gave Clinton margins as big or even bigger than those listed above.
One survey, the USC Dornsife/Los Angeles Times “Daybreak” poll, at least deserves credit for bucking the trend by showing Trump was still very much in the running – but even here the results skewed considerably the actual returns, just in the opposite direction, giving Trump 48% to 43% for Clinton in a two-way match up.
Overall, none of the polls, whether four-way or two-way, came within four cumulative percentage points of the actual result, an effective tie in the popular vote, with both candidates receiving 48%.
Newspapers and other big news orgs, which based their predictions on these polls, and generally made no secret of their disdain for Trump, now face a crisis of credibility with their readers, who could be forgiven for simply disregarding poll results in future contests.
In fact, in an ironic twist it may be the news media’s strident opposition to Trump that is to blame for the discrepancies.
While it’s hard to know for sure, the near-unanimous condemnation of Trump by news media may have made some poll respondents reluctant to indicate support for his candidacy when canvased by strangers on the phone.
On that note, some pundits — not to mention Trump himself — have drawn parallels between his victory and the British vote this summer for “Brexit.” It's another case where expert and elite opinion was so strongly against the final result that it seems likely poll respondents said one thing but ended up voting another.
Erik Sass is editor of Publishers Daily. You can reach Erik at eriksass@gmail.com.
Data is only as good as the questions, the sample and the analysis. The fault was in not recognizing a significantly changed context and with the two most disliked candidates in history adjusting for, what I like to say is, "consumers lie"! If the polls had focused on change and the customer instead of the brand maybe they would have been more accurate but not in traditional polling terms. Let's also remember that Trump's own polls had him losing as well?