By Jerry Kroth
Social scientists have long known that releasing poll information early, before polls have closed, has two effects: first it decreases voter turnout by about 12 percent,[1] and it increases the bandwagon effect, where people hop on and vote for the winner, by about 8 percent. [2]
On the morning of Super Tuesday, before anyone had voted, the Associated Press released a story that Hillary Clinton had already won. She was the "presumptive presidential nominee" and the victor. AP had made that announcement because of a super delegate count and decided she already beat Sanders.
Other media outlets then piggy-backed on this story, and virtually every American woke up that morning to headlines that Hillary had won-and remember, that is before anyone voted on Super Tuesday.
What a surprise! By the time you had your morning coffee and went off to the polls, you already knew Mrs. Clinton was the winner. Did that bias the election? Did it discourage people from voting? Did it create a "bandwagon effect?"
If one looks carefully at the percentage totals for Clinton versus Sanders totals for those primary states, it is clear the so-called "landslide" victory of Clinton on that day was fully within this margin of bias created by the bandwagon and voter turnout effects.
In other words, the AP story determined the outcome of this election.
Strong words? Well, let's look at the data.
Three days before the election, a Yougov poll showed Clinton leading Sanders by two points in California. But after the Associated Press released its story, Clinton beat Sanders not by two points but by 13! Hillary got an 11 point "bump."
From somewhere.
The same effect happened in New Mexico. Sanders was ahead of Clinton by a wide margin 54 to 40 percent. [3] By Super Tuesday, the situation reversed and Clinton beat Sanders 51.5 to 48.5. That surprising result gave Hillary an additional 13 points. Surprise! A 13 point "bump."
In New Jersey, poll results just before Super Tuesday showed Clinton leading sanders 54 to 40 percent [4] but on election day she beat him 63 to 36, another unexpected 9 point "bump" in Hillary's favor.
In South Dakota, a poll showed Sanders ahead of Clinton by 6 percentage points [5] just a few weeks before the primary, but on Super Tuesday Hillary pulled another rabbit out of her hat and beat Sanders by two points; an 8 point "bump" for Clinton.
Those are the only states where we can calculate pre-post results. Hillary got an unexpected 9 points in New Jersey, 8 points in South Dakota, 13 points in New Mexico, and 11 points in California. All unexpected. All unpredicted. All quite different from polls held just days before Super Tuesday.
And all very suspicious!
If one tries to rebut these findings alleging they all are within the margin of error for polls, then Sanders should have had just as many spurious bumps as Clinton. Didn't happen! All went to Hillary. The skewing is not random! The statistical anomalies are consistently prejudiced toward Hillary.
Sixteen European countries ban reporting election results before voting occurs, and in the UK, reporting poll data on the day of the election is forbidden. [6]
All for good reason.
Serious attention should be paid to declaring these primaries invalid. Furthermore, the possibility of investigating media entities, in particular Gary Pruitt, CEO of the Associated Press, for any alleged collusion with the Clinton campaign should be aggressively pursued. Even if there is no corporate media complicity, it can still be argued that the AP's desire for an early morning scoop determined, biased and corrupted this entire election.
Jerry Kroth, Ph.D. is Associate Professor Emeritus Santa Clara University. He may be contacted through his website, collectivepsych.com
Notes
[1] http://www.politics.as.nyu.edu/docs/IO/2797/FrenchVoting_13.pdf
[2] http://rap.sagepub.com/content/1/2/2053168014547667
[3] http://www.nj.com/politics/index.ssf/2016/05/whos_leading_in_polls_as_nj_primary_approaches.html
[4] http://www.nj.com/politics/index.ssf/2016/05/whos_leading_in_polls_as_nj_primary_approaches.html
[5] https://www.isidewith.com/poll/801555698/9333341
[6] https://aceproject.org/ace-en/topics/me/mea/mec03c
