Skip Navigation
Fellows

Is it Right to Report Results Before the Polls Close?

A new company intends to post election results before the polls close. Just because it can be done, should it?

October 3, 2016

The views expressed are the author’s own and not neces­sar­ily those of the Bren­nan Center for Justice.

Around lunch­time on Elec­tion Day 2004, the first wave of leaked exit polls stunned journ­al­istic insiders. Instead of the tight elec­tion that had been fore­cast, the early exits had John Kerry demol­ish­ing George W. Bush in a land­slide. Even South Caro­lina (which had last gone Demo­cratic in 1976) was tilt­ing Kerry’s way.

As you may have guessed, there were seri­ous meth­od­o­lo­gical prob­lems with the 2004 exit polls. But, at the time, there was a near certainty among the cognoscenti that Kerry had won the elec­tion. In fact, journ­al­ist Mickey Kaus memor­ably called Kerry “the seven-hour pres­id­ent.”

In the 12 years since that disaster, exit poll results have remained as hush-hush as Donald Trump’s tax returns. The rise of social media has made this bank-vault secrecy on Elec­tion Day imper­at­ive since any leaked numbers would reach millions of voters through Twit­ter and Face­book within minutes.

But all that is about to change on Novem­ber 8.

A private company called VoteCastr—with a bipar­tisan bent and an impress­ive collec­tion of data mavens—has announced that it will provide a form of elec­tion returns from seven swing states in real time as Amer­ica votes. Slate, which published leaked exit poll numbers in 2000 and 2004, will trum­pet the VoteCastr calcu­la­tions to the world.

Journ­al­ist Sasha Issen­berg, a friend and the author of an acclaimed book on campaign tech­no­logy, The Victory Lab, explained VoteCastr’s innov­at­ive meth­od­o­logy in a lengthy article in Slate. Accord­ing to Issen­berg (who is a director of the company), VoteCastr will monitor minute-by-minute turnout in key precincts and then match it to models predict­ing the likely candid­ate share of the vote in these precincts.

Using a Clev­e­land precinct as an example, Issen­berg wrote, “So when a tracker visits Denison Element­ary School and reports via a smart­phone app that 78 people have voted by 11:15 am, VoteCastr can determ­ine how well each candid­ate is faring. By blend­ing this inform­a­tion with reports from dozens of other precincts across the state, VoteCastr’s stat­ist­ical models can predict who is, at that very moment, winning Ohio and by what margin.”

Just because you tech­nic­ally can do some­thing, should you do it?

The argu­ment that Slate editor Julia Turner makes is that the Elec­tion Day embargo on exit poll data or other voting inform­a­tion “is pater­nal­istic toward voters and puts journ­al­ists in the awkward and unfa­mil­iar posi­tion of conceal­ing inform­a­tion from their read­ers.”

Issen­berg makes a more subtle point: Since campaigns view the elect­or­ate through the lens of precinct-by-precinct target­ing, voters deserve to know how they have been profiled. Turnout models help victori­ous candid­ates under­stand what their elec­tion means. As Issen­berg puts it, “A candid­ate who is sworn in think­ing he won by turn­ing out a party base is likely to govern differ­ently than one who cred­its a last-minute swing from late-decid­ing inde­pend­ents.”

Although I am grimly recon­ciled to the ines­cap­able real­ity of real-time elec­tion returns this year while the polls are still open, I remain uncon­vinced by the argu­ments put forward by VoteCastr’s cheer­ing squad.

Even in the age of Trump, seri­ous public­a­tions (includ­ing Slate) frequently make “pater­nal­istic” decisions about what their read­ers need, instead of what they merely crave. It is why, say, the New York Times does­n’t publish porno­graphy even though it might boost third-quarter earn­ings. Or why no respons­ible public­a­tion prints the iden­tit­ies of CIA agents or rushes to name the victims of a tragedy before next-of-kin are noti­fied.

Moreover, after the knot­ted 2000 results, the TV networks and the Asso­ci­ated Press have been right­fully cautious in making Elec­tion Night projec­tions. I am hard pressed to see why walling off the exit-poll analysts from the network news teams on Elec­tion Day has under­mined journ­al­ism or anything else.

I agree with Issen­berg that voters should under­stand how pres­id­ents inter­pret their mandate. But I am baffled why voters need to have that inform­a­tion—on a precinct-by-precinct basis—at 2:38 on the after­noon of Elec­tion Day. The data analysis from VoteCastr will be just as valu­able to voters, polit­ical scient­ists and campaign profes­sions on the morn­ing of Novem­ber 9.

The roots of the current Elec­tion Day reti­cence by the networks date back to the 1980 campaign when Pres­id­ent Jimmy Carter collapsed in the final week. Infuri­at­ing House Speaker Tip O’Neill, Carter conceded to Ronald Reagan just after 6:00 p.m. Pacific Time, while the polls were still open on the West Coast. Veteran Demo­crats like Oregon’s Al Ullman, the chair­man of the House Ways and Means Commit­tee, blamed their defeat on Carter making the earli­est conces­sion speech since the inven­tion of radio.

As the New York Times poin­ted out in its article on VoteCastr, the data does not support the claim that there was a dramatic drop off in 1980 turnout on the West Coast. But that may have been because Carter appeared doomed by Elec­tion Day or that the know­ledge of his conces­sion speech spread much more slowly in a 20th century news envir­on­ment. Even without stat­ist­ical evid­ence, it is impossible to believe that anyone would be spurred to vote by being told that the pres­id­en­tial elec­tion had already been decided.

Since its meth­od­o­logy is unproven, VoteCastr’s Elec­tion Day projec­tions prob­ably will not reduce turnout on Novem­ber 8.  But, if success­ful, VoteCastr and its likely compet­it­ors will become a staple of Amer­ican polit­ical life by 2020. As a result, getting voters to the polls would become an even more daunt­ing chal­lenge. About the best you could hope for in the future is that these early Elec­tion Day returns would be published with a large italicized warn­ing: “Spoiler Alert.” 

Walter Shapiro is an award-winning polit­ical colum­nist for Roll Call who is cover­ing his tenth pres­id­en­tial campaign. He has also worked for two news­pa­pers (USA Today and The Wash­ing­ton Post), two news weeklies (Time and News­week), two monthlies (Esquire and The Wash­ing­ton Monthly), and two online magazines (Salon and Slate). He has also been a colum­nist for Yahoo! News. He is the author of “One-Car Cara­van: On the Road with the 2004 Demo­crats Before Amer­ica Tunes In,” a chron­icle of the early skir­mish­ing for the pres­id­en­tial nomin­a­tion, published by Publi­cAf­fairs in 2003. Shapiro teaches a polit­ical science seminar on the news media and the 2012 campaign at Yale. And he is work­ing on a book about his con-man great uncle who cheated Hitler. He can be reached at walter­sha­piro@y­mail.com and followed on Twit­ter @MrWal­ter­Sha­piro.

(Photo: Think­Stock)