Exit Polling around the World

Your weekly excerpt from one of my books. This week: "The Hidden History of the War on Voting"

The Republican War against your right to vote. Over the next few months the entire book will be here, new chapters posted every Sunday, for subscribers to read at no cost. If you want to get a physical book to mark up or share with others, just click on the picture above, visit your local bookstore, or check your favorite online seller.

Exit Polling around the World

Most Americans, when they hear the name Viktor Yushchenko, vaguely remember a Ukrainian politician lying in a hospital bed with a severely disfigured face, the result of poisoning with a deadly form of dioxin, the toxic ingredient in Agent Orange. But it was exit polls that got him there.

Exit polls are polls taken outside of voting stations or polling places, where people who have already voted are questioned as they’re leaving. They’re considered far more accurate than other types of pre- or post-election polling because they don’t rely on people answering their phones, returning a mailed inquiry, or asserting that they intend to vote when they may well not.

In a clean election environment, it’s safe to assume that nearly 100 percent of the people polled actually voted, and history shows that such polls are typically (outside the United States) accurate to within a fraction of a point, or a point or two at most, depending on how many people are polled.

Exit polls are held in such regard that when, in the 2007 Kenyan election, Mwai Kibaki (the government’s choice) was declared the election winner over Raila Odinga—who the exit polls reported had easily won—riots broke out, and, to quote the Carter Center, “more than 1,000 died and some 600,000 fled their homes.”77 In response, the Carter Center went to Kenya to monitor the re-vote (which was also severely marred by fraud).

Similarly, back in 2004 in Ukraine, Yushchenko—the reformer outsider candidate—was well ahead in the regular polling against Viktor Yanukovych. When the election returns came in, however, the government election commission reported that Yanukovych had won the election by 49.5 percent to 46.6 percent for Yushchenko. When it became widely known, however, that exit polling done by three different organizations concluded that voters had actually turned out for Yushchenko 54 percent to 43 percent for the guy the government said had won, people took to the streets in what was called the Orange Revolution.78

The Washington Post said in an editorial, “Despite the government’s brazenly unfair campaign, a majority of Ukrainians voted for . . . Yushchenko [and] authorities then tried to steal the election.”79

The US government, along with many European allies, declared outrage at the election fraud, proven by (among other events) the exit polls. They suggested that the election result tampering was orchestrated by pro-Russian supporters of Yanukovych. To quell the riots, a new election was called, and as Yushchenko began to fall ill from the dioxin poisoning, he was elected the new president of Ukraine.

In Germany, exit polls have been used for years to functionally call elections as soon as the polls close, even though hand counting of paper ballots can take days. They’re rarely

off by more than a fraction of a single point. They’re considered so reliable and so important that the German government criminalized releasing even preliminary results before the polls close; when two Twitter users leaked exit polls 90 minutes before the polls closed in 2009, it provoked a national scandal.80

Similarly, exit polls are routinely used to call elections all over the world, where paper ballots are almost universally used and thus can take days to count. A quick summary of AP headlines shows the reporting trend in the United States: “Exit polling indicates Peruvians vote to fight corruption,”81 “Poland: Exit poll gives centrists edge in key mayoral races,”82 “Exit polls suggest Irish voters have repealed abortion ban,”83 “Exit polls: Dutch vote on spying law too close to call.”84 And that’s just the Associated Press.

The British Broadcasting Corporation, a network sponsored by the government of Great Britain, where exit polls are also used to report election results before the paper ballots are counted, routinely uses exit polls all over the world to call elections.

A quick search finds elections being called by the BBC, in just the past three years, in Italy, the UK, Israel, Japan, India, the Netherlands, Haiti, Tunisia, France, Ireland, Bolivia, Bulgaria, Australia, Indonesia, Crimea, Portugal, Macedonia, Ecuador, the nation of Georgia, Kosovo, Latvia, Greece, Argentina, and even the Moscow mayor’s race in Russia.85

Every single one turned out to accurately call the election it headlined.

Exit Polls in the US and Red Shift Explained

We used to use exit polls that way in the United States too. Techniques were tried and refined on a variety of local bases around the country for decades through the mid-20th century, but Warren Mitofsky fine-tuned it from an art into a science and conducted the first real acid test of the technique in 1967, accurately calling the Kentucky governor’s race that year. Throughout the 1970s, virtually every election was called by the networks based on exit polls, and the technique was considered noncontroversial.

In the 1980 presidential election of Carter versus Reagan, the East Coast exit polling results were reported by NBC News hours before the West Coast polls had closed (the exit polls showed Reagan had won), producing bipartisan outrage; the networks promised to tighten up their handling of exit poll data, which was virtual news gold.86

Everything was going well until the 2000 election, when the exit polls clearly showed Al Gore winning the presidency (including in Florida, which, it turns out, he did win when all the ballots were counted by the news organizations a year after the election), and the networks called the election for Gore before all the states had reported their counts.

While election exit polls are still the gold standard worldwide, since that 2000 election they seem to have gone to hell in the United States.

In the 2004 presidential election, exit polls called John Kerry the clear winner with a margin of more than two million votes, even winning handily in Ohio, but this time the networks held back.

As ABC News reported in a postmortem of their reporting on the exit polling of the 2004 election:

“The exit poll estimates in the 2004 general election overstated John Kerry’s share of the vote nationally and in many states. There were 26 states in which the estimates produced by the exit poll data overstated the vote for John Kerry by more than one standard error, and there were four states in which the exit poll estimates overstated the vote for George W. Bush by more than one standard error. The inaccuracies in the exit poll estimates were not due to the sample selection of the polling locations at which the exit polls were conducted. We have not discovered any systematic problem in how the exit poll data were collected and processed.”87

The exit polling companies, in the four years since 2000, had developed a new strategy to report their polls—unique to the United States in its widespread use—in which they’d “adjust” their results to reflect what the individual states reported as the actual vote.

ABC News’ postmortem noted, “[T]he final exit poll data used for analysis in 2004 was adjusted to match the actual vote returns by geographic region within each state.”88 That “final” and “adjusted” data purported to show that John Kerry had won by only about a half-million votes, and he’d lost the decisive state of Ohio, which became the reporting the networks went with.89

In 2004, fully 22 states experienced what has now come to be called “red shift”—where the polls are “wrong” but almost always in a way that benefits Republicans.

For example, in the 2016 election, the exit polls showed Hillary Clinton carrying Florida by 47.7 percent to Trump’s 46.4 percent, although the “actual” counted vote had Trump winning by 49.0 percent to 47.8 percent. Trump gained 2.5 percentage points . . . somehow.90

In North Carolina, exit polls showed Clinton winning 48.6 percent to 46.5 percent, but the votes that were counted turned out with Trump’s 49.9 to Clinton’s 46.1, a red shift of 5.9 percentage points for the GOP.91

Pennsylvania’s exit polls showed that Clinton won 50.5 percent to Trump’s 46.1 percent, but when “eligible” votes were counted, Trump carried the state 48.8 percent to Clinton’s 47.6 percent—a red shift of 5.6 percentage points.

In Wisconsin, it was Clinton beating Trump in the exit polls 48.2 percent to 44.3 percent, but the “real” count put Trump over the top at 48.8 percent to 47.6 percent, a red shift of 5.1 percentage points.

Perhaps even more interesting, in states without a Republican secretary of state, there is virtually no shift at all, either red or blue, and hasn’t been ever. The election results typically comport with the exit polls in those states.

Given that red shift began to explode across the American electoral landscape in a big way with the 2000 election, and it continues to favor the candidates of one party in a way not seen in any other developed nation that does exit polling, a number of theories have evolved to explain it.

Warren Mitofsky, who’d been doing exit polling since the 1960s and invented the modern technique in the 1970s, found himself and his firm terribly embarrassed with the results of the 2000, 2002, and 2004 elections when the red shift numbers were enough to throw critical elections to Republicans. He came up with the theory of the “shy Republican voter,” which postulated that, for some reason, Republican voters were just simply embarrassed to tell exit pollsters that they’d voted for a Republican.

That theory was widely reported in the media and became the go-to excuse for adjusting exit poll numbers by changing them to conform to state-reported results after the 2004 election.

Few people buy it, however, particularly since there are no similar examples in any other nation in the world, even where a winning leader may otherwise be seen as a war criminal or buffoon. Exit polls—except when there’s clear fraud—are the single most accurate way to measure an election outside of counting actual ballots.

Voting Machines, Hacking, and Red Shift

Given that much of the red shift that America has seen in the past three decades exploded after the passage in 2002 of the Help America Vote Act (HAVA), which allocated billions of dollars to the states to buy electronic voting machines from private corporate vendors, many people alarmed by widespread red shift were quick to blame the machines.

And, indeed, they are easy targets.

The 2002 senatorial election in Georgia, done entirely on electronic voting machines that produced no paper or receipts, was severely marred by accounts of lost memory cards containing votes from largely urban areas and produced a result that flipped the polls upside down.

War hero Max Cleland, who’d left three limbs in Vietnam and was nationally famous and popular (and ahead in the Georgia polls by five points a week before the election), was defeated by eight points by Saxby Chambliss, a Vietnam War–era draft dodger who’d run a bizarre campaign questioning Cleland’s patriotism.92

The Georgia governor’s race that year saw a similar reversal of poll versus outcome results favoring the Republican challenger, Sonny Perdue, who was seven points down in the polls but beat incumbent Democrat Roy Barnes with a 16 percent swing on Election Day, something unheard of in modern politics absent a last-minute scandal (and there was none).93

Similarly, in the 2018 Georgia election, the Republican lieutenant governor candidate, Geoff Duncan, beat Democrat Sarah Riggs by 123,172 votes. Inexplicably, the Georgia electronic voting machines—which still don’t have any audit ability or paper trail—registered slightly over 160,000 voters who simply chose not to vote for either of the lieutenant governor candidates. When Politico investigated, “the Georgia Secretary of State’s office did not respond to repeated requests for comment.”94

And when a lawsuit was filed against Secretary of State Brian Kemp (who both ran the election and ran successfully against Stacey Abrams for governor) to access the actual votes, a server was mysteriously wiped clean in a way that prevented even the NSA from recovering its data.95

Frank Bajak reported for the AP, “A computer server crucial to a lawsuit against Georgia election officials was quietly wiped clean by its custodians just after the suit was filed, the Associated Press has learned.” Bajak said, “It’s not clear who ordered the server’s data irretrievably erased.” The lack of data effectively killed the lawsuit, and when the AP repeatedly inquired of the agency that wiped the server, “It did not respond to the AP’s question on who ordered the action.”96

Howard Dean rather famously hacked into a Diebold election computer tabulator and changed the results of an election in 90 seconds on CNBC while filling in for Tina Brown on her Topic A show on August 8, 2004. When he was made chairman of the Democratic National Committee, the video was pretty much scrubbed off the internet, although recuts of it pop up from time to time.97,98

There was also, to add flames to the conspiracy fire, the simple reality that the two biggest voting machine vendors in the 2000s were banking giant Diebold, whose CEO, Wally O’Dell, famously wrote a leaked 2004 letter promising to “deliver Ohio for George W. Bush,” and Election Systems & Software, which was started by two Christian end-times-rapture-believer brothers and then passed through other GOP-connected hands over the years.99

Hacker conventions, year after year, have featured demonstrations of how easy it is to hack a wide variety of voting machines used in the United States—in 2018, the hack of a clone of the Florida election system was accomplished in less than 10 minutes by an 11-year-old.100

It’s an example of why Ireland, after experimenting with American voting machines for one election, sold its $80 million worth of machines for scrap metal (for a mere $79,000); it refused to resell them as voting machines, taking a huge loss on the deal, so that there was no chance any other country would buy them and make the mistake of using them in an election.101 As the New York Times has documented, among others, our intelligence services are worried about how easily foreign governments (particularly Russia, China, and North Korea) can get into most states’ election systems.102 The states themselves, however (at least those controlled by Republicans; California famously decertified all of its machines in 2004), continue to stonewall or refuse to change to more secure systems.103

And while it’s not hard to believe that in a state with a centuries-old tradition of election fraud (mostly by white people against black people) like Georgia, “losing” memory cards with votes on them104 or even “patching” machines in the weeks before the election without notifying anybody (both things that are well documented)105 could have thrown an election, it’s harder to conceive of it as a multi-decade national conspiracy.

On the other hand, voter suppression very much has been at the core of a multi-decade effort by the GOP and may well explain red shift as much as hacked or rigged machines. We’ll circle back to that in the next chapter.