By day Sam Wang works as a professor of molecular biology and neuroscience at Princeton University, a job that has given him some national attention through his bestselling books, “Welcome to Your Brain” and “Welcome to Your Child’s Brain.”

By night Wang has what he calls “a little hobby,” tracking and statistically analyzing presidential opinion polls. That work, updated at his Princeton Election Consortium website (, has brought him even more national attention since he began analyzing poll data in 2004. That year he predicted the electoral vote total exactly: Bush 286, Kerry 252.

In 2008 he called the race for Obama, 364-174. The actual result was 365-173.

In the 2012 election, while Gallup and other pollsters were either calling the election for Mitt Romney or describing the race as too close to call, Wang called the race correctly in 49 out of 50 states. His statistical analysis of the polls almost exactly matched the final popular vote outcome, with Obama beating Romney 51.1 to 48.9 percent. In 10 close races for U.S. Senate, Wang predicted each outcome correctly. His results put him in the same league as (and sometimes slightly better than) Nate Silver, who runs the highly publicized website.

This year, in what surely has been the most surprising, tumultuous, dramatic, and emotional election season ever, with one of the candidates operating totally outside all political norms, Wang has remained steadfast in his confidence of the polls and how he evaluates them. “This has been a very strange year,” Wang told a group of Princeton University alumni on October 5. “But despite all the emotion, this has been one of the least suspenseful and least volatile elections.”

On October 5, Wang’s aggregated data — based on state polls only rather than less accurate national polls — showed Clinton leading by roughly 3 percent in the polls of the popular vote, with a 91 percent chance of winning the electoral college. On Tuesday, November 8, after brutal headlines about Donald Trump’s sexual boasting and the FBI’s re-awakening of the Clinton e-mail controversy, Wang’s Princeton Election Consortium median showed Clinton over Trump, 307 electoral votes to 231. Wang’s website estimated that Clinton would win the national popular vote by 4 percent. Wang gave her a 99 percent chance of winning the Electoral College vote.

The correct pronunciation of Wang’s name rhymes roughly with “wrong” — as in Wang is almost never wrong. If that’s the case can we then assume that Wang’s prediction will turn into a guaranteed victory for Hillary Clinton and that the country will be looking forward to its first woman president when this paper gets circulated on Wednesday, November 9?


In fact, in 2012 while many in the media were celebrating Wang as the new gold standard in assessing public opinion polls, the Princeton professor was scratching his head over something he actually did get wrong that year — his prediction that Democrats would regain control of the House of Representatives. In analyzing that mistake Wang’s post-mortem led to a conclusion: that gerrymandering of congressional districts following the 2010 census had been more partisan than most people ever realized. And that conclusion led Wang to create a three-part test that would allow a judge to determine if a district has been configured to deliberately to benefit one party over the other. (See excerpt at right from David Daley’s new book, “Ratf**ked,” showing how Wang’s work has influenced the anti-gerrymandering movement.)

So if Wang is wrong in this tumultuous year, he will not only eat a bug (as he promised to do in 2012 if Romney had upset Obama), but he will surely go back to the statistical drawing board, to figure out where and what he and the collective public opinion polls had missed. And if he is right about Clinton, he will still be looking carefully at the down-ballot races. As of election day the Princeton Election Consortium was predicting an 82 percent chance of Democrats taking back control of the Senate or earning a 50-50 split. The website predicted that, while Democrats were likely to beat Republicans by about 1 percent in the nationwide congressional vote totals (about the same as 2012), Democrats would remain a minority in Congress. The Princeton Election Consortium cited a Cook Political Report that projected an eight-seat gain for Democrats, still far from a majority.

So no matter what the outcome of the national election, expect Wang to continue his work on the gerrymandering issue, which he shares with the public at

To Wang public opinion polls are not just numbers that television newscasters use to highlight the horse race between candidates on the nightly news. Wang views the opinion polls as instruments to measure a set of data, in the same way a thermometer measures someone’s temperature. If the measurement is off, then the instrument needs better calibration.

Wang is just 49 years old, but data and mathematics have been obsessions since he was a kid — one of two children of Taiwan immigrants who came to California in the 1960s and went into real estate development. When he was about eight years old Wang came across a book about mathematics. He devoured that book and snapped up as many more as he could. He graduated in 1986 (at the age of 19) with a B.S. in physics and then earned his Ph.D. in neuroscience from Stanford.

Wang came to Princeton as an assistant professor of molecular biology and neuroscience in 2000. His research today is focused on understanding how the cerebellum affects cognitive function. According to a university press release, Wang’s lab “is exploring the theory that the cerebellum plays a teaching role during sensitive periods of development and plays an important role in organizing the rest of the brain.”

Because the cerebellum is the most commonly aberrant brain region in people with autism, Wang wants to know “what patterns of inactivation of the cerebellum are linked with social deficits.” The research could show, he says, “one way by which brains can become autistic. It could also make some contribution to understanding where autism comes from generally.”

His research, Wang continues, “requires analytical techniques for dealing with large amounts of data.” In that sense his day job informs his “little hobby” dealing with election polls.

“When faced with complex data, the difficulty is to extract an understandable, simple meaning from a large data set,” Wang says. “My general approach to data analysis, to the extent possible in neuroscience, is to take all the observations we make in the lab and try to come up with some relatively simple fact that can be stated about the data. The thing that’s in common in these other areas is that I’m just using tools that I use in the course of my research.”

As Wang told the Princeton alumni meeting in early October, he does not just gather and assess the polling data to provide entertainment for the nightly news. “Politics are important to our lives,” he says. “I use the math to allow people to allocate their resources.” In a close election for the Senate or the House of Representatives, a relatively small number of people can have a major impact. The Princeton Election Consortium identifies close races in both houses of congress — for example, in the eighth district of Pennsylvania, which includes all of Bucks County and a portion of Montgomery County.

Says Wang: “It doesn’t matter if you are a Democrat or a Republican. If it’s close you can have a major influence.”

While all that’s true at the congressional level, the opposite seems to prevail in the presidential race. There used to be a great deal of volatility in presidential races, says Wang, but that has not been the case since 1992, when Ross Perot’s strong showing as an independent caused substantial voter realignment.

Since then, Wang says, “something has happened in our politics to make us much more fixed in our opinions.” And, as wildly different as Donald Trump is from Mitt Romney, their position in the polls compared to their Democratic Party opponent is roughly the same. “It’s indicative of the intense voter polarization.”

A month later, at a lecture presented prior to the Princeton-Penn football game on November 5, the Saturday before the election, Wang greets an audience that includes more than a few Hillary supporters who are looking for reassurance. The October 28 letter from FBI director James Comey is still casting a shadow over the Clinton campaign.

Nate Silver’s has lowered its chances of a Clinton victory to 65 percent. Not so great by Wang’s standard, who says he views chances of 75 percent or less as highly uncertain — “worse than if you were playing Russian roulette.” Of all the various organizations charting the national and state election polls, Wang’s is the highest at 99 percent. But at the November 5 event, even Wang admits his model might be too aggressive, and by tweaking its methodology a little, Clinton’s chances might drop to 91 or 92 percent. He nevertheless sticks with the higher number.

For now Wang’s attention is still focused on the Senate races in Pennsylvania, North Carolina, New Hampshire, Nevada, Indiana, Missouri, and Wisconsin, where the margins are extremely close. He is also focused on the gerrymandering issue — the one that won’t go away no matter who wins on November 8.

Wang’s November 5 audience includes a New York lawyer who is part of a Common Cause lawsuit challenging a gerrymandered Congressional district in North Carolina. Wang, who has devised a three-part statistical test to enable a judge or other arbiter to determine if a district is unfairly configured, is expected to be a star witness in that case. Wang is hoping that courts of law will buy his analytical approach and help make Congress truly representative. As Wang says, democracy used to mean that “voters would choose their representatives. Now representatives are choosing their voters.”

But, he says, “the math can clear up a lot of confusion.”

How Re-Districting

Became a Dirty Trick

In his new book David Daley, editor-in-chief of Salon magazine, traces the Republican Party’s current control of the House of Representatives to the party’s dismal showing in the 2008 national election and its realization that by controlling state legislatures it could also control the inevitable re-districting that would accompany the 2010 Census. This coordinated effort was facilitated by contributions that flowed as a result of the Citizens United case, decided by the Supreme Court in January, 2010.

The title of Daley’s book, “Ratf**ked, The True Story Behind the Secret Plan to Steal America’s Democracy,” comes from the political term made popular by Bob Woodward and Carl Bernstein in “All the President’s Men.” Herewith an excerpt:

Sam Wang has sleuthed out the impact of gerrymandering as well as anyone. His real passion is good government. . . . In 2012 he called 49 of 50 states, but more impressively, predicted the exact two-candidate percentages for Barack Obama and Mitt Romney, 51.1 to 48.9. For that, the Washington Post’s Wonkblog named Wang the best election modeler of the year, ahead of even the amazing Nate Silver, then with the New York Times. . .

Accolades or not, Wang, wasn’t impressed with his own work. He had argued throughout the fall that Democrats had a shot to take back the House in 2012 if they carried the aggregate popular vote, as seemed likely given Democratic turnout in a presidential election year. Many experts told Wang that his models missed the impact of redistricting after 2010, but Wang kept arguing that Democrats had a shot. He underestimated the influence of the new maps. Democrats earned 1.4 million more votes than Republicans in congressional races nationwide, but captured just eight additional House seats. Republicans maintained control, John Boehner retained his Speaker’s gavel. Wang had gotten the numbers right but the politics wrong.

“The election happened and I was proved wrong,” he says. You don’t graduate from Caltech with a physics degree at 19 by being wrong very often. The error — more embarrassing, in that he hadn’t seen the influence of gerrymandering — made Wang dig deeper into what he’d missed. How was it possible that he predicted the numbers correctly but got the results so wrong? The post-2010 gerrymander, his analysis revealed, was historic and different from others in the modern era. “It appeared to be the case that there was something about the districts that did not allow the Democrats a chance to retake the House. I wanted to know what that was.”

That difference, of course, was REDMAP. “Through artful drawing of district boundaries, it is possible to put large groups of voters on the losing side of every election,” Wang explained, sharing his research in a New York Times op-ed piece called “The Great Gerrymander of 2012.”

Then he applied his background in neuroscience and statistical research to try to understand his error and the historic aberration that, for only the second time since World War II, had prevented the party with the most overall votes from capturing the House. Wang wanted to determine, as scientifically as possible, whether the artful line-drawing was the reason Republicans had kept control.

He started with the idea that a party that wins more than half the votes ought to win at least half the seats. That low bar of basic representative fairness, he observed — in which the partisan interests of the state’s voters are reflected by the officials sent to Washington— was not cleared in five states: Arizona, Michigan, North Carolina, Pennsylvania, and Wisconsin. Percentages won’t ever match exactly, of course, but the ideal is what happened in Colorado, where 51.4 percent of the 2012 congressional vote went to Republicans, electing a delegation that favored the GOP 4-3.

Then Wang broke out the math. He wanted to calculate something he called the appropriate seat breakdown of each state — what the delegation would look like if the lines were not twisted. He set up a control group: random combinations of districts from around the country that added up to the same statewide vote totals. These districts, he argued, represented what would have happened if that state had districts that looked like those across the country. Keep in mind: these control districts are still the post-census, GOP-dominated lines. But Wang’s model revealed the big con.

Start with Pennsylvania. Wang had his computer run 1,000 simulations for outcomes with a 50.7 Democratic/49.3 Republican vote. The median result? 9.7 GOP seats, 8.3 Democratic seats. The actual result in Pennsylvania? 13-5 to the Republicans. The 13-5 result came up once in the 1,000 simulations. The five Democrats won with an average of 76 percent of the vote, the 13 Republicans with 59 percent. That outcome, Wang found, would arise by chance far less than 1 percent of the time. “In other words,” he wrote, “gerrymandering’s contribution to Pennsylvania’s partisan outcome was about five times as large as the effect of overall structural advantages.” Even simpler: the odds against Pennsylvania voters sending 13 Republicans and 5 Democrats to Congress were nearly 1,000 to one.

The numbers suggested to Wang that something had happened to skew the delegation. “This math doesn’t tell you what that something is,” he tells me. “That’s for somebody else to prove. People who are aware of the political process can say, ‘There was a partisan process of drawing boundaries. There was partisan intent.’ All I’m providing is a forensic standard that says, ‘Something happened! Why don’t you go take a look.”’

Wang kept looking, and found troublingly skewed results from the nine most egregious states: Arizona, Michigan, North Carolina, Pennsylvania, and Wisconsin, along with Texas, Ohio, Illinois, and Indiana. Six of those states had been redistricted by Republicans, one by Democrats, one by an independent commission and one by Republicans with input from the Justice Department.

Wang concluded in his Times piece, first, that Republican-controlled redistricting led to a swing in margin of at least 26 seats, almost as large as the 31-seat majority of the new Congress. Second, that in those nine states, the net effect of both parties’ redistricting combined was a swing of 11.5 seats toward the GOP. If all of the lines had been drawn by nonpartisan commissions, Wang argued, it would have led to a swing of at least 23 seats toward the Democrats — or 222 for the GOP and 213 for the Democrats in 2012.

That would have created a very different Congress. If Speaker Boehner had held such a narrow majority, he would likely have been forced to govern in the spirit of compromise. “It would have been a much more closely divided House, which would have changed the flavor of politics in the ensuing year.”

Perhaps because one of Wang’s specialties is autism, he recognized the patterns that political scientists and journalists missed. He doesn’t seem particularly motivated by which party runs Congress, so long as it is the one that earns the most votes.

“I think politics is very important,” he says. “I can’t contribute in the domain of writing speeches, but given these really imperfect mechanisms in democracy, it’s good to try to find ways to make them move toward a better ideal case. . . I think that a technical person could maybe make a contribution there.”

A technical precedent is actually exactly what Wang has in mind. If he can pull it off, Wang might be able to do nothing less than return the people’s House to the people. Here’s his thinking. The Supreme Court has refused, time and again, to get in the way of a partisan gerrymander. In Davis v. Bandemer, in 1986, the Supreme Court found that a partisan gerrymander was justiciable, but said they had no standard by which to strike one down. Eighteen years later, in Vieth v. Jubelirer, Justice Scalia tried to call time, arguing, in his majority opinion, that proponents of reform had had nearly two decades to determine a standard and, none having emerged, it was time to conclude that drawing lines was simply a political process, and the courts should move out of the way.

Not so fast, wrote Justice Kennedy. Kennedy did not believe the Vieth plaintiffs had found a partisan gerrymander. But he welcomed future efforts to determine what one was. He wrote:

“The rapid evolution of technologies in the apportionment field suggests yet unexplored possibilities. . . Technology is both a threat and a promise. On the one hand, if courts refuse to entertain any claims of partisan gerrymandering, the temptation to use partisan favoritism in districting in an unconstitutional manner will grow. On the other hand, these new technologies may produce new methods of analysis that make more evident the precise nature of the burdens gerrymanders impose on the representational rights of voters and parties. That would facilitate court efforts to identify and remedy the burdens, with judicial intervention limited by the derived standards.”

Wang read an invitation in those words. “He appears to be saying here, basically, that a technological approach is making this situation worse, but could also be the source of a judicial standard that he could sign on to,” Wang says. “He’s saying it is justiciable, and I’m waiting for some nerd to tell me what the standards are. I think Kennedy’s position here could be interpreted as being the Democrats’ best route to getting a level playing field.

“I have a feeling,” Wang says with a slight smile, “that it can be solved mathematically. I’m sort of surprised nobody’s come up with it yet.”

That’s right: he’s attempting to write a theorem to detect ratfucking. Wang grabs a napkin and begins to sketch a graph. Along one axis, the proportion of the vote. Along the other, the number of seats a party wins. Political scientists call this the seats-votes curve. Win 30 percent of the national vote, you won’t win many seats. Win 70 percent, as Democrats did in the 1930s and 1940s, you’ll carry a huge majority. His argument is revelatory. For too long, the courts and the media have tried to find gerrymandering by studying the boundaries and the shapes of districts. They’ve been looking in the wrong place. If a court wants to judge partisan asymmetry, he says, they need to compare the results from every district in the state. . .

“If Democrats win 51 percent of the vote and they only get 30 percent of the representation, as per Pennsylvania, that’s a problem,” says Wang. “I have to come up with the statistical criterion for how far away from the line is a foul. What I’m trying to do is cook up a forensic tool where a judge could figure out what happened and draw a bright line saying, ‘Don’t do that.’”

What other options do we have, Wang asks sincerely, other than putting faith in math and courts? I explain what seems to be the Democratic strategy heading into 2020: raise millions, copy what the Republicans did in 2010, keep their fingers crossed. A bank shot.

“That strikes me as something that could have worked in 2006, but not now,” says Wang. “The horse has left the barn. And not to be too thumb-sucking about it, but it’s not actually a step in the direction of good government.”

Excerpted from Ratf**ked by David Daley. Copyright 2016 by David Daley. Used with permission of the publisher, Liveright Publishing Corporation, a division of W. W. Norton & Company Inc. All rights reserved.

Facebook Comments