Published: Dec. 1, 2012 By

Romeny obama Debate

Two CU professors reveal the factors contributing to presidential wins since 1980.

Ken Bickers has a crystal ball. But it’s not shaped like a ball. Flat and rectangular, really — like a computer screen. Well, actually, it is a computer screen. Still, when the CU-Boulder political science professor stares into it, he says the future is often as clear as a Colorado cobalt sky.

Last summer, Bickers used his crystal ball to predict that Mitt Romney would send President Barack Obama packing by winning 325 Electoral College votes when only 270 are needed for the White House keys. His crystal ball shattered after Obama won 303 electoral votes to Romney’s 206 (not counting Florida). The results surprised him and many political scientists and presidential historians. History has shown when the economy is not doing well incumbent presidents usually lose re-election.

“Our estimate is that the President’s campaign efforts, specifically in the battleground states, lifted his vote share in those states by about 5-6 percentage points,” Bickers says.

Ronald Reagan1980
Ronald Reagan wins by a landslide with 489 electoral votes over incumbent Jimmy Carter’s 49 and leads the popular vote by nearly 10 percentage points.
Ross Perot1992
Independent candidate Ross Perot receives 18.9 percent of the popular vote but no electoral votes. Bill Clinton defeats incumbent George H.W. Bush with 370 
electoral votes.Bill Clinton1996
Incumbent Bill Clinton receives 379 electoral votes to Bob Dole’s 159, winning the popular vote by nearly 9 percentage points. Ross Perot nets 8.4 percent of the popular vote.
George Bush2000
Al Gore wins the popular vote, but George W. Bush wins the election by 5 electoral votes.
Barack Obama2008
Barack Obama wins over John McCain with 365 electoral votes and leads the popular vote by more than 7 percentage points.
Mitt romeny2012
CU professors predict Mitt Romney will win more electoral votes than incumbent Barack Obama. But Obama wins with 303 electoral votes (without Florida’s results).

In the battleground states he says that was enough to produce a win for Obama in all of them, except North Carolina, and to produce close to a tie in Florida.

So where exactly did Bickers come up with his model and why did it work for every election, except for this one, since 1980?

The story begins in 2008 as workers swept up confetti at Invesco Field at Mile High Stadium where President Obama accepted his party’s nomination for president.

Bickers and Michael Berry, an assistant professor of political science at CU-Denver, had spoken with several reporters who asked if they thought hosting the Democratic National Convention would win Obama Colorado’s Electoral College votes.

“At the time we could only speculate but thought it made for an interesting research question,” Berry says.

After the 2008 election, the two men crunched large sets of data from states in each election cycle from 1948 to 2008. The result? Absolutely no convention effect for either party.

But after that project, the two men began to do what professors often do: Think. Specifically, they began thinking about how they might apply all the data they amassed to predict presidential elections.

Berry and Bickers were familiar with other models and the many factors used to project the national popular vote. They say they didn’t deliberately set out to build something that would differ from existing models, but since the president is elected through the Electoral College, Bickers says it made more sense to construct a model that would let them make state-by-state forecasts.

In essence, the two men see the presidential election as 51 separate contests — the 50 states and the District of Columbia — not just one national competition.

One model used state forecasts in 2008, but it only counted change in personal income. Bickers and Berry say that since jobs are often a large focus in campaigns, they opted to add state and national unemployment data to the model.

They also considered party voting in prior presidential elections — again at the state level — and how long a party has been in power. But they say economics is key and reminiscent of political pundit James Carville’s famous line from the 1990s, “It’s the economy, stupid.” In fact, Bickers calls the economic variables “the heart of the model.”

“The surprise to me was the success of the Obama campaign in neutralizing the economic conditions.”

— Political science professor Kenneth Bickers

So the model carefully considers unemployment rates in each state from late spring and summer of the election year and compares changes in real per capita income — people’s disposable income — from when the current president was elected to the first quarter of his last year in office.

“It’s like the question Ronald Reagan asked in the 1980 election: ‘Are you better off now than you were four years ago?’ ” Bickers says. “The percentage change of income has fallen in more than half of the states. In a normal economy, it would be many fewer states.”

“The surprise to me was the success of the Obama campaign in neutralizing the economic conditions.”

— Political science professor Kenneth Bickers

After they completed the model, Bickers and Berry put it to the history test, applying it to the last eight presidential elections beginning in 1980. The federal government lacked state data they needed prior to 1980, so they stopped at that election year. The model passed with flying colors, correctly classifying the winner in each election, including the photo-finish 2000 election when George W. Bush edged Vice President Al Gore.

Bickers says that particular election illustrates the power of too much time in office, another variable they deem important. President Clinton had served his limit.

“Voters look at how long a party has been in office, and sometimes that works against the in-party,” he says. “It happened in 2008 when Sen. [John] McCain ran for a third consecutive presidential term for Republicans.”

But Bickers adds the economy was starting to sag due to the dot-com bubble in 2000.

“Tougher economic times were starting to ripple through some states before the election,” he says.

In 1992 the economy and incumbency factors were again at play. Republicans had held the White House for 12 years, and the economy was growing tepid in many parts of the country.

But 1992 also showed Berry and Bickers that a strong third-party candidate can throw a wrench in their model. While the model correctly predicted Clinton, in that election they also missed 10 states.

“That’s 20 percent, basically, and it shows just how significant Ross Perot ended up being,” Bickers says. “We’ve puzzled over third-party candidates. We’re essentially saying people can vote for a Democrat or Republican, but they can vote for others. They did for Perot in 1992 and 1996 — our model misses a lot in those circumstances.”

The two did offer themselves a little wiggle room with the 2012 election because of the amount of time before the election and the number of states where the presidential race seemed close this summer. The latest data they could use came in September before the Nov. 6 election. Things can happen in two months. And some states were near a 50-50 split in late summer.

Minnesota in particular was a head-scratcher for Bickers.

“Minnesota hasn’t voted for a Republican in a long time, but our model has Romney winning it,” he says, which ended up being incorrect.

The model predicted Obama would lose nearly all of the states considered swing states in the election — North Carolina, Virginia, New Hampshire, Colorado, Wisconsin, Minnesota, New Mexico, Pennsylvania, Ohio and Florida. In fact, he only lost North Carolina, and Florida was not tallied at press time.

Two other states viewed by some as swing states — Michigan and Nevada — were predicted to stay in Obama’s column. This prediction was correct.

After releasing their prediction in the summer of a Romney win, Berry and Bickers say they were glorified and vilified, at once being called the smartest men alive and then accused of trying to sway the election.

For the record, Bickers preferred not to share his party affiliation, and Berry said he’s unaffiliated.

As for any lessons learned from their prediction project, both say they came away a little wiser.

Bickers says he learned that when there’s a big thing in the election, there really is a big thing. That thing in this election was the economy.

“The surprise to me was the success of the Obama campaign in neutralizing the economic conditions,” Bickers says. “By doing that, they were able to shift focus to other issues, such as immigration and women’s rights — issues where the Democrats enjoy healthy advantages over Republicans.”

He also thinks early negative ads in the battleground states painted Romney as someone who couldn’t be trusted with the economy.

“And Romney’s stance on immigration may have played a heavy hand in his loss,” Bickers says. “The tone of the Republican nomination process put him in a box, particularly with the Latino voters, that he couldn’t find a way out of — his kind of harsh, strident position on illegal immigrants.”

Photo courtesy Craig F. Walker, The Denver Post