UC Berkeley Press Release
Study explores Iraq impact on U.S. presidential race
BERKELEY – Contrary to current conventional wisdom, deaths and injuries of American troops in Iraq did hurt the election efforts of President George Bush while gay marriage ban initiatives in 11 states had no measurable impact, say two researchers at the University of California, Berkeley.
"These findings don't mean the war was or was not justified, but that there was a political cost," said David Karol, a UC Berkeley acting assistant professor of political science. "What this shows is that it cost him votes."
Edward Miguel, assistant professor of economics, teamed up with Karol to assess how voters responded to the rising number of U.S. soldiers killed and wounded while on duty in Iraq.
On Election Day, the number of U.S. soldiers killed in Iraq hovered around 1,100, and the number of wounded was approximately 8,500, according to the report. Every state in the country has recorded military deaths or injuries on battlefields in Iraq.
The researchers, both in the College of Letters & Science, compared the number of votes Bush received in 2000, with his 2004 election results broken down state by state and county by county. Then, they correlated that data with Pentagon information about the number of dead and wounded per capita, along with the location of the hometown or assigned military bases of the U.S. dead and wounded, as well as casualty coverage in 210 major media markets.
Their results indicate that Bush would have garnered 53 percent of the vote instead of 51 had there been no U.S. dead or injured -- no small matter in such a statistically close contest.
Controlling for factors such as changes in the state unemployment rate or population, the proportion of residents serving in the armed forces, and how close deaths or injuries came to the election, Miguel and Karol determined the key variable was how many people from a state died or were injured in Iraq. While the statistical approach comparing states allows the researchers to gauge the impact of additional Iraq war deaths and injuries, it does not capture overall nationwide trends in support for Bush related to the war.
For example, while Bush gained support across the country, the researchers found that Bush generated less support than in the past in states with high death and injury tolls such as Vermont and South Dakota. Vermont had the highest per capita casualty rate, 7.8 deaths and injuries for every 100,000 people.
The war had no impact on voters in the Deep South, while Northern voters reacted sharply and negatively, they said. The researchers also found that no one state overshadowed the rest, with results remaining the same even if vote tallies from any state were disregarded.
"Had there been very few deaths, Bush would have won pretty decisively. He would have won four more states, the battleground states of New Hampshire, Wisconsin, Pennsylvania and Oregon. And he would have won a lot of electoral votes -- 328," said Miguel, who typically studies the intersection of crime, violence and development in foreign countries.
"That's essentially what it looked like a year ago," Miguel said. "Bush was way up in the polls, and everyone thought he was invincible."
But around April, the United States suffered an upsurge in casualties, which have continued to climb and reached approximately 1,200 by mid-November, the report said.
Because Bush won, there's a tendency to say that the war didn't matter, or that it actually helped him by boosting his image as a strong leader, said Karol.
"But those are actually premature conclusions," he said. "It may be that the war had an effect, and he was elected despite it. Or if he lost, that might not have been why he lost. If you see localized variations in the change of Bush support from state to state over the four years, and that is correlated with the level of casualties that that state has suffered, that makes a more persuasive case -- that the casualties had an effect on the election. That is what we found."
Yet another finding of the study was that the Bush vote appears unaffected by the presence of gay marriage ban initiatives in 11 states, or by the call-up of military reservists. Voters opposing gay marriage lined up with Bush early, so the ballot measures to halt gay marriage failed to generate new support for the president, the researchers said.
The 2004 election provided a unique test case, said Karol, because the nation has not seen in more than 100 years a president running for re-election in the midst of a controversial war that he initiated. "The only comparable cases," he said, "are maybe William McKinley and the Philippine Insurrection in 1899-1902, and before that you have to go back to James Madison."
President Harry Truman opted not to run for re-election while the Korean War continued. Likewise for Lyndon Johnson and the Vietnam War, said Karol. During publicly supported conflicts such as World War II and the Civil War, presidents sought and won re-election, he said.
"Here we have the largest number of casualties since Vietnam," Karol said. "So it's a better way to learn about how the political system processes casualties, and what the political consequences of them may be."
The researchers pointed out that the data shows that even relatively modest war casualties continue to carry political costs.
The study also adds a new wrinkle to ongoing debate since the 1990s about whether the United States has become risk-adverse as some speculated after a quick U.S. pullout of troops from Somalia, a reluctance to commit ground troops in Kosovo, and a military retreat in Haiti, Karol said.
"When we started on this project, we had no idea what was going to happen, and we thought it was going to be interesting either way," said Karol. "We were watching (the election results) like everyone else."
The study is critical for U.S. foreign policy "to the extent that there are costs to invasions like Iraq, and politicians in the future may have to take them into consideration when they make foreign policy choices," said Miguel. "Just understanding whether there is an effect or not, and how big it is and how it works, is important for foreign policy and scholars as well."
A summary of the study's results is on the Web at http://emlab.berkeley.edu/users/emiguel/iraq.shtml. Related research is ongoing.