Xbox/YouGov Panel on Romney’s First Debate Victory & Obama’s Subsequent Debate Rebound (Syndicated on the Huffington Post)
DavidMRothschild on October 29, 2012 @ 9:23AM
Most political polls and pundits declared Governor Romney the runaway victor of the first presidential debate, while President Obama was generally credited with winning the second and third debates. However, what matters is not winning or losing debates, but whether the candidates won or lost votes because of the debates. Standard forecasting methods do not provide a clear answer to whether debate performances are accompanied by actual changes in voter preferences (i.e., whether the electorate cumulatively shifts their support one way or another around the events). A comprehensive analysis of the data from the YouGov/Xbox poll shows that Romney made a sizable gain after the first debate and that Obama cut back into that gain after the second and third debates. Overall, across the 30 days that included the ups and downs of the four most important preset events of the campaign, Romney had a net gain of 1.4 percentage point relative to Obama.
Since September 28, the YouGov/Xbox poll has conducted nearly 400,000 interviews with Xbox LIVE users. The opportunity to interview many of the same people repeatedly creates a “panel” of users who are asked how they intend to vote at different points in time. This process affords us the ability to draw insights that are not possible by looking at the aggregate numbers from typical national polls, which are subject to substantial sample variation and other noise. Specifically, we can measure whether individual voters’ opinions are actually changing over time. The data -- comprised of around 20,000 paneled users for each week -- indicate that voters’ preferences have clearly shifted.
In the wake of the first debate, Romney benefited from a 1.5 percentage point improvement among Xbox users, while Obama lost roughly half a point. After the vice presidential debate, Romney’s improvement continued, albeit more modestly, as he gained just over half a percentage point, while Obama recovered some of the ground he had lost. The second and third president debates saw further gains for Obama of a little more than half a point and a little less than half a point after each of these, respectively, while Romney remained relatively flat during the same time. In every single week, the number of paneled Xbox LIVE users who said they were undecided or voting for another candidate declined.
What has been the cumulative impact of these changes over the last month? Overall, Romney has improved roughly 2 percentage points with Xbox LIVE users, and those gains occurred almost entirely during the period between the first two debates. President Obama has made up some of the lost ground due to his strong showing in the second and third debates. He now stands roughly 0.6 percentage points ahead of where he started in late September, though there is still a net shift of approximately 1.4 percentage points in Romney’s direction over this period.
Mike Malecki, Doug Rivers, and Brian Stults (YouGov) contributed to the data work and writing of this article. In the interest of disclosure, I help run the Xbox/YouGov poll in my capacity as an economist at Microsoft Research.
This article is syndicated on the Huffington Post.
DavidMRothschild on October 26, 2012 @ 12:29PM
There is a lot of concern in the media and political circles about the effects of poll results on voters. In late September, when Obama was dominating the polls, Republicans latched onto the idea that the polling industry was skewing polls in Obama's favor to give him the air of inevitability. When Gallup's likely voter model later gave Romney a 7 point lead in mid-October, everyone started wondering not only what this meant about the election, but if could affect the election.
Finally, in a quiet period during the morning after the final presidential debate, the price of the contract for Mitt Romney to win the election escalated rapidly on Intrade, only to retreat back down almost as quickly. Had someone tried to manipulate the contract to make it appear that Romney won the debate or that the election was suddenly tied? Would that be a rational use of money?
To answer the burning question, researchers have long observed that people often conform to majority opinion (i.e. during every election, some people jump on the bandwagon and shift their preference to the leading candidate or the most popular policy).
During elections, and major public policy events, much of the media coverage focuses on the "horse race," or fluctuations in support for a candidate or policy. Reporting on public opinion not only affects support, but levels of engagement: donations, volunteering and turnout. These bandwagon effects can make polls self-fulfilling prophecies; the predictions of the polls come to pass because the polls not only measure public opinion but also influence public opinion and engagement.
While numerous studies have documented the existence of the bandwagon effect in the political domain, very few have attempted to understand the underlying mechanisms of why people conform to prevailing popular opinion. Researchers generally posit two psychological mechanisms underlying conformity: (1) people's desire to adopt the majority position so as to feel liked and accepted or believe they share the prevailing opinions of their community (i.e., social acceptance); and (2) people learn from the "wisdom of crowds," or assume that other people did the research so their collective wisdom indicates something about the quality of the candidate or platform (i.e., social learning).
Research that I conducted with Neil Malhotra of Stanford University also examines a third mechanism: polls reveal information about the likelihood of a policy passing or the election of a candidate, so people resolve cognitive dissonance ("Policy X is going to pass, but Policy X makes me unhappy") by switching to the side they believe is going to win.
We conducted an experiment on a nationally representative sample of voters in 2011 which simultaneously tests these three mechanisms to assess which are more powerful drivers of the bandwagon effect, or conformity to majority public opinion in politics. We do not suggest that these three mechanisms are collectively exhaustive or mutually exclusive, but due to their prevalence as broader psychological phenomena, they offer a promising direction to begin investigation.
Confirming past studies, we find strong evidence of an overall bandwagon effect; people become more supportive of policies that have more general support. We further find that both social acceptance and social learning drive the bandwagon effect. However, the effect of social learning is significantly and substantially stronger than that of social acceptance. Thus, the main reason that people conform to majority opinion in the political domain is that they believe there is information about the quality of the candidates or policies to be learned from mass support. We find no evidence for the third mechanism -- that people conform because they want to reduce cognitive dissonance related to not supporting a candidate that will likely win or policy that will likely be implemented.
The idea of the bandwagon effect seems disheartening for democracy if conformity pressures silence minority opinions. However, this research has given us a reason for optimism; people seem to be conforming not only because they see normative value in being part of the majority but also because they believe that there is information in collective opinion. Citizens want to be informed and the collective wisdom of their fellow citizens is just one source of information on which they have learned to rely.
This article is syndicated on the Huffington Post.
Indiana Senate race now leaning Democratic after Mourdock’s abortion comment (Originally posted on Yahoo!'s "The Signal" Blog)
DavidMRothschild on October 25, 2012 @ 5:32PM
We switched Indiana to a possible pickup for Democrats earlier this month, but at the time it had the longest odds for Democrats of the five Republican-leaning seats still in play.
There's a lot of daylight between these two suffering Republican campaigns in Indiana and Missouri. While Missouri's Akin's claim represents a fringe, discredited theory about rape and pregnancy, Indiana's Mourdock's position finds support among some in his party. The Republican Party's platform does not mention rape or incest when discussing abortion, and Rep. Paul Ryan has stated that "the method of conception does not change the definition of life" (though he dutifully defers to Romney's more moderate position now that he's on the ticket). Slate estimates that 12 to 15 of the 33 Republican senatorial candidates share this position.
A majority of Americans continue to believe that abortion should fall in the area between always legal and always illegal. Curiously, while public opinion on this point has not shifted much through time, polls have found that Americans are now more likely to identify themselves as "pro-life" than "pro-choice." Mourdock's statement is damaging not because it is factually incorrect, like Akin's, but because it exposes rifts among abortion opponents that no pragmatic Republican should want surfaced in this election.
DavidMRothschild on October 24, 2012 @ 1:16PM
Since mid-February most states have drifted slowly toward their initial favorite. Even in early May, many states provided the whiff of promise with a 10 or even 20 percent likelihood of being picked off by the trailing candidate. This has not been a year of surprises. As you run the map from May 1 to October 23, you see states fleeing to the safe, dark colors until only a handful of the true swing states remain standing.
By shifting the shading cutoff at the bottom of the map, you can see how sparse those swing states really are. If you define a swing state as one where the underdog candidate has at least a 35 percent chance of winning, only three states make the cut: Colorado and Virginia, which both lean slightly toward Romney, and New Hampshire, which is 63.9 percent likely to go to Obama. Iowa and Ohio join the fun right below that 65 percent point for Obama.
We will update this map regularly over the next two weeks. One thing we know for certain: The light colors will get sparser and sparser as we approach the day of judgment.
The election is over! (Results embargoed two weeks) (Originally posted on Yahoo!'s "The Signal" Blog)
DavidMRothschild on October 23, 2012 @ 2:01PM
We'll know in a few days how much "Monday Night Football," Game 7 of the National League Championship Series, and Anything-Else-but-a-Foreign-Policy-Lecture detracted from the TV audience Monday night. The Signal does not particularly care about this factor, because the final debate was always destined to have a small impact. Three reasons:
* There are not many undecided voters left. In most national polls, undecided voters account for 2 to 3 percent of potential voters. This is plenty enough to sway an election, but these 2 to 3 percent are typically not voters engaged enough to be watching debates.
* For a nation that just wound down a seven-year war, is still fighting an 11-year-war, and faces the prospect of further military intervention in the Middle East, foreign policy still ranks low on the concerns of most American voters.
* The implications of this debate have only two weeks to etch themselves into a campaign narrative that has narrowed in focus to only a few states.
All told, this election is probably over. We're just not allowed to open the envelope for another two weeks. Take it away, Ohio.