About That Swing

by publius

I generally agree with the Drum/Yglesias argument that, in examining exit polls, we should look at a given group’s swing relative to the national swing. For instance, under this assumption, Kevin argues that the “weekly churchgoing” Obama swing isn’t particularly newsworthy because the swing (+10) is almost perfectly consistent with the broader national swing. The implication is that Obama wasn’t uniquely appealing to this group.

I want to push back a bit with respect to the churchgoers. Actually, I have questions rather than critiques. And if it’s simple statistical ignorance on my part, please let me know.

First, the swing isn’t the only relevant number, right? It’s also important to look at the overall size/turnout of the subgroup as well. Let’s say, for instance, that Obama adopted a deliberate “base deflation” strategy with respect to religious voters and that we’re assessing its success. By base deflation, I mean that Obama both reached out to weekly churchgoers and avoided hot-button cultural issues that drive them angrily to the polls.

Let’s then say that this group swings +10, but that the overall turnout of the group drops considerably. In this situation, Obama’s strategy seems vindicated (and the swing becomes more newsworthy), even if the swing is in line with the national numbers. Thus, before we claim there’s nothing to see here, I think we need to see the size of the pie too.

I spent a few minutes digging around some 2004 vs. 2008 exit polls from swing states, but it’s hard to do because the questions don’t always match from year to year. In Colorado, however, there’s at least some evidence of this. In 2004, self-identified white “evangelicals” were 26% of voters – and these voters went 86-13 for Bush. In 2008, evangelicals were 21% of voters – and they went 76-23 for McCain. So not only did Obama get a +10 swing, he also benefited from the relatively lower turnout. (Again, I don’t know if these numbers check out in other swing states – I’m just posing the question).

Second, I think some +10 swings are more equal than others. I mean, Drum is probably right in that the most obvious explanation is that weekly churchgoers are actually a fairly diverse group and thus got caught up in the national Obama tide. But that said, it’s worth thinking about whether a +10 swing for this particular subgroup is a more significant accomplishment. After all, these people are not merely sensitive to the culture wars, they’ve also been subjected to lots of “he’s a Muslim” garbage. It’s possible that Obama’s religious outreach – and his deliberate avoidance of culture war hot buttons – helped allow this group to take its rightful place in the +10 national tide toward Obama. Or maybe it's just a national swing.

I think we need a more competent numbers person to answer this stuff. But it’s worth thinking about because it has important implications for how and whether Dems should do evangelical outreach.

Comments

[tomeck]

Drum's comment doesn't make sense to me. That 10% overall swing is made up of many parts that are far over or under 10%.
Beyond that, this represents a significant inroad into an opponent's base, which is far different from the population in general.

Chuck Tanner, former White Sox and Pirates (I think) manager used to like to day "Each win counts for 2" about games against pennant opponents late in the season, meaning you get a win and your opponent gets a loss. Likewise, convincing an undecided puts you up one vote, but convincing someone who could be expected to support your opponent puts you up one and the other guy goes down one.

A plus 10 swing among weekly church-goers is significant.

[Warren Terra]

Also, these numbers often aren't normalized - thus, to take the most obviously silly example, Obama only increased his share of the Black vote by 5% - but considered another way, the Republican share of the Black community's vote was halved. Obama gained a smaller share of the Black electorate than he did of the total electorate simply because there were so few remaining still to be gotten. Similarly, the 6% swing in the Jewish vote is slightly more impressive when you consider that the Republicans, after all the Nazi imagery they invoked, their dire warnings about Iran, and their insinuations about Muslim associations, saw their Jewish support drop not merely by 6% of the Jewish vote but by about a quarter of what they got four years ago. When considered in these lights, Obama may have only upped his shared of the Evangelical vote by the same amount as he did the Dem share of the total vote - but as a percentage of the 2004 Dem electoral vote it might be considered a bit bigger.

[Gary Farber]

Tangentially, Steve Schmidt suddenly gets honest.

[Gary Farber]

And elsewhere in sweet music: Obama Positioned to Quickly Reverse Bush Actions; Stem Cell, Climate Rules Among Targets of President-Elect's Team.

Transition advisers to President-elect Barack Obama have compiled a list of about 200 Bush administration actions and executive orders that could be swiftly undone to reverse White House policies on climate change, stem cell research, reproductive rights and other issues, according to congressional Democrats, campaign aides and experts working with the transition team.

Ain't it bootiful?

[ThatLeftTurnInABQ]

I imagine that some politically minded spreadsheet jockeys will data mine the results from this election until we have a variety of competing explanations for the 2008 election results, but I think we are looking at trees and missing the forest here.

The fact that Obama's campaign was able to generate a swing to their advantage across a wide range of demographic categories is the real story - because it shows that the campaign was succesfull at breaking out of the confines of group identity politics. They appealed to voters first and foremost as individuals rather than as an impersonal collection of demographic factoids. This should come as no surprise if you've been following the articles posted by Sean Quinn at 538 or Al Giordano at The Field about the grass-roots level organizing techniques used by the campaign, or if you listened to Obama's speeches and took seriously his rhetoric about national unity and overcoming barriers that divide us.

This campaign wasn't about swinging "key demographic group X". It was about appealing to people as, well, people. The key demographic in this case was citizens of the United States of America.

And in my mind that is perhaps their greatest accomplishment in this campaign. The conventional practice of chopping up our body politic into pieces and then fighting over bits of the corpse to see who can stitch together a bigger Frankenstein monster (albeit one that is in most cases missing a few limbs or some crucial organs) is IMHO one of the most poisonous legacies of the politics of the past. Good riddance, and a hearty thank you to everyone who worked so tirelessly to try something different this time.

[Blar]

According to CNN's national exit polls for 2004 & 2008, the share of voters who identified as "White Evangelical/Born-Again" increased from 23% in 2004 to 26% in 2008. But what's the margin of error on these things?

Also, I'd be careful about weaving in particular campaign issues. There were also particular campaign issues happening in 2004, when we were hearing garbage about Kerry being denied communion, and anti gay marriage measures were on the ballot in a bunch of states.

[John Spragge]

The heritage of evangelical Christianity includes William Wiberforce; Contemporary evangelicals include Ron Sider and Jim Wallis. Outreach by the Left to evangelicals could include, first, the acknowledgment of the pivotal role the Evangelical movement has played in social change, and the manner in which Evangelicals have begun to return to their roots.

[MandyW]

Shouldn't we be a little hesitant about drawing broad conclusions about detailed demographic breakdowns from exit polls, as if every number, down to the last decimal point, were Revealed Truth from On High?

On Election Day at FiveThirtyEight.com, Nate Silver clearly explained Ten Reasons Why You Should Ignore Exit Polls. Exit polls provide useful information about the broad composition of the electorate, but it's a mistake to detect "trends" when the size of the variance is less than the inherent margin of error in the source data.

And on Friday at DailyKos, a blogger named Shannika crunched the numbers to show that Facts Belie the Scapegoating of Black People for Proposition 8. Basically, she did the math using Census data to show that there just aren't enough black people of voting age in California to have had the decisive effect in passing Prop 8 that could be inferred by assuming that CNN exit-poll data was gospel truth.

Along with ThatLeftTurninABQ, one of the things I most applaud about the Obama campaign was the degree to which they did not address Americans as a conglomeration of individual interest groups. They helped us all remember that civic life isn't a zero-sum game, in which every "win" for one group means a "loss" for another group. We're all in this together, and we need to work together to get out of the current mess we're in.

[Bernard Yomtov]

I'm not sure how it fits into the discussion, but the "base deflation" and the swing could really be the same thing.

Stay away from the hot-button issues and you drive down the hot-button voter turnout. The remaining Evangelicals may not support Obama any more than they supported Kerry, but because a number of anti-(Obama and Kerry) voters didn't show up in 2008 it looks like there was a swing to Obama.