1/28/2008

On Electoral Behavior and the Credibility of Polling

The polling industry was rocked when Hillary Clinton won a come from behind victory in the New Hampshire primary despite the fact that the question raised by almost every poll taken immediately beforehand was not whether she would lose, but rather by how much. Pundits commonly talked about the Bradley effect, in which voters lie to pollsters about their willingness to support a candidate of color only to abandon this candidate in the privacy of the voting booth because of their own unspoken prejudices. Other pundits looked for other possible causes for Clinton's "silent surge." These reasons ranged from being emotionally-based (e.g., Clinton's crying) to institutionally-based (e.g., Clinton did better among Democratic establishment-types) to psychosocially-based (e.g., the male candidates were slighting Clinton much like the way many women feel the men in their professional and personal lives slight them). All of these explanations had some legitimacy.

The sheer margin of Barack Obama's surprisingly strong finish has emerged as the dominant storyline coming out of the South Carolina primary. However, there's another storyline that warrants further examination--the fact that the polls got it wrong again. According to Real Clear Politics, most major polls taken before the primary had Obama winning by anywhere from 7 to 15 points. (Obama ended up winning by 28.) And the Rolling Stone was warning that the high percentage of undecideds could spell potential disaster for Obama.

So what happened this time, especially since the polling for Clinton and Edwards turned out to be far more accurate?

Explanation 1: There is something of a "reverse Bradley effect" in which voters who really do support Obama tell pollsters they don't simply because they don't want to contribute to the popular media/political storyline about how diverse Obama's supporters are or how much Obama is relying on the Black vote. As I wrote about here, there was a very real possibility that pundits, the media, and (almost certainly) the Clinton campaign would try to spin Obama's victory as the inevitable result of an electorate that was simply too difficult (e.g., too Black) for any other candidate to overcome. If this is indeed what's going on, then that would make it even more difficult to accurately poll Obama in the future. Who are the nonsupporters saying "yes" to Obama out of political correctness, and who are the true supporters saying "no" to Obama out of political strategizing?

Explanation 2: Voters concluded at the last minute that the Clinton campaign did not deserve their vote. Exit polls showed that more voters, including more White voters, thought that Hillary Clinton had run an unfair campaign. Had the election taken place a few days later, perhaps this disdain could have been reflected in the polls. So if this scenario explains what happened in South Carolina, then the polls were right all along and simply suffered from the fact that this disdain on behalf of the voters was a lagging indicator.

Explanation 3: John Edwards is being used as a repository for hidden votes. The South Carolina press was particularly bullish on Edwards and speculated that he could make a real run for second place. Could this perceived surge in Edwards' support really have been a reflection of this hidden Obama vote? Edwards performed miserably among Black voters despite aggressively courting them in his campaign ads. Were Black voters feigning support for Edwards because they didn't want to inflate Clinton's numbers? Obviously, if Clinton's polling displayed an upward trajectory, she would spin that as having "cross-racial" appeal or simply being a stronger candidate overall than Obama. This, in turn, would fuel "is Obama in trouble?"-types of stories. Similarly, one of Clinton's perceived advantages was how she could lock up the women's vote. However, she lost women to Obama by 24 points and tied Edwards among men. Were men feigning support for Edwards even though they were really for Obama? Were these phony Edwards supporters also gaming the system?

Explanation 4: Voters are sick of polling and campaign advertisements and are simply telling the pollsters and campaign workers what they want to hear in order to get them off the phone as quickly as possible. In the days leading up to the primary, I would commonly receive about 10-15 calls a day from campaign workers asking me who I was voting for or reminding me to vote. I also received countless other recorded messages from the candidates themselves or other celebrities telling me why I should support candidate X. It irritated me to no end, as these calls commonly interrupted my dinner, study time, conversations with my wife, or other phone calls. And they clogged up my answering machine and mailbox as well. So whenever someone from the Smedley campaign, for example, called me about voting, I would commonly tell them I was already planning to vote for Smedley and didn't need to be convinced to vote for him or reminded to vote in general. That would usually cut the conversation short and allow me to get back to whatever I was doing before I was interrupted for the umpteenth time. I can only wonder how many thousands of other voters felt the same way. If this frustration is as real as I suspect it is, then polling would obviously be skewed.

A good task for further study would be to develop a means by which these kinds of voting behaviors can be credibly assessed. Come Super Tuesday, the consequences for getting these polls wrong could be a lost election, millions of dollars wasted, and hundreds of man-hours lost pursuing ineffective campaign and advertising strategies and tactics. Each of the four most probable nominees (McCain, Romney, Obama, and Clinton) has a unique demographic characteristic that could potentially benefit or hamper their electibility (age, religion, race, and gender, respectively). And each of the four has an ideological or political vulnerability that they must compensate with (mistrust among conservatives, flip-flopping on several key conservative issues, a thin political resume, and a polarizing approach to politics, respectively). And of course, they all have unique strengths as well (appeal among independents, an aura of competence, the ability to inspire new voters, and nostalgia of a relatively popular presidency, respectively).

For the sake of the polling industry and a better understanding of modern electoral behavior, how these variables interact with modern campaigning and political warfare begs further scrutiny. The challenge, however, lies in going beyond superficial thinking (e.g., "Obama should win the Black vote. McCain should do well with independents.") and figuring out how all these variables interact with each other on a more complex level.

6 comment(s):

Nikki said...

Anthony this made me laugh a little bit because in the '04 election I was working as a volunteer for Bush and making those very phone calls. I felt a little queesy everytime someone did answer the phone and I had to actually talk to them. I got yelled at a lot in that campaign. Clintons cooked their own goose in my opinion. Passionate Obama supporters had a fire lit and showed up big time!! I was glad to see it.....great post. Nikki

Anthony Palmer said...

Nikki,

I think that these campaign communications are only going to become more and more intrusive in future elections. I think there's nothing wrong with grassroots calls. They can be annoying sometimes, but that's just a part of democracy. It's the robocalls that really got under my skin. Do these people really think that just because Celebrity X endorses someone, I should too? Obama's staff was guilty of that and Clinton's staff was guilty of that. I didn't receive as many robocalls from the Edwards campaign. Maybe that was because they didn't have those endorsements to tout, or maybe it was because they wanted to keep the campaign humble. I don't know. But thanks for the comment.

Schenck said...

EDWARDS IS OUT! EDWARDS IS OUT! EDWARDS IS OUT!

Schenck said...

http://www.washingtonpost.com/wp-dyn/content/article/2008/01/30/AR2008013001069.html?referrer=email

Silence Dogood said...

Nikki, what a nastaligic feeling reading your comment brought back for me - I have made a lot of those phone calls in the past for friend's and relative's campaigns (none for presidential). When I was younger I remeber feeling sooooooo much anxiety about doing it.

Thomas said...

Hi Anthony. Do you think the Obama campaign downplaying of the Florida vote will be somehow harmful towards his campaign? It can appear that he is just playing "politics as usual" instead of standing up for people and their right to have their votes count.