Margin Of Error: Breaking down the polls

Margin Of Error: Breaking down the polls

Early poll may reflect voter discontent

Posted June 24, 2014

The most recent Public Policy Polling (PPP) survey has Kay Hagan leading Thom Tillis, 39 percent to 34 percent in the North Carolina Senate race. I saw two things about the results that are worth discussing.

First, the poll clearly captures weakly held sentiments at this point in the campaign. Thus, a lot can change. Over 25 percent of likely voters are either undecided or say they will support someone else. Interestingly, in 2008, Hagan never polled ahead of Dole in their contest until late August, which would be two months from now. So, the outcome is clearly in doubt. Margin of Error logo Margin Of Error: Breaking down the polls

Second, PPP finds Sean Haugh, the Libertarian candidate for Senator, with 11 percent support. In 2010, when Richard Burr won reelection, the Libertarian Party candidate captured only about 2 percent of the vote. In 2008, when Hagan first won, the Libertarian candidate won about 3 percent of the vote. Double-digit support for third party candidates in a U.S. Senate race does not have any recent precedence.

When I looked into the dynamics of the 2010 race, Beitler, the Libertarian, was similarly polling at 10 percent in a late June PPP poll. He also consistently polled around 6 percent until late September, and then trended downwards toward his final actual result of 2 percent in November.

I think we are seeing a similar distribution of early polling preferences, and the voting results won’t match them.

Why? It’s a combination of some voters not being sold on either candidate right now, so they take the alternative option, or others who really would prefer to vote for a Libertarian but recognize that their most disliked major party candidate might win if they don’t vote for their second-most preferred candidate, a Democrat or a Republican.

How might these voters break? PPP says more Haugh supporters would opt for Hagan over Tillis, if they had to pick one. That distribution slightly favoring Hagan as the second preference is unlikely.

Most Libertarians will vote for a Republican if they can’t/won’t vote for their own candidate. So, what’s going on? Is the polling bad? Are people lying? No and no.

I think I can explain this by referencing the 2008 Democratic presidential primaries, pitting Barack Obama against Hillary Clinton. When Obama started winning states, pollsters uncovered a substantial percentage of Democratic respondents who said they’d rather vote for Republican John McCain if Obama beat Clinton. These Clinton supporters told pollsters they’d defect, or sit out the general election, if she lost to Obama.

Looking back at even earlier polls and voting outcomes, it is clear that Clinton voters were simply momentarily upset. In the end, Democrats would unite behind Obama, spurred in part by the healing process of the nominating convention.

This dynamic repeats itself over and over. Democrats, it turns out, really do prefer Democrats to Republicans; and Republicans vote for most any Republican rather than a Democrat. While the polling generated hand wringing and media navel gazing, it was unreasonable to assume Clinton supporters would really vote for McCain over Obama, and indeed they did not.

I think to a lesser extent this is happening with North Carolina Republican voters who are more ideologically conservative than Tillis. These voters preferred a Republican nominee other than Tillis, such as Brannon. Right now, far from the election, they can express their discontent in a poll with no real consequences. Come November, Tillis is overall clearly closer to Republican voters’ political preferences than Hagan, and in the end they will break for him accordingly.

This poll should be a reminder that early polling often captures more than true voting intentions. Some people really are uncertain, and others are discontent. This doesn’t mean election polls are misleading or pointless. Indeed, the PPP survey had a number of other interesting tidbits that capture the mood of the electorate. Yet, we should be cautious about taking the results of candidate preferences too literally this far from November.


Please with your account to comment on this story. You also will need a Facebook account to comment.

Oldest First
View all