This week saw another poll about the Republican presidential primary — this one by NBC (pdf) — that got plenty of discussion. We learned, for example, what percentage of Republicans “could see themselves supporting” each of 14 different candidates. I am not going to tell you those percentages, because they really don’t matter. You’re better off ignoring the early primary polls. Here’s why.
Let’s start with the obvious. We already know that, this early in the campaign, presidential general election polls have essentially no relationship to the outcome. And those polls at least enable people to draw on a powerful factor, their partisan loyalties, that will influence their ultimate decision.
So why would we think that early presidential primary polls are any more reliable? After all, these are polls about races where voters know even less about many of the candidates and cannot draw on their party identification to cast a vote.
Here’s another reason to ignore the polls. By now, I think a lot of people get that winning the support of party leaders — not voters — is the key to winning the invisible primary. Political science research has shown that party leader endorsements are correlated with voters’ opinions about the candidates before the primaries begin, and are correlated with how many delegates the candidates win at the party conventions and thus who is nominated. Party leaders thus shape the field even before Iowans gather to pick corn and presidents. Just ask Mitt Romney.
But what are these party leader endorsements based on? In The Party Decides, a team of political scientists — including The Monkey Cage’s own David Karol — studied the primaries from 1980-2004 and found that endorsements were not correlated with early polls — or on early fundraising or media attention. They write:
Thus, party insiders go their own way, regardless of polls, media, and fund-raising.
Moreover, they found that once endorsements were taken into account, there was only a small and statistically insignificant relationship between early polls and how many delegates the candidates won. Endorsements appeared to matter much more.
The claim is not that endorsements are all that matters, or that money or media attention are completely irrelevant to the process. But these results should lead us to put very little weight on how the candidates are polling almost 11 months before the Iowa caucus. These early polls don’t really predict the outcome, and they don’t really predict the choices of the party leaders, which actually will help us predict the outcome.
Of course, you can use early polls to extract a “snapshot” of where opinion is now. Nevertheless, a lot of steam goes out of headlines like “Voters Are Divided Over Jeb Bush Candidacy” if today’s snapshot tells us next to nothing about where opinions will be in 2016.
As Jon Bernstein rightly says, ignore those polls!