What Trump's Poll Numbers Mean
We know that the relationship between such early polls and the eventual nomination results is weak. Doing well in June surveys, more than a year before the party conventions, doesn’t predict anything. But that doesn't explain why Trump is leading a dozen other Republican candidates.
Dan Drezner tweeted that it’s “1) He got an announcement bump; 2) The GOP field is so fragmented that 10% will get you second.”
I’d put name recognition first. Although political analysts and party actors have been steeped in 2016 for months or even years, plenty of Republican voters barely know who Scott Walker and Marco Rubio are at this point. Of course, name recognition isn’t a real asset for a presidential candidate. By the time the Iowa caucuses come around, voters there will know quite a bit about every candidate who runs a real campaign there.
As for the announcement bump, Drezner is correct. Hardly any voters have fixed candidate preferences at this point, so if asked, they’re likely to say they support (or oppose) any candidate whose name has been in the news lately. We can’t predict what the information context will be in various states right before Republicans vote in primaries. We can say that by then it will depend on recent events and on the orchestrations of party actors -- with favored candidates reaping positive publicity.
The short-term bumps will come and go as reactions to events wax and wane (such as instant reactions to debates or the results of a recent primary). But the input of party actors will be fairly constant once they decide whom they are supporting.
Then there’s the 10 percent factor. With about 15 or 16 candidates and no dominant front-runner, random variation in responses can have a big impact on the order of results in any particular survey.
Take that Iowa Quinnipiac poll. Walker leads with 18 percent. Trump is tied for second with Ben Carson at 10 percent, just a tick ahead of Ted Cruz and Rand Paul at 9 percent, with Jeb Bush at 8 and Marco Rubio at 7. Even if these numbers reflected real choices by voters (as opposed to just name recognition and reactions to current information), there isn't any real difference between second place and sixth.
Early polls are a trap. They give the illusion of being objective measures based on hard data, but neither national nor early-state polls tell us anything about what’s going to happen next year. That goes for Hillary Clinton's early lead in general-election polls, too, by the way.
Political scientists have found that high-profile endorsements are a much better predictor than early polling, or even early fundraising, for who eventually gets the nomination. Prediction markets can also be a useful hint about what's going on, suggesting that plenty of information is available beyond public-opinion polls for those know how to read it.
Reporters feel they have to cover these polls anyway, and that's OK -- especially since Republicans are foolishly using them as the qualification barrier for who gets to take part in their early debates beginning next month. Just don't expect more from these surveys than they can offer.
That's why I also don't take seriously the polling on which candidates voters will or won't consider supporting. It, too, can be artificially inflated (or deflated) by recent news.
Party actors can send a strong signal if they converge on one choice. But even if they haven't done that by the beginning of the primaries and caucuses, they will have made decisions, pro and con, about many candidates. Add it up, and it affects what voters will be hearing about the various contenders.
So why is Walker first? His name recognition might be a bit higher because he's governor of Wisconsin, a next-door state, and he had a surge of positive information a few months back. Don't mistake the number of people who pick him in a poll with firm commitments by future caucus-goers.
This column does not necessarily reflect the opinion of Bloomberg View's editorial board or Bloomberg LP, its owners and investors.
To contact the author on this story:
Jonathan Bernstein at firstname.lastname@example.org
To contact the editor on this story:
Katy Roberts at email@example.com