I’ve been critical of the uses and abuses of polls in political discourse but recently I retweeted something from political scientist Kevin Collins “the only poll on presidential candidate preference worth reading right now.” Have I gone squishy on the uselessness of early presidential preference polls? Not really.

Professor Collins was linking to a piece by Huffington Post’s Mark Blumenthal and Ariel Edwards-Levy titled We Polled True Activist Grassroots Republicans, And Here’s Who They’re Backing.

That’s right, HuffPost and its partner YouGov scrubbed its Internet panels for Republican activists, kicking out the hoi polloi. So much for the voice of the citizen. That’s good because Blumenthal and Edwards-Levy at least recognize political science research that shows that early polling does not correlate with the eventual party nominees, but endorsements of party activists does heavily influence the party’s choice. The HuffPost report is consonant though with other bits of evidence including party endorsements and polling of the broad population; party activists haven’t settled on a candidate on the GOP side. Here’s something else I like about the HuffPost piece: Blumenthal and Edwards-Levy are right upfront that their method is not a prediction about who will win but a reflection of where party activists – the most important constituency in the party – are.

One thing we can see is that activists are ambivalent at least about Jeb Bush. Bush started well – his early money raising and appeal to insiders forced Mitt Romney out, remember? But his subsequent stumbles have loosened the dynastic grip and he will have to earn the trust of party activists.

All this is consistent with Cohen, Karol, Noel and Zaller, The Party Decides: Presidential Nominations Before and After Reform, the enormously influential study that shows that endorsements are far more significant in the nomination process than early polls or money. More on that soon as I’m teaching the book for the first time this fall.

Also on polling, Professor John Sides of the Washington Post’s themonkeycage blog got into it with journalists on Twitter and the blog the other day. Here’s one of the tweets that seem to have set off journalists:

Some of my readers will remember that I’ve suggested that early polls be accompanied with the disclaimer, “For entertainment purposes only.” Professor Sides recommends: “A new poll is out today, but history shows that it tells us little about who will actually win the nomination/election.”

One journalist tweeted back that of course polls are important, Fox is using them to choose candidates for their debate! Exactly, replied Prof. Sides: Fox is using a virtually useless yardstick. How can we do better?

Fear not though journalists, we can learn some things from early polls (just not who the nominees will be). Professor Sides points to information about the national mood, immigration, ISIS, etc. I’d add a few more. A Washington Post-ABC News poll found that Donald Trump’s support comes largely from less-educated Americans with strong negative views on immigration. A New York Times piece analyzed the seeming lack of harm Mr. Trump suffered from his statements about Senator John McCain. Such reporting is useful and informative; it’s just swamped by horse race coverage.

So is Professor Sides correct that journalists have learned nothing about early polls from history and political science? I think his insight is best expressed by lyrics from Kris Kristofferson’s “Shipwrecked in the Eighties”:

Like an old Holy Bible you clung to through so many seasons
With the rules of survival in words you could still understand
When they prove something wrong you believed in so long you go crazy