Uncertainty Doesn't Make Polling Worthless

(Bloomberg View) -- On Tuesday, voters in Alabama will either put Republican Roy Moore or Democrat Doug Jones in the U.S. Senate. 

Who will win? No one knows. As Harry Enten reported last week, the polls now show a clear, but not decisive, lead for Moore.

But if there's ever an election that the polls may get wrong, it's this one. The main problem is that it's impossible to know who will vote, and therefore no way to know how to interpret polling results. SurveyMonkey's Mark Blumenthal shows that perfectly reasonable adjustments for expected turnout can lead to estimates of anything from a 9-percentage-point lead for Moore to an 8-percentage-point lead for Jones based on the same underlying numbers (see also Charles Franklin's discussion). 

Pollsters have always had to use complicated techniques to estimate turnout because it turns out that if you just ask people whether they will vote, they'll get it wrong -- in part because people don't like to admit to not voting, in part because people aren't necessarily good at predicting their own actions. That's always been a problem with polling, and the much lower response rates for traditional surveys nowadays don't help. The electorate is even harder to estimate in special elections. Add to that the peculiar nature of this particular special election, and it gets even more difficult. Franklin cautions us against saying that pollsters are just guessing at the turnout, so I'll just say that whatever they do will either be based on a lot of untestable assumptions or on projecting regularities forward from extremely limited data. Or both.

On top of all of that is the possibility that late-breaking developments after the last polls are taken could sway the electorate. In fact, given that voters may well tune in later for special-election media coverage than for regular general elections, it wouldn't be surprising if voters decide later than usual. I don't know whether Donald Trump's enthusiastic campaigning at the end will bring partisan Republicans around, or that Alabama's Republican senior Senator Richard Shelby's decision to go on national television to make clear he hadn't voted for Moore might give some Republicans permission to do likewise (or just stay home). But I wouldn't count it out, and if it matters, we won't know from polls conducted before those things happened.

It gets worse. I'm not a great believer in theories of "shy" voters who supposedly may lie to pollsters because they feel pressure against admitting to plans to vote for a socially undesirable candidate. But if it's ever going to happen, this Alabama Senate special is as good a contest for it as we'll ever find. But it might work in either direction! At least among Republicans, that is. It's easy to imagine a Republican who plans to vote for Moore but doesn't want to admit to supporting a candidate who is an accused child molester. It's also, however, easy to imagine Alabama Republicans who can't stomach Moore and plan to vote for Jones but don't want to admit to voting for a Democrat. 

The punchline? None of this means that polls are phony, or worthless, or that polling averages are bad ideas. Survey research is a tool, and like all tools it has both uses and limitations. Polling averages are helpful, but only to the extent that the polls themselves do what we want. Just because the results in Alabama may well be far off from the final polling averages is no reason to throw out perfectly good tools when the need for them arises. 

1. At the Monkey Cage, John Sides talks with Molly Reynolds and Sarah Binder about the agreement to keep the government open for two weeks -- and the next step.

2. Jeremy Pressman at the Monkey Cage on the Jerusalem declaration and the Trump administration's plans for the region. 

3. Sam Rosenfeld on how parties change over time and why it would take more than a Trump election to change the parties.

4. Seth Masket and Sherry Zaks on why the Masterpiece Cakeshop case is a genuinely difficult constitutional puzzle. 

5. David Wessel at Brookings on Ronald Reagan's tax cuts

6. Greg Sargent on Roy Moore and Al Franken

7. And my Bloomberg View colleague Tyler Cowen on Washington-area cuisine and immigration. All I can say from my occasional visits is that Washington is a far better place to eat than it was when I lived there -- but the improvement isn't nearly as dramatic as the improvement in those parts of the nation that had frightfully few good options until very recently but now have lots of great choices, both eating out and via the local supermarket. 

Get Early Returns every morning in your inbox. Click here to subscribe.

This column does not necessarily reflect the opinion of the editorial board or Bloomberg LP and its owners.

Jonathan Bernstein is a Bloomberg View columnist. He taught political science at the University of Texas at San Antonio and DePauw University and wrote A Plain Blog About Politics.

To contact the author of this story: Jonathan Bernstein at jbernstein62@bloomberg.net.

For more columns from Bloomberg View, visit http://www.bloomberg.com/view.

©2017 Bloomberg L.P.