Can Poll Respondents Predict the Future?

We're on the verge of a government shutdown, and the political question of the hour is: If a shutdown happens, who comes out ahead politically?

As is often the case, polling constitutes the only empirical evidence we have to try to assess. And the polling we have is based on similar, future-tense questions, with different polls offering different results. The Wall Street Journal and NBC say President Obama. The Washington Post and ABC say it's even. So does Gallup.

This brings up a broader question about polling reliability: Can poll respondents accurately predict future opinions?

Not always.

I asked several polling experts what they thought about this, and former CBS polling director Kathy Frankovic pointed to an example from the Clinton administration -- when Republican impeachment proceedings loomed -- when the public said they would feel one way before an event happened, and in the end felt another way entirely.

Frankovic said, via email:

CBS News asked a national sample in the week leading up to the House impeachment vote what they wanted IF the House voted to impeach Bill Clinton, and a sizable majority said they would then want him to resign.  We called many of those same people back AFTER the impeachment vote was taken, and only about a third supported resignation.

Huffington post polling editor Mark Blumenthal sees opinions as fickle, given that different phrasings have elicited different responses.

Recommended Reading

Blumenthal, via email:

As a general rule, when small, subtle changes in wording produce strange or unexpected differences, the culprit is usually fleeting reactions that are not grounded in strongly held opinions -- what pollsters call "non-attitudes." In this case, the difference made by changing "responsible" to "blame" probably tells us that it is simply too soon to anticipate how Americans will react to a government shut down, should it occur.

Gary Langer of Langer Research, who heads up polling for ABC, cautions that it's just tough to know:

You can't really know who to blame for something until it's happened, you've picked through the wreckage and assessed who's at fault. At the same time, in a situation like this, results to such questions aren't entirely spurious, since there's a very strong partisan element, and partisan predispostions are pretty well fixed.

I'd also warn against assuming that, for all Americans, there'll be "blame" to lay. Note the following from our last ABC/Post poll. We very intentionally asked about "responsibility," not "blame" (and tried to focus it on the then-current state of play); then we tested positive/negative reactions to the prospect of a shutdown. Surely it would be problematic to try to assess "blame" among the three in 10 Americans who said it'd be a good thing.

Drop-down bar thumbnail credit: Gage Skidmore/Flickr