Tuesday, October 23, 2012

Ignore Those Polls! (Or At Least Be Wary)

Remember, most people do not have strong opinions about most policy options. Therefore, the way they answer survey questions about issue positions can be massively affected by current news events, opinion leadership, question wording, and pretty much everything else. The effect probably grows larger the less people know -- and think -- about the issue. Something that's a long-standing policy question, or something that is part of ordinary life, might produce relatively stable views; those issues which are rarely in the news, or highly abstract, are likely to produce less stable answers. That is, answers which are even more highly dependent on recent information flows and on question wording.

Example: both Conor Friedersdorf and Dan Larison had posts up before the debate last night emphasizing the reluctance of the American people to get any "more involved" in "Middle East leadership changes." That's from a recent Pew poll. But in early September, just before the attacks in Benghazi, Pew found strong pluralities approving of the intervention there (44/33) and of Barack Obama's overall handling of Libya (49/32). 

In other words, current opposition to the abstract notion of US "involvement" in "leadership changes" turns out to utterly fail to predict reactions to something that sure appears to fit that characterization. At least, when things were going well -- or, I should say, at least when no apparent bad news was visible.

For candidates, I think this suggests that for low-salience issues such as these, there's probably a way to phrase things so that they don't sound objectionable to most voters. That is, if you want to, say, intervene in Syria in a particular way, there's probably a way to describe the underlying principle that would support such a move that can produce polling majorities. Unfortunately, there's also a way to describe it that will produce polling majorities against the policy.

For presidents, the most obvious point is that what matters on low salience items such as these isn't so much how something polls going in, but whether the policy actually works or not. Or at least avoids damage that would show up on the front pages of newspapers, especially damage clearly linked to the policy. Of course, that's probably true for all policies; presidents who believe, for example, that policies which poll badly will yield strong economic growth should almost certainly pursue those "unpopular" policies. At least as long as there's enough time for payoffs before Election Day.

All of which is to say that if Friedersdorf and Larison believe that, say, an intervention in Syria would be bad policy, then that's the best electoral basis for opposing it. Not what people told Pew about some abstract question about intervention.


  1. About polls right now: James Fallows made a great point earlier today, that this election is turning into a good test of the "Moneyball hypothesis" in terms the pundits saying Romney is doing well right now while the prediction models, like 538, are showing a small to large Obama advantage, depending on whose model you look at.

    "The "pros" tell us that Romney is catching up, the quants say he is falling behind...we have an exceptionally clear case of people judging from their experience, their "bones," their personal instinct, etc that things are going one way (like veteran scouts saying that a prospect "looks like a Big Leaguer"), while data (on-base efficiencies in one case, swing-state polls in another) point in the opposite direction."


    1. He's trying to get it close enough to steal.

  2. Daniel Larison replies:


    The first paragraph summed up: even after it was a fait accompli, and at low cost, the American people didn't care.


Note: Only a member of this blog may post a comment.

Who links to my website?