Pollster.com

On issue polls I look a the margin of error, and multiply it by 3 (or 4 if its Rasmussen). If the results are outside of that, then, I see the poll as having significance. Otherwise it goes into my mental circle file.

First of all, I would not have dropped the "next to" from the quotation in the title.

I think Klein is a bit confused here. The problem he actually highlights isn't that polling doesn't convey useful insight into the content of people's "mixed feelings." It's that the ginormous headlines don't.

I think Klein is right. Nobody ever asks the right question on a complex issue, like health care. Asking "Do you favor Obama's health care proposal, or not?" is of course ridiculous. The President only proposed outlines and a 2000 page House bill so far produces the only details. Who can know what's in it?

A better question (which I don't recall ever being polled) would be one like this: "Currently, there are government-run health insurance plans like Medicare and Medicaid for certain Americans and there are privately run health insurance plans like Blue Cross/Blue Shield, Humana, United Health Care and others for other Americans. If a plan was put forward to expand health care coverage, which type of plan would you trust more to pay for procedures that you and your physician think should be done to improve your health. A government run plan like Medicare or Medicaid, or a private plan, like Blue Cross or the others mentioned above?"

A question like would at least offer some semblance of an intelligent response from most people since it gives them a frame of reference to opine on.

Mark L is right. As noted above, I restored "next to" in the headline. Apologies to all.

@nelcon1551: Do you have a hunch what result that question might produce? And don't the actual proposals (the bill passed in the House and pending in the Senate) offer a choice of either type of plan?

Personally, as an individual now covered by Medicare after decades in high paying jobs with high quality private health plans, I would vote strongly for Medicare. It is by far the best and most efficient and effective medical insurance I've ever had.

But since most people aren't over 65 and know Medicare only by "reputation" or through relatives, I'm not sure what the response would be. Also, Medicaid might be a wild card in the mix. So my answer is that I'm not sure what the public split on the question would be, but at least those polled would have a good frame of reference for their viewpoint.

As for the current bills in Congress, they don't necessarily offer an open choice of plans. I believe that only if a private plan is unavailable to a certain individual, then there would an open choice to make.

I've given this a think over night, and I think I am in agreement. There is a stark difference with issue polls as opposed to candidate-centered-polls. Candidate polls generally give your a very limited and concrete set of responses: Would you vote for X or Y?; Do you approve of how Y is handling her job?; etc. Issue-based polls, as alluded to, have results that are subject to wild fluctuations in conjunction with the language of the question. Since there's too much variation in responses to these types of polls, it strikes me as very difficult for a pollster to claim that their language is "correct" when there is not a credible null hypothesis against which to judge the results (because of said extreme variation in the extant issue polls). As Klein suggests, the real problem is the reporting done on these polls with erratic results, where outlets will latch onto one poll's results and report is as fact. Maybe I'm just in the can for John Stewart, but I think that cable news is the main problem here while polling methodologies and question verbiage are largely inside baseball.

I agree with Klein. Cable news have a habit of creating a narrative around a single poll.

issue polls are not at all useless. The different wordings precisely show how people respond to different nuances. We need more, rather than less of these nuances.

This is especially the case, obviously, if polls also disaggregate how attitudes change depending on the knowledge and experience of various groups, and pollster.com has been doing a great job at that.

That issue polls are abused in sensationalist ways is a different story, and does not make issue polls less valuable in themselves.

We do polling in the South Caucasus, and while many preferences are latent and inarticulate, understanding what people think of ongoing reform initiatives is one important dimension (not sufficient, but necessary) to their eventual success.