Monday, October 26, 2009

How Closely Have People Been Following the Debate?

The Kaiser Family Foundation survey has been asking a question this year that attempts to gauge how closely people have been following the debate over health care reform. There is always the possibility that some respondents will overstate how closely they are following the debate, in order to present themselves as well-informed, civically engaged citizens. For that reason, it is helpful to obtain independent confirmation of a survey's general findings, from sources external to the survey itself.

It recently occurred to me that an independent source of information on people's interest level in a particular news story is the amount of web-searching they do on the story. Google Trends is one site that provides data on how much searching has taken place on an issue. Accordingly, I typed in "health care" to Google Trends and then did a screen capture of the results, shown below (see blue trendlines). I also manually superimposed the trendline (in red) for the percent of respondents who said they were following the health care debate "very closely" in the Kaiser polls.


The two trends -- web-searching and poll responses -- are on different absolute scales, but I don't think that matters. What's important, to my mind, are the arcs of the two trends. For the most part, they seem consistent with each other; both trends rose from July to August, and fell from September to October. One discrepancy is that, whereas the Kaiser poll showed a slight increase in interest from August to September, web-searching suggested a decrease from the August to the September peak. I also did a screen capture of Kaiser's summary table, for those interested in greater detail (note that, whereas the graphic trendlines go chronologically from left to right, Kaiser's data table goes chronologically from right to left). You can click on the graphic to enlarge it.

Saturday, October 24, 2009

CBS News polling director Sarah Dutton examines trends in support for a public option, including the effects of question-wording (hat tip).

Friday, October 23, 2009

Pollster.com's Mark Blumenthal examines conservatives' criticism of question wording on the public option in a recent poll.

Thursday, October 22, 2009

Andrew Gelman uses 2000 and 2004 polling data (2008 are not available) to examine support for federal spending to expand health-insurance coverage, along various respondent demographic dimensions (primarily income, age, and state of residence). If you think some of my graphics are funky, take a look at Andrew's!

Friday, October 16, 2009

Sen. Landrieu, the Public Option, and Free Lunches

Sen. Mary Landrieu, a relatively conservative Democratic U.S. senator from Louisiana, apparently is not on board for direct implementation of a public option, but would support a trigger to implement one automatically if certain conditions were not met (e.g., if health insurance prices exceeded a certain level after some amount of time had passed).

Within this context, Landrieu was asked in an MSNBC television interview why she opposed a public option when it enjoyed high support in public-opinion polls. Here is her answer (as transcribed by a Daily Kos discussant):

"I think that when people hear 'public option,' they hear 'free health care.' Everybody wants free health care. Everybody wants health care they don't have to pay for. The problem is that we as government and business have to pick up the tab, and as individuals. So I'm not at all surprised that the public option has been sold as free health care. But there is no free lunch."

(Here is a link to where the video is available.)

Landrieu makes a fair point. In fact, it's a point that can be examined empirically by comparing polls whose question-wordings do or do not explicitly mention that consumers would have to pay for (or purchase, etc.) a health-insurance policy under the public option, if they wanted to join such a program. If the polls whose wordings do not mention the pay aspect (i.e., that possibly imply the public option is a "free lunch") tended to be the only ones that generated majority (or at least plurality) support, then perhaps Landrieu would have a point.

One batch of polls that quickly shoots down the Landrieu thesis is the set of state-specific surveys commissioned by Daily Kos and conducted by Research 2000. The DK/R2K polls have consistently included the phrase "...that anyone can purchase," and even in moderate-to-conservative states such as Arkansas, Kentucky, and Montana, higher percentages of respondents have expressed support for a public option than opposition toward one.

One national poll that has used the "purchase" terminology when assessing support for the public option is that by The Economist/YouGov ("Do you favor or oppose having a 'public option' which would allow individuals to purchase health insurance coverage from the government?"). Although this poll comes out weekly, it does not always ask about the public option. I have plotted support levels for the last five Economist/YouGov polls, below (original reports: July 26-28, Aug. 16-18, Aug. 23-25, Sept. 13-15, Oct. 11-13).


Again, support for a public option seems pretty solid, even when the polling item makes clear that a public option is not a "free lunch" and instead would have to be purchased by interested individuals.

There is one area where I think Sen. Landrieu's claim may have some merit, however. It is true that polling results showing "super-high" (59-77%) support for the public option have not used question-wordings that make clear the program would entail a cost.

Wednesday, October 14, 2009

Pollster.com's Mark Blumenthal examines the polling -- and there isn't a lot of it -- on the American public's knowledge of what, specifically, is the "public option."

Friday, October 9, 2009

Economist Poll (But Not Others) Confirms Recent Surge (10/9)

Following up on my previous post, the just-released Economist/YouGov poll (in the field October 4-6) shows 52% support for President Obama's and the Democrats' health care reform plans, up from 49% in the same poll a week before. This finding is consistent with a recent upward trend from three other polling outfits.

Not all new polls show such high support, however. Pew Research Center (September 30-October 4) shows a stunningly low 34% of respondents favoring reform, whereas Quinnipiac University (September 29-October 5) pegs support at 40%.

Wednesday, October 7, 2009

Support for Reform May Be Rising, Following Mid-September Swoon (10/7)

Over at Pollster.com, Brendan Nyhan has reviewed trends in support for health care reform in the weeks following President Obama's speech to Congress. Nyhan acknowledges some short-term upticks in support, but concludes that in the long run, support has not changed all that much.

I decided to pursue this question too, but by breaking out the support levels from particular polls individually. The following chart (which you may click on to enlarge) is a bit busy! A couple of things to note are that the widths of the color-coded rectangles correspond to the days a given poll was in the field, and that the vertical gray bar represents the day of Obama's speech.


The trends in support appear similar to those in Nyhan's analysis. As can be seen in the above graph, polls from the Economist/YouGov (here and here), Rasmussen, and FOX News all show mid-September slides, probably a fall-off after the post-speech bounce.

However, Rasmussen, Gallup, and AP-GfK all appear to show early-October resurgences (a little less clear in the case of AP-GfK, as it doesn't have a mid-September data point). The Economist/YouGov poll from late September shows stabilization (although not gain) after its earlier polls showed a drop in support. The next Economist/YouGov poll, which should be out shortly, potentially could corroborate other polls showing an October rise in support.

Whereas Nyhan identified a trend of rising opposition to Obama's and the Democrats' reform plans, some of the newer polls (not graphed here) appear to suggest otherwise. Gallup showed opposition (i.e., the percent of respondents who would advise their members of Congress to vote against) falling from 40% in mid-September to 36% in early October, while AP-GfK found less disapproval of Obama's handling of health care in October (47%) than in September (52%).

Monday, October 5, 2009

Nate Silver suggests that unusually low levels of support for health care reform in FOX News polls may stem from question-order effects.

Sunday, October 4, 2009

Comments on Resurgent Republic Analysis (10/4)

The right-leaning political-analysis group Resurgent Republic recently issued a brief synopsis of national polls, arguing that a trio of issues -- consumer costs, increases to the federal deficit, and reduced quality -- are responsible for holding down support for President Obama's and the Democrats' proposed health care reform legislation (hat tip). I, too, have examined similar issues (here and here), and there are many ways to look at them.

Survey items about cost and quality generally offer respondents three choices: will reform plans under consideration make things better (i.e., reduce costs, raise quality), make things worse, or leave things unchanged? The thrust of my argument today is that the "leave unchanged" category offers useful information and should be incorporated into any analysis of poll results. Here are some examples of what I'm getting at:

Resurgent Republic (RR) says: “The CBS/[New York] Times September poll also found that only 19% believe the proposed changes will affect the cost of health care for the better.”

That figure is absolutely correct. But, if one looks at the full CBS/NYT report, one sees that in addition to the 19% who are optimists, 27% of respondents say costs will get worse, 22% say it will stay the same, and 28% claim not to know enough about the issue to comment. Among those offering an opinion, 60% thus feel health care costs will either get better or stay the same:

(19 + 22)/(19 + 27 + 22)

RR: “Only 14% believe the proposed health care changes will increase the quality of their own health care (CBS/Times 9/19-23).”

Percentages for the other answer choices were 30% worse, 32% same, and 21% don’t know enough. Thus, by the same calculation method as shown above, 61% of those offering an opinion say the quality of their own care will either get better or stay the same.

RR: “In the ABC/[Washington]Post surveys in August and September, 2x the number of voters believe health care costs will worsen instead of improve should health care reform pass (ABC/Post 9/10-12 and 8/13-17).”

Indeed, as documented at Polling Report's compendium of health care surveys (go to page 2 of PR's health care page), RR's statement is spot-on. For the September ABC/WP poll, 20% believe their health care costs will change for the better, 40% for the worse, and 38% feel these costs will stay about the same. Because there are so few "don't knows" in this polls, I won't bother to subtract them out. What we get is that 58% of respondents feel their costs will either come down or stay the same.

Just a few more. In the September 11-13 CNN/Opinion Research Corp. poll (results of which I found at Polling Report), the findings were as follows:

"From what you know of the health care reforms which the Administration is working on, do you think the amount you pay for medical care would increase, decrease, or remain the same?" Fifty-one percent replied that these costs would either decrease (16%) or remain the same (35%), whereas 47% thought they would increase.

"From what you know of those health care reforms, do you think you and your family would, in general, be better off, worse off or about the same?" Sixty-four percent said they'd either be better off (21%) or about the same (43%), whereas 35% said they'd be worse off.

In fairness to RR, they do on occasion frame their statements in a way that suggests a large percentage of respondents either think things will get better or stay the same: “NBC/WSJ surveys find that a plurality of voters (36%) believe their health care quality will worsen as a result of the President’s plan.”

There we have it. Consistent majorities -- oftentimes around 60% -- feel health care costs and quality will either get better or stay the same. Admittedly, the "stay the same" category is driving much of these results. Still, even if many American consumers feel their health care costs and quality will only stay the same (and not improve), that arguably is a good deal for a program that will provide health insurance for all or most of the 47 million Americans who currently lack it.