Sunday, January 23, 2011

Parsing Question-Wordings Used in Surveying on ACA Repeal

Recently, a number of columnists have remarked upon the lack of consistency (and in some cases, lack of precision) in how various pollsters have attempted to assess public opinion on repealing the health care reform (HCR) law enacted last March (see Jon Terbush, Greg Sargent, and Frank Newport).

My aim in this posting is to distill these writers' comments into a graphical format, depicting polling results obtained this month. First, though, I classify different polling outfits' question-wording formats into different categories (links to reports of a particular poll's results are given only if they're not available on Polling Report's HCR compendium).


Only a dichotomous repeal/uphold item.

CNN/Opinion Research Corporation

NBC News/Wall Street Journal (respondents could choose whether they "strongly" or "not so strongly" held their views toward repealing or upholding HCR law; I combined the two pro-repeal and two pro-retention options in the graph)


Rasmussen (respondents could choose whether they "strongly" or "somewhat" held their views toward repealing or upholding HCR law; I combined the two pro-repeal and two pro-retention options in the graph)

Dichotomous repeal/uphold item, followed by item(s) to probe in greater detail (e.g., repeal all or only part of HCR law)

ABC News/Washington Post

CBS News/New York Times

Item directly asks respondents to choose between four main options: repeal all, repeal parts (i.e., make law do less), keep law as is, or expand law.

AP/GfK ("What would you prefer Congress do with the new health care law? Leave it as is. Change it so that it does MORE to change the health care system. Change it so that it does LESS to change the health care system. Repeal it completely.")

McClatchy/Marist ("Which one of the following comes closest to your opinion about what Congress should do with the 2010 health care law? Let it stand. Change it so it does more. Change it so it does less. Repeal it completely.")


Each segment along the horizontal axis at the bottom of the graph represents a different poll (identified by sponsor/pollster and dates in the field). Above each entry, you can find the percentages of respondents endorsing a particular option signified by the color of a given dot. (You can also click on the graph to enlarge it.)

Let's look first at the blue and orange dots, which represent, respectively, percentages favoring repeal and retention in polls that examine the issue dichotomously (either with only a dichotomous item, or a dichotomous item followed by a probe for further detail). In other words, these items do not permit distinctions between full or partial repeal, or keeping the law as is or perhaps expanding it.

As Sargent says, "When pollsters offer respondents a straight choice between full repeal and leaving the law as is, more end up supporting repeal." Actually, this trend is more apparent for polls taken earlier in the month than later, although on the available information, we cannot separate the effects of polls' timing vs. those of the outfits conducting the poll.

The polls that use what I think is the best format -- AP/GfK and McClatchey/Marist -- both show expanding the scope of the HCR law (green) to be more popular than full repeal (red), status quo (brown), or partial repeal (lavender). In the polls shown in the graph, full repeal -- which is what passed the Republican-controlled House on January 19 -- tended to draw between 20-30 percent support.

In his January 19 column, Newport alluded thusly to an additional survey: "A USA Today/Gallup poll question we asked over this past weekend found that if given a number of options about what to do with the bill, 13% of Americans would keep the bill just as it is, and 32% would repeal in entirely. The rest, 53%, would opt to make minor or major changes in the bill." I overlooked this poll when I was creating my graph. The 32% support for full repeal is at the high end of the aformentioned range; also, the phraseology regarding preferences for major/minor changes is different from other polls' wordings.

[This entry replaces and expands upon my January 21, 2011 posting.]

No comments: