Science1 publisher2 min readPublished
An AAMC poll puts conservative support for health equity at 65% when the question skips a definition
Phillip Alberti told a National Academies roundtable that support for health equity moved with the question's wording. He quoted the figure for one subgroup answering one version. The size of the movement stays unquantified.
The Scientist · Science desk
What happened
- Phillip Alberti of the AAMC's Center for Health Justice told a National Academies roundtable on population health that polling he conducted found most American adults favor efforts to achieve health equity.
- Asked whether they support the goal of health equity with no definition offered, 65 percent of self-identified conservatives said they did, and Alberti argued the partisan difference is smaller than commonly described.
- The same panel discussed STAT's reporting over the past year on health equity researchers who said they have felt political pressure not to speak publicly about their work.
Compiled by The ScientistSomething wrong?How this is made
Why it matters
- constraint One version's percentage can set a floor for conservative agreement. It leaves the wording effect unmeasured, and the size of that effect is what any communications advice would rest on.
- decision For a researcher weighing whether to keep the phrase in a grant abstract, the operative question is what a funder's keyword screen flags, not what a majority of adults tell a pollster.
- precedent If the AAMC's next release reports each question version with its own sample, the claim that the partisan gap is overstated becomes checkable by outsiders; a topline-only release leaves it resting on the person who ran the poll.
A framing effect is a difference between two numbers. STAT's newsletter quoted one of them. "When we asked a representative sample of U.S. adults ... 'Do you support the goal of health equity' without any definition, 65% of self-identified conservatives said 'I do.' So I think even though there was a difference, it's not as big as maybe we are being told it is," Alberti said [3]. Alberti did not state the sample size, the field dates, or the percentages for the other versions of the question [9].
The usual way to measure a wording effect is a split ballot. One panel, respondents randomly assigned to two or three versions of the same item in the same field period, each version's percentage reported next to the number of people who saw it. Randomization is what lets you attribute the gap to the words instead of to who happened to answer. A four-point gap and a twenty-four-point gap sit behind the same summary sentence, and they point in different directions for anyone deciding whether to keep a phrase in a grant abstract.
The 65 percent also describes a subgroup. It covers self-identified conservatives, and roughly a third of that group sits outside the number [8].
There is a second limit in the item itself. It asks about a goal [3]. A respondent who agrees that health equity is a worthwhile aim has not been asked whether to fund a program, accept a tradeoff, or keep a particular office open, and those are the questions behind the political pressure health equity researchers have described to STAT [5].
The pressure runs through a different channel than public opinion. STAT reported on Wednesday that some researchers are suing the National Institutes of Health over its screening of grants for specific keywords [7].
Alberti said the AAMC will have more data in coming weeks about how trust in health care institutions has changed [6]. Trust in institutions is a separate measure from support for a goal.
What to watch
- Whether the AAMC's promised data on trust in health care institutions arrives with per-question percentages and sample sizes attached.
- Whether the AAMC releases the full questionnaire and crosstabs for the health equity item, including the framings Alberti left unquantified.
- Whether another pollster fields the same undefined-goal wording and lands near 65 percent among conservatives.