Polling vouchers: the choice of words can be taxing

Gov. Greg Abbott reminded the public and, more pointedly, state legislators Tuesday that he intends to call the legislature into another special session in October to “vote in favor of school choice.” In addition to adding an extra turn or two to the torqued-up tension between the House and Senate, Abbott’s resumption of the battle over vouchers will also reignite debates over whether “the public” supports such measures.

As they struggle to claim the cloak of public approval for their respective causes, advocates both for and against voucher or voucher-like programs will resume a familiar pattern (at least to non-partisan pollsters) of praising or criticizing public opinion polling that either supports or undermines their efforts to claim a public mandate (or at least a Republican one). In doing so, both sides will heavily rely on the same argument when responding to survey results they don’t like, to the effect of: well, wouldn’t the results have been different if you had asked that question differently?

We should note that the tone in which this “question” is delivered from advocates ranges from brittle disingenuity (“I wonder what happen if you changed the wording to…”) to hostile condescension (“If you weren’t so completely in the tank for their side, you’d have asked that by including…”). Whatever the tone and intent, the answer is surprisingly straightforward.

Yes. The results would be different if you asked a different question – because it’s a different question.

These kinds of responses from advocates happen with a lot of issues, of course. But we see this argument raised frequently (and with a lot of ferocity) in response to polling assessing voters’ attitudes toward school vouchers. 

Some of this frequency and ferocity has to do with the nature of the issue and the coalitions it mobilizes, while also reflecting the decisions pollsters necessarily make about question wording. With all this in mind, we decided to test our sense of whether and how much the common wordings of school voucher questions, in our surveys and those of others, produce varied results. Results from this experiment nicely illustrate the difference wording can make, while also illuminating why different wordings might elicit different interpretations and associations that lead to variations in results.

In August 2023 UT/TXP polling, we randomly asked half of our respondents whether they support or oppose “establishing a voucher, educational savings account (ESA), or other ‘school choice’ program in Texas” while asking the other half whether they support or oppose “redirecting state tax revenue to help parents pay for some of the cost of sending their children to private or parochial schools?” In setting up the survey this way, we know that any difference we find in response to these two questions would be due to differences in the question, as respondents were randomly assigned to see only one version of the question and not the other.

Here is the full wording of the two forms of the question (with links to results and several cross tabulations):

• "Do you support or oppose redirecting state tax revenue to help parents pay for some of the cost of sending their children to private or parochial schools?"
• "Do you support or oppose establishing a voucher, educational savings account (ESA), or other “school choice” program in Texas?"

Overall support didn’t differ dramatically based on question wording. Asked about establishing a voucher or voucher-like program, 52% of Texans expressed support, while 45% expressed support for redirecting state tax revenue to private school tuition. 

While the difference wasn’t large, there are some interesting differences hiding in this 7-point gap in support. For example, 29% of voters strongly support “establishing a program,” compared to 21% who strongly support “redirecting tax revenue,” with a similar gap emerging among Republicans (41% vs. 32% strongly supporting), the primary audience for these policies. Not surprisingly, including the reference to “tax revenue” in the question reduced strong support among both Republicans and voters overall.

Larger differences exist, however, when looking at the share of voters opposed to each proposition. While only 27% of voters say that they oppose establishing a voucher, ESA, or other school choice program, when asked about redirecting tax revenue to help parents pay for some of the cost of sending their children to private or parochial schools, opposition increases to 42%, nearly identical to the level of support. And while the share of Democrats in opposition increases from 43% to 54%, the share of Republican opposition more than doubles, from 14% to 33%.

Given the impact that location has on school quality, especially the range of options available to parents, it’s not surprising to find that geography also matters. Among suburban voters, 49% support establishing a voucher program with 31% in opposition; but when asked about redirecting revenue to help parents pay for the cost of private education, the results flip, with 49% expressing opposition compared to 39% who express support. Among rural voters, the focus of much of the retail politicking engaged in by the governor in support of these policies during the regular legislative session, 61% support establishing a voucher program (18% opposed); while near equal shares support (43%) and oppose (39%) redirecting tax revenue to fund such a program.

A fair way to interpret these results is to acknowledge that the establishment of a voucher program, likely or especially couched in terms of “choice,” is more popular than specific aspects of what that program would actually entail. It’s these specific aspects that often cause fits for pollsters in trying to write good, unbiased questions attempting to measure support or opposition for voucher programs — themselves rarely clearly defined in the legislative process when proponents and opponents start competing to describe how public opinion favors their position. The flipside of this difficulty is that it’s not terribly difficult for advocates to write questions likely to produce results that favor their position (such as by using references to “failing schools” or “depriving schools of needed funds”), another reminder to consider the source of publicly-released poll results.

It shouldn’t be surprising that the specifics of the voucher policy probably matters when measuring these attitudes. Any proposal entails a wide range of consequential policy choices.

For example: are funds available for secular, private schools, parochial schools, homeschooling supplies, or some combination (and which combination - who is included and who is left out)? What is the source of the funding, and how is that funding described (e.g. taxpayer funds, the current public education budget, the budget of local schools, “savings accounts”)? Who is eligible for the program being considered (e.g. all students, students below a certain income threshold, students from schools identified as “failing”)? Would the proposed program provide enough funding to cover the whole cost of private education, or only part of the costs? Will the schools receiving the funds be expected to provide certain services, and/or meet certain standards? What kind of accountability measures follow the money, if any? It’s easy to imagine how these aspects might make a difference when assessing voters’ reactions to a proposed school choice program, and might even be likely to make a difference to voters with vastly different predispositions on the subject.

Another deceptively simpler decision that a pollster must consider when constructing a question about such programs is what to even call them. While “vouchers” have been a common description of these efforts for decades, the term has developed a negative connotation (on the ideological left) — an issue often raised by school voucher proponents in recent years, who don’t want their preferred program to be called a voucher. At the same time, the seemingly innocuous term “school choice” might generate less native resistance precisely because it, too, is a term of art  designed to achieve the highest level of support (likely tested with polling), or at least to minimize resistance (Who’s against more choice?). This is to say nothing about how to describe an educational savings account.

Given all of these possible considerations, it should be easy to imagine collecting very different responses to a question gauging support for an educational savings account funded with a tax credit for private school tuition, let’s say, versus a universal school voucher program that takes formula funding from public schools and transfers it to individual parents for use on any qualified educational expense. And as the survey experiment described above seems to illustrate, it’s easy to describe public sentiment very differently, with opposition significantly increased in voters’ responses to the question that describes some of the mechanics of a voucher program compared to a grab bag of terms often used to describe these types of programs with no further detail. Results to different questions will, in fact, vary, and therefore, it’s especially important to consider not only the source of public perceptions and the specific question wording used, but also the source of any criticism — frequent though it may be.