Fiscal Policy Surveys and the "Mueller Test"

GMU Professor Bryan Caplan has posted some survey data on farm subsidies. My co-blogger and resident survey expert has already weighed in on the comments of Caplan's blog about the survey design (more on that aspect in a minute). Caplan presents the survey as showing that Americans overwhelmingly favor the status quo in farm subsidies, and that this fact holds true over both farming and non-farming states, once again refuting the self-interested voter hypothesis.

I would add, though, that several questions in the survey present an answer to Caplan's challenge. First, a solid majority of almost two-thirds opposes subsidies to large farms, defined as 500 acres or more (aside: how many Americans have any conception of how big an acre is?). And of those that support subsidies to small farms, the majority would only do so in "bad years," meaning that once again close to two-thirds of Americans oppose farm subsidies on a regular annual basis (aside: only half of those surveyed know that farm subsidies are actually given out on a regular annual basis). Also notably, less than one-quarter of Americans support subsidies to tobacco farmers, yet they persist.

Now back to the survey design. There seems to be a larger problem with the design of this and most other surveys, outside of the "framing issue." For evidence, I would encourage you to read Eva Mueller's 1963 QJE article "Public Attitudes Toward Fiscal Programs." Mueller's survey design asked respondents many broad questions on fiscal policy, but then adding an interesting twist: asked the same questions again (only to those that wanted to spend more), but mentioned that taxes would have to be increased to provide the additional funding. The results are very revealing.

Keep in mind with all of these figures that the follow-up tax question (I'll call it the "Mueller Test") was asked only to those who first favored more spending on a program, those that we might view as more "fiscally liberal" to begin with. Of the fourteen broad spending categories investigated, five had clear majorities favoring more spending, and two more were within three percentage points. When the Mueller Test was applied, not a single program had majority support for increased spending. Remember, that's among those that had previously supported increased spending!

The category that fared the best was education, with 41% of the original 60% of "increase spending" types passing the Mueller Test. Of note for one of my colleagues, only 26% supported increased spending on space exploration (more supported less spending), and only 14% of the big spenders still supported it when the Mueller Test was applied. And this was in November 1961, just four years after Sputnik, and with Kennedy starting to ratchet up his support for putting a man on the moon.

To wrap this up, yes, it is true that this data was collected in the early 1960s. But first, this presents an interesting intertemporal response to "Caplan's challenge," since it seems certain that all the categories asked about have been vastly increased in real terms since the early 1960s, yet none had majority support when the Mueller Test was applied. And more broadly, we should be skeptical of any survey that does not use the Mueller Test. It seems easy to include this in the survey. So why not? Maybe you won't get the answer you want.