First, overall government spending passes Jeremy's Mueller test. The GSS asks:
If the government had a choice between reducing taxes or spending more on social programs like health care, social security, and unemployment benefits, which do you think it should do? (We mean all taxes together, including social security, income tax, sales tax, and all the rest.)
The public comes down about 60/40 in favor of more spending. (For similar questions where "keep the status quo" is on the list of answers, that's normally the median position; but I can't quite remember where I've seen those questions).
Second, the Mueller effect could be one of two things: An information effect or a question-wording effect. What's the difference? An information effect is when you tell the respondent something he doesn't already know (or pose a hypothetical that he doesn't take for granted). A question-wording effect is when you redescribe something known to make it sound better or worse.
Take for example the question: "Abortion stops a beating heart. Should abortion be legal?" I strongly suspect that you will get unusually pro-life answers. But is that because respondents do not know that fetuses have beating hearts? Or is it because you've worded the question to get a pro-life result? I say the latter.
In contrast, you could ask people "Would you favor or oppose raising the minimum wage if it resulted in fewer jobs available to low paid workers in this country?" Gallup once did just that; support for the minimum wage sharply falls. Since I'd never even heard about disemployment effects before I started studying economics, I say that this is an effect of information, not mere redescription.
I think the Mueller test is a lot more like the abortion question than the minimum wage question. Maybe the public doesn't realize that more government spending requires more taxes. But even I doubt that the public is that stupid.
Still, I'm open to persuasion. A dissertation on the Mueller test sounds like a great idea to me. Any takers?