Scott Sumner  

Are most Americans radical libertarians?

California is Broke(n)... Rugged Communitarians...

This survey by Stanford University suggests the answer is yes:

Screen Shot 2017-09-06 at 3.09.07 PM.png Notes: The surveys asked whether individuals agreed or disagreed with the statement "I would like to live in a society where government does nothing except provide national defense and police protection, so that people could be left alone to earn whatever they could." This question wording is from Page, Bartels and Seawright (2013). Cell probabilities above give the percent that either somewhat or strongly agreed.
Based on the numbers for Democrats and Republicans, it looks like around 50% to 55% of Americans claim to support an extreme minimal state, which means Milton Friedman and Friedrich Hayek were more socialist that the typical American.

Or perhaps (as I've argued elsewhere) there is no such thing as "public opinion". People are like electrons; you can't measure them without changing their positions. Lots of people gravitate to the "wouldn't it be nice" answers. As in, "wouldn't if be nice to have a simple flat tax you could put on a postcard". Or wouldn't it be nice to have free health care for all. Or wouldn't it be nice to spend more on education. Or have lower taxes. Or pay off the national debt. Or deport the "bad hombres". Or have the government leave me alone.

Most people don't have views that are internally consistent, so their "views" on public policy issues are strongly shaped by the wording of the polls. Next time a progressive tells you that "the polls show" that Americans agree with the Democrats, show them this survey.

Most politicians do not have a set of views that are entirely internally consistent, although they are a bit more so in places like the UK, where elections actually deliver governments with the power to enact legislative agendas. I recall reading that candidate Trump once confided to a friend that the best strategy was to just promise anything the public wanted to hear. Thus (unlike other politicians) he promised to pay off the entire national debt in 8 years. And he would do so not by raising taxes or cutting spending, but through "trade".
Indeed he promised to slash almost everyone's taxes while also boosting spending. That doesn't sound too painful. Nor does building an expensive wall with Mexico's money. Trump outflanked other politicians by not having even a tiny bit of lingering guilt about promising anything that sounded appealing. But when people criticize Trump for doing this, the average person just thinks, "all politicians do that", which to some extent they do.

PS. The most accurate polls are where there is no "wouldn't it be nice" position. Such as, "Who do you plan to vote for in next week's election?" Those force the public to make a choice between two unpleasant alternatives.

Comments and Sharing

COMMENTS (10 to date)
E. Harding writes:

Sumner, you definitely seem correct in this matter. Maybe ~40% of the public gives a totally different answer depending on the question asked.

Stephen Gradijan writes:


Those numbers almost certainly would be even higher if the statement was instead:

“I would like to live in a society where government does nothing except provide environmental protection, national defense and police protection, so that people could be left alone to earn whatever they could.”

Mark writes:

I think it'd be more accurate to describe people's opinions as being random variables, the distribution of which depends on the topic. If it's an issue one knows a lot about (or has thought a lot about, even if they've reached the wrong conclusion), they will be pretty consistent about what they believe. If it's something they don't know or think they know about, the distribution may be basically uniform: depends entirely on their current interlocutor.

In any case, abstract questions and answers are pretty much meaningless I think. I'm more interested in how many people want to increase/decrease the minimum wage or capital gains tax.

Miguel Madeira writes:

"Cell probabilities above give the percent that either somewhat or strongly agreed."

The "somewhat" includes people who are not "radical libertarians" (probably Friedman and Hayek will answer something like "somewhat agree").

Uday writes:

Unfortunately, this problem is found all around the world. Politicians have found it all too easy to promise unicorns and ponies without any of the strings attached. People can't have their cake and eat it too.

Andrew_FL writes:

The number overstate Republican support for limited government because of what I'd call "Tribal Desirability Bias"

It's Social Desirability Bias, except tribal.

Scott Sumner writes:

Stephen, Especially higher for Dems.

Mark, Even questions on taxes depend on the wording. I can write a question that makes it seem the public wants to cut taxes on the rich. That question would read: "What should the highest tax rate be?" They will pick a number below the current top tax rate. Ot I can ask "Should the rich pay more taxes?" And the public will say yes. It's all in the wording.

Miguel. I suppose that's plausible. But even so, the numbers are surprisingly high, especially for Dems.

Andrew, I agree.

martin writes:

[Comment removed. Please consult our comment policies and check your email for explanation.--Econlib Ed.]

Mark writes:


I guess my hope would be that the more specific the policy question (and the more sterile the language), e.g. asking about a specific provision of the ACA or Dodd-Frank instead of about 'healthcare' or 'big banks', would lead those without fully formed opinions to recuse themselves by saying they don't know/don't care/weakly agree or disagree.

If someone doesn't have a stable opinion about enough such 'detailed questions', then I don't think we should care about whether, in the abstract, they identify as libertarian or conservative or socialist or whatever. Answers to these general questions are, imo, really only useful inasmuch as they are predictors of more particular opinions.

Of course, you could measure how invariant public opinion is on a subject just by seeing how much results vary with the wording of questions. Doing so would allow us to see the difference between "50-70% of people support X with a weighted average of 55%" and "20% to 90% support X with average 55%." Would need a much bigger sample size though than is normally used in polls.

foobarista writes:

One other problem with polls is pollsters often make highly subjective interpretations of poll results that have little to do with the actual questions asked in the poll.

Once I read the details of a poll that claimed that some huge percentage of Americans don't "believe in evolution". It turned out that the actual poll questions were about religious belief, and the people doing the analysis essentially inferred that if you were anything other than a hard-core atheist, you didn't "believe in" evolution.

This would be a shock even to biologists and paleontologists, who aren't uniformly atheists; Darwin himself was quite religious.

After this, I realized that polls asking anything other than the most specific questions are crap.

Comments for this entry have been closed
Return to top