Americans Don’t Think That Using Nuclear Weapons is a Taboo

by Erik Voeten on November 17, 2011 · 12 comments

in Public opinion,War

Why have nuclear weapons not been used after they seemingly effectively led to the capitulation of Japan in World War II? One important explanation advanced by Nina Tannenwald and others is that first usage has become stigmatized as unacceptable. Indeed, the argument goes, the normative prohibition against first use has become so strong that it is like a “taboo.”

A key characteristic of a taboo is that beliefs about wrongness cannot be easily manipulated by shifting peoples’ perceptions of costs and benefits of violating the normative prohibition. A new paper (pdf, non-gated) by Scott Sagan, Daryl Press, and Benjamin Valentino shows that the moral aversion to using nuclear weapons is much weaker than the taboo argument suggests, at least among the American public. The authors asked survey respondents whether they approved of or even preferred the use of nuclear or conventional weapons in two scenarios: the discovery of an Al Qaeda nuclear lab in Syria and retaliation against an Al Qaeda attack on a cruise ship that had led to mass American fatalities. The retaliation was targeted at a base in Syria. Both attacks would lead to Syrian civilian casualties. (Syria was chosen because at the time of the survey, Americans were largely indifferent about the country).

Randomly selected groups of survey respondents were told that the nuclear and conventional attacks would either be equally effective or that the nuclear attack had a greater chance of success. Everything else was held constant. When both options are equally effective, only a relatively small proportion of respondents prefers nuclear weapons, presumably to “send a message.” Yet, when nuclear weapons are portrayed as having a 90% success rate while conventional weapons hit the target with only 70% of the time, a (small) majority of Americans prefers nuclear weapons to conventional ones. This effect is even greater when the discrepancy in success rates widens further. Moreover, among the people that still prefer conventional weapons, most say that they do so not out of moral aversion but because they are concerned that first usage sets a dangerous precedent.

Obviously, these are results of one survey among the general public. They may not accurately convey how elites or other publics think. Especially the President has a unique set of considerations on this question. What the results do convey, however, is that a President may very well be able to muster majority support for nuclear usage by appealing to the effectiveness of the latest generation of nuclear weapons. Perhaps we ought not to be surprised by this but it is sobering and important to realize nonetheless.


Andrew Gelman November 17, 2011 at 12:40 pm

A minor comment, but since I’m here . . . I like how the numbers are presented in the blog post (“70%”, “90%”). I don’t like how the numbers are presented in the graph (“18.9%”, “47.9%”, etc.). The fractions of a percentage point don’t mean anything. Even if n=1000 they wouldn’t mean anything. But certainly not with n=157! Also, next time I’d recommend a line plot rather than a bar plot. A line plot allows comparisons along as well as between the lines. Don’t get me wrong—I like the graph. I just think it could be improved.

LFC November 17, 2011 at 10:26 pm

This exercise has more than a bit of an air of unreality or irrelevance, since no matter what percent of the public favored it, no sane President would order first use of nuclear weapons. Official US nuclear doctrine, unless I’m much mistaken, abjures first use. Admittedly it’s frightening to see the result (which I noted in glancing through the paper) that 72 percent of respondents approved a nuclear strike on the hypothetical al-Qaeda base from which the attack on a cruise ship had been launched (where the hypothetical base contained no WMD). I’m inclined to think these survey results are more of an indictment of the US educational system than anything else.

Dylan November 18, 2011 at 1:37 am

“Official US nuclear doctrine, unless I’m much mistaken, abjures first use.”

You are mistaken, but might be functionally right under Obama as of 2010:

First use was a key NATO defensive plan against the Soviet Union.

LFC November 18, 2011 at 8:45 am

Thanks. (I guess I was thinking of the pledge not to use against non-nuclear-weapon, non-proliferating states, which of course is different than blanket no-first-use.)

Erik November 18, 2011 at 9:38 am

I agree (and note in my post) that the President may well have different considerations but this remains an untestable assumption. Doctrines can change. There was never a doctrine on preventive war until the Bush administration invented one to justify the Iraq war. The same can happen with nuclear arms. This isn’t a short-term issue of what Obama may do. What it does tell us is that we shouldn’t count on the American public to reign in a President who may be tempted. Constraints have to come from some other source.

Talleyrand November 18, 2011 at 10:20 am

The conclusion drawn in this post/paper about the weakness of the moral aversion to nuclear weapons is misleading. It’s like saying that the taboo on adultery doesn’t exist because people are more likely to commit adultery when the person is very attractive and they won’t get caught. This study says that if you start talking to people about the costs and benefits of using nuclear weapons, they will think about them in terms of costs and benefits. This disproves the hypothesis that people generally do not think about nuclear weapons in terms of costs and benefits!!!11!

Plus, saying that people give a pragmatic reason for their choice ignores the fact that people generally do not provide moral reasons when there is a pragmatic alternative. And how does it constitute an equilibrium in a prisoner’s dilemma if Syria has no nuclear weapons? The “rationalist-functionalist” basis for a prohibition on first-use does not make sense in the context of a non-nuclear adversary. The paper hilariously cites Schelling who has himself written about the nuclear taboo.

It is baffling the extent to which some scholars will go to strawman any constructivist arguments. Here, in a case where nukes are twice as effective and there is no chance of retaliation, a third of respondents still oppose their use. Proof that the taboo is weak!

Erik Voeten November 18, 2011 at 11:54 am

Adultery is not a taboo, cannibalism is. What this shows is that people consider nuclear usage more like adultery than like cannibalism.

You also misunderstand what they did: different people were randomly assigned to read identical reports that differ only in success rates. So people were not explicitly primed to think about this in utilitarian terms. And note that nearly half the people in the equal effectiveness condition thought nuclear usage acceptable.

This is not about constructivism. This is about not fooling ourselves that a president would never use nuclear weapons because the public would never accept it. (Tannenwald’s argument is much broader. As I note in my post, elites may think differently.)

Talleyrand November 18, 2011 at 12:39 pm

I can accept your point about a president not being held back from using nukes by the public. That makes sense given the data that Sagan et al provide and anecdotal evidence of some of the discussion of nuclear weapons in the media.

I also accept that use of nuclear weapons is less viscerally disapproved of than cannibalism. It is a slightly disingenuous point, however, given that Sagan et al have a “constructivist” hypothesis, i.e. taboo, and a “rationalist-functionalist” hypothesis, i.e. fear of retaliation. They then rejected the former in favor of the latter. If nuclear use is more like adultery, then neither of these two hypotheses captures what is going on. Adultery still involves normative prohibition and moral condemnation of perpetrators etc. And the problem with it is not purely the fear that you will also be cuckolded.

And not only do I not misunderstand what they did, you seem to misunderstand the point. They presented the reports _in terms of success rates_. That is already presenting the information in utilitarian terms.

Erik Voeten November 18, 2011 at 12:46 pm

Then I don’t understand your point. If people are so easily swayed by simply pointing to success rates, then how important is the normative prohibition? I am not saying the normative prohibition isn’t important (it sure influences my own thinking) but surely any elite who would intend to go nuclear would make a utilitarian argument. If this so easily sways large segments of the public, then the normative prohibition is there but not all that strong.

Talleyrand November 18, 2011 at 2:34 pm

When explaining the tradition of non-use, you don’t actually need the norm to be that strong to be a crucial part of any explanation. But that is irrelevant to your point about the constraining role of the opinions of the public on the president.

I think it would be interesting to do another experiment (one the authors do not report) with open-ended answers and where the nuclear option is not mentioned or hinted at. Then see how many people volunteer that as a serious possibility. If a good amount of people volunteer nukes as a cheap way of dealing with the problem, something I have encountered anecdotally in private discussion, then that would be evidence against a normative prohibition. Also, maybe presenting a scenario where a nameless/neutral country has used nuclear weapons in pursuit of some self-interested policy goal and then open-endedly asking for reactions and reasons for those. If almost no one then said that it was ‘wrong’ or ‘barbaric’ or whatever, then there would be evidence against the existence of a norm.

Sebastian November 18, 2011 at 2:57 pm

I think Tannenwald’s use of the term “taboo” is part of the problem. When she says “taboo” – we think incest or cannibalism. But Tannenwald actually means “[a] normative prohibition on nuclear use has developed in the global system, which, although not (yet) a fully robust norm, has stigmatized nuclear weap- ons as unacceptable weapons of mass destruction” (p. 434 in her IO article:
On p. 436 she further explains how her use of “taboo” differs from its common usage and is inspired largely by the fact that policy makers use the term.

I’ve never been happy with this conceptual choice on her part, but still it’s pretty clear that “it is not like cannibalism” is indeed not a particularly relevant critique of Tannenwald’s argument).

Erik November 18, 2011 at 7:25 pm

Sebastian: it is a perfectly fair critique of one part of her argument (not the elite part). I suspect she herself would be shocked to learn that almost 50% of respondents would think nuclear force acceptable even in the absence of a clear utilitarian advantage. She has expressed concern that the new generation of nuclear weapons would erode moral aversion. Surely this paper provides evidence against the notion that domestic public public opinion is a strong constraining force, which IS a part (though not ALL) of her argument.

Comments on this entry are closed.

Previous post:

Next post: