Keywords

What do people think about nudges? The simplest answer is that in multiple nations, strong majorities approve of them. They like educative nudges. They also like noneducative nudges . They favor both (with some important qualifications to come).

Surveys in many nations confirm these claims. Perhaps surprisingly, the same essential pattern of results can be found in the USA, Canada, Australia, the UK, Germany, Denmark, France, Italy, Canada, and Hungary. Importantly, there are differences across nations. For example, somewhat lower approval levels are found in Denmark and Hungary (and also Japan); within Europe, approval levels are highest in Italy; and in South Korea and China, approval levels are highest of all (by far). But across many nations, the basic tale is one of continuity. We might fairly speak of a consensus among democratic nations—and with some significant qualifications, some broadly similar patterns hold among undemocratic nations as well.

Most of my discussion will focus on the USA, in part because I have compiled a great deal of data there, and in part because the principles that explain people’s judgments in the USA can be found in many other nations as well (certainly in Europe). As we shall see, Americans tend to be highly supportive of nudges of the kind that have been seriously considered, or acted on, by actual institutions in recent years. This support extends across standard partisan lines; it unifies Democrats, Republicans, and Independents. So long as people believe that the end is both legitimate and important, they are likely to favor nudges in its direction. This is an important finding, a kind of green light for policymakers, because it suggests that most people do not share the concern that nudges, as such, should be taken as manipulative or as an objectionable interference with autonomy . (We shall find a few yellow and red lights as well.)

In fact, we will not find, in the various national surveys, higher levels of support for System 1 nudges or for System 2 nudges. Everything depends on the area and the direction of the nudge—not on what kind of nudge it is. Revealingly, Americans are far more negative about mandates and bans, even when they are taken to have perfectly legitimate ends. Many people do care about freedom of choice as such, and they will reject many well-motivated policies that do not allow for that kind of freedom. It follows that there is a high degree of skepticism about mandates and bans, taken as such, whereas with nudges, what matters is what those tools do—not the mere use of them.

People are most likely to oppose those nudges, whether educative or not, that (1) promote what they see as illicit goals or (2) are perceived as inconsistent with either the interests or values of most choosers. A third source of opposition, one that counts against some default rules , is that people do not want choice architects to produce economic or other losses by using people’s inertia or inattention against them. As we shall see, these are the three principal grounds on which people reject particular nudges. When nudges promote legitimate goals, are consistent with people’s interests and values, and do not impose losses, it is highly likely that strong majorities will support them.

A Principled Public?

I devised a nationally representative survey involving numerous nudges —thirty-four, to be exact. The survey was administered by Survey Sampling International and included 563 Americans, with a margin of error of plus or minus 4.1% points.

From the responses, it is clear that both System 1 and System 2 nudges—default rules , warnings, and public education campaigns—attract bipartisan support in the USA, unless people disapprove of their ends, or think that they are inconsistent with choosers’ values and interests. Importantly, several of the policies tested here can be counted as highly tendentious and arguably manipulative. Nonetheless, they too attracted majority support, with the single (and highly exotic) exception of subliminal advertising (which, surprisingly, receives substantial minority support in the context of efforts to combat smoking and overeating). It follows that Americans are reluctant to reject nudges as unacceptably manipulative. Their evaluations are dominated by their assessment of the legitimacy of the underlying ends.

As we will see, political divisions sometimes affect the level of support, because Democrats are more favorably disposed toward certain health and safety nudges than Republicans. (As I will emphasize, it would be easy to devise nudges that Republicans like more than Democrats.) And in cases that raise strong partisan differences, such divisions will map onto nudges as well. But across a wide range, clear majorities of Democrats and Republicans (and also Independents) are in full agreement about what they support and what they reject.

Popular Nudges

In the period between 2005 and 2016, the federal government in the USA adopted or promoted a large number of nudges . Three of the most prominent include (1) mandatory calorie labels at chain restaurants; (2) mandatory graphic warnings on cigarette packages (struck down by a federal court of appeals 1 ); and (3) automatic enrollment in savings plans, subject to opt out. We can see the first as a System 2 nudge, because it is educative; the second and third are more naturally classified as System 1 nudges. The nationally representative sample found substantial majority support for all three policies, including support for (3) regardless of whether it consists of federal “encouragement” of automatic enrollment or a federal mandate for automatic enrollment , imposed on large employers.

About 87% of Americans favored calorie labels and 74% favored graphic warnings . Both policies had strong majority support from Democrats, Republicans, and Independents. Overall, 80% and 71% respectively approved of encouraged and mandatory enrollment in savings plans. Here as well, all three groups showed strong majority support (Table 2.1).

Table 2.1 American attitudes toward four prominent nudges
Full size table

Three educational campaigns—classic System 2 nudges —attracted widespread approval. Respondents were overwhelmingly supportive of a public education campaign from the federal government to combat childhood obesity (82% approval, again with strong support from Democrats, Republicans, and Independents). Similarly, they were highly supportive of a public education campaign from the federal government designed to combat distracted driving, with graphic stories and images (85% approval). About 75% of people favored a federal education campaign to encourage people not to discriminate on the basis of sexual orientation, though here there was a noteworthy division across party lines (85% of Democrats, 57% of Republicans, and 75% of Independents).

Three other educational campaigns attracted majority support, but at significantly lower levels and with only minority approval from Republicans. About 53% of Americans favored a federal requirement that movie theaters run public education messages to discourage people from smoking and overeating. Democrats showed higher approval ratings than Republicans (61% as opposed to 41%, with Independents at 51%). By a very small majority (52%), Americans supported a public education campaign, by the federal government itself, to encourage people to give money to the Animal Welfare Society of America (a hypothetical organization) (59% of Democrats, 34% of Republicans, and 55% of Independents; party was a statistically significant factor). This latter finding seems surprising; it could not easily be predicted that respondents would want their government to design a campaign to promote donations to an Animal Welfare Society.

About 57% of people supported an aggressive public education campaign from the federal government to combat obesity, showing obese children struggling to exercise, and also showing interviews with obese adults, who are saying such things as, “My biggest regret in life is that I have not managed to control my weight,” and “To me, obesity is like a terrible curse.” This question was designed to test people’s reactions to a tendentious and arguably manipulative campaign, which might have been expected to receive widespread disapproval, as it did not. Indeed, one of the goals of the question was to establish such disapproval—but it was not found here. There was a significant disparity between Democrats (61% approval) and Independents (60% approval) on the one hand and Republicans (47% approval) on the other hand; the difference between the views of Democrats and Republicans was statistically significant (here and elsewhere, percentages may not total 100 due to rounding) (Table 2.2).

Table 2.2 American attitudes toward five educational campaigns
Full size table

Most Americans were also supportive of multiple efforts to use choice architecture to promote public health and environmental protection. In recent years, there has been considerable discussion of “traffic lights” systems for food, which would use the familiar red, yellow, and green to demarcate health rankings. 2 We can see this as a System 2 nudge insofar as it is designed to provide information in simplified form. In the USA, the national government has shown no official interest in these initiatives, but with respondents in the nationally representative survey, the idea attracted strong support (64%).

There was also majority approval of automatic use of “green” energy providers, subject to opt out—perhaps surprisingly, with support for automatic use of green energy whether it consisted of federal “encouragement” (72%) or instead a federal mandate on large electricity providers (67%). In these cases, there were significant differences across partisan lines, but majorities of Democrats, Republicans, and Independents were all supportive.

Most respondents were in favor of a System 2 nudge, requiring companies to disclose whether the food they sell contains genetically modified organisms (GMOs) (86% approval). Similarly, there was strong majority support (73%) for a mandatory warning label on products that have unusually high levels of salt, as in, “This product has been found to contain unusually high levels of salt, which may be harmful to your health.” Perhaps surprisingly, most respondents (but not most Republicans) approved of a System 1 nudge in the form of a state requirement that grocery stores put their most healthy foods in prominent, visible locations (56% approval; 63% from Democrats, 43% from Republicans, 57% from Independents). Respondents also supported a state requirement that people must say, when they obtain their driver’s licenses, whether they want to be organ donors (70% approval; 75% from Democrats, 62% from Republicans, 69% from Independents). For all of these policies, the differences between Democrats and Republicans were statistically significant.

Five other forms of choice architecture , most of which count as System 1 nudges and all of which might be expected to be far more controversial, nonetheless obtained majority support. The first would list the name of the incumbent politician first on every ballot. It might be expected that this pro-incumbent nudge would be widely rejected, because respondents might not want the voting process to be skewed in favor of incumbents, and because any effort to enlist order effects might be seen as manipulative (as indeed it should be). But a bare majority (53%) approved of this approach, perhaps because most people believed that it would promote clarity, perhaps because they did not see the risk of bias from order effects.

There was also majority approval (53%) for the approach, adopted in Oregon and California, of automatically registering eligible citizens as voters, subject to opt-out. Interestingly, most Republicans (61%) rejected this approach. One reason might be that they believe that people who do not take the time to register to vote ought not to be counted as voters. Another reason is that they might believe that Oregon’s approach would favor Democrats. Yet another reason is that they might believe that such an approach would increase the risk of fraud.

By a modest majority, most people (58%) also approved of an approach by which women’s last names would automatically be changed to that of their husbands, subject to opt-out. This approach obtained majority support from Democrats, Republicans, and Independents. This result is especially noteworthy in view of the fact that an approach to this effect would almost certainly be unconstitutional as a form of sex discrimination, even if it tracked behavior and preferences. 3 We might expect a difference between men and women on this question, but notably, 58% of both groups approved of this approach.

Finally, there was majority support for a federal labeling requirement for products that come from companies that have repeatedly violated the nation’s labor laws (such as laws requiring occupational safety or forbidding discrimination). About 60% of participants supported a System 2 nudge , with a significant difference between Democrats (67% approval) and Republicans (50% approval). There was also majority support for federally required labels on products that come from countries that have recently harbored terrorists. This approach attracted 54% approval—56% from Democrats, 58% from Republicans, and 49% from Independents (Tables 2.3, 2.4).

Table 2.3 American attitudes toward environmental and public health nudges
Full size table
Table 2.4 American attitudes toward some potentially provocative nudges
Full size table

Unpopular Nudges

By contrast, 12 nudges were widely disapproved. Of these, seven were noneducative nudges in the form of default rules . Two of these defaults were designed so as to be not merely provocative but also highly offensive, in the sense of violative of widely held principles of neutrality and thus illicitly motivated. Strong majorities took them exactly as they were designed.

Under the first, a state would assume that people want to register as Democrats, subject to opt out if they explicitly say that they want to register as Republicans or Independents. Of course, a default rule of this kind should be taken as an effort to skew the political process (and it would certainly be unconstitutional for that reason). 4 The overwhelming majority of people, including three-quarters of Democrats, rejected this approach (26% total approval; 32% of Democrats, 16% of Republicans, and 26% of Independents, with statistically significant differences between Democrats and Republicans). The second was a state law assuming that people are Christian, for purposes of the census, unless they specifically state otherwise. Such a default rule could also be seen as an attempt to push religious affiliations in preferred directions (and it would similarly be unconstitutional). 5 Here too, there was widespread disapproval (21% overall approval; 22% of Democrats, 27% of Republicans, 17% of Independents).

The third unpopular default rule (completing the set of unconstitutional nudges ) involved a state law assuming that upon marriage, husbands would automatically change their last names to that of their wives, subject to opt out (24% total approval; 28% of Democrats, 18% of Republicans, and 23% of Independents). Interestingly, there was no gender disparity here (just as with the question that involved the opposite defaults); 24% of both men and women approved. With the fourth, the federal government would assume, on tax returns, that people want to donate $50 to the Red Cross, subject to opt out if people explicitly say that they do not want to make that donation (27% approval; 30% of Democrats, 20% of Republicans, 28% of Independents). The fifth was identical but substituted the Animal Welfare Society for the Red Cross. Not surprisingly, that question also produced widespread disapproval (26% approval; 30% of Democrats, 20% of Republicans, and 25% of Independents). Somewhat surprisingly, and revealingly, the numbers were essentially the same for two charities, even though it might be expected that presumed donations for the Red Cross would be more popular.

With the sixth, a state government assumed that state employees would give $20 per month to the United Way, subject to opt out. It might be expected that because a state government and state employees were involved, approval rates might grow. But they did not (24% approval; 26% of Democrats, 17% of Republicans, and 25% of Independents). With the seventh, a majority (64%) disapproved of a federal requirement that airlines charge people, with their airline tickets, a specific amount to offset their carbon emissions (about $10 per ticket), subject to opt out if passengers said that they did not want to pay. Interestingly, a strong majority of Democrats (57%) disapproved of this approach, although the number for Republicans was significantly higher (75%) (Table 2.5).

Table 2.5 Unpopular defaults
Full size table

The five other unpopular nudges involved information and education; four of them should be counted as System 2 nudges. With the first (and most extreme), a newly elected president adopted a public education campaign designed to convince people that criticism of his decisions is unpatriotic and potentially damaging to national security. There was overwhelming disapproval of this campaign (23% approval; 24% of Democrats, 21% of Republicans, 22% of Independents). What is perhaps most noteworthy here is not majority disapproval, but the fact that over one-fifth of Americans, on essentially a nonpartisan basis, were in favor of this most unusual and quite alarming public campaign.

With the second, the federal government adopted a public education campaign designed to convince mothers to stay home to take care of their young children. Over two-thirds of respondents rejected this nudge (33% approval; 33% of Democrats, 31% of Republicans, 34% of Independents). The third (the only System 1 nudge in the group, if it can be called a nudge at all) involved a government requirement that movie theaters run subliminal advertisements to discourage smoking and overeating. Here too, there was majority disapproval (41% approval; 47% of Democrats, 42% of Republicans, 35% of Independents). It is noteworthy and surprising, however, that over two-fifths of people actually supported this requirement.

With the fourth, the federal government would require all products that come from a Communist country (such as China or Cuba) to be sold with the label, “Made in whole or in part under Communism.” Slightly over half of the respondents disapproved of this requirement (44% approval; 47% of Democrats, 43% of Republicans, 42% of Independents). With the fifth, a majority (59%) also rejected a public education campaign from the federal government informing people that it is possible for people to change their gender from male to female or from female to male, and encouraging people to consider that possibility “if that is really what they want to do.” There is yet another surprise here, which is that this pretty adventurous campaign was endorsed by 41% of respondents; note that approval rates differed between Democrats (49%) and Republicans (29%), with Independents landing between the two (38%) (Table 2.6).

Table 2.6 Unpopular education campaigns and disclosure
Full size table

Why Are Some Nudges Unpopular?

Two Dominant Principles

What separates the approved nudges from the rejected ones? Two principles seem to explain high disapproval rates. First, people reject nudges that they take to have illegitimate goals. In a self-governing society, it is illegitimate to attempt to convince people that criticism of a public official is unpatriotic. Certainly in the USA, nudges that favor a particular religion or political party will meet with widespread disapproval, even among people of that very religion or party. Of course, we could imagine a nation in which favoritism on the basis of religion or party would attract widespread support and might be seen as analogous to a default rule in which women’s last name changes to that of their husband (which was approved, it will be recalled, by a majority of respondents here). In such a nation, a default rule in favor of the most popular party or the dominant religion might be taken to track people’s preferences and values, and not to be a violation of the governing conception of neutrality at all.

The general principle justifies a prediction: Whenever people think that the motivations of the choice architect are illicit, they will disapprove of the nudge , whether it involves education or not. To be sure, that prediction might not seem terribly surprising, but it suggests an important point, which is that people will not oppose, for example, default rules and warnings as such; everything will turn on what they are nudging people toward. 6 By contrast, we will see that mandates do run into some opposition simply because they are mandates. When there are partisan differences in judgments about nudges , it is typically because of partisan disagreement about whether the relevant motivations are legitimate. Resolution of such disagreements would of course depend on judgments having nothing to do with nudging as such.

Second, people oppose nudges that are inconsistent with the interests or values of most choosers. The most direct evidence is the finding that while most people support automatic name change for women, they reject automatic name change for men. The evident reason is that the former generally tracks people’s interests and values, while the latter countermands them. Note that we could easily imagine a population that would reverse these results. Suppose that one believes that automatically assuming that wives take their husbands’ last names undermines sex equality, and automatically assuming that husbands take their wives’ last names promotes sex equality. For those who have these beliefs, and are committed to sex equality, reversing the majority’s views might seem attractive. And indeed, many respondents, though far less than a majority, did support a default rule to the effect that men would take their wives’ last name.

Any default rule, of course, is likely to harm at least some people. For any such rule, some people will want to opt out, for good reason, and some of those people will not do so, perhaps because of inertia and procrastination. This point is a potential objection to default rules in general. By itself, however, that fact is not enough to produce public opprobrium. Recall that majorities approve of automatic voter registration and automatic enrollment in pension plans and green energy , apparently because most respondents think that those nudges are in most people’s interests. Recall too that most respondents support public education campaigns designed to combat obesity and discrimination on the basis of sexual orientation. By contrast, most people oppose public education campaigns to encourage women to stay at home and to inform people that they can change their gender, apparently on the ground that those campaigns are inconsistent with what people regard as prevailing interests and values.

To be sure, there is an ambiguity in these findings. Do respondents reject nudges that are (a) inconsistent with their own interests or values or (b) inconsistent with the interests or values of most choosers? On this question, the findings here do not provide a clear test. When respondents reject nudges, they probably believe that the nudges that are inconsistent with their own interests or values are also inconsistent with the interests or values of most choosers. It would be interesting to pose questions that would enable us to choose between (a) and (b).

When people are deciding whether to favor default rules , the size of the group of disadvantaged people undoubtedly matters. If a default rule harms a majority, it is unlikely to have much appeal. If the disadvantaged group is large, but not a majority, people might reject a default rule and favor active choosing instead.

No Losses by Default!

Most respondents appear to accept a third principle: Before certain losses can occur, people must affirmatively express their wishes. The principle, whose boundaries remain to be tested, forbids the state from taking certain goods by default.

Whether this principle is triggered must depend, of course, on a theory of entitlement, from which any account of “losses” will flow. To give an easy case, people do not question the idea that thieves must give back what they have stolen, nor do most people object to some kind of tax system, probably on the ground that people are not entitled to all of their pre-tax income. We could imagine harder cases—as, for example, with adjustments in how to calculate benefits under a social security program, where losses and gains might not be self-evident and might be subject to framing effects. It is relevant that most respondents favor a state requirement that when obtaining their driver’s licenses, people must indicate whether they want to be organ donors (and thus favor active choosing ), even though most Americans reject a default rule in favor of being an organ donor.

In the questions here, the question of entitlement is not especially complicated. If a default rule means that people will end up giving money to specified charities (subject to opt out), they will lose something that they own. And indeed, strong majorities of people reject automatic charitable donations of diverse kinds. A likely concern is that as a result of inertia, procrastination, or inattention, people might find themselves giving money to a charity even though they do not wish to do so. We might therefore complement the third principle with a fourth and narrower one, which can be seen as a specification: Most people reject automatic enrollment in charitable giving programs, at least if they are operated by public institutions.

The case of carbon offsets can be understood in similar terms. While it does not involve a charitable donation, and instead might be seen as an effort to prevent a harmful act (pollution), most people insist on active consent. We do not yet know the exact boundaries of apparent public skepticism about default rules that would give away people’s money without their active consent, but there is no doubt that such skepticism exists.

Political Disclosures

We have seen that people generally favor System 2 disclosures that, in their view, bear on health and safety (salt content, GMOs). At the same time, the results leave open the question whether and when people will favor mandatory disclosures that involve issues of value, harm, or morality associated with the production of a product rather than the health and environmental effects of a product itself. Americans seem closely divided on that question. With repeated violations of the nation’s labor laws, and nations that harbor terrorism, such disclosure achieved majority support—but products coming from Communist nations did not. People might well demand a certain threshold of egregiousness, in terms of the behavior of those who produce a good or service, before they will want to require disclosure of that behavior. On this question, partisan differences are to be expected, because people will disagree about whether the relevant threshold has been met, and about what it exactly is.

Manipulation

It is tempting, and not inconsistent with the data, to suggest that people’s reactions to nudges also show the influence of a fifth principle: People reject nudges that they regard as unacceptably manipulative. The subliminal advertising finding can be taken as support for this principle. But what counts as unacceptable manipulation?

Most people are in favor of graphic warning labels on cigarettes. They like default rules (if consistent with people’s values and interests). A majority favors mandatory cafeteria design to promote healthy eating. People approve of a graphic campaign to discourage distracted driving. With respect to obesity, a majority favors a somewhat tendentious public education campaign, one that could plausibly be characterized as manipulative. No one likes manipulation in the abstract, but aside from cases that cross the line into deception, there do not appear to be many cases in which people are willing to reject nudges as unacceptably manipulative, at least if they have legitimate ends and are taken to be in the interest of most choosers. 7 To be sure, more work would be valuable on this question. Undoubtedly, we could devise a variety of examples that would trigger the fifth principle.

Politics and Partisanship

What is the role of partisan differences? Democrats and Republicans will sometimes disagree, of course, about whether the goals of a particular nudge are illicit, and they will also disagree on occasion about whether a nudge is consistent with the interests or values of choosers. To know whether a goal is illicit, and to assess interests and values, it may be necessary to take substantive positions on contested questions, and for that reason, political disagreements can lead to sharp differences.

For example, those who disapprove of abortion will be especially likely to support nudges that are designed to discourage abortion; those who do not disapprove of abortion will be unlikely to support such nudges. Imagine an antiabortion nudge in the form of a law requiring pregnant women seeking abortions to be presented with a fetal heartbeat or a sonogram. We can predict, with a high degree of confidence, that Democrats would show lower approval ratings than Republicans. My own study, on Amazon’s Mechanical Turk, finds exactly that. The precise question asked people whether they approve or disapprove of a “state requirement that pregnant women must see a sonogram of their fetus, and hear its heartbeat, before proceeding to have an abortion.” About 28% of Democrats approved, but 70% of Republicans did so. (Interestingly, only about one-third of Independents approved, essentially the same as Democrats.) With respect to a public education campaign informing people that they can change genders, the significant difference between Democrats and Republicans should not come as a big surprise.

But there are other divisions as well. Even when majorities of Democrats, Republicans, and Independents support a particular initiative, the level of support is sometimes higher within one group than within another. Consider, for example, a state requirement that grocery stores place healthy foods in conspicuous places; support is higher among Democrats and Independents than Republicans. (It would of course be easy to design nudges that would show an opposite pattern, as with nudges that influence people in directions that are most favored by Republicans.) And even if the underlying end is broadly shared—as it is, for example, in the area of public health—at least some people are skeptical of government nudges, taken as such, and will therefore disapprove of them even if they do accept the legitimacy of the end and do not think that the nudge is inconsistent with choosers’ interests or values. 8

Some Republicans, and undoubtedly some Democrats and Independents, appear to support another principle: There should be a rebuttable presumption against nudging , at least if the government can avoid it. The survey does not provide conclusive evidence that a subgroup embraces this principle, but it is highly suggestive. Many people reject graphic health warnings on cigarette packages (26%), an educational campaign for childhood obesity (18%), an educational campaign for distracted driving (15%), and a traffic lights system for food (36%). It is reasonable to infer that those who oppose such nudges agree that they have legitimate ends and are in the interest of most choosers, but nonetheless do not favor government intervention. Other evidence supports that inference.

It is important to see that the strength of any anti-nudge presumption will vary with the particular issue, with partisan affiliations, and with competing views about the role of government. In some of the cases, Republicans are more skeptical of particular nudges than are Democrats. With calorie labels and childhood obesity campaigns, for example, there are significant differences in the levels of support within the two groups, even though majorities of both are supportive. But Republicans are hardly skeptical of nudges as such. I have observed that in some cases, Republicans are undoubtedly more enthusiastic about particular nudges than are Democrats, as in the case of the antiabortion nudge. The fact that few such cases are found here is an artifact of the specific questions. If the issue involved automatic enrollment in programs by which high-income earners automatically receive capital gains tax benefits, we can predict, with great confidence, that Republicans would be more supportive than Democrats. There is good evidence to this effect. 9

Nudges vs. Mandates

We have said that many people are skeptical of mandates and bans, even if they have legitimate ends. In other words, people are inclined to reject coercion as such, and on this count, mandates and bans face objections that do not apply to nudges. Because people care about agency and want to maintain it, they will be skeptical of initiatives that eliminate it, even as they acknowledge that in some domains, coercion is amply justified.

To test that proposition, I used Amazon’s Mechanical Turk (with 309 participants) to explore people’s reactions to three pairs of initiatives with exactly the same goals, taking the form of either nudges or mandates. The initiatives involved savings (with a 3% contribution rate); safe sex education; and education about intelligent design. In each of those domains, people were asked to evaluate a nudge (in the form of a default rule , with opt-out for employees and parents) and also a mandate. If the end is what matters, of course, we would expect to see similar levels of approval and disapproval for both. But if agency matters as such, nudges would receive far higher levels of approval than mandates.

That is exactly what I found. In all cases, the nudge was far more popular than the mandate (and received majority support), and indeed, in all cases, the mandate ran into majority disapproval. So long as people could opt out, the savings initiative received 69% approval; same-sex education, 77% approval; and intelligent design, 56% approval. But as mandates, approval levels for all three fell dramatically to 19%, 43%, and 24%, respectively. The difference between approval rates ranged from a whopping 50% for savings to a still-whopping 32% for intelligent design. That is strong evidence that agency greatly matters.

It follows that many people do oppose mandates as such, even when they are enthusiastic about the underlying ends and are supportive of nudges that are designed to promote those ends. We have seen that majorities of Americans have no general view about nudges; their assessments turn on the principles identified here. With mandates, people do have a general view, and it is unfavorable. Of course, it is also true that people support mandates of various kinds, especially when harm to others is involved (as in the case of the criminal law and many regulatory requirements, such as restrictions on air pollution and dangerous foods). My goal here is not to map the boundaries of approval for mandates, but simply to show that people distinguish between choice-preserving and choice-restricting approaches, valuing as they do agency as such.

Europe

With Lucia Reisch of Copenhagen Business School, I conducted similar surveys in six diverse nations in Europe: Denmark, France, Germany, Hungary, Italy, and the UK. 10 The countries were chosen to represent different cultural and geographic regions, as well as different socioeconomic regimes and political traditions: a Nordic welfare state (Denmark); a market economy with a deep, historically grounded distrust of paternalism (Germany); a Central European post-socialist country (Hungary); two Southern European countries with different political regimes, problems, strengths, and experience with nudging (France and Italy); and the UK, a country that has spearheaded nudging as a policy tool worldwide since 2010 and hence has had several years of debate on the pros and cons of nudging.

To adjust for the European setting and also to attain a representative sample in six countries, the number of items was reduced to 15. We picked thirteen from the US survey and added two additional interventions that had been recently discussed in European politics: (1) requiring supermarket chains to keep cashiers free of sweets and (2) requiring canteens in public institutions to have one meat-free day per week (acknowledging that this requirement goes beyond a nudge ).

Central Findings

To our surprise, the major findings are strikingly close to what emerges from the US data. The underlying principles seem to be the same, and in general, there is broad support, throughout the six nations, for 12 of the 15 nudges that we tested—and broad opposition, throughout those nations, to the remaining three nudges . In that respect, we find a substantial consensus among disparate nations. Here again, the simplest lesson is that if people believe that a nudge has legitimate goals and think that it fits with the interests or values of most people, they are overwhelmingly likely to favor it. As in the context of the USA, this lesson applies to both System 1 and System 2 nudges.

Two of the three rejected nudges run afoul of a principle on which there is apparently a European as well as American consensus: The government should not take people’s money without their affirmative consent, even for a good cause. With respect to both charitable donations and carbon offsets, a default rule is unacceptable because it offends that principle. Like Americans, Europeans also reject a nudge that is unambiguously manipulative: a subliminal advertising campaign in movie theaters, designed to convince people not to smoke and overeat. The results are given in Table 2.7.

Table 2.7 Approval rates for 15 nudges in six European countriesa
Full size table

It will be readily apparent that of the six nations, Italy and the UK are most favorably disposed toward the tested nudges. In Italy, only one nudge (N 14: sweets-free cashier zones in supermarkets) is less popular than in most of the other countries. Similarly, the UK is in the top ranks of approval eleven out of fifteen times. (France and Germany cannot be so clearly ranked.) It is worthwhile to note that Italy is hardly known to have a tradition or recent history of antipathy to paternalistic interventions, and it is possible that the experience of the UK, involving many uses of behavioral science, has influenced public opinion.

Differences—and (the Bigger Story) Continuities

The basic story here is that Europeans often agree with one another and with Americans, but if we dig deeper, we will find some noteworthy differences where we might not expect to find them—and even more surprising similarities. Perhaps most important: Both Hungary and Denmark are consistently less favorably disposed toward nudges in general. It is not as if educative or noneducative nudges are especially troublesome for citizens of both nations. Instead, a wide variety of nudges attract something like 10% lower approval rates. Why is that? The answer probably has something to do with judgments about public officials—and perhaps human agency as well.

The case of Hungary is not especially puzzling. In that nation, there is widespread distrust of social institutions, which has been below the Organisation for Economic Co-operation and Development (OECD) average for a long time. The legacy of Communism may lead Hungarians to question or fear government in general. At the same time, Hungary is the country (from our subset) with the highest corruption index. Moreover, it is below OECD level in voting in national elections. The Hungarian findings also cast a broader light on differences, within nations, with respect to nudges: Citizens who distrust their government, or government in general, will be less likely to approve of nudges, even if they approve of the particular ends that those nudges would promote. Some of the American data can be explained in this way. While Republicans, as such, do not oppose health and safety nudges, a small subset of people, with libertarian leanings, seem to do so. 11 Hungarians are apparently like that.

With respect to Denmark, the findings are more difficult to explain. That nation is not exactly known for its distrust of government or for its firm opposition to anything that smacks of paternalism . Denmark has traditionally maintained one of the highest levels of trust in government among all OECD countries. But while trust in politicians on the communal and regional levels has remained high, there was a decline in trust in national politicians and government in the period before the survey. Some controversial health-related interventions in Denmark (including a tax on foods with high levels of saturated fats) might have contributed to the findings.

Notably, however, the survey did not produce clear differences across party lines within Europe. One of the main findings, and among the most surprising, is that party affiliations are not correlated in any systematic way with support for the tested nudges. Within countries, however, there are some weak correlations and several overall patterns. (1) In France, Green Party and left-wing supporters are more favorably disposed toward the relevant nudges . (2) In the UK, people who have voted for populist parties are particularly skeptical toward information nudges. (3) Within all countries, European liberals are somewhat less inclined to favor health nudges. (4) Within all countries, Green Party voters are somewhat more inclined to favor environmental nudges (not surprisingly). But these findings should be taken with considerable caution in light of our rough measurement of political preferences (most recent vote) and the clustering of political parties in Europe.

With respect to demographic differences, only one characteristic seems to be correlated with people’s attitudes toward the tested nudges: gender. Women favor such nudges more than men do, with a less pronounced (but still significant) gender divide in France and Denmark. With respect to other demographic characteristics, however, there are no relevant correlations. Here again, the results suggest that it is the aim that the government wants to achieve with the nudge that determines approval, and that as the cases of Denmark and Hungary show, country differences can matter.

Beyond the USA and Europe

To broaden the picture, Reisch and I have also collected data in a diverse collection of nations: Canada, Australia, Brazil, South Africa, Japan, South Korea, China, and Russia. Not surprisingly, Canada and Australia look very much like the USA, Germany, the UK, Italy, and France, with broadly similar patterns. (A puzzling exception is that 52% of Australians support subliminal advertising.) Brazil is also broadly similar, with two important exceptions: 66% of Brazilians favor default donations to the Red Cross, and 66% actually favor subliminal advertising. South Africa is close to Brazil, with these two anomalies as well (50% support for the Red Cross donations, 61% support for subliminal advertising).

In general, Russia is similar to South Africa and Brazil, but without those two anomalies and with significantly lower levels of support for sweet-free cashier zones and one meatless day per week (55% and 49%, respectively). By contrast, Japan looks a lot like Hungary and Denmark, with consistently lower levels of approval, including majority opposition to a distracted driving campaign (but otherwise with the same basic patterns).

China shows spectacularly high levels of support for all of the tested nudges (over 90% for 10 of the 15), with only one noteworthy exception, which is just 58% support for active choosing with respect to organ donation. It is tempting to think that in China, a “yes” answer is usually thought to be appropriate to questions asking about support for government action, and so respondents answer “yes,” even on an anonymous survey—which suggests that, in that nation, our survey might not be capturing people’s actual opinions. But South Korea, a generally free and democratic nation, also shows spectacularly high levels of support for the tested nudges, comparable to those in China. In fact, strong majorities of South Koreans favor all fifteen nudges (including subliminal advertising and default donations to the Red Cross), with support levels frequently exceeding 80% or even 90% (75% in the case of subliminal advertisements and 62% in the case of default donations to the Red Cross, compared to 90% and 83% for China, respectively).

A great deal could be said about the intriguing national differences, but that would take me far beyond the subject here. Despite those differences, the more important story is one of continuity. With the stated qualifications, there appears to be something like a consensus, across democratic nations, about which nudges deserve support and (to a somewhat lesser extent) which do not—and the most important distinction between the two depends on identifiable principles, involving the legitimacy of the underlying end, and consistency with the interests and values of choosers.

Notes

  1. 1.

    See R.J. Reynolds Tobacco Co. v. U.S. Food & Drug Admin., 823 F. Supp. 2d 36 (D.D.C. 2011), aff’d on other grounds, 696 F.3d 1205 (D.C. Cir. 2012).

  2. 2.

    See Anne Thorndike et al., Traffic-Light Labels and Choice Architecture , 46 American Journal of Preventive Medicine 143 (2014) (Thorndike 2014).

  3. 3.

    See Craig v. Boren, 429 U.S. 190, 200–204 (1978).

  4. 4.

    In principle, the problem would be most interesting in an area in which the default rule tracked reality. If most people are, in fact, Democrats, is it clearly objectionable if a city or state assumes that its residents are Democrats for purposes of registration? The answer is almost certainly yes; political affiliations should be actively chosen, not assumed by government. This principle almost certainly has constitutional foundations (though it has not been tested): If a voting district consisted of 80% Democratic voters, it would not be acceptable to assume that all voters intend to register as Democrats. But I am aware that this brief comment does not give anything like an adequate answer to some complex questions about the use of “mass” default rules that track majority preferences and values. For a more thorough discussion, see Cass R. Sunstein, Choosing Not To Choose, 77 (2015a) (Sunstein 2015a).

  5. 5.

    Here as well we could imagine interesting questions if the default rule tracked reality. If most people in a city or state are Christians, is it so clearly illegitimate to presume, for purposes of the census, that most of the people in that city or state are Christians, subject to opt out? But with respect to religion, as with respect to politics, there is a strong social and constitutional norm in favor of official neutrality, which would be violated even if a particular default reflected majority preferences and values.

  6. 6.

    The striking findings of “partisan nudge bias” are fully consistent with this claim. David Tannenbaum, Craig Fox & Todd Rogers, On the Misplaced Politics of Behavioral Policy Interventions (2014), https://pdfs.semanticscholar.org/0418/4cd62d265d49b300b60528fb0e36692964a8.pdf (Tannenbaum et al. 2014).

  7. 7.

    On the idea of manipulation, see generally Cass R. Sunstein, The Ethics of Influence (2015b) (Sunstein 2015b).

  8. 8.

    Tannenbaum, Fox & Rogers, supra note.

  9. 9.

    Id.

  10. 10.

    Lucia Reisch & Cass R. Sunstein, Do Europeans Like Nudges ? Judgment & Decision Making 310 (2016). Interested readers might consult that essay for a detailed account; I offer only a few highlights here (Reisch and Sunstein 2016).

  11. 11.

    See Janice Jung & Barbara A. Mellers, American Attitudes Toward Nudges, 11 Judgment and Decision Making 62–74 (2016).