What Do People Like When They Say They Like Nudges?

In his article, Do People Like Nudges?, Cass Sunstein argues that most people do not have a general view, either positive or negative, about nudging. Their attitude towards particular nudges depends on whether they approve of the ends towards which the nudges are directed.

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, because it suggests that most people do not share a concern that nudges, as such, should be taken as manipulative or as an objectionable interference with autonomy.

This is hardly a contentious finding. After all, most people will spend very little time thinking about nudging at all, so it's not surprising they do not share a concern that nudges might undermine rational autonomy.

Nevertheless, it is true that when people are asked, their responses demonstrate broad support for nudges that are perceived to be directed towards legitimate ends and to be in line with the interests and values of the majority. This is borne out by the data we've collected from To Nudge Or Not To Nudge, our on-going, interactive experiment.

For example, we've found broad support for compulsory calorie labels (73%); mandatory warnings on cigarette packaging (75%); and the use of framing and loss aversion to increase charitable donations to a range of charities (with approval scores from 55% to 66%).

Our data also confirms Sunstein's contention that whether or not people will approve of a particular nudge tends to be determined by whether they approve of the thing it aims to bring about. This is best illustrated by a couple of questions we asked about abortion. One question asked whether people would approve or disapprove of abortion clinics being legally required to provide adoption counselling prior to going ahead with a termination. It also indicated there would be no opt-out provision for clients. Another question asked people whether they agreed or disagreed with the statement, Abortion should always or normally be legal.

If Sunstein is right, then the responses to these questions will be inversely correlated; that is, the more strongly a person considers that abortion should be legal, the more opposed they will be to compulsory adoption counselling. And this is precisely what the data shows (N=350):

nudging abortion

We also found a similar pattern when it came to a nudge that favoured an environmental charity. If people had previously indicated that they doubted the occurrence of anthropogenic climate change, they tended to disapprove strongly of the charity using a framing technique to push prospective donors towards higher donations.

None of this is particularly surprising. People were never going to be wildly enthusiastic about nudges that aim at what they take to be a bad outcome. This doesn't add up to an indictment of nudging in general, of course, which means given a different issue and outcome, you might find broad support for nudging.

However, this does lead to the rather troubling thought that when people respond to these sorts of questions, they might not be thinking much about nudging at all, but rather about the outcome of the nudging process, and whether it lines up with their interests and values. The fact people are willing to endorse a nudge does not entail that they have given the mechanism of the nudge any thought at all; or, if they have, that it has any particular salience in the moral calculus that results in the endorsement.

An interesting possibility here is that people are being nudged not to think about nudging in and of itself when they are asked questions about the ethics of nudging. If questions are always couched in terms of outcomes, and if these outcomes are normally salient to the respondents, then this might have the effect of attenuating the process of nudging in the assessment people make of the moral legitimacy of any particular nudge. If something like this does occur, while it would remain true that there is broad support for nudging (in the circumstances detailed by Sunstein), this support would be predicated upon a focus on outcomes, and a desire to bring about outcomes that are favored.

This point, in and of itself, does not run contrary to the Sunstein position. However, it does raise a possibility that would undermine the claim that there is broad support for nudging; namely, that people would respond differently to questions about the ethics of nudging, if these were couched in such a way that put the mechansim of the nudge, rather than its outcome, at front and center. If we're asked about subliminal advertising, we tend to reject it as legitimate, even if we approve of the outcome, presumably because we don't like the idea that we're being manipulated via an unconscious process. It would be interesting to see how people would respond to questions about framing and loss aversion, for example, if the unconscious aspects of the efficacy of these techniques were laid bare in questions about their moral legitimacy.