When a Nudge is not Enough: Complex Challenges and Norm Interventions

Recently, there has been a fierce debate over whether ‘nudges’, or “freedom-preserving” light-touch behavioral interventions, are effective at changing behavior. First popularized in Sunstein and Thaler’s 2008 book Nudge, the approach consists of subtle changes to the set of choices individuals have in order to alter their behavior in predictable ways. It would not be an exaggeration to say that Nudge took the world by storm upon publication, inviting the attention of big consulting firms like McKinsey and BCG, spurring the creation of behavioral science consulting firms like Ideas42, and being instrumental in the formation of government “Nudge Units” in the UK, US, Australia, Canada, Singapore, and others. Indeed, the publisher’s page for the book proudly proclaims that over 2 million copies of the book have been sold, and makes clear both authors’ laurels since the original publication, including a Nobel Memorial Prize in Economic Sciences for Richard Thaler and the Holberg Prize for Cass Sunstein. Thaler even had a cameo in the 2015 film The Big Short, and Sunstein was tapped to head the Obama Administration’s Nudge unit, followed by being appointed chair of the World Health Organization’s Technical Advisory Committee in 2020 (he has since been replaced by Professor Susan Michie from University College London). Nudges, it seemed, were not only here to say, but growing in their relevance and importance to major institutions across the world.

Perhaps for this reason, it was unsurprising to many when a recent meta-analysis in the journal PNAS ^[1]Mertens S, Herberz M, Hahnel UJJ, Brosch T. The effectiveness of nudging: A meta-analysis of choice architecture interventions across behavioral domains. Proc Natl Acad Sci. 2022;119(1):e2107346118. … Continue reading suggested that nudges have an average effect size of d = 0.43, which is fairly large in the social science literature. For some context, that’s about 12% larger than the effect of text message interventions to increase physical activity, ^[2]Smith DM, Duque L, Huffman JC, Healy BC, Celano CM. Text Message Interventions for Physical Activity: A Systematic Review and Meta-Analysis. Am J Prev Med. 2020;58(1):142-151. … Continue reading and 37% larger than the effect of mindfulness-based interventions to reduce depression in adolescents. ^[3]Dunning DL, Griffiths K, Kuyken W, et al. Research Review: The effects of mindfulness-based interventions on cognition and mental health in children and adolescents – a meta-analysis of randomized … Continue reading Indeed, some researchers found this effect “implausibly large”, and re-analyzed the data using several methods to correct bias. Their results? After correcting for publication bias in the literature (the tendency for an overrepresentation of studies with statistically significant results), the overall effect of nudges drops down to somewhere between d = –0.01 and d=0.08, an essentially zero overall effect. ^[4]Maier M, Bartoš F, Stanley TD, Shanks DR, Harris AJL, Wagenmakers EJ. No evidence for nudging after adjusting for publication bias. Proc Natl Acad Sci. 2022;119(31):e2200300119. … Continue reading ^[5]Szaszi B, Higney A, Charlton A, et al. No reason to expect large and consistent effects of nudge interventions. Proc Natl Acad Sci. 2022;119(31):e2200732119. doi:10.1073/pnas.2200732119 Does this mean that nudges have no effect at all on behavior change?

Indeed, some researchers found this effect “implausibly large”, and re-analyzed the data using several methods to correct bias. Their results? After correcting for publication bias in the literature (the tendency for an overrepresentation of studies with statistically significant results), the overall effect of nudges drops down to somewhere between d = -0.01 and d=0.08, an essentially zero overall effect.

No. As pointed out by different behavioral scientists, ^[6]Hallsworth M. Making Sense of the “Do Nudges Work?” Debate. Behavioral Scientist. Published August 2, 2022. Accessed September 28, 2022. … Continue reading ^[7]Fox CR, Doctor JN, Goldstein NJ, Meeker D, Persell SD, Linder JA. Details matter: predicting when nudging clinicians will succeed or fail. BMJ. 2020;370:m3256. doi:10.1136/bmj.m3256 nudges can and do work for some behaviors, and under certain conditions, but don’t expect large-scale effects in terms of  individual behavior change. Using data from two US-based Nudge Units which use nudges to improve take-up of various government services – BIT North America and the Office of Evaluation Sciences – a study in the journal Econometrica found an overall effect of nudges amounting to a 1.4 percentage point change, on average. ^[8]DellaVigna S, Linos E. RCTs to Scale: Comprehensive Evidence From Two Nudge Units. Econometrica. 2022;90(1):81-116. doi:10.3982/ECTA18709 This is a pretty small effect in terms of a percentage point change, but can really matter on the scale of millions, within which Nudge Units often operate. For instance, the Nudge Unit programs assessed in the study cover over 23 million people, and a 1.4 percentage point change equates to over 322,000 people taking up government services who may not have done so otherwise. Nudges have their place, but expectations about their effectiveness and the range of real-world challenges for which they are useful should be tempered.

Some have argued that binary “effect or no effect” thinking is the wrong way to think about nudges, and what really matters is identifying the conditions under which they do work. ^[9]Hallsworth M. Making Sense of the “Do Nudges Work?” Debate. Behavioral Scientist. Published August 2, 2022. Accessed September 28, 2022. … Continue reading We agree with that, and think that the principle applies to any behavioral intervention. Indeed, behavioral science is said to be undergoing a heterogeneity revolution, in which there is increasing recognition that most behavioral intervention effects are extremely varied, and that we should be learning from the variation. ^[10]Bryan CJ, Tipton E, Yeager DS. Behavioural science is unlikely to change the world without a heterogeneity revolution. Nat Hum Behav. 2021;5(8):980-989. doi:10.1038/s41562-021-01143-3 However, one aspect of the conversation about nudge effectiveness that we have not seen discussed as much in public fora is that nudges are far from the only approach to behavior change. Rather, nudges are just the tip of the iceberg of intervention approaches that social and behavioral sciences have to offer. They’re useful, cost-effective, and very popular, to be sure, but also limited, fallible, and not appropriate for all contexts. ^[11]Ewert B. Moving beyond the obsession with nudging individual behaviour: Towards a broader understanding of Behavioural Public Policy. Public Policy Adm. 2020;35(3):337-360. … Continue reading

However, one aspect of the conversation about nudge effectiveness that we have not seen discussed as much in public fora is that nudges are far from the only approach to behavior change. Rather, nudges are just the tip of the iceberg of intervention approaches that social and behavioral sciences have to offer. They’re useful, cost-effective, and very popular, to be sure, but also limited, fallible, and not appropriate for all contexts.

To those unfamiliar with the social and behavioral sciences, nudges may have seemed to come from nowhere, and imbued with a near-magical quality and potential to transform policy. You’re telling me that just rewording a letter slightly can increase tax compliance by 15%? Amazing! If you felt or still feel that way about nudges, we get it. When used appropriately, they can deliver serious results for little to no cost. But we should always keep in mind that they are not a panacea for all behaviors in all contexts. Rather, they should be understood as one of the many tools that can be used to achieve policy impact…for certain behaviors and under certain conditions. This may seem obvious to those with training in the social and behavioral sciences, but the prominence of nudge interventions has overshadowed other areas in the field. As behavioral scientists David Gal and Derek Rucker have pointed out, the popularity of nudge interventions can lead to an “impoverished view of the relevance of behavioural science to the real world.” ^[12]Gal D, Rucker DD. Experimental validation bias limits the scope and ambition of applied behavioural science. Nat Rev Psychol. 2022;1(1):5-6. doi:10.1038/s44159-021-00002-2 In short, there is more to behavior change than nudging; we shouldn’t limit ourselves when approaching behavior change problems.

In short, there is more to behavior change than nudging; we shouldn’t limit ourselves when approaching behavior change problems.

Ok, there’s more to behavioral science and behavior change than nudging. But, what? We illustrate some of these considerations by looking at interventions to change social norms (that’s sort of our specialty!).

The term “social norms” means many things to different people. Under a commonly used approach developed by behavioral theorist Cristina Bicchieri, ^[13]Bicchieri C. Norms in the Wild: How to Diagnose, Measure, and Change Social Norms. Oxford University Press; 2017 a behavior is said to be driven by a “social norm” when it is conditional on your expectations about how commonly that behavior is performed, and how it is approved, among a group of people whose opinions matter to you. If your behavior isn’t driven by these considerations, then it’s probably not due to a social norm. If I drink excessive amounts of alcohol because my friends do it and reward it, I’m acting out of a social norm. However, if my drinking is typically done on my own without considering what my friends do, then it’s not due to a social norm (and is more likely driven by alcohol dependence).

One of the most common social norms interventions – the “Social Norms Approach” ^[14]Berkowitz AD. An overview of the social norms approach. Chang Cult Coll Drink Socially Situated Health Commun Campaign. 2005;1:193-214. – is essentially a nudge that has achieved policy-relevant behavioral changes with effects that surpass those of significant economic incentives. ^[15]Sunstein CR. The Council of Psychological Advisers. Annu Rev Psychol. 2016;67(1):713-737. doi:10.1146/annurev-psych-081914-124745 It involves providing information (often using percentages and graphs) about the frequency or desirability of a behavior among relevant others to correct misperceptions (e.g. You mean my friends don’t actually binge drink that much? Binge drinking is actually not that common in my community?). The idea is that the recipient of such information would modify their behavior to become more aligned with the true state of things. For example, a study to promote towel reuse at hotels left door hangers in guest’s doors with the sentence “75% of the guests who stayed in this room (#xxx) participated in our new resource savings program by using their towels more than once. You can join your fellow guests in this program to help save the environment by reusing your towels during your stay.” ^[16]Goldstein NJ, Cialdini RB, Griskevicius V. A Room with a Viewpoint: Using Social Norms to Motivate Environmental Conservation in Hotels. J Consum Res. 2008;35(3):472-482. doi:10.1086/586910 This message achieved a towel reuse rate of about 50%.

Pretty impressive for a door hanger, right? But this doesn’t mean that the Social Norms Approach nudge always works for every scenario. In fact, a review of 66 studies using this approach to reduce alcohol misuse found little to no evidence of its effectiveness. ^[17]Foxcroft DR, Moreira MT, Santimano NMA, Smith LA. Social norms information for alcohol misuse in university and college students. Cochrane Database Syst Rev. 2015;(12). … Continue reading More worryingly, such nudges can backfire. For instance, a study using this type of Social Norms Approach nudge to encourage greater purchasing of sustainably-sourced seafood in a German supermarket found the opposite effect – it resulted in a significant drop in the purchase of sustainably-sourced seafood. ^[18]Richter I, Thøgersen J, Klöckner CA. A Social Norms Intervention Going Wrong: Boomerang Effects from Descriptive Norms Information. Sustainability. 2018;10(8):2848. doi:10.3390/su10082848 One reason might be that advertising that fewer people are doing something (e.g. 28% of shoppers are buying sustainably-sourced seafood) may communicate that it is normative to NOT do that behavior. Nudge with care, people.

While this is one consideration of applying nudges (in this case, social norms nudges), such approaches are limited in a number of ways, ^[19]Yamin P, Fei M, Lahlou S, Levy S. Using Social Norms to Change Behavior and Increase Sustainability in the Real World: a Systematic Review of the Literature. Sustainability. 2019;11(20):5847. … Continue reading and we offer three practical considerations to remedy such limitations for the development of effective and sustainable social norms change interventions.

Accurate Measurement Matters.  Norms are based on expectations around what relevant others do and expect you to do in a particular situation. Social norms interventions hinge on identifying whether the target behavior(s) in question are actually driven by social norms, or by other factors. If they are indeed driven by social norms, it becomes essential to identify the reference group (i.e. the people whose opinions exert an influence on your beliefs and behavior) with as much specificity as possible. This approach involves determining whether people care about what others do or think regarding a particular behavior, the relevant beliefs and expectations related to that behavior, and how relevant such beliefs are for different group members. Accurate measurement not only helps accurately diagnose the behavioral problem, but also guides the development of interventions to change the target behavior(s). This has a strong bearing on what sources of normative information are relevant and believable, specific behavior change techniques used, and deciding the appropriate intervention content. Nudges and interventions using generic messages and intervention techniques have low chances of success, as they rarely address the actual particularities of the population they aim to transform.

Accurate measurement not only helps accurately diagnose the behavioral problem, but also guides the development of interventions to change the target behavior(s). This has a strong bearing on what sources of normative information are relevant and believable, specific behavior change techniques used, and deciding the appropriate intervention content.

Complex Challenges Need Complex Interventions. Nudging might be the right tool to influence fairly simple discrete choices (i.e. either this OR that, but not both), but is rarely the best option when more complex challenges are involved. Nudges involve subtle alterations to individual choices. But what if the problem involves more than discrete choices? Think about what it takes to reduce gender-based violence, reduce the symptoms of depression, increase compliance with traffic laws, or even getting people to build and regularly use a toilet when they have never done so. In such cases, using only a short message or alteration to a set of choices simply won’t do it for most people. This is in part because such problems involve a series of choices, expectations, beliefs that interact with factors related to the environment, community, and culture. In such cases, interventions  that “make sense” and engage people, that are based on participatory processes and a sense of group agency instead of external impositions, ^[20]Yamin P, Lahlou S, Ortega Gonzalez S, Sáenz A. The Power of Narratives in Social Norm Interventions: A Study of the Civic Culture Interventions of Antanas Mockus in Bogotá, Colombia. Published … Continue reading that get closer to the context where behaviors actually happen, and that rely on social interaction processes, ^[21]Yamin P, Fei M, Lahlou S, Levy S. Using Social Norms to Change Behavior and Increase Sustainability in the Real World: a Systematic Review of the Literature. Sustainability. 2019;11(20):5847. … Continue reading can get closer to affecting the psychosocial and ecological factors that give rise and sustain such complex problems.

In such cases, interventions  that “make sense” and engage people, that are based on participatory processes and a sense of group agency instead of external impositions, that get closer to the context where behaviors actually happen, and that rely on social interaction processes, can get closer to affecting the psychosocial and ecological factors that give rise and sustain such complex problems.

Context Matters. Norms and behaviors are embedded in complex social reproduction processes and power relations with which nudges can rarely engage. For instance, norms that support gender-based violence often reflect deeply entrenched attitudes and beliefs about the acceptability of violence and the roles that women are allowed and expected to play in society. Changing these beliefs requires consideration of historical, cultural, and community factors. This work is challenging but effective approaches exist. ^[22]Doyle K, Levtov RG, Barker G, et al. Gender-transformative Bandebereho couples’ intervention to promote male engagement in reproductive and maternal health and violence prevention in Rwanda: … Continue reading Physical/material (such as available infrastructure and assets), psychological/individual (such as skills and emotions), and social/institutional elements (such as networks and formal and informal regulation processes) are often ignored in interventions, as are the power and gender dynamics embedded in both the current state of things and proposed change initiatives. It would be unfair to expect nudges to address this confluence of behavioral determinants, but different frameworks from across the social and behavioral sciences do address this (such as Cislaghi and Heise’s dynamic framework for social change ^[23]Cislaghi B, Heise L. Using social norms theory for health promotion in low-income countries. Health Promot Int. 2019;34(3):616-623. or Lahlou’s Installation Theory ^[24]Lahlou S. Installation Theory: The Societal Construction and Regulation of Behaviour. Cambridge University Press; 2018. ).

Physical/material (such as available infrastructure and assets), psychological/individual (such as skills and emotions), and social/institutional elements (such as networks and formal and informal regulation processes) are often ignored in interventions, as are the power and gender dynamics embedded in both the current state of things and proposed change initiatives.

While we agree that nudges are often an attractive and cost-effective approach to guide behavioral public policy, we encourage practitioners to not limit themselves to just one tool from the wider social and behavioral science toolkit. As we have illustrated with interventions to change social norms, many considerations exist with complex behavioral problems. However, effective interventions exist and can be developed by focusing on three of these: measurement, complexity, and context. Accurate measurement results in more relevant data for intervention design and data analysis. Multicomponent and tailored interventions can get closer to addressing specific psychological and ecological factors that are common with complex problems. Finally, understanding the context of behavior can provide a roadmap for understanding the factors that hinder and facilitate behavior change.

About the Authors

Shaon Lahiri is a postdoctoral researcher in the Philosophy, Politics and Economics (PPE) program, and a postdoctoral fellow in the Center for Social Norms and Behavioral Dynamics (CSNBD) at the University of Pennsylvania. He completed his PhD in Social & Behavioral Sciences in Public Health at the Milken Institute School of Public Health at The George Washington University. His research focuses on the influence of social norms on various health behaviors, and how to change social norms. shaonl [at] sas.upenn.edu 

Paulius Yamin is the outgoing Managing Director at the Penn Center for Social Norms and Behavioral Dynamics, and is a PhD graduate and Visiting Fellow at the Department of Psychological and Behavioral Science, London School of Economics