NUDGE UNITS AND DATA-DRIVEN NUDGING

Thaler and Sunstein, 2008

The past decade has witnessed a global trend of using nudging to intractable policy problems. To date, nudging has been applied in a range of different spheres such as public health, taxation and environmental protection; it is considered as particularly useful when conventional policy tools based on hard economic incentives and strict legal requirements have failed. Since establishment of the UK Behavioural Insights Team (BIT) in 2010, many nations including developed, emerging and developing ones have founded nudge units to support their policies and programmes. International institutions such as the World Bank, OECD and EU have also followed suit. According to OECD data, the number of nudge units around the world, both inside and outside governments, and multinational ones, has reached over 200 and is continuing to grow.

At the same time, policy makers are becoming increasingly interested in using digital tools to influence behaviour. Based on a large number of datasets, data-driven automated systems have been able to establish correlations between multiple variables of people’s behaviour, thus providing policy makers with better understanding and even predictions about citizens’ behaviour. These results constitute an increasingly important development in combining nudging with digital technology and data science. For example, the BIT added a Data Science team in 2017 to integrate the latest methods from data science and machine learning in policy making. Data-driven nudging, which is also known as hypernudging, develops more accurate predictive models that can identify citizens’ most common biases and behavioural inclinations, thereby systematically hypernudging them to choose more wisely (Yeung, 2017). With its networked, real-time updated and pervasive nature, data-driven nudging is very powerful in fostering behavioural change and is reshaping public policy.

There are already some case studies about the implementation of data-driven nudging in public sectors, such as Boston’s Safest Driver Competition and experiments with sensors in streets to increase safety in Eindhoven (Netherlands). Here are the details of just one example. In Durham, North Carolina, the use of data-driven nudging was tested in 2018 which encouraged citizens to choose public transportation over private vehicles. The city, partnering with several large employers, provided personalised route maps with bike, bus and walking routes from individuals’ homes to work through email, together with information about potential benefits to be derived by opting for alternatives to driving such as trip times, savings in gas costs and gains in physical activity. Messages drawing on and fostering social norms were also included in the email such as “Driving downtown is so 2017”. Six months later, employees participating in the data-driven nudging pilot project were 12% more likely to use alternative methods of transit than those who were not nudged.

However, despite potential benefits to be derived in this way, data-driven nudging has not been widely used in public policy making due to two major constraints. Firstly, the lack of IT infrastructure hinders the application of data-driven nudging in developing countries, such as South Africa. Secondly, the introduction of data-driven nudging in the area of public policy raises both legitimacy and ethical concerns.

Most criticisms suggest that nudging is manipulative and infringes on citizens’ autonomy. Furthermore, there is often a lack of transparency in nudging as it tends to be more effective when people are unaware of being nudged. In this case, people are not given a fair opportunity to resist, ignore or reject nudges, but are nevertheless being subjected to unwitting persuasion. According to the criticism, nudging organisations employ various strategies to steer people in specific ways and hence individuals’ resulting actions are not genuinely their own, but reflect the choice architecture (Hausman and Welch, 2010). Since data-driven nudging is able to influence cognitive and behavioural patterns with greater accuracy, individuals are thus left with even less room to retain their autonomy (Viljanen, 2017). Data-driven nudging may further draw on “dark patterns” which promote a certain behaviour with design practices and thereby disadvantage users (Rieger and Sidners, 2020).

Legitimacy concerns about data-driven nudging centre much on privacy. With digital technology and data science, personal data are continually collected and algorithmically processed. Digital technology has now been used in a wide variety of contexts, leaving less room for individuals’ choices. Moreover, data collection involves not only users themselves, but also non-users (e.g., users authorising an app to use the address book with information about others) leading to worries about privacy interdependence (Pu and Grossklags, 2016). Data-driven nudging thus raises questions as to whether the systematic data collection, processing and analysis of personal data have been authorised, thereby infringing individuals’ right to informational privacy. For instance, according to a study from the London Underground about tracking users throughout their journey using WiFi, there were even plans to generalise the tracking and sell data to third parties (Privacy International, 2017).

Finally, data-driven nudging may trigger a power shift. Surveillance capitalism produces a new form of power (Zuboff, 2015), with big data leading to a power shift between individuals and data controllers (Ulbricht and Grafenstein, 2016). Data-driven nudging for public policy usually requires cooperation between the public sector and private companies. For instance, government departments collaborate increasingly with telecommunications providers. As a result, the “corporate-state nexus” formulated in data-driven nudging will further shift power away from individuals.

We have collected data about nudge units in eight countries and found policy makers have already grasped the great potential of data-driven nudging. In the next step, our empirical research will provide a detailed analysis about the integration of big data technology in the design of nudge projects in the real world.

The blogs published by the bidt represent the views of the authors; they do not reflect the position of the Institute as a whole.

• Hausman, D. M., & Welch, B. (2010). Debate: To nudge or not to nudge. Journal of Political Philosophy, 18(1), 123-136.
• Privacy International. (2017). Smart cities: utopian vision, dystopian reality, last accessed on May 26, 2021.
• Pu, Y. & Grossklags, J. (2016). Towards a model on the factors influencing social app users’ valuation of interdependent privacy. Proceedings on Privacy Enhancing Technologies, 2016(2), 61-81.
• Rieger, S. & Sinders, C. (2020). Dark Patterns: Design mit gesellschaftlichen Nebenwirkungen. Policy Brief from Stiftung Neue Verantwortung, last visited on May 26, 2021.
• Thaler, R. H. & Sunstein, C. R. (2008). Nudging: Improving decisions about health, wealth, and happiness. New Haven, CT: Yale University Press.
• Ulbricht, L. & von Grafenstein, M. (2016). Big data: big power shifts? Internet Policy Review, 5(1).
• Viljanen, M. (2017). A cyborg turn in law?. German Law Journal, 18(5), 1277-1308.
• Zuboff, S. (2015). Big other: surveillance capitalism and the prospects of an information civilization. JInf Technol, 30, 75–89.