People often make decisions that do not align with their own welfare, creating opportunities for paternalistic interventions (Camerer et al. 2003, Thaler and Sunstein 2003). These interventions raise two fundamental questions: should individual freedom to choose be constrained to promote welfare? And who should determine what constitutes an improvement in welfare? These questions have shaped an extensive normative literature, both in economics and the social sciences, more broadly (Nussbaum 2001, Sen 2002, Le Grand and New 2015, Schubert 2016).
Previous research highlights the complexities surrounding paternalistic interventions and the public’s attitudes toward them. Survey-based studies suggest that people often support soft interventions, such as providing information or adjusting choice architecture, and it has been argued that people are sceptical of hard paternalistic measures that constrain freedom outright (Sunstein and Thaler 2008). Other studies point to a general distrust in the ability of governments to implement paternalistic policies effectively, particularly hard measures (Kuziemko et al. 2015). The concept of libertarian paternalism, which advocates for interventions that preserve freedom while promoting welfare through choice architecture manipulation, has gained significant traction among policymakers and business leaders (Sunstein and Thaler 2008, OECD 2017, Saito et al. 2021, DellaVigna and Linos 2022).
Our study (Bartling et al. 2023) examines paternalistic preferences in a controlled experimental setting. By observing real decisions rather than hypothetical responses, it sheds new light on how people weigh the tension between enhancing welfare and preserving freedom. This approach offers critical insights for policymakers navigating these trade-offs in designing effective and publicly acceptable interventions.
Study 1: Hard versus soft interventions
To explore people’s paternalistic preferences, we implemented a novel experimental design in two large-scale studies with 14,000 US participants. Participants acted as third-party ‘spectators’ who could make decisions that impacted the welfare of another individual, the ‘stakeholder’.
Stakeholders were individuals who faced choices with real monetary consequences. Specifically, stakeholders had to choose between two bonus options:
- A safe option offering a guaranteed $4.
- A risky option offering $10 or $0 with equal probability.
In the experiment, we focused on stakeholders who preferred the safe option.
The choice environment was sometimes presented in a non-transparent way, such that stakeholders would be mistaken about the odds of the risky option, and therefore mistakenly prefer it to the safe option. Absent an intervention, the stakeholder would make their choice between the two options in a non-transparent choice environment, and thus make a mistake and choose the less preferred option. This experimentally controlled situation created opportunities for spectators to intervene.
In study 1, spectators were randomly assigned to one of two treatments that differed in the type of intervention they could implement:
- Hard intervention: Remove the risky option from the stakeholders’ choice set, directly assigning them the safe option.
- Soft intervention: Provide stakeholders with accurate information about the odds, allowing them to make an informed decision while preserving their freedom to choose.
In both treatments, the spectator could abstain from intervening, in which case the stakeholder would end up with the less-preferred option.
The experiment also manipulated the source of the stakeholders’ mistake to explore its impact on intervention choices. In treatments with ‘internal’ mistakes, spectators were informed that stakeholders had miscalculated the odds of the risky option. In treatments with ‘external’ mistakes, spectators learned that stakeholders received incorrect information about the odds. This treatment allowed us to investigate whether the source of the stakeholder’s mistake influenced spectators’ willingness to intervene.
Key findings: Study 1
In study 1, a large majority of spectators in the soft treatment chose to intervene (86% in the internal versus 88% in the external source condition) (Figure 1). In contrast, only about one-third of spectators decided to intervene in the hard treatment (34% in the internal and 30% in the external condition). Thus, the nature of the intervention appears to be of great importance for people’s willingness to intervene, while the source of the stakeholder’s mistake did not significantly influence the spectators’ decision.
Figure 1 Spectator decision depending on source of error (study 1)
Notes: The figure shows the share of spectators that intervene by treatment. The left panel shows the share of spectators intervening in the hard (remove risky option) and soft (provide accurate information about the odds) treatments when the mistake reflects the stakeholder’s mistake in calculations (‘internal’ source). The right panel shows the share of spectators intervening in the hard and soft treatments when the mistake reflects the stakeholders receiving misleading information (‘external’ source). The black bars indicate standard errors.
Study 2: Exploring welfare judgements
Study 2 expanded on this design by introducing a welfare evaluation condition, where spectators directly determined whether stakeholders would receive the safe or risky option. This allowed us to compare spectators’ intervention decisions with their welfare evaluations, providing deeper insights into whether spectators prioritised aligning outcomes with stakeholders’ true preferences or preserving their autonomy.
Key findings: Study 2
First, results from study 2 independently replicate our findings from study 1. While 83% of spectators assigned the soft treatment intervened, only 35% of those assigned the hard treatment intervened (Figure 2). In the welfare evaluation treatment, 70% of spectators directly allocated the safe option to stakeholders, showing that the large majority of the participants respect the preferences of the stakeholder.
Figure 2 Spectator decisions by treatment (study 2)
Notes: The left bar shows the share of spectators choosing the hard treatment (remove risky option). The middle bar shows the share of spectators allocating the stakeholder’s preferred safe option in the welfare treatment. The right bar shows the share of spectators intervening in the soft treatment (provide accurate information about the odds). The black bars indicate standard errors.
Using a theoretical framework outlined in the paper, we estimate that about one-third of the spectators act as libertarian paternalists, who respect the preferences of the stakeholder but give strict priority to the freedom to choose, while a bit more than half of the spectators act as welfarists, who do not assign any weight to the freedom to choose.
Among the welfarists, we identify two types: one group of welfarists respect the stakeholder’s freedom to choose, while the other is paternalistic in the sense that they consider the risky option to give the highest welfare to the stakeholder.
Discussion and policy implications
This study provides causal evidence on how the type of intervention – hard or soft – influences people’s willingness to intervene in others’ decision-making. Across two large-scale samples of the US population, we found that only about a third of spectators chose hard interventions, which restrict freedom by removing options, while a large majority implemented soft interventions, which provide information without limiting choice. These patterns were consistent regardless of whether the stakeholder’s mistake arose from internal miscalculation or external misinformation, and they held across diverse groups in society.
To connect these findings to broader policy debates, we asked spectators in a survey after the experiment if they agreed that people sometimes make harmful choices, and a large majority did. When asked whether the government can improve lives by restricting freedom of choice, many agreed, though scepticism about hard paternalistic policies was significant, particularly among Republican-leaning participants.
Despite these differences in responses regarding government paternalism, political affiliation did not substantially influence behaviour in the experiment, suggesting that political disagreements over paternalistic interventions are more about people having different beliefs about the government’s ability to implement such policies than about fundamental differences in their paternalistic preferences (also see Luyten et al. 2020).
Our framework offers tools for further exploring paternalistic preferences across different domains and contexts. Future research should investigate cultural differences and their influence on attitudes toward paternalistic policies. Comprehending people’s paternalistic preferences is essential for designing interventions that effectively balance autonomy and welfare and understanding the varying support for paternalistic policies across the globe.
References
Bartling, B, A W Cappelen, H Hermes, M Skivenes, and B Tungodden (2023), “Free to Fail? Paternalistic preferences in the US”, SSRN Working Paper 4449814.
Camerer, C F, S Issacharoff, G Loewenstein, T O’Donoghue, and M Rabin (2003), “Regulation for conservatives: Behavioral economics and the case for asymmetric paternalism”, University of Pennsylvania Law Review 151(3): 1211–54.
DellaVigna, S, and E Linos (2022), “RCTs to scale: Comprehensive evidence from two nudge units”, Econometrica 90: 81–116.
Kuziemko, I, M I Norton, E Saez, and S Stantcheva (2015), “How elastic are preferences for redistribution? Evidence from randomized survey experiments”, American Economic Review 105(4): 1478–508.
Le Grand, J, and B New (2015), Government paternalism: Nanny state or helpful friend?, Princeton, New Jersey: Princeton University Press.
Luyten, J, S Tubeuf, and R Kessels (2020), “Public preferences for prioritising a COVID-19 vaccine”, VoxEU.org, 25 November.
Nussbaum, M (2001), “Symposium on Amartya Sen’s philosophy: Adaptive preferences and women’s options”, Economics and Philosophy 17: 67–88.
OECD (2017), “Behavioural insights and public policy: Lessons from around the world”, Technical Report, OECD Publishing.
Saito, T, S Sasaki, and F Ohtake (2021), “How to nudge COVID-19 vaccination while respecting autonomous decision making”, VoxEU.org, 13 December.
Schubert, C (2016), “A note on the ethics of nudges”, VoxEU.org, 22 January.
Sen, A (2002), Rationality and freedom, Harvard University Press.
Sunstein, C R, and R H Thaler (2008), Nudge: Improving decisions about health, wealth, and happiness, New Haven: Yale University Press.
Thaler, R H, and C R Sunstein (2003), “Libertarian paternalism”, American Economic Review 93(2): 175–79.
