ESSAYS on NUDGING Way to Influence Someone Without Violating His Or Her Freedom of Choice

Home , Nudge theory, Parimutuel betting

THE GOOD PLACE Gustav Almqvist Google “how to make better decisions” and you get 5 million hits. Blog posts, TED talks and self-help books that promise to improve our lives dramatically as long as we follow their easy steps towards becoming better decision makers. A popular concept within the same discourse is known as choice architecture or THE GOOD PLACE nudging. Nudge, a book published by two scholars in 2008, stands out as it offers us a way to make better decisions for someone else. A nudge is a gentle ESSAYS ON NUDGING way to influence someone without violating his or her freedom of choice. It must be for a good cause, but must not involve money. Graphic warnings on cigarette packages, the power saving mode on your TV, and the “open here” symbol at the back of your bag of chips are all nudges. The pedometer on your smartphone is a nudge.

Judgment and decision making (JDM) research has normative, descriptive and prescriptive sides. Nudging has become a cornerstone of prescriptive JDM. Over the past decade, it has spread to business and public administration. Many governments now use it. This dissertation critically discusses the ideological assumptions behind nudging and empirically investigates when it may be unpopular, unnecessary or Big Brother-esque. That is, when not to nudge.

Building on secondary data analyses and surveys, the dissertation contains four articles. Articles 1 and 2 investigate the Swedish public support for nudging. Article 3 studies an alleged bias among horseracing bettors. Article 4 looks into predictions from Big Data. The dissertation concludes that Swedes are cautiously positive towards nudging; that horseracing bettors generally Gustav Almqvist are not biased; and that it remains to be seen whether Big Data leads to Big (Brother) nudging. The dissertation generally warns against unrealistic expectations of nudging.

GUSTAV ALMQVIST is a researcher at the Center for THE GOOD PLACE THE GOOD Media and Economic Psychology and the Center for Sports and Business at the Stockholm School of Eco- nomics Institute for Research (SIR) and a teacher at the Department of Marketing and Strategy at the Stock- holm School of Economics. He lives in Stockholm.

ISBN 978-91-7731-182-9

DOCTORAL DISSERTATION IN BUSINESS ADMINISTRATION STOCKHOLM SCHOOL OF ECONOMICS, SWEDEN 2020 THE GOOD PLACE Gustav Almqvist Google “how to make better decisions” and you get 5 million hits. Blog posts, TED talks and self-help books that promise to improve our lives dramatically as long as we follow their easy steps towards becoming better decision makers. A popular concept within the same discourse is known as choice architecture or THE GOOD PLACE nudging. Nudge, a book published by two scholars in 2008, stands out as it offers us a way to make better decisions for someone else. A nudge is a gentle ESSAYS ON NUDGING way to influence someone without violating his or her freedom of choice. It must be for a good cause, but must not involve money. Graphic warnings on cigarette packages, the power saving mode on your TV, and the “open here” symbol at the back of your bag of chips are all nudges. The pedometer on your smartphone is a nudge.

ISBN 978-91-7731-182-9

DOCTORAL DISSERTATION IN BUSINESS ADMINISTRATION STOCKHOLM SCHOOL OF ECONOMICS, SWEDEN 2020

The Good Place

Essays on Nudging

Gustav Almqvist

Akademisk avhandling som för avläggande av ekonomie doktorsexamen vid Handelshögskolan i Stockholm framläggs för offentlig granskning tisdagen den 15 december 2020, kl 13.15, sal Torsten, Handelshögskolan, Sveavägen 65, Stockholm

The Good Place

Essays on Nudging

The Good Place

Essays on Nudging

Gustav Almqvist

Dissertation for the Degree of Doctor of Philosophy, Ph.D., in Business Administration Stockholm School of Economics, 2020

The Good Place: Essays on Nudging © SSE and the author, 2020 ISBN 978-91-7731-182-9 (printed) ISBN 978-91-7731-183-6 (pdf) Front cover illustration: © iStockphoto, 2020 Back cover photo: SSE, 2020 Printed by: BrandFactory, Gothenburg, 2020 Keywords: Nudge, choice architecture, judgment and decision making, bounded rationality, heuristics, favourite-longshot bias, big data

The Good Place is divided into distinct neighbourhoods. Each one contains exactly 322 people, who have been perfectly selected to blend into a blissful harmonic balance […] in each one, every blade of grass, every ladybug, every detail has been precisely designed (Schur & Goddard, 2016)

Fitter, happier. More productive. Comfortable. Not drinking too much. Regular exercise at the gym, three days a week. [---] Eating well. No more microwave dinners and saturated fats. A patient, better driver. A safer car […] Sleeping well […] Careful to all animals […] Keep in contact with old friends, enjoy a drink now and then [---] Car wash, also on Sundays [--- ] An empowered and informed member of society […] Tires that grip in the dark […] Calm, fitter, healthier and more productive (Radiohead, 1997)

Foreword

This volume is the result of a research project carried out at the Department of Marketing and Strategy at the Stockholm School of Economics (SSE). The volume is submitted as a doctoral thesis at SSE. In keeping with the policies of SSE, the author has been entirely free to conduct and present his research in the manner of his choosing as an expression of his own ideas. SSE is grateful for the financial support provided by the Jan Wallander and Tom Hedelius Foundation and the Tore Browaldh Foundation, which has made it possible to carry out the project.

Göran Lindqvist Hans Kjellberg Director of Research Professor and Head of the Stockholm School of Economics Department of Marketing and Strategy

Preface

I took my first psychology course at Stockholm University 15 years ago. First up was a class in social psychology, where my friends and I were assigned to perform a field experiment. We had no idea what a field experiment was, but would soon learn that it required us to manipulate something in the outside world to see if that had any effect on people. My friends were into fitness and so wanted that as part of the experiment design (I was not, but caved in). We ended up spending two days in a subway station at rush hour. Day one, we registered what proportion of subway passengers voluntarily took the stairway instead of the escalator. It was a very small one. Day two, we placed ourselves just in front of the escalator, holding a hand-made sign saying: “Take the stairs, live longer!” Again, we counted the proportion. It had suddenly become larger. A computer program named SPSS later told us that we had discovered a statistically significant increase in the proportion of subway travellers taking the stairs after having been exposed to our message. In those days, there was no catchy word for tricking people into doing something they normally would not do, supposedly for their own good. Nowadays there is one. Nudging. Had someone told me back then that you could build an entire ideology around manipulations like ours, I probably would not have believed it. For deep down, we never thought that people really needed our help. Neither did we expect it to actually benefit their health to take the stairs that one time. Nor that they would have appreciated it, had we been standing there every morning. Those very same doubts are this dissertation’s raison d'être. x

Acknowledgments

I am incredibly thankful for having been given the opportunity to write a doctoral dissertation, especially at the Stockholm School of Economics (SSE), where I first studied as an undergraduate. Epstein (2019) likens writing a book to running the 800 metres–at its worst halfway through. Writing a dissertation is more like the marathon. Exhausting throughout. So there are many people I want to thank for helping me reach if not the finish line then at least, at the time of writing, the home-straight. First and foremost, I want to express an enormous debt of gratitude to my supervisors. At SSE, Associate Professor Patric Andersson, Professor Richard Wahlund and Assistant Professor Emelie Fröberg. At Uppsala University, Associate Professor Håkan Nilsson. Patric. Your knowledge and patience have been invaluable to me these past several years. You have taught me everything I know about golf and most of what I know about research. I will never forget the time you have invested, and faith you have shown, in me. That includes research, teaching and lunches at Man on the Moon. I owe you a lot. Richard. Your passion and intellectual curiosity has been inspiring, and your support invaluable. As Department Chair, you have been the voice of reason during Kafkaesque times. If it had not been for you, I would never have made it. Thank you for everything. Emelie. Thanks for excellent advice while keeping my spirits up. I owe you some thirty fifty cups of Sosta’s coffee by now. Much obliged. Håkan. I am very grateful for your guidance. I also acknowledge your ex- pertise on football, although we disagree whether Arild Stavrum was off- side when scoring his championship-winning goal in 1999. Again, thanks. Thank you also to the several great professors at SSE, many at the De- partment for Market and Strategy, from which I have learned so much. You include Pär Andersson, Micael Dahlen, Anna Dreber Almenberg, Hans Kjellberg, Magnus Söderlund, Örjan Sölvell and Udo Zander. I am also thankful to Professor Karl Wennberg, for whom I have studied, for help and inspiration; and Professor Emeritus and former PhD student representative Claes-Robert Julander for critical insights on office politics. My gratitude also goes to Mats Jutterström and Maria Grafström at the Stockholm Centre for Organizational Research (SCORE) for giving me my xi first job in academia as a research assistant. The same to Johan Söderholm at the Stockholm School of Economics Institute for Research (SIR), for first letting me know about the opening that would later turn into this job. I must also take the opportunity to thank some amazing friends that I have acquainted at SSE over the past few years. Kajsa Asplund, smart, inspiring and kind; David Falk, a great guy; Enrico Fontana, a brilliant story- teller; Henrik Glimstedt, an expert on life and wine; Per Henrik Hedberg, a true intellectual; Dante Holmberg, an expert on life and beer; Gabriel Karlberg, empathic, fun and innovative; Leon Nudel, curious and knowledgeable; Sofie Sagfossen, life-coach and strategist; Adam Åbonde Garke, a sincere person and researcher. As for my PhD cohort–you all made strong impressions on me, and I will fondly remember the time we spent together. The same goes for the rest of my friends in the office corridor, most recently Claes Bohman, Agneta Carlin, Peter Hagström, Dan Hjalmarsson, Claire Ingram Bogusz, Patrik Regnér, Mattias Svahn and Agnieszka Zalejska Jonsson. It has been great getting to know all of you! To my office roommates Emre Yildiz, Sergey Morgulis-Yakushev and Nurit Nobel– thanks for the company, and for tolerating my notoriously untidy desk. I also want to thank professors Gerd Gigerenzer, at the Max Planck In- stitute for Human Development in Berlin and Peter Juslin, at the Depart- ment of Psychology at Uppsala University, for inviting me to present my work before your respective research groups. I remain deeply honoured. My mock defence opponent, professor emeritus Henry Montgomery, generously shared his insights on how the dissertation could be improved. Thank you to the students I have taught, for keeping me on my toes. It was with great pride that I first stepped into an SSE classroom as teacher, and the feeling has not left me since. To the organizations that have supported my research–Demoskop, Novus and the Swedish Meteorological and Hydrological Institute–yet another thank you. Special thanks to Svenska Handelsbanken, the Jan Wal- lander and Tom Hedelius Foundation and the Tore Browaldh Foundation for financing my work. Stiftelsen Louis Fraenckels Stipendiefond kindly financed my participation at a conference last year. Last, but not least. Thank you to my family and friends, many of which have contributed to this work. I could not have done it without you.

Contents

Part One CHAPTER 1. Introduction ...... 1 Choice architecture vices ...... 3 Unpopular nudging ...... 3 Unnecessary nudging ...... 3 Big nudging ...... 4 Purpose ...... 6 Research questions ...... 6 Intended contributions...... 7 Outline ...... 7 Reading guide ...... 7 CHAPTER 2. Background ...... 9 Judgment and Decision Making (JDM) ...... 9 Normative JDM ...... 10 Descriptive JDM ...... 11 Prescriptive JDM ...... 13 CHAPTER 3. Ideologies ...... 17 Paradigms ...... 17 Bounded rationality ...... 18 Worldviews ...... 21 Risk and uncertainty ...... 21 Heuristics ...... 23 Analogies ...... 29 CHAPTER 4. Previous research ...... 33 Carrots, sticks, sermons, and nudges ...... 33 The public support for nudging ...... 33 Favorite-longshot bias ...... 38 xiv

Background ...... 38 Mini-case: The Elit race and Sweden Cup ...... 41 Big nudging ...... 44 Man versus machine ...... 44 CHAPTER 5. Methodology ...... 47 Methods ...... 47 Secondary data analyses ...... 47 Survey designs ...... 47 Data analyses ...... 52 Limitations ...... 53 Disclosures ...... 53 Conflict of interest statement ...... 53 Data ...... 53 CHAPTER 6. Abstracts ...... 55 Article 1: Cautious support for nudging among Swedes in representative sample ...... 55 Article 2: Carrots, sticks, sermons or nudges? The Swedish public support for traditional and behavioral tools of government ...... 56 Article 3: Pari-mutuel betting on horseracing: The favourite- longshot bias revisited ...... 57 Article 4: Uncertainty and Complexity in Predictions from Big Data: Why Managerial Heuristics will Survive Datafication ...... 58 CHAPTER 7. Discussion ...... 59 Conclusions ...... 59 Point of departure ...... 59 Findings ...... 59 Impact ...... 60 Final remarks ...... 65 Eutopia ...... 65 Dystopia ...... 66 Epilogue ...... 67 Health ...... 67 Wealth ...... 68 Happiness ...... 68 xv

Part Two CHAPTER 8. Cautious support for nudging among Swedes in representative sample ...... 73 Introduction ...... 73 Method ...... 76 Respondents ...... 76 Questionnaire ...... 77 Results ...... 78 Descriptive statistics ...... 78 Multivariate analyses ...... 82 Discussion ...... 85 Appendix ...... 86 CHAPTER 9. Carrots, sticks, sermons or nudges? The Swedish public support for traditional and behavioral tools of government ...... 89 Introduction ...... 89 Study 1 ...... 92 Method ...... 92 Results ...... 96 Conclusions ...... 97 Study 2 ...... 98 Method ...... 98 Results ...... 105 Conclusions ...... 115 General discussion ...... 115 Appendix ...... 117 CHAPTER 10. Pari-mutuel betting on horseracing: The favorite- longshot bias revisited ...... 125 Introduction ...... 125 Method ...... 126 Data ...... 126 Results ...... 128 Study 1: Win bets ...... 128 Study 2: Pick 4-8 ...... 130 Conclusions ...... 138 xvi

Appendix ...... 139 CHAPTER 11. Uncertainty and Complexity in Predictions from Big Data: Why Managerial Heuristics Will Survive Datafication ...... 141 Introduction ...... 141 Predictions From Big Data ...... 143 Prediction...... 143 Big Data ...... 145 The Bias/Variance Dilemma ...... 147 The Curse of Dimensionality ...... 149 Further examples ...... 150 Managerial Heuristics ...... 152 Discussion ...... 154 Conclusions ...... 155 References ...... 157

Part One

Chapter 1

Introduction

There is something about nudging. The idea itself is not new. But the name is. For a decade, essentially the same concept had been called anti-anti paternalism (Jolls, Sunstein & Thaler, 1998), asymmetric paternalism (Camerer et al., 2003) or libertarian paternalism (Thaler & Sunstein, 2003). Under these alias- es, few had taken notice. It took a book editor recommending Thaler and Sunstein (2008) to rename it Nudge for things to really take off. Today, nudging is more than a buzzword. It has outgrown the research field I am in–judgment and decision making (JDM)–to become an ideology affecting public administration in many countries (OECD, 2017; Whitehead et al., 2014). Its goal is to align individuals’ choices with their own or society’s best interests. In practice, choice architects envision a society where people save for retirement (Thaler & Benartzi, 2004), hold diversified stock portfolios (Benartzi & Thaler, 2001), smoke less and cut back on fast-food while donating to charity and using green energy (see Sunstein, 2017a). The rationale behind nudging–or choice architecture–is that people sometimes make poor decisions because of intuitive thinking (Kahneman, 2003), cognitive shortcuts (Tversky & Kahneman, 1974), mental illusions (Thaler, 1980) or limited self-control (Thaler & Shefrin, 1981) and therefore need help. When such mistakes occur in a predictable manner they are called cognitive biases. Biases are deviations from norms like rules of logic. They may be caused by heuristics: rules of thumb that simplify decision-making. Nudg- es exploit such heuristics through “any aspect of the choice architecture that alters people’s behavior in a predictable way without forbidding any 2 THE GOOD PLACE options or significantly changing their economic incentives” (Thaler & Sun- stein, 2008, p. 6); by how the choice is structured and described (Johnson et al., 2012). This narrative helped Richard Thaler win The Sveriges Riksbanks Prize in Economic Sciences in Memory of Alfred Nobel (The Committee for the Prize in Economic Sciences in Memory of Alfred Nobel, 2017). I appreciate Thaler’s work. But he was awarded the prize based on a biased reading of the JDM literature (Almqvist, 2018a).1 Right or wrong, others criticize nudging for threatening personal integrity (Helbing et al., 2017), benefitting choice architects themselves (Berg, 2014), confusing its root- cause and effect (Loewenstein & Chater, 2017), being myopic (Frischmann, 2019), self-contradictory (Lodge & Wegrich, 2016), unethical (Rebonato, 2012), manipulative (Wilkinson, 2013) and based upon a biased selection of, sometimes misinterpreted, research (Berg & Gigerenzer, 2007; Gigeren- zer, 2015; 2018) under questionable assumptions about people’s preferences (Sugden, 2017) and opportunity costs (Berg & Davidson, 2017).2 The debate for or against nudging has been highbrow and fierce. From now onward, I try a less polemic way. I infer principles for when nudging may be inappropriate. That is, when not to nudge. It does not make for as good a TED talk, but remains interesting for several reasons. First, choice architects promote state nudging for its cost-effectiveness (Benartzi et al., 2017). If taxpayers’ money is the concern, then knowing when nudging is redundant is also important. Some nudges fail (Sunstein, 2017d), others backfire (Mols, Haslam, Jetten & Steffens, 2015). Secondly, nudges can be annoying (Damgaard & Gravert, 2018) wherefore they must not be over- used. Thirdly, it would be undemocratic to nudge in disregard of public opinion. Fourthly, unrealistic faith in nudging puts alternatives–like boosting people’s capability to make better decisions by themselves–at an unfair dis- advantage (Hertwig & Grüne-Yanoff, 2017). Whether to nudge or boost is something that must be considered on a case-by-case basis (Hertwig, 2017). Lastly, it was a good story. One I enjoyed writing.

1 For a more nuanced discussion of Thaler’s contributions, see Earl (2018). 2 Also see Madi (2020). There have been different reactions to the criticism. It has been downplayed (Gärdenfors, Johannesson, Molander, Sjöström & Strömberg, 2017) and belittled (Thaler, 2017), but also constructively discussed (Sunstein, 2015a, 2015b, 2016a, 2017b, 2017c, 2018). CHAPTER 1 3

Choice architecture vices

Vitruvius published De Architectura in ancient Rome (Pollio, 1914). It included virtues for architecture.3 Nudge contains equivalents to the Vitruvian virtues, this time for choice architecture. Its golden rule is to nudge when people “do not get prompt feedback” (Thaler & Sunstein, 2008, p. 74) on their decisions. Nudge also borrows a concept from Rawls (2009)–publicity– which here means for the government to openly defend its actions before the citizens. The narrative however becomes more interesting once we turn it around and consider the opposite of virtues: vices. I will consider three choice architecture vices–unpopular nudging, unnecessary nudging and big nudging–and investigate one research question for each.

Unpopular nudging

A prerequisite for nudging is the choice architect’s responsibility for it, which is what Thaler and Sunstein (2008) mean by publicity. But nudging must be evaluated on the same terms as its alternatives (Weimer, 2020). In the end, it will be up to the voters whether or not they approve of nudging and, if so, within which areas. Unpopularity could spell the end for it. Until recently, choice architects had not surveyed the public opinion of nudging. Consequently, there is state nudging all over the world, but still relatively few surveys on the public support for it. For example, the Swe- dish government has considered it for several years (Ramsberg, 2016), although little is known about its domestic approval (but see Hagman, 2019).

Unnecessary nudging

In order for nudging to lead to better decisions, the status quo must be suboptimal. If there is no cognitive bias, there is no need to nudge. So far, everyone would agree. The problem is the false positives. Illusory biases. They signal reasoning errors when really there are not any, and may lead to unnecessary nudging. A warning-sign for illusory biases is when the golden

3 That it should last (firmitas), work (utilitas) and be appreciated (venustas). 4 THE GOOD PLACE rule is violated. When there actually is frequent, reliable feedback to learn from. To Hogarth (2001), this is what defines kind learning environments. Take making a living as a cab driver. It is a kind learning environment as customers come and go all day. There are predictably both slow and busy hours, peaking at rush hour. Camerer, Babcock, Loewenstein and Thaler (1998) yet suggest that cab drivers would be biased in deciding their job hours. By setting a daily income target they supposedly fall victim to a heuristic called mental accounting (see Thaler, 1985). However, it has since been found that cab drivers are not biased in choosing job hours. More extensive analyses, using data from Uber, show they quickly learn it almost perfectly (Sheldon, 2016).4 The bias seems to have been illusory. Gigerenzer (2015; 2018) has long argued that the prevalence–and consequences (Arkes, Gigerenzer & Hertwig, 2016)–of many biases have been exaggerated in previous research, which would be a foundation problem for nudging. But he only has a handful convincing cases thus far. Another kind learning environment, suitable for such an investigation, is the racetrack. Also there have cognitive biases been suggested (Thaler & Ziemba, 1988). Tversky and Kahneman (1981) and Camerer (2000) all use the racetrack to prove the generalizability of their theories. As does Thaler (2015). The racetrack accordingly offers a case for an empirical investigation of the rationale behind nudging. With so much at stake (pun intended), a renewed assessment of alleged heuristics and biases at the racetrack is called for.5

Big nudging

There is widespread belief that robots will replace humans in many indus- tries. The umbrella term for the computational techniques some of these robots will use is artificial intelligence (AI). Those critical of humans naturally prefer AI. Thaler believes in it (Javetski & Koller, 2018). As does Kahne- man–Thaler’s mentor and fellow Nobel laureate–whose advice at the Na- tional Bureau of Economic Research (NBER) conference in 2017 was to “replace humans by algorithms whenever possible”. Slovic–their mutual

4 Ironically, Uber nudges its drivers to choose worse hours, to smoothen supply (Scheiber, 2017). 5 In addition, nudges are common in betting. Only they usually trick people to bet more, not less. So-called dark nudging (Newall, 2018). Thaler (2018) calls counter-productive nudges sludges. CHAPTER 1 5 friend–sees AI as the next frontier of JDM research (P. Slovic, personal communication, September 18, 2018). A related opinion is that more (that is, big) data always leads to better decisions (McAfee & Brynjolfsson, 2012). At the NBER conference, Kahneman found it “very difficult to imagine that with sufficient data there will remain things that only humans can do”. New technology enables more nudging (Weinmann, Schneider & vom Brocke, 2016). While some welcome this development; others fear it. An alarmist scenario of large-scale, data-driven manipulation is called big nudging (Helbing et al., 2017; Puaschunder, 2017).

The new, caring government is not only interested in what we do, but also wants to make sure that we do the things that it considers to be right. The magic phrase is "big nudging", which is the combination of big data with nudging. To many, this appears to be a sort of digital scepter that allows one to govern the masses efficiently, without having to involve citizens in democratic processes […] citizens could be governed by a data-empowered “wise king”, who would be able to produce desired economic and social outcomes almost as if with a digital magic wand (Helbing et al., 2017).

However, both optimists and pessimists may overestimate Big Data and AI. It remains difficult for algorithms to predict and influence many characteristics and behaviours. Persuading undecided voters through micro-targeted ads, for example (Sumpter, 2019). Big nudging needs to be demystified. Digital nudging does not outperform conventional one (Hummel & Maedche, 2019). But the digital sphere can be difficult for policy-makers to regulate, and for some internet users to navigate, which increases the risk for nudges to be misused (Reisch, 2020).

  

The choice architecture vices are straightforward and transcend partisan- ship. Whether you are for or against nudging in general, you probably dis- approve of unpopular, unnecessary and big nudging. The choice architecture vices accordingly enable a fair, critical discussion on nudging. 6 THE GOOD PLACE

Purpose

As mentioned in the preface, I know firsthand that nudging can work. The purpose of this dissertation is to investigate its approval and the rationale behind it.

Research questions

Someone once said that academia is like the quiz show Jeopardy! You come up with questions to answers you already have. Here are my best attempts. On unpopular nudging In the unpopular nudging category, the main question reads: How popular is nudging among Swedes? It is complemented by sub-questions on how this popularity compares to that in other countries and traditional ways in which the government can influence people; and whether it differs across nudges. Another sub-question is how individuals’ approval of nudging relates to their political party preferences and ideological views. The answers to these research questions are found in Articles 1 and 2. On unnecesarry nudging In the unnecessary nudging category, the main question reads: How calibrated are odds 6 and race outcomes in pari-mutuel horserace betting in Sweden? Its sub- questions concern differences in calibration across different types of bets. The answers to these research questions are found in Article 3. On big nudging In the big nudging category, the main question reads: Why is it that predictions from big data must not outperform managerial heuristics? It is accompanied by related empirical and conceptual sub-questions. The answers to these research questions are found in Article 4.7

6 Here, odds are the distribution of aggregate bets (cf. Andersson & Nilsson, 2015). 7 Article 4 belongs to the science popularization genre while Articles 1, 2 and 3 are research articles (see De Oliveira & Pagano, 2006). Business administration research at SSE has a practitioner-oriented history (Engvall, 2009), motivating the inclusion of a science popularization article with managerial implications. CHAPTER 1 7

Intended contributions

With this dissertation, I hope to contribute to the JDM literature. Research contributions must of course be scientifically interesting. They can then fall into one of four categories: revelatory, consolidatory, incremental or replicatory contributions (Nicholson, LaPlaca, Al-Abdin, Breese & Khan, 2018). Revelatory contributions can come from problematizing the assumptions in previous research. Consolidatory contributions are made through literature reviews. This dissertation’s first part–the so-called kappa–seeks to make revelatory and consolidatory contributions. Incremental contributions address gaps in the literature. Replicatory contributions empirically double- check prior research. This dissertation’s second part–the articles–aims for incremental and replicatory contributions. More specifically, the intended contributions follow from the dissertation’s purpose: to investigate the approval of nudging, and the rationale behind it. Articles 1 and 2 focus on the approval of nudging; the kappa, Article 3 and Article 4 on different aspects of the rationale behind it.8

Outline

The outline of the dissertation is as follows. Chapter 2 consists of an introduction to JDM. Chapter 3 problematizes JDM ideologies. Chapter 4 features a literature review of previous research. Chapter 5 covers the methodology. Chapter 6 summarizes the articles. Chapter 7 contains the discussion. Chapters 8, 9, 10 and 11 correspond to the respective articles.

Reading guide

I recommend you read the whole thing. However, if you are an especially restless reader–eager to get straight to the point–you can skip Chapters 2 and 3 and jump directly to Chapter 4. You will miss out on some background and problematization, but the dissertation can be read independently of those. It is your choice. I won’t nudge you.

8 Articles 3 and 4 accordingly do not address nudging directly.

Chapter 2

Background

Judgment and Decision Making (JDM)

The governing bodies of JDM research are the Society for Judgment and Decision Making (SJDM) and the European Association for Decision Mak- ing (EADM) respectively. They both identify as interdisciplinary academic organizations dedicated to the study of normative, descriptive, and prescriptive theories of JDM. They arrange one conference each: the SJDM annual meeting in North America and the biennial SPUDM (Subjective Probability Utility and Decision Making) in Europe. The latter is the older of the two, dating back to Hamburg in 1969. JDM’s normative, descriptive and prescriptive perspectives apply a di- verse set of theories and relate to different types of research questions (Bell, Raiffa & Tversky, 1988). They are also evaluated differently.

Descriptive models are evaluated by their empirical validity, that is, the extent to which they correspond to observed choices. Normative models are evaluated by their theoretical adequacy, that is, the degree to which they provide acceptable idealizations or rational choice. Prescriptive models are evaluated by their pragmatic value, that is, by their ability to help people make better decisions (Bell, Raiffa & Tversky, 1988, p. 18, italics in original).

One could argue all three of normative, descriptive and prescriptive JDM are needed in order to help people make better decisions (Baron, 2004). They also reflect JDM’s history chronologically. Its normative phase began with the probabilistic revolution (Fox, Erner & Walters, 2014; Gigerenzer et al., 1989; Krüger, Gigerenzer & Morgan, 1987), which “made concepts such as 10 THE GOOD PLACE probability, chance, and uncertainty indispensable for understanding nature, society, and the mind” (Gigerenzer, 1991a, p. 83).

Normative JDM

In the early 18th century, a thought-experiment, the St. Petersburg Paradox, had begun to circulate among philosophers and mathematicians. It was one of these peculiar cases where logic and common sense disagree. The St. Petersburg Paradox came across as subversive, not because of its theoretical or practical applications, but because it highlighted a mismatch between probability theory and the prescribed behaviour of reasonable men (Das- ton, 1980). Daniel Bernoulli therefore did what every reasonable man would do under such circumstances. He changed the theory. Probability theory had previously relied on the notion of mathematical expectation. This was the idea that rational decision-making implied maximization of an objective expected value (the product of an outcome and its probability). It had been around since 1654, when the two mathematicians Pascal and de Fermat solved a gambling scenario on behalf of the Chevalier de Méré (Bernstein, 1996). An advantage of this traditional approach had been its consistency with logical virtue, as reflected among contemporary intellectuals. In Laplacian terms, probability theory had been considered common sense reduced to calculus (Daston, 1988; Gigerenzer et al., 1989). Bernoulli (1954) was about to become among, if not the, first to pro- pose maximization of moral expectation or expected utility (EU). His innovation was to extend the old theory so that the EU of a choice now became the product of an outcome’s subjective value and its probability. Some 200 years later, by the time of 1952’s seminal decision making conference in Paris, neoclassical economics was at the height of its powers. EU theory had grown into a formalized framework postulating logically coherent preferences (von Neumann & Morgenstern, 1947) and derivable utility functions (Friedman & Savage, 1948). EU theory remained a normative and not descriptive model. The correspondence between the leading economists of the time reveals they considered EU theory a normative exercise (Moscati, 2016). The obvious exception being Friedman (1953). CHAPTER 2 11

Normative JDM would come in for internal criticism (Allais, 1953a, 1953b; Ellsberg, 1961). But by then, an extension of EU theory had already developed out of psychological experiments on probabilistic inferences.

Descriptive JDM

Independently of neoclassical economics, another research stream emerged from the judgment (J) stream of JDM (Goldstein & Hogarth, 1997). Similarly to visual perception, it was theorized there was a general psychophysical law regulating the ability to discriminate between probabilities. Experiments both in the laboratory (Preston & Barrata, 1948) and in the field (Griffith, 1949) supported this view. These ideas became the subjective expected utility (SEU) theory (Edwards, 1955). In Psychological Bulletin, Edwards (1954) introduced the normative JDM theories to a wider audience. He concluded, “these topics represent a new and rich field for psychologists, in which a theoretical structure has already been elaborately worked out and in which many experiments need to be performed” (p. 411). Edwards’ call would be answered in the 1970’s through the heuristics and biases program. At its forefront were the Israeli psychologists Tversky and Kahneman (see Tversky & Kahneman, 1971, 1973, 1974, 1981, 1983, 1985, Kahneman, Slovic & Tversky, 1982; Kahneman & Tversky, 1972, 1973, 1979, 1984). They studied cognitive heuristics that, while generally effective, also could “lead to severe and systematic errors” (Tversky & Kahneman, 1974, p. 1124). Some of the original heuristics were labelled representativeness, availability, and adjustment and anchoring. Thaler (1985) complemented them with mental accounting. An extensive list of biases and fallacies would follow. It included overconfidence, framing effect, confirmation bias, status-quo bias, hindsight bias, base rate ne- glect, loss aversion, narrative fallacy, conjunction fallacy, endowment effect, gambler’s fallacy, hot-hand fallacy and planning fallacy (Kahneman, 2011; Kahneman, Knetsch & Thaler, 1991; Fischhoff, Slovic & Lichtenstein, 1977). See Table 1 below for some examples of alleged biases and fallacies. There are literarily hundreds of similar cases on Wikipedia (Gigerenzer, 2016). 12 THE GOOD PLACE

Table 1. Alleged biases and fallacies.9

Bias/ fallacy Empirical example Overconfidence10 Car driving. A vast majority of drivers think of themselves as better than the median (Svensson, 1981). Framing effect11 Mortality. A 10 % chance to die sounds worse than a 90 % chance to survive (Wilson, Kaplan & Schneiderman, 1987). Confirmation bias Forensics. A confession–true or false–affects the interpretation of other evidence against a suspect (Kassin, Dror & Kukucka, 2013). Status quo bias Health insurance. Most keep whatever plan they happen to have; preferences aside (Samuelson & Zeckhauser, 1988). Hindsight bias Diagnostic medicine. Physicians find patients’ diagnoses obvious after the fact (Arkes, Wortmann, Saville & Harkness, 1981). Endowment effect12 Auctions. First give half a group of people some mugs and ballpoint pens for free, then have the other half bid for them, and there will be a bid-ask spread (Kahneman, Knetsch & Thaler, 1991). Gambler’s fallacy13 Roulette. After long sequences of consecutive red or black numbers, most players bet against the streak (Croson & Sundali, 2005). Hot-hand fallacy14 Roulette. Players are more inclined to bet again after a win than after a loss (Croson & Sundali, 2005). Planning fallacy Infrastructure projects. Railways practically always cost more, and deliver less, than originally projected (Flyvbjerg, 2009).

The heuristics and biases program would later receive a nemesis: Gerd Gigerenzer (see Gigerenzer, 1991a, 1991b, 1996; Gigerenzer, Hell & Blank, 1988; Gigerenzer, Hoffrage & Kleinbölting, 1991). To Gigerenzer, there were never any biases or fallacies, only unrepresentative tasks and patroniz- ing ideals on Kahneman-Tversky’s behalf. For example, Gigerenzer and Hoffrage (1995) showed that certain biases disappeared when information was presented in a more understandable way (also see Cosmides & Tooby, 1996). Then how could they be hardwired into the human brain? Instead, Gigerenzer thought of heuristics as fast and frugal (quick, yet accurate) parts of a cognitive toolbox, tailored to fit specific environments (Gigerenzer & Goldstein, 1996; Gigerenzer, Todd and the ABC Research Group, 1999).

9 But see Gigerenzer (2015). 10 But see Juslin (1994) and Olsson (2014). 11 But see Kühberger (1995). 12 But see Plott and Zeiler (2005). 13 But see Ayton and Fischer (2004). 14 But see Miller and Sanjurjo (2018). CHAPTER 2 13

Prescriptive JDM

In Greek mythology, Prometheus was a rebellious Titan who stole the fire from Mount Olympus to give to mankind along with art and science (Aischylos, 1903). He also effectuated the first ever nudge. According to the Theogony (Hesiod, 1914), Prometheus had been ap- pointed peace mediator in a conflict between the Olympians and mankind. Reconciliation required man to sacrifice in Zeus’ honour. The titan, who secretly sided with the humans, allowed Zeus to decide the token of appre- ciation himself. From a slaughtered ox, he gave Zeus the choice between the beast’s horn (worthless) and its meat (valuable). But not before he had planned an ingenious way to influence him. To nudge Zeus into choosing the horn, Prometheus began by turning the stomach inside out. This hid the beef under disgusting guts. Meanwhile, he presented the horn covered in delicious fat. “Very noble Zeus, greatest of the gods who are for always,” he said, “choose whichever of these the spirit in your breast bids you” (Hesiod, 1914, lines 548-549). As he had hoped, Zeus went by the exteriors, and surrendered to the Trick at Mekone. More recently, a nudging renaissance began when Sunstein (1997; Jolls, Sunstein & Thaler, 1998) questioned the legal system. It would continue with attempts to improve people’s stock portfolios (Benartzi & Thaler, 2001, 2002, 2003; Newall & Love, 2015), increase employees’ retirement saving rate (Thaler & Benartzi, 2001, 2004), reduce poverty (Bertrand, Mul- lainathan & Shafir, 2004, 2006), promote healthier life styles (Loewenstein, Brennan & Volpp, 2007), reduce obesity (Olvier & Ubel, 2014), decrease energy consumption (Allcott & Mullainathan, 2010), prevent job discrimi- nation (Bohnet, van Geen & Bazerman, 2015), recommend better products (Goldstein, Johnson, Herrmann & Heitmann, 2008) and improve government (Sunstein, 2013), to name some initiatives (also see Egan, n.d.). Kahneman has characterized nudging as follows: “There is no overarching theory. It is not big. It is interventions that cost essentially nothing and that achieve small but reliable results” (Nelson, 2015). Accordingly, there are many different ways to nudge. Table 2 below contains the most common types (Sunstein, 2014), along with some homemade examples. 14 THE GOOD PLACE

Table 2. Ten types of nudges (Sunstein, 2014). Homemade examples.

Type of nudge What it might involve Defaults Auto-play (HBO, Netflix), power saving mode (TV) Simplifications Executive summaries, “open here” (packages) Social norms “Trending” (Twitter, Spotify), flight shame15 Ease, convenience Swish (payments), single sign-on (passwords) Disclosures Alcohol by volume, effective interest rate Warnings Traffic signs, images on cigarette packages Pre-commitments RSVP, restaurant reservations, letters of intent Reminders Snooze, “people you may know” (Facebook) Pledges New Year’s resolutions, oaths of citizenship Feedback Screen time stats (smartphones), pedometers

Hummel and Maedche (2019) conducted a meta-analysis on nudging’s effectiveness. It included 100 empirical articles reporting a total of 308 effects with p-values. The meta-analysis revealed that the studies’ average (median) relative effect size was 55 % (21 %).16 Almost two thirds of the effects were statistically significant (p < .05).17 One of the more effective nudges was Duflo, Kremer and Robinson’s (2011) program to increase Kenyan farmers’ use of fertilizer. It resulted in a 15-percentage point increase, from a sample proportion of 28 % to 43 %. Across the 10 types of nudges in table 1, average (median) relative effect sizes ranged from 7 % (7 %) to 107 % (50 %). Defaults turned out to be the most effective type of nudge. However, replication studies have taught us that the social sciences often report false positives and inflated effect sizes (Camerer et al., 2016, 2018; Open Science Collaboration, 2015). Many groundbreaking discoveries in psychology–several of which featured in Kahneman’s (2011) Thinking, fast and slow–were not true. This caused Kahneman (2012) to raise his concerns over the state of affairs in psychology research.

15 Flight shame is the result of anti-flying group pressure on social media (“flygskam” in Swedish). 16 The relative effect size is calculated as the ratio between the averages of the treatment group and control group on the dependent variable, minus one. It measures the relative increase from the treatment. 17 In this subset (n = 190), the average (median) relative effect size was 77 % (39 %). CHAPTER 2 15

JDM is unlikely to be an exception. The estimated replicability of papers published in Judgment and Decision Making–from z-curve analyses18 of their test statistics and p-values–is on par with that of psychology journals in general (Schimmack, 2018). Another reason why some results are irreproducible are questionable research practices. The other year, a prominent nudging researcher was widely criticized for dishonest data analyses (so-called p-hacking) and saw many of his papers retracted (van der Zee, 2017). He was later found guilty of academic misconduct (Kotlikoff, 2018).

  

Nudging can evidently mean many things (for a discussion, see Hansen, 2016).19 And its effectiveness should not be taken for granted. However, one could argue that its essence is neither the nudges themselves nor their effect but the rationale behind it. An ideology bringing normative, descriptive and prescriptive ideas together as one. The next chapter will problematize that ideology.

18 A z-curve analysis uses the distribution of p-values in a population of studies to estimate its actual effect size and expected replication rate. 19 Pretty much anything, says Gigerenzer (2015).

Chapter 3

Ideologies

In this chapter, I problematize JDM ideologies. Ideologies are “general and abstract social beliefs, shared by a group, that control or organize the more specific knowledge and opinions” (van Dijk, 1998, p. 49). I discuss them in two steps. The first concerns how knowledge is organized. The second compares and contrasts rivalling worldviews.

Paradigms

Ideologies in academia are called paradigms. Paradigms organize and censor scientific discoveries to protect the status quo (Kuhn, 1962). They rely on underlying philosophical assumptions (Alvesson & Sandberg, 2011; 2013). JDM is somewhat of an “orphan field, lacking a dedicated and exclusive academic home” (Gilovich & Griffin, 2010, p. 2). As a consequence, “JDM research is not ‘paradigmatic.’ There is no single, universally endorsed, overarching theoretical framework that researchers use to organize and guide their efforts” (Goldstein & Hogarth, 1997, p. 3). This means there are several JDM paradigms.20 The rationale behind nudging reads bounded rationality (Battaglio Jr, Belardinelli, Bellé & Cantarelli, 2019).

20 The same paradigm can contain several research programs. Here, a research program is simply a pro- fessional network of researchers sharing similar ideas. Not necessarily one in Lakatosian terms. 18 THE GOOD PLACE

Bounded rationality

JDM theorists typically define rationality from either coherence or correspondence criteria (Hammond, 1996). The former are normative whereas the latter are practical. Coherence requires logical thinking. From this perspective, to comply with von Neumann and Morgenstern’s (1947) postulates, apply Bayes’ theorem, or subscribe to the rules of Euclidean geometry always qualifies as rational. Correspondence, in contrast, emphasizes resonance in the external environment. It concerns the outcomes. Long-run relative frequencies thus matter more than does the correct application of probability theory. Dunwoody (2009) complements coherence and correspondence with pragmatic rationality criteria. These are functionalistic and relates to goal attainment. An aspiration level of sorts. Coherence, correspondence and pragmatism are originally philosophies of truth. They are complementary with Simon’s (1957, 1976) distinctions between substantive, bounded and procedural rationality. The first of these de- notes the best possible outcome preceded by the optimal course of action. It is the sort of global rationality traditionally attributed to the omniscient Economic man (Simon, 1987). The last two refer to how decisions actually are made (the Administrative man). Substantive rationality implies optimization–the weighting and adding of all available information to arrive at the perfect solution–as the linkage between coherence, correspondence and pragmatism. Conversely, bounded rationality is subject to constraints that prevent optimization. Like when someone settles for a satisfactory, rather than optimal, solution to a problem (Simon, 1955, 1956). While bounded rationality is the negative of substantive rationality, procedural rationality is positively defined. It refers to the actual processes through which decisions are made (Simon, 1976). Accordingly, bounded rationality and procedural rationality are not the same; they are complementary. Although the latter was Simon’s favourite, it is the former that has become his legacy. Many economists have taken bounded rationality to heart not despite, but because it was more vaguely operationalized (Barros, 2010). Two research programs have dominated bounded rationality (Gigeren- zer, 2006). The first program argues that bounded rationality really is sub- CHAPTER 3 19 stantive rationality in disguise, considering mental transaction costs of information. This view is called optimization under constraints (see Arrow, 2004). The second program is behavioural economics (BE).21 It catalogues deviations from substantive rationality as biases and uses these to create more refined utility functions (Camerer & Loewenstein, 2004). While historically there has been little integration between economics and psychology (Schumpeter, 1954), BE has supposedly reunified the two (Camerer, 1999). The World Bank (2015) concludes: “Economics has thus come full circle. After a respite of about 40 years, an economics based on a more realistic understanding of human beings is being reinvented” (p. 5). To some, the essence of these discoveries seems to be that people are predictively irrational. As Ariely (2009) tells Harvard Business Review, “irra- tionality is the real invisible hand that drives human decision making”. Human beings, he explains, “are motivated by cognitive biases of which they are largely unaware [---] incapable of making good decisions“. If so, most of us need to be de-biased (see Fischhoff, 1981) or nudged.22 However, I cannot help finding such descriptions overly negative. When Katona (1975) pioneered BE (also called economic psychology or psychological economics) his goal was not to ridicule. It was to “discover and analyze the forces behind economic processes” (p. 9). Likewise, Simon (1989) did not merely ask whether or not people lived by neoclassical principles. His quintessential question read “how do people reason when the conditions for rationality postulated by the model of neoclassical economics are not met?” (p. 377, emphasis added), the answering of which requires more than refut- ing economic man. This is why Thaler’s (2015) eagerness to ridicule (ideal- ize) humans (econs)23 might be missing the point. Simon (1986) formalized four tenets of procedural rationality. Accord- ing to these, models of procedural rationality: (i) consider the actual decision process(es), that is, the how; (ii) neither involve utility nor optimization as concepts; (iii) study observables (either to the researchers or the actors themselves); and (iv) enable good, testable predictions.

21 See Angner and Loewenstein (2012) and Sent (2004) for historical recapitulations. 22 De-biasing explores ways in which to eliminate biases; while choice architects see them as chronic. 23 Econs are a made-up species that thinks in accordance with economic theory. 20 THE GOOD PLACE

As BE does not follow these rules, some suggest it has become a neoclassical wolf in psychological clothing. Indeed, it is grounded in the same axioms as its predecessor (Berg, 2010). Moreover, like Friedman (1953) before it, BE has come to rely on as-if models (Berg & Gigerenzer, 2010). JDM is said to involve weighting and integrating information according to a complex scheme. This model is then defended at all costs, even at the price of psychological realism (Selten, 2001). Such shortcomings would make BE incompatible with procedural rationality. Figure 1 sums up the bounded rationality paradigm, and positions the dissertation’s articles within it. What the figure essentially says is that the bounded rationality paradigm has focused on diagnosing and treating humans for not thinking like econs.24 To run an example through it, remember the cab drivers from the introduction (Camerer et al., 1998). Normatively, BE said they ought to maximize their expected income per hour worked. Descriptively, they were accused of using a heuristic instead, leading to suboptimal outcomes. Prescriptively, they would therefore have benefitted from de-biasing or nudging. In theory, that is.

Figure 1. The bounded rationality paradigm. Articles positioned accordingly.

24 Or explaining the seemingly human as subconsciously econ: the repair program (see Selten, 2001). CHAPTER 3 21

  

Going forward, a sincere interest in how decisions are made requires cir- cumventing the invisible hands of both Smith (1759) and Ariely (2009) to ensure safe passage through the strait between the neoclassical Scylla and the behavioural Charybdis. This calls for a different worldview.

Worldviews

The philosophy of the nature of the world is called ontology and is a com- pound from Greek. It comes from ὄντος (being) and λογία (logic). I prefer the more mundane term worldview myself. A worldview contains specific knowledge and opinions about the world, and stems from an ideology.

Risk and uncertainty

Small worlds What many JDM theories from Bernoulli through Kahneman-Tversky have in common is their reliance upon hypothetical lotteries where outcomes and probabilities are known in advance. Following Savage (1954), these can be denoted small worlds. Technically, a small world requires all possible outcomes to vary according to a known probability distribution. This allows calculating expected values (flip a fair coin, and it occurs in a small world). Small worlds were always simplifications. But no alternatives could have enabled preferences of outcomes and perceptions of probabilities to be so elegantly derived. To have neoclassical economists work on utility while psychologists addressed probabilistic inferences was also a functional division of labour, not least as it avoided the difficulty of “simultaneously axiomatizing utility and probability” (Samuelson, 1952, p. 670).25

25 This had been for JDM what Heisenberg’s indeterminacy principle was for quantum mechanics, so the two were typically studied in isolation. Ramsey (1931) and Savage (1954) were exceptions. 22 THE GOOD PLACE

Large worlds Most believe that probabilities exist in an objective sense; while others in- sist there are only subjective degrees of belief (Arrow, 1951). Either way, the small world epistemology–where probabilities are known by physical design–seldom holds up in the real world. I have previously developed a classification tree to illustrate how and why (see Figure 2 below). It suggests life is characterized by uncertainty. Following Knight (1921), these are conditions in which not all consequences or probabilities are known for sure. Instances where “there is no scientific basis on which to form any calcula- ble probability whatever” (Keynes, 1937, p. 213). When the information at hand is ambiguous (Ellsberg, 1961), fuzzy (Luce & Raiffa, 1957) or vague (Ar- row, 1992; Keynes, 1921). To Savage (1954), these are large worlds. In figure 2, I suggest there are six types of worlds. They can be classi- fied with three questions. The first is if outcomes vary or not. The second is whether their probability distribution is known or not. The third is how ambiguous the probability estimates are. These worlds range from absolute certainty (think of sunrise, death or taxes) to radical uncertainty (knowing nothing whatsoever. Leif Erikson setting foot in America, say). But of most practical relevance are arguably the worlds in between those extremes. Two of them are risky: a priori probability (playing dice) and statistical probability (car insurance). The others are uncertain: black swan (the stock market) and true uncertainty (almost everything else). Many JDM theories stem from small worlds that provide a benchmark against which judgments can be evaluated. However, only in a small world can such a gold standard exist. In the psychophysicist’s laboratory all mag- nitudes of interest are measurable with precise instruments. This is also the case in the casino, where probability theory reigns.26 But outside Rouletten- burg, even the lotteries are uncertain.27

26 For example, the European roulette offers its players the odds 36-1 for a bet on any individual number (each with a 1/37 chance). In this type of setting, notions like the gambler’s fallacy (Kahneman & Tversky, 1972) or the hot hand fallacy (Gilovich, Vallone & Tversky, 1985) may apply. 27 Modern slot machines reveal potential payoffs but never the underlying probability function. The state lottery’s numbers are all equally likely to be drawn, but the winnings also depend on the numbers’ relative popularity (not randomly selected). In sports and horse lotteries, there are no probabilities in advance. And for all your banker’s arguments, also the stock market is uncertain. None of these scenarios translate into roulette lotteries (see Anscombe & Aumann, 1963). CHAPTER 3 23

Figure 2. On risk and uncertainty. Adapted from Almqvist (2016).

Heuristics

While small worlds resemble platonic ideals, in all of their perfection, decisions under true uncertainty–whom to marry, what to have for dinner, where to go on holiday, how to vote, which entrepreneurial venture to un- dertake or stock to buy–are altogether different propositions. They require a large world lens and consequently other theories than do psychophysical laboratory experiments. Simon (1990) concluded “human rational behaviour is shaped by scis- sors whose two blades are the structure of task environments and the computational capabilities of the actor” (p. 7). Indeed, all organisms–Homo sapiens included–work in tandem with their environments. Humans’ rationality has nonetheless been evaluated differently to that of all other animals (Einhorn & Hogarth, 1981). A serious acknowledgement of heuristics may help to reconnect (wo)man with the universe in which (s)he actually lives. 24 THE GOOD PLACE

Ants Black garden ants (Lasius niger) live in colonies of tens of thousands individuals. Their nests are complex systems of underground chambers whose walls consist of pillars built piece by piece from soil. Considering the nest’s volume relative to the ants’ size and workforce, this task is much like building one of the Egyptian pyramids. Not only do the ants manage a project of such difficulty, they do so at merely a fraction of the time. Friedman (1953) could very well have said that the ants acted as-if they were secretly instruct- ed by an architectural mastermind, an experienced developer and a phar- aonic boss, all at once. In reality, they are not. The process is completely decentralized (all the queen ever does is to lay eggs). Ants lack intelligence and cannot adhere to formal logic. And yet they successfully design, develop and rebuild their own towns out of nothing. The explanation for this remarkable achieve- ment reads heuristics. Or here, following Khoung et al. (2016), building rules. They involve a motion rule, a pick-up rule and a stacking rule. Black garden ants are simple creatures. Their workers’ jobs are always the same. They must constantly be in motion, repeatedly collect building material (pick something up every 30 seconds), and then dispose of it somewhere else. Movement occurs uniformly and random (this is the motion rule). During collection, the ability to smell pheromones–which other ants leave behind as trace–enables them to prioritize materials that other ants have previously been in contact with. They are thus more inclined to pick up and carry such materials (this is the pick-up rule). These are then dropped off somewhere, preferably stacked on top of a pile. However, the pile’s height must not exceed that of the ant itself. If it does, the materials are placed somewhere else (this is the stacking rule). Together, these three simple building rules explain each ant’s behaviour, their thousands of workers’ organization, and the nest’s eventual physical structure. Not only do the building rules enable the original construction, they also ensure continuous rebuilding as the colony expands. That is how a set of designated heuristics allows the black garden ants to build and maintain their very own equivalents to the great pyramids. While the building rules are simple, their contextual function is rather advanced. It is reliant upon the evaporation rate of pheromones, which in- CHAPTER 3 25 turn is temperature-dependent. In warmer climates, pheromones evaporate faster. Without such traces, the picking-up rule becomes less effective. As a result, more pick-ups occur randomly. The pillars then become fewer with more space in between, and the nest’s volume consequently grows larger. Interestingly, these dynamics enable the ants to optimize the nest’s size relative its local climate. It indeed seems as if Simon (1996) was right to con- clude that while we–as organisms–are relatively simple, “[t]he apparent complexity of our behavior over time is largely a reflection of the complexity of the environment in which we find ourselves” (p. 53). Humans Humans use heuristics too. Keynes (1937) and Katona (1953) called them conventional judgments and habits respectively. But it would be a mathemati- cian, Pólya (1945), who introduced the first theory of heuristics. Their name comes from the Greek word for discovery (heuriskein). In a similar general sense, Pólya considered heuristics an umbrella term for applied problem-solving. Simon (1956) suggested more specific cognitive heuristics. Like recognition, heuristic search and pattern recognition. Next up were Tversky and Kahne- man. Their heuristics stemmed from psychological experiments, but remained somewhat vaguely operationalized, as Gigerenzer (1996) would stress in his ferocious critique (but see Tversky & Kahneman, 1996). Gigerenzer revisited Simon to say that heuristics constitute a toolbox of cognitive rules of thumb, which ignore part of the information at hand (Gigerenzer & Goldstein, 1996; Gigerenzer, Todd & the ABC Research Group, 1999). Such heuristics are typically sequential and involve a search rule, a stop rule and decision rule (Gigerenzer & Gaissmaier, 2011). A problem for the Gigerenzians is that there still is relatively little evidence that their heuristics are used in practise (Newell, 2005, 2011; Pohl, 2011). The views on heuristics have accordingly differed across the Kahne- man-Tverskian, Pólyan and Simonian-Gigerenzian schools of thought. Ta- ble 3 below sums up some of their key differences. 26 THE GOOD PLACE

Table 3. Different perspectives on heuristics.

Kahneman-Tverskian Pólyan Simonian-Gigerenzian Operationalization Weak Semi-strong Strong Mental origin Subconscious schemata Conscious tactics A cognitive toolbox accessed intuitively for logical problems of rules of thumb Key heuristics Availability Imitate Satisficing Representativeness Make a list Recognition Anchoring & adjustment Draw a picture Take-the-best Affect heuristic28 Guess & check Fluency Mental accounting Work backwards Pattern-recognition

Mini-case: Who Wants to Be a Millionaire For an example of how a heuristic of the Simonian-Gigerenzian kind may help to solve a large world task, consider the curious case of Rick Rosner. Back in year 2000, the then stripclub-bouncer (later Giga Society29 member) participated in the 124th televised episode of the American quiz show Who Wants to be a Millionaire. For USD 16,000 he was asked the following question: Which capital is located at the highest altitude above sea level?

A. Mexico City B. Quito C. Bogota D. Kathmandu

The question is tricky. Rosner would even take it to court–literarily–as the alternatives did not include La Paz, the Bolivian administrative capital, whose altitude otherwise would have been the highest (the right answer now reads Quito). But what his lawsuit failed to address was why he got it wrong. Cognitively, it happened not once but twice. Facing this type of task, you either know the answer or not. Following Gigerenzer, Hoffrage and Kleinbölting (1991), in case you do know it, this

28 See Finucane, Alhakami, Slovic and Johnson (2000); Peters, Västfjäll, Gärling and Slovic (2006). 29 A club exclusive for those with a one-in-a-billion IQ. Rosner’s is 192. CHAPTER 3 27 comes from evoking a local mental model (local MM). A local MM links to semantic information stored in long-term memory. Like when schoolchil- dren recapitulate what they have learned in, say, history class. In the present case, a local MM would either confirm one of the alternatives as the world’s highest-located capital, or first know each altitude individually and next compare these among the options. When you do not know the answer, there are probabilistic mental models (PMM) instead. These may trigger an intuitive response or require conscious cognitive processing. To illustrate how a PMM works, imagine the question had read: Which out of the following mountains is the highest?

A. Mont Blanc B. Barre des Ecrins C. Grand Casse D. Mount Teide

Alternatives A through C are the top-summits in France while D represents Spain’s highest mountain. The former three are all located in the Alps; the last is a volcano on Tenerife. Their respective elevations range between roughly 3,700 and 4,800 metres. Although few are aware of these facts, most still find this question easier than Rosner’s. Why? Because not only is Mont Blanc the highest among them, it is also by far the most recognized. Consciously or not, we realize that familiarity and elevation are positively correlated in this category. We know of Mount Blanc because it is a very high mountain. Familiar mountains are generally higher than those less recognised. This kind of inference comes from evoking a PMM. It will not yield a certain conclusion, but a qualified guess. PMMs utilize cues in the environment. Recognition is perhaps the most prominent thereof (Gigerenzer & Goldstein, 2011). 30 For example, Hoffrage (2011) ran experiments in which participants made pairwise choices such as; which city has the largest population: Detroit or Milwau- kee? To his surprise, Germans found it easier to predict the relative sizes of American than domestic cities, and vice versa. The less his subjects knew,

30 But see Newell (2005, 2011) and Pohl (2011). 28 THE GOOD PLACE the better their judgments became. The explanation is that recognition only worked for the less knowledgeable group. Those who actually knew of both cities could not go by familiarity and performed worse as a result. Returning to Rosner, he was unlucky in that he happened to face an unconventional pairing of criterion (elevation) and reference class (world capi- tals). Although the recognition heuristic is good with city populations (Gigerenzer & Goldstein, 1996; Goldstein & Gigerenzer, 1999, 2002), it is less accurate for elevation.31 So Rosner got thrown off the hot seat because of a mismatch between the task at hand and not one but two mental models. He first failed his question the local MM, and then the PMM, way. For all of his lifelines, what he really would have needed was a good heuristic. Rosner’s case can also be looked at from a Kahneman-Tverskian perspective. His eventual guess was Kathmandu. Perhaps he used, say, the representativeness heuristic to reach the wrong conclusion? Well, maybe he did, maybe he did not. We cannot really know. The representativeness heuristic means “one evaluates subjective probability by the degree of correspondence between the sample and the population, or between an occurrence and a model” (Kahneman & Tversky, 1972, p. 451). That is, start with an unknown selection of features–each given some unknown weight–of what you see. Next compare them to an equally unknown selection of features–each given some unknown weight–of what you had imagined. And then rate their consistency on an unknown scale. Subconsciously. There are too many unknowns to solve such an equation. Was it something about the country (Nepal, Mexico, Ecuador, Colombia), the weather (snow, heat, rain, constant spring), or the proximity to a very high mountain (Mount Everest, Pico de Orizaba, Chimborazo, Pico Cristóbal Colón)? And against which prototypical high-altitude city was the comparison made? Surely not the Peruvian town La Rinconada–the highest located set- tlement in the world–as it is practically unheard of. The representativeness heuristic has intuitive appeal, but is not quite precise enough.

31 Recognition helps you pick the winners in Wimbledon matches (Serwe & Frings, 2006) but unfortunately, contrary to suggestions (Borges, Goldstein, Ortmann & Gigerenzer, 1999), it cannot help you pick the right stocks (Andersson & Rakow, 2007). CHAPTER 3 29

  

The Gigerenzians argue the only way to tackle heuristics is through precise operationalizations. For example, I would never open an account in an unknown bank. Now compare inferences A and B below, both of which could have preceded that decision. Then inference A (from the recognition heuristic) is preferable to B (from the representativeness heuristic) because it is more parsimonious, falsifiable, and does not lead to circular reasoning.

A. I will not put my money in a bank whose name I do not recognize

B. I will not put my money in a bank whose characteristics are not representative of my cognitive representation of a trustworthy bank

Analogies

The controversy over heuristics reflects conflicting analogies of man. Man- as-computer versus man-as-organism (Almqvist, 2018a). Man-as-computer is a computeresque model of the mind. It contains the old idea from psychophysics that the mind is a receptive apparatus that perceives probabilities as if these were sensory stimuli like sound or light (see Preston & Barrata, 1948). Prospect theory (Kahneman & Tversky, 1979; Tversky & Kahneman, 1992) is the most renowned theory of this type (but see Millroth, Nilsson & Juslin, 2019). The man-as-computer analogy also includes the dual-process view, according to which the human brain processes information in distinctly different ways (Evans & Stanovich, 2013; but see Keren, 2013; Kruglanski & Gigerenzer, 2011; Osman, 2004). In an intuitive System 1 or an analytical System 2. The former has been portrayed as flawed software, susceptible to fallacies (see Kahneman, 2003). Man-as-computer is said to compensate for the much-too-simple neoclassical model of man (The Committee for the Prize in Economic Sciences in Memory of Alfred Nobel, 2017). But this reasoning is fragile. In fact, the opposite objection–over-complexity–has been raised against neoclassical economics (Selten, 2001), BE (Berg & Gigerenzer, 2010) and prospect theory (Brandstätter, Gigerenzer & Hertwig, 2006) alike. Neither Simon (1956) 30 THE GOOD PLACE called for complexity. On the contrary, he asked: “How simple a set of choice mechanisms can we postulate and still obtain the gross features of observed adaptive choice behavior?” (p. 129, emphasis added). The alternative analogy is man-as-organism (also see Fiedler & Juslin, 2006). It is compatible with procedural rationality. There, heuristics belong to an adaptive toolbox–whose effectiveness depends on their fit with the environment (Gigerenzer & Goldstein, 1996; Gigerenzer, Todd & the ABC research group, 1999; Gigerenzer & Selten, 2001). Studies of procedural rationality–how heuristics meet correspondence or pragmatic criteria–have been labelled ecological rationality (Todd & Brighton, 2016). The partisan divide between the two camps is evident. Although both are indebted to psychology’s cognitive revolution, they are different schools of thought (Fiedler & von Sydow, 2015; Heukelom, 2005). Kahneman was trained in perception-research, Gestalt psychology and psychophysics, which in-turn influenced how he and Tversky modeled human judgment. The mind became an information-processing system, whose circuits and software it was natural to problem-search for biases. Gigerenzer, on the other hand, was always more influenced by Simon, Brunswick and others who subscribed to an ecological analogy and empha- sized representative design (see Brunswik, 1955; Dhami, Hertwig & Hoffrage, 2004; Hammond, 1954; Hoffrage & Hertwig, 2006). Cognition then becomes domain-specific and adaptive (Tooby & Cosmides, 1992, 2005).

Table 4. Conflicting analogies. Man-as-computer versus man-as-organism.

Man-as-computer Man-as-organism Spokesperson(s) Kahneman & Tversky Gigerenzer Emphasis Mind Environment Setting Small world Large world Design Systematic Representative Paradigm Bounded rationality Procedural rationality32 Heuristics Error-prone Fast and frugal

32 Although Gigerenzer technically defines his work as a research program on bounded rationality. CHAPTER 3 31

Kahneman and Tversky (1972) originally set as goals to study how people evaluate uncertainty, but also “the conditions under which the various heuristics work” (p. 452). The worldview argued for in this chapter thinks of heuristics and the environment as two sides of the same coin. Although they can be studied individually, what matters most is their interaction. Rudimentary as it may be, my worldview enables understanding normative, descriptive or prescriptive facets of JDM without having to make unrealistic assumptions about the environment (small worlds). Instead, in line with the man-as-organism analogy, it proposes that humans evaluate uncertainty from cues in their environments (large worlds) using heuristics (procedural rationality). It is a straightforward attempt to capture how people reason when economic theory does not apply.

  

If choice architects’ ideology does not hold water, their programmatic am- bitions suffer. Simon (1985) concluded, “there is plenty of evidence that people are generally quite rational; that is to say, they usually have reasons for what they do” (p. 297). Of course, that does not imply that all of people’s decisions will be good, all the time (there is plenty of evidence of the opposite too). Nor that nudges could not be effective. But if we seriously consider a universe in which it is very difficult to identify optimal courses of action before hand, then arguing your right to overrule others’ decisions–whatever their reasons might be–becomes a much harder sell.

Chapter 4

Previous research

This chapter provides traditional reviews of research in three domains. It begins with citizens’ approval of state nudging, continues with alleged biases among horserace bettors, and ends with big nudging in the digital era.

Carrots, sticks, sermons, and nudges

With societal life comes both authority and liberty (Mill, 1860). Traditional- ly, governments use three types of tools to influence citizens: carrots, sticks and sermons. Carrots mean economic incentives. Sticks are mandates. Ser- mons refer to information (Bemelmans-Videc, Rist & Vedung, 2011). These days, there are also nudges (OECD, 2017; Whitehead et al., 2014). While there is some research on how carrots and nudges do in terms of relative effectiveness (Benartzi et al., 2017), less is known about how carrots, sticks, sermons and nudges compare when it comes to public support.

The public support for nudging

Former British Prime Minister David Cameron liked nudging (Halpern, 2015). As did former U.S. President Barack Obama. He implemented it throughout public administration and hired Sunstein for a job in government. What Obama’s successor, Donald Trump, thinks about nudging is unknown. Back in 2016, I asked the then President elect what he thought about Executive Order 13707, in which Obama had told federal agencies to nudge. He never answered. Instead his staff mistook me for a supporter, wherefore I strangely now subscribe to the Trump Make America Great Again 34 THE GOOD PLACE

Committee’s newsletter. Since President Trump’s inauguration, I have been receiving e-mails on an almost daily basis. They often involve nudging. There are defaults for how much money to donate, and the choice set is structured to make certain options look better than others. To boost voter- turnout in the midterm elections, recipients were asked to sign a pledge months in advance, promising to vote republican. Still convinced that state nudging is something we should implement without a thorough debate? Me neither. The academic debate on state nudging is heating up (Meder, Fleischhut & Osman, 2018) and the political debate may just be getting started. For example, it has long been suggested that changing from an opt-in to an opt- out rule may increase organ donations (Johnson & Goldstein, 2003).33 But like in all public policy, effectiveness is not all there is. When Germany considered it in 2018, Peter Dabrock, Chairman of Germany's Ethics Council, rejected the idea, calling it “compulsory organ contribution”. Between-nudge comparisons Nudges differ in popularity depending on their purposes (Hagman, Anders- son, Västfjäll & Tinghög, 2015), perceived effectiveness (Djupegot & Han- sen, 2020; Reynolds et al., 2019), transparency (Loewenstein, Bryce, Hagmann & Rajpal, 2015), intrusiveness (Jung & Mellers, 2016), and who their architects are (Tannenbaum, Fox & Rogers, 2017). If nudges come across as illicit or radical, they also receive lower support (Sunstein, 2015c). There can be large differences in public support across nudges. For example, 9 out of 10 Germans would support a public education campaign, aimed at parents, to fight child obesity. However, only 1 in 4 would support a default donation (subject to opt-out) to charity upon tax return (Reisch & Sunstein, 2016). Table 5 below exemplifies more differences in support, now in the U.S. and for a different set of nudges. Their public support range between 20 % (using an optical illusion to speed down traffic) and 83 % (providing better information to retirement savers).

33 Which, however, is not necessarily the case (see Noyes et al., 2019). CHAPTER 4 35

Table 5. The public support for nudges in the U.S. (Jung & Mellers, 2016).

Nudge Support /opposition (%) Retirement programs must provide customers with information about their 83/8 projected monthly income at specified ages of retirement Credit card companies must provide customers with spending alerts if 71/16 they are close to a spending limit (via mail, email, or text message) Notifications to voters by mail, email or text messages right before elec- 60/23 tions, to tell them exactly how to get to the polls Default displays in grocery stores that make healthy foods especially 58/31 conspicuous and easier to reach Default food orderings in school cafeterias, with salads and lower calorie 58/25 foods coming first, to promote healthy choices Default settings on social media to post information and photos to friends 58/20 and not the public at large (unless people choose to opt out) Automatic enrollment into a retirement savings plan (unless employees 54/23 choose to opt out) Use of graphic warnings with photographs of the effects of smoking on 48/32 cigarette packages Automatic enrollment into a medical plan with basic coverage at col- 38/41 leges and universities (unless the students choose to opt out) Default that people obtaining drivers licenses will become organ donors 33/51 under hopeless medical conditions (unless they opt out) Use of one-click opportunities to solicit charitable donations when check- 28/61 ing out at the grocery store Government-based websites that allow people to track their energy us- 26/51 age, credit card bills, health care expenditures and cell phone bills Use of increasingly narrower white lines on roadways that create the 20/54 visual illusions of speeding up to control vehicle speeding

Between-country comparisons There have been a handful surveys on the public support for nudging internationally. They cover the U.S. (Jung & Mellers, 2016), Sweden (Hagman et al., 2015; Hagman, Erlandsson, Dickert, Tinghög & Västfjäll, 2019) and other parts of Europe (Reisch & Sunstein, 2016; Reisch, Sunstein & Gwozdz, 2017) and the world (Sunstein, Reisch & Rauber, 2017). Figure 3 below summarizes the average public support for nudging in China, Brazil, South Africa, South Korea, Russia, Australia, Italy, the U.K., Canada, France, Germany, the U.S., Japan, Hungary and Denmark. 36 THE GOOD PLACE

Figure 3. International comparison. Mean percentage support for 15 nudges across 15 countries (Reisch & Sunstein, 2016; Sunstein, Reisch & Rauber, 2017).

100 95 90 85 80 75 70 65 Average support (%) 60 55 50 UK Italy USA Brazil China Japan Russia France Canada Hungary Australia Denmark Germany South Africa South Korea

Sunstein, Reisch and Rauber (2018) find three levels of nudge support internationally: overwhelmingly pro-nudge nations (China and South Korea), principled pro-nudge nations (Brazil, South Africa, Russia, Australia, Italy, the U.K., Canada, France, Germany and the U.S.) and cautiously pro-nudge nations (Japan, Hungary and Denmark). The national support for nudging may be affected by the domestic political debate (Sunstein, Reisch & Rauber, 2017) and trust in public institu- tions (Sunstein, Reisch & Kaiser, 2019). Other than that, prior studies give little guidance as to why the support for nudging differs across countries. No one can really explain why, say, Americans seem to like nudges more than Danes. How Sweden compares to the above countries, using the same questionnaire, and consequently which category Sweden belongs to, remains unknown. Beforehand, it is difficult to know what to expect. On the one hand, one could speculate Sweden’s political heritage would increase the tolerance for state paternalism, as supporters of big government tend to be more pro-nudging (Pedersen, Koch & Nafziger, 2014). On the other hand, Swedes are individualistic, which is negatively correlated with the support for nudging (Hagman et al., 2015; Jung & Mellers, 2016). CHAPTER 4 37

Between-people comparisons The support for a nudge is generally higher when people sympathize with its goal or identify with the choice architect (Tannenbaum, Fox & Rogers, 2017). Certain traits, attitudes and political opinions also explain the support for nudging (Hagman et al., 2015; Jung & Mellers, 2016). As do moral intuitions (Nilsson, Erlandsson, Västfjäll & Tinghög, 2020). For their M.Sc. thesis, two of my students–Nudel and Wiik (2017)–looked into the support for nudging. They discovered that the support for nudging was driven by cultural cognition (see Kahan, 2012) in egalitarianism and collectivism and mediated by opinions about the nudge’s effectiveness and purpose. Interestingly, previous research has failed to establish a conclusive relationship between political party preferences and the support for nudging. In some studies, party preferences have been treated as categorical variables (Reisch & Sunstein, 2016). Alternatively, they have been complemented by a self-assessment of political ideology on a Likert type scale ranging from “very liberal” to “very conservative” (Sunstein, Reisch & Rauber, 2018). A different approach–that I prefer–is to go by the political parties’ positions in two dimensions. The first dimension is economic: the left-right spectrum. Sweden is suitable for such an analysis as its political landscape is one of the most left-right oriented in the world (Oscarsson & Holmberg, 2016). The second dimension is cultural: the GAL-TAN scale. It contrasts green, alternative and libertarian (GAL) values with traditional, authoritari- an and nationalistic (TAN) ones (Hooghe, Marks & Wilson, 2002).

  

Articles 1 and 2 investigate the Swedish public support for nudging and its relationship to socio-demographics, political party preferences and ideological views. Article 1 compares Sweden with the rest of the world, asking the same questions as in previous surveys internationally. Article 2 is then one of the first studies to measure the support for nudging against relevant benchmarks, namely the traditional tools of government: carrots, sticks, and sermons (see Ly & Soman, 2013). 38 THE GOOD PLACE

Favorite-longshot bias

Horse lotteries are different from lotteries of the casino type (Anscombe & Aumann, 1963). They occur in a large world. More than a million Swedes bet the horses a regular year, gross (net) spending some SEK 14 (4) billion in the process. In doing this they also fall victims to an alleged bias. The favorite-longshot bias (FLB) stipulates the market generally underbets favorites (likely winners) and overbets longshots (unlikely winners) in pari-mutuel betting on horseracing. For Win bets, Thaler (2015) suggests the returns of bets on favourites and longshots respectively differ by a factor of six.34

Background

The pari-mutuel (tote) system was first introduced in Belle Époque-Paris (Canymeres, 1946) and to this day constitutes the typical organization for betting on horse racing. The tote has circumvented middlemen (bookmak- ers) to enable direct competition among bettors. In the pari-mutuel, total bets sum to the bet pool. A fixed proportion of the bet pool, the track take, is then deducted in order to compensate the betting-organizer. What remains post track take is divided among the winning bettors, proportionate to their respective bets (see Ali, 1979; Eisenberg & Gale, 1959). Griffith (1949) pioneered psychological studies at the racetrack. As he considered it a market for probabilistic predictions, his primary interest was whether the estimates implicit in the tote odds were consistent with the long-run relative frequencies of wins. Calibration between subjective probabilities (bet fractions) and objective probabilities (win fractions) operationalized bettors’ accuracy in probabilistic inference. His data confirmed bet fractions did predict outcomes, but they also revealed an anomaly. Favor- ites tended to be under-bet and longshots over-bet. That anomaly is FLB. Meaning “the expected returns per dollar bet increase monotonically with the probability of the horse winning” (Thaler & Ziemba, 1988, p. 163).35

34 A Win bet refers to a single event (picking the winner) and thus belongs to the simpler bets. 35 But see Bjorkman and Bukszar (2003); Busche and Walls (2000). CHAPTER 4 39

Griffith assumed betting would “touch upon fundamental philosophies of probability and mathematical expectation” (1949, p. 293) wherefore his aim was to discover a general psychology of probabilities. He related FLB to the psychophysical probability function proposed by Preston and Baratta (1948), who in-turn had drawn heavily on an analogy to visual perception. Griffith would inspire more research over the next several decades. Both replications of FLB (Ali, 1977; Fabricand, 1965; Griffith, 1961; McGlothlin, 1956; Rossett, 1965; Snyder, 1978; Weitzman, 1965) and statistical modeling of race outcomes (Ali, 1977, 1979, 1998; Bacon-Shone, Busche & Lo, 1992a, 1992b; Harville, 1973; Isaacs, 1953; Lo, 1992; Lo & Bacon-Shone, 1994; Lo, Bacon-Shone & Busche, 1995; Stern, 1990). There are two schools of thought to explain FLB.36 The first comes from BE. Its theories have included cognitive biases (Griffith, 1949, 1961; Metzger, 1985; McGlothlin, 1956), mental accounting (Thaler & Ziemba, 1988) and bragging rights (Canfield, Fauman, & Ziemba, 1987). The second school of thought stems from microeconomics. It advocates neoclassical utility models (Ali, 1977; Friedman & Savage, 1948; Jullien, & Salanié, 2000, 2008; Kanto, Rosenqvist & Suvas, 1992; Mayer, 1989; Rossett, 1966; Quandt, 1986; Snowberg & Wolfers, 2010; Weitzman 1965) or informa- tional economics (Asch, Malkiel & Quandt, 1982; Asch, Malkiel & Quandt, 1986; Asch & Quandt, 1987; Figlewski, 1979; Hurley & McDonough, 1995; Ottaviani & Sørensen, 2008, 2009, 2010, Plott, Wit & Yang, 2003; Sobel & Travis Raines, 2003). There have also been attempts to combine the two (Coleman, 2004).

[FLB] is consistent with two bettor populations. One is risk-averse, knowledgeable about winners, backs favourites, believes in the gambler's fallacy, and has a positive expected return. The other, a larger group is risk loving, backs longshots, believes in hot hands, and has a significant, negative expected return (Coleman, 2004, p. 315).

36 For different approaches, see Camerer (1998) and Sung, Johnson and Dror (2009). 40 THE GOOD PLACE

The stock market analogy “Metaphor and analogy can be helpful, or they can be misleading. All depends on whether the similarities the metaphor captures are significant or superficial” (Simon, 1962, p. 467). By the early 1980s, the FLB program had settled for the analogy of an “archetypical securities market” (Hausch, Ziemba & Rubinstein, 1981, p. 1436). Placing a bet was likened to stock valuation (Ali, 1979; Quandt, 1986) or option trading (Hodges, Tompkins & Ziemba, 2003). Some saw similarities between handicappers at the races and brokerages at the stock market (Losey & Talbott, 1980). Racetracks were said to provide “rich experimental settings for testing asset pricing theories” (Gulati & Shetty, 2007, p. 49) and “an excellent forum for a psychological investigation of investor risk behavior” (Hausch, Lo & Ziemba, 2008, p. 7). Bettors were thought to apply the same methods as traders: seeking arbitrages (Hausch & Ziemba, 1990) and undertaking fundamental and technical analyses (Hausch, Ziemba & Rubinstein 1981).

A bettor using a fundamental or handicapping strategy attempts to determine which horses, if any, have probabilities of winning (or placing or whatever) that exceed the market-determined odds by an amount sufficient to overcome the track take. Technical systems require less information and use only current betting data. Bettors using a technical system attempt to find inefficiencies in the market and bet on such ‘overlays’ when they have positive expected value (Thaler & Ziemba, 1988, pp. 164).

Eisenberg and Gale (1959) described the fundamental analysis, where “each [bettor], after careful study of the form sheets, the condition of the track, and other relevant material, has arrived at an estimate of the relative merits of each of the [horses] which he expresses in quantitative terms” (p. 165). While early racetrack betting was isolated to individual races and single or disjunctive events, the market has gradually come to involve more exotic (complex) bets on conjunctive events and multiple races. As a consequence, the modern racetrack features bets where both fundamental and technical analysis realistically would be impossible. Consider the sophistication of an algorithm to use technical analysis for a common exotic bet like the Trifecta (pick the top-3 horses, in exact order) in a race of n starters. See Table 6. CHAPTER 4 41

Table 6. 10 step algorithm to bet the Trifecta (technical strategy).

Step For bettor to do (in real-time) 1 Collect Win odds for all n horses from the tote board37 2 Add back track take 3 Inverse adjusted odds to derive implicit Win probabilities 4 Identify all n*(n-1)*(n-2) Trifecta permutations 5 Apply the Harville (1973) formula to 4 to estimate probabilities 6 Collect Trifecta odds for all permutations38 7 Divide 6 by 5 to construct odds-probability ratios 8 Maximize 7 with regard to expected value 9 Consider the Kelly (1956) criterion on 8 to decide optimal bet size(s) 10 Place bet(s)

To see bettors as traders also implies that they are out to beat the market. This is of course true in the sense that betting is fueled by hopes of winning. However, as the track take vastly exceeds the transaction costs of the stock market (track takes in Sweden range between 15% and 35%), betting is first and foremost a recreational activity whose expected value is known to be negative. Not the type of market where arbitragers come to realize free lunches. Most bettors use the market rather than try to outsmart it. Finding the most likely winner implies betting with, not against, it.

Mini-case: The Elit race and Sweden Cup

The FLB program assumes that FLB is caused by the bettors’ dispositions and thus stable. It supposedly applies to exotic bets (Koivuranta & Korho- nen, 2018). The following mini-case will problematize those claims, using another exotic bet: the Quinella (pick the top-2 finishers, in any order). The Elit race is among the most prestigious trotting races in the world. It is arranged on the last Sunday of May at Solvalla racetrack, outside of Stockholm. The day before, on the Saturday, Solvalla also hosts races. One

37 Data from the (separate) Win pool, and will be subject to constant changes. 38 Data from the (separate) Trifecta pool, and will be subject to constant changes. 42 THE GOOD PLACE of them is the Sweden Cup, often referred to as the mini-Elit race. It takes place at the same track, features the same number of horses (8), is the same distance (1,609 meters) and attracts largely the same attendants as its prestigious big brother. If FLB is stable for exotic bets, then it should be found also for the Quinella (the single-race exotic bet with the highest turnover) and, of course, be reproducible from one day (the Saturday) to the other (the Sunday). Let us see whether that is actually so. In a race of 8 starters, there are 28 possible Quinella combinations. Comparing the Quinella odds with their respective probabilities implicit in the Win odds can be used as an approximation of FLB. The Harville (1973) formula enables this calculation across all combinations.39 FLB says bets’ expected return (utility) decrease (increase) as a function of odds. To test this, I collected the aggregates of all Quinella bets and Win bets on Sweden Cup and the Elit Race in 2017 (turnover: SEK 3,678,984) from the bookmaker’s web page. The Quinella odds and the implied probabilities from the Win odds could then be compared. The spread between the two would be FLB (assuming calibrated Win odds). Again, if FLB was to be found on the Saturday, it should manifest again on the Sunday. Figure 4 below presents FLB for the Quinella betting on the Elit race and Sweden Cup. Similarly to Friedman and Savage (1948), it derives a utility function from the expected return of bets at different probabilities. Here by comparing the actual Quinella odds and the implied Quinella odds from the Win pool. Local risk aversion means giving up expected net return (ex- tracting utility) at a certain probability level. Local risk seeking is the opposite. Risk neutrality implies no relationship between utility and probability. The graph’s upper area is the domain of risk aversion. Its lower area is the domain of risk seeking. Utility is measured as the difference in expected net return. The horizontal line equals risk neutrality. Critically–contrary to what the FLB program postulates–the two utility functions are not the same. They are mirrored. Separated only by a day, the bettors behave as complete opposites of each other. So FLB is clearly unstable. How can this be? Surely people cannot see their utility functions change over night?

39 The Harville formula is not perfect, but it remains a very good approximation (Ali, 1998). CHAPTER 4 43

Figure 4. FLB for all Quinella bets on the Elit race and Sweden Cup in 2017.

Sweden Cup Risk neutrality

Utility Elit race

0 0,01 0,02 0,03 0,04 0,05 0,06 0,07 0,08 0,09 0,1 Probability

In fact, the reason is very simple. Like on any exotic bet, Quinella bettors typically submit several combinations. In doing this, they stick to quite similar budgets and roughly the same numbers of combinations every time. All else being equal, keeping system-sizes fixed, the probability distribution of the race then determines market efficiency. An unusually uneven race would see bettors under-bet short odds. An unusually even race would conversely see them under-bet long odds. The bias could be predicted and explained by how (un)even the race is. Not the bettors’ disposition. Indeed, that was the case in 2017. The Elit race featured Bold Eagle, one of the heaviest favourites ever, assigned a 75 % chance in the Win pool. Sweden Cup, in contrast, was remarkably open, with 4 horses having between 17 and 22 % chance of winning, again according to the Win pool. In the first instance, FLB was found because the systems, on aggregate, included too many combinations. In the second instance, FLB reversed because the systems were generally too small.

  

BE has taken deviations from strong market efficiency as evidence that bettors would be irrational. But that does not have to be the case. Article 3 investigates FLB; beginning with Win bets, and then studying some of the most difficult exotic bets there are. 44 THE GOOD PLACE

Big nudging

Big Data is a buzzword for especially large, fragmented datasets processed in real-time (Laney, 2001). It is a natural consequence of digitalization, destined to transform marketing (Erevelles, Fukawa & Swayne, 2016). Accom- panying it is a methodological paradigm, featuring new computational models (Gandomi & Haider, 2015; Mayer-Schönberger & Cukier, 2013). Alarmists warn that Big Data will lead to exploitation of the masses (Puaschunder, 2017) and replace democracy (Helbing et al., 2017) with totalitarian control through social media. Big nudging. Affecting consumption and general elections alike. But how, exactly, would it work? There are two ways for Big Data to turn into big nudging. One is if Big Data allows algorithms to figure new things out (before people are nudged into doing what they suggest). Like Alexa, Amazon’s virtual assistant, does. The other is if Big Data enables choice architects to learn how to increase the susceptibility to nudging. Like Cambridge Analytica, the infamous political consulting firm, tried to do. There are still relatively few academic studies on digital nudging. But those to date do not indicate that it would be any more effective than conventional one (Hummel & Maedche, 2019).

Man versus machine

If computers have already learned how to beat man in chess (IBM’s Deep blue), poker (Facebook’s Pluribus), Jeopardy (IBM’s Watson), go (Google’s AlphaGo) and Quake III (Jaderberg et al., 2019)–then surely, as Thaler and Kahneman have said, all that remains is to outsource as many decisions as possible to algorithms, lean back and reap the rewards? Well, not quite. Because–unlike chess, poker, Jeopardy, go and Quake III–the human universe mostly consists of large worlds, where outcomes are uncertain. Situations that human decision makers find difficult are often difficult for algorithms too (Rich & Gureckis, 2019). Not for lack of sophistication, but because there is not enough reliable data to learn from. So- called wicked learning environments (Hogarth, 2001). CHAPTER 4 45

An old misperception in forecasting was that advanced prediction models always outperformed simple ones. Big Data has seen it resurface.

[T]his situation is the same with what was happening in statistical literature in the late 1970s and 1980s. At that time, it was thought that forecasting methods were of superior accuracy simply because of their sophistication and their mathematical elegance (Makridakis, Spiliotis & Assimakopoulos, 2018a).

In reality, they often do not (Makridakis, Hogarth & Gaba, 2010). General- ly, “[s]tatistically sophisticated or complex methods do not necessarily provide more accurate forecasts than simpler ones” (Makridakis & Hibon, 2000, p. 452). In my work, I have had the privilege to interview profession- als in several applications of forecasting. Experts who predict the weather for the Swedish meteorological and Hydrological Institute (SMHI), model the economic development for Sweden’s central bank (Sveriges Riksbank), or pick winners at the racetrack. They would all agree with Albert Einstein’s alleged saying that everything should be made as simple as possible, but not simpler.40 Using prediction models that balance simplicity and complexity. Before Big Data, businesses went by heuristics (Artinger, Petersen, Gigerenzer & Weibler, 2015). And heuristics must not be underestimated just because they are simple. They may outperform more refined alternatives (Goldstein & Gigerenzer, 2009). For example in predicting repurchas- es among digital customers in retail (Wübben & Wangenheimat, 2008). Big Data and AI leverage on economies of scale (Agrawal, Gans & Goldfarb, 2018) but will still not have to spell the end for managerial heuristics. Already Einhorn and Hogarth (1981) questioned whether there would ever come a day when human decision making became redundant: “Given the complexity of the environment, it is uncertain whether human responses or optimal models are more appropriate. Furthermore, we know of no theory or set of principles that would resolve this issue” (p. 59). As a case in point, consider Goldman Sachs’ attempt to predict the outcome of 2018’s football World Cup in Russia. Leading up to the tour-

40 What Einstein actually said was this: “It can scarcely be denied that the supreme goal of all theory is to make the irreducible basic elements as simple and as few as possible without having to surrender the adequate representation of a single datum of experience”. 46 THE GOOD PLACE nament, the bank proudly announced that had put its most refined prediction model through millions of iterations to come up with the most likely outcome of the competition. It expected Brazil to defeat Germany in the final. In a client note, its international research team reportedly said they were “drawn to machine learning models because they can sift through a large number of possible explanatory variables to produce more accurate forecasts than conventional alternatives”. They assumed, a priori, that higher model complexity (more variables) would guarantee better forecasts. So how did they do? Poorly. Not only had both their anticipated final- ists been eliminated by the semi-finals (Germany already at the group stag- es), their predictions for the elimination stage had also been subpar. Out of the 16 (from 32) teams to advance, they only got 11 correct. In comparison, when my colleague and I surveyed our students at the course The psychology of judgment and prediction, and asked the same question to them and their friends (n = 33), several of which admitted knowing nothing about football, they still managed to get 15 out of 16 right. Some of them simply went by country familiarity or teams they recognized as coming from “footballing nations”. A sign test confirmed that the students outperformed the bank’s AI by a statistically significant margin (z(16) = 2.00, p = .046). With the semi-finals coming up and its original bets already lost, Gold- man Sachs decided to give it one last go. Blaming tournament surprises (never the prediction model) for their performance, they now expected Belgium and England to win their respective games against France and Croatia, with Belgium to emerge victorious from the final. How it turned out this time around? France won the title, beating Croatia in the final.

  

To unlock Big Data’s potential, or assess big nudging’s threat, the focus must change from the data themselves to the performance of whatever model–computerized or not–trying to predict something in an actual environment. Article 4 provides a non-technical illustration of mathematical- statistical and methodological issues central to this and provides a 7-point checklist for businesses to leverage on Big Data for predictive analytics.

Chapter 5

Methodology

In this chapter, I highlight some methodological fundamentals. For complete details, see the respective articles.

Methods

Secondary data analyses

For Article 3, a medium-sized dataset (N= 185,198) of betting odds and results over five years (2011-2017) was obtained from the head of a database originally compiled for commercial reasons. The data analyses and computer simulations were all done using Microsoft Excel. Article 4 included analyses of two large datasets (each N= 1,533,695) of temperature forecasts and observations across 26 Swedish weather stations (1994-2017). SMHI provided the data. SMHI also helped with the most precise calculations (to prevent rounding errors). The remaining analyses were done in Microsoft Excel.

Survey designs

Articles 1 and 2 were survey-based. In both cases, the target population was adult Swedes. Data were collected through survey companies and the sampling frames consisted of their web panellists. Working with survey companies helped with recruiting and data-collection while enabling well-powered surveys with representative samples. 48 THE GOOD PLACE

Table 7. Survey respondents. Selected sample frequencies (%).

Survey A (N=1,032) Survey B (N=5,916) Survey C (N=595) Gender Men 521 (50) 3054 (52) 307 (52) Women 511 (50) 2856 (48) 280 (47)

Age (years) 18 - 29 219 (21) 1054 (18) 66 (13) 30 - 49 358 (35) 1708 (29) 213 (36) 50 - 64 251 (24) 1472 (25) 156 (26) 65 - 89a 204 (20) 1677 (28) 67 (13)

Education Primary 187 (18) 738 (12) 57 (10) Secondary 454 (44) 2169 (37) 210 (35) Higher 369 (36) 2834 (48) 328 (55)

Living area Largest citiesb 380 (37) 1647 (28) 134 (23) Other 651 (63) 4269 (72) 462 (78)

Political camp Left (S+MP+V) 303 (29) 1965 (33) N/A Centre-right (M+KD+C+L) 337 (33) 2002 (34) N/A Sweden democrats (SD) 127 (12) 1078 (18) N/A a The upper age inclusion criteria were 79 years of age (survey A), 89 (B) and 70 (C) respectively. b Refers to Stockholm, Gothenburg or Malmo in survey A and B; but only Stockholm in survey C. Survey A The first survey was done together with the Swedish marketing research company Novus. It was a custom survey, featuring an adapted version of a 15-item questionnaire developed by Reisch and Sunstein (2016), found in table 8 below. Two of the original 15 nudges were excluded. One–using subliminal advertising–because it is illegal. The other–enforcing meat-free days in public canteens–because it is not a nudge but a mandate. The survey was a custom survey whose final sample included 1,032 respondents CHAPTER 5 49

(50 % female), representative of the target population. Controls included gender, age, education, occupation, job sector, marital status, household income, living-region, and political party preference. Respondents were asked whether or not they approved (yes/no) of the 13 nudges (see Table 8), presented in random order. Data-collection took place in early Decem- ber 2017. The subsequent data analysis was done in IBM SPSS.

Table 8. Selected survey A items. Adapted from Reisch and Sunstein (2016).

Item Nudge 1 Requiring calorie labels in chain restaurants 2 Requiring traffic light labels signalling healthiness of food 3 Encouraging defaulting customers into green energy providers 4 Requiring active choice on organ donation upon obtaining driver’s license 5 Requiring choice architecture for healthy food in grocery stores 6 Public education campaign with vivid pictures against distracted driving 7 Public education campaign for parents to fight childhood obesity 8 Default-charging airline customers a carbon emission compensation fee 9 Requiring warning labels on food with high salt content 10 Default to donate 50 Euro to the Red Cross upon tax return 11 Information campaigns in movie theatres against smoking and overeating 12 Requiring energy providers to default customers into green energy 13 Requiring sweet-free cashier zones in supermarkets

Survey B The second survey was administered by the Swedish market research company Demoskop. It was of the omnibus type, whereby extra questions were added to an already existing questionnaire. Its final sample contained 5,916 respondents (48 % female), representative of the target population. Con- trols included gender, age, education, income (absolute and self-assessed income category), living area and political party preference. Respondents were presented with a vignette, reading: “People’s behaviour can be influenced in several ways. How appropriate do you find it for 50 THE GOOD PLACE government agencies to use the following methods?” They then rated to what extent they approved of five different policy tools–three traditional (information, positive economic incentives, negative economic incentives) and two behavioural (nudging)–on a 4-point Likert scale, ranging from “completely inappropriate” to “very appropriate”. They were also asked within which, if any, from eight areas they approved of governmental influence–from environment issues down to family life–and if governmental information was appropriate in pursuing societal and individual goals respectively. Data-collection began in late February and was concluded by mid-March 2017. The dataset consisted of aggregate response data tabulat- ed by socio-demographics and political party preferences. Microsoft Excel was used for the data analyses. Survey C The third survey was supplied by the Swedish market research company Norstat. It was a custom survey. Its sample contained 595 respondents (47 % female). Controls included gender, age, education, occupation, income, marital status, and cultural cognition using Kahan’s (2012) 6-item questionnaire. Respondents rated their support for different tools of government: information, taxes, subsidies, nudges and mandates; in general and across five applications: flying, diet, green energy, mortgages, and COVID-19. See Table 9 below. Data collection took around one week and was concluded by April 6th 2020. The data analyses were performed in IBM SPSS and R.

CHAPTER 5 51

Table 9. Carrots, sticks, sermons, and nudges featuring in survey C.

Flying Sermon: Information about the flight industry’s climate impact upon booking a ticket Carrot: Subsidies on alternative, more environment-friendly modes of transportation Carrot: Tax-increase on airline tickets Nudge: Default climate compensation fee added to ticket price* Stick: Restricting the number of airline trips per household and year Diet Sermon: Education for parents on how to make healthier food for their kids Carrot: Child allowance complemented by healthy food-stamp Carrot: Tax-increase on snacks and fast-food Nudge: Grocery stores required to make unhealthy food and candy less visible Stick: Ban against fast-food and candy advertising Green energy Sermon: Information campaigns to promote green energy Carrot: Subsidies on green energy Carrot: Tax-increase on non-green energy Nudge: Default customers to green energy providers* Stick: Ban against non-green energy Mortgages Sermon: Educational campaigns on the benefits of amortization Carrot: Amortization made tax deductible Carrot: Amortization required for interest costs to be tax deductible Nudge: Retail banks recommends amortization to all customers by default* Stick: Mandatory amortization on all mortgages COVID-19 Sermon: Information web pages and apps for tracking the virus’ spread Carrot: Subsidies to small enterprises affected by the crisis Carrot: Fines on large social gatherings Nudge: Self-quarantining upon showing symptoms* Stick: Temporary city curfews *Subject to opt-out 52 THE GOOD PLACE

Data analyses

Many results in the articles involve sample proportions (p)̂ expressed as percentages (%) or decimals. They will interchangeably be reported with standard errors (SEs), relative standard errors (RSEs), confidence intervals (CIs) or tests of statistical significance. As a complement, Table 10 below contains 95% CI margins of error (+/-) across sample sizes (n) and values of p.̂ Some readers might find it useful later on.

Table 10. Margins of error (+/-), in %, at different values of n and p.̂ For example, the 95 % CI margin of error at p ̂ = .70 and n = 200 is +/- 6.35 %.*

p ̂ n .50 .55 .60 .65 .70 .75 .80 .85 .90 .95 50 13.86 13.79 13.58 13.22 12.70 12.00 11.09 9.90 8.32 6.04 100 9.80 9.75 9.60 9.35 8.98 8.49 7.84 7.00 5.88 4.27 200 6.93 6.89 6.79 6.61 6.35 6.00 5.54 4.95 4.16 3.02 400 4.90 4.88 4.80 4.67 4.49 4.24 3.92 3.50 2.94 2.14 800 3.46 3.45 3.39 3.31 3.18 3.00 2.77 2.47 2.08 1.51 1,600 2.45 2.44 2.40 2.34 2.25 2.12 1.96 1.75 1.47 1.07 3,200 1.73 1.72 1.70 1.65 1.59 1.50 1.39 1.24 1.04 0.76 6,400 1.23 1.22 1.20 1.17 1.12 1.06 0.98 0.87 0.74 0.53 12,800 0.87 0.86 0.85 0.83 0.79 0.75 0.69 0.62 0.52 0.38 25,600 0.61 0.61 0.60 0.58 0.56 0.53 0.49 0.44 0.37 0.27 51,200 0.43 0.43 0.42 0.41 0.40 0.38 0.35 0.31 0.26 0.19 102,400 0.31 0.30 0.30 0.29 0.28 0.27 0.25 0.22 0.18 0.13 204,800 0.22 0.22 0.21 0.21 0.20 0.19 0.17 0.15 0.13 0.09 *Change (Δ) in margin of error is inversely proportional to the square root of Δn. Generally, p ̂ and 1 - p ̂ have the same margins of error, so for p ̂ < .50 simply read the table as 1 - p ̂ instead.

CHAPTER 5 53

Limitations

In retrospect, there are several things I could have done differently. Be that case-selections, methods or analyses. For example, I never did any experiments or mixed methods. International generalizability suffers as my work is based entirely on Swedish data. For Articles 1 and 2, I used surveys. In the first case (survey A), my co- author and I adapted a questionnaire developed by Reisch and Sunstein (2016), which enabled us to do international comparisons. In the second case (surveys B and C), we developed two of our own. Operationalization is challenging in surveys on the public support for nudging. Only a small proportion of the general public is familiar with the concept, and the question- naires are typically constrained to certain policy areas, so the survey responses arguably often say more about the approval of that particular set of nudges and policy goals than the general policy tool as such. Survey B asked general questions while survey C featured specific nudges covering five different policy goals. A shortcoming of survey B was its reliance on aggregate data. On reflection, it was probably reasonable to do all three surveys. Together, their range mitigated some of their individual limitations. To evaluate predictions (Articles 3 and 4), I used secondary data. Dif- ferent, or more granular, data would have enabled some additional analyses.

Disclosures

Conflict of interest statement

I have no conflicts of interest to declare.

Data

Second party data have been used throughout. They have been microdata (Articles 1 and 4), aggregate data (Article 3) or a combination of both (Arti- cle 2). As for statistical disclosure control (SDC), a principles-based SDC has been used, with respondents’ confidentiality given highest priority.

Chapter 6

Abstracts

Article 1: Cautious support for nudging among Swedes in representative sample

Co-authored paper.41 Presented at SJDM 39 (New Orleans, USA). Condi- tionally accepted for publication in Behavioural Public Policy. International surveys indicate that citizens generally are positive towards state nudging. Less is known about differences in the support for nudging across socio-demographics and political party preferences, a research gap identified in the literature. This paper investigates these relation- ships through a population representative survey in Sweden. It also analyses differences in the support for nudging across political party preferences in two ideological dimensions: the economic left-right and cultural GAL-TAN spectra. Data were collected in December 2017 through a web survey adapted from Reisch and Sunstein (2016). The respondents (N= 1,032) were representative of the population ages 18-79 years with regards to gender, age, education, job sector, household income, living-region, and political party preferences. Sweden was found to belong to the cautiously pro- nudge nations (along with Japan, Hungary and Denmark), contrary to hy- potheses in previous research. Differences in the support for nudging were found along the economic left-right and GAL-TAN spectra. Nudges’ support, controversy and politicization are compared, analyzed and discussed.

41 Co-authored (equal contributions) with Associate Professor Patric Andersson (SSE). 56 THE GOOD PLACE

Article 2: Carrots, sticks, sermons or nudges? The Swedish public support for traditional and behavioral tools of government

Co-authored paper.42 Presented, in part, at SPUDM 25 (Haifa, Israel). Nudging has become common in public administration across the world. The rationale behind its implementation is that it can be an effective complement to the traditional means of governmental influence. Its effectiveness aside, relatively little is known about citizens’ opinions about nudging. Few studies have directly compared the support for nudging with that for traditional alternatives. This paper investigated how the support for nudging compared to that for the traditional tools of government: information, economic incentives, and mandates. Additional research questions included whether the support for nudging was affected by political ideologies and within which, if any, policy areas citizens approved of government influence. Study 1 consisted of responses to an omnibus survey involving 5,916 respondents. Recruiting and data-collection was undertaken in Feb- ruary-March 2017. Study 2 was a custom survey (N= 595) whose data were collected in March-April 2020. The samples were representative of the adult Swedish population when it comes to gender, age, education, income, living-area, occupation, marital status, and political party preferences. The respondents rated to which degree they approved of different tools of government, both in general and for specific policy goals. In line with prior research, the results revealed that the general public supported nudging. However, nudging did not compare favorably to the traditional tools of government. Moreover, the support for nudging was influenced by political ideology and party preferences. Individualists and those to the political left were more in favor of nudging than were collectivists, liberals and conservatives. Some policy areas were found more controversial than others when it came to governmental influence.

42 Co-authored (equal contributions) with Associate Professor Patric Andersson (SSE). CHAPTER 6 57

Article 3: Pari-mutuel betting on horseracing: The favourite-longshot bias revisited

Single-authored paper. Presented at SJDM 40 (Montreal, Canada). The pari-mutuel system is the typical organization of horserace betting. It has become known for a market anomaly known as the favorite-longshot bias (FLB), which stipulates the more favored the horse; the better on average is the bet. This paper analyzes more than 5 years of Swedish harness races (n= 19,521). Study 1 investigates whether there is a FLB for Win bets. Study 2 then analyzes a unique subset of so-called exotic (complex) bets to compare subjective probabilities (bet fractions) and objective probabilities (win fractions). In total, the analyses involved 185,198 starters (up to 10 starters per race). In line with previous research, the results revealed that there is a FLB for Win bets in Sweden (semi-strong market efficiency). But FLB is very small. The vast majority of the betting turnover is actually close to strong market efficiency in terms of expected net return. So the bettors, on aggregate, do well in terms of calibration. Also the exotic bets have rea- sonably calibrated bet fractions and win fractions. Further analyses theorize how differences in the number of selected horses per race can drive spreads in bet fractions for complex bets. FLB can thus appear, disappear or re- verse for simpler reasons than suggested in previous research. 58 THE GOOD PLACE

Article 4: Uncertainty and Complexity in Predictions from Big Data: Why Managerial Heuristics will Survive Datafication

Single-authored paper. Published in Andersson, P., Movin, S., Mähring, M., Teigland, R. & Wennberg, K. (Eds.) (2018), Managing Digital Transformation. Stockholm: Stockholm School of Economics Institute for Research (SIR). Big Data is a buzzword for especially large datasets processed in real- time by computers. One of its main applications in business is predictive analytics. But what constitutes a good prediction? Which dynamics determine the predictability of the future? And are extensive information and sophisticated models really pre-requisites to anticipate the behavior of digital customers? These and similar questions are answered in this paper. For illustrative purposes, analysis of two large weather forecasting datasets (each N= 1,533,695) provided by SMHI shows that exponential increases in data usage could have a relatively small effect on prediction accuracy. The paper argues that simple managerial heuristics may yet survive, even in this day and age of Datafication. A 7-point checklist for businesses to leverage on Big Data for predictions is also provided.

Chapter 7

Discussion

Conclusions

Point of departure

The purpose of this dissertation was to investigate two things: the approval of nudging, and the rationale behind it. I laid out three choice architecture vices–unpopular nudging, unnecessary nudging and big nudging–and set out to investigate one research question for each. The first research question was how popular nudging was among Swedes; the second how calibrated odds and race outcomes were in Swedish horserace betting; and the third why predictions from Big Data must not outperform managerial heuristics. I problematized the bounded rationality paradigm’s normative, descriptive and prescriptive sides, and proposed that humans use cues in their environments (large worlds) in face of uncertainty and make decisions through ecologically rational heuristics (procedural rationality).

Findings

On the first question, the answer read it depends. In absolute terms, Swedes are cautiously positive towards nudging. There is majority support for it, whether you ask them about specific nudges or the method in general. But in relative terms, Swedes are not all that positive. Neither when their approval of specific nudges is compared to that in other countries nor 60 THE GOOD PLACE when the support for the method as such is measured against traditional alternatives. Item for item, Swedes’ support for nudging most closely resembles that in another cautiously pro-nudge country: Denmark. The approval of nudging also depends on socio-demographics, political ideology and party preferences. The last finding is important, as prior research “had found no systematic correlation along approval [of nudging] and party affil- iations” (Sunstein, Reisch & Kaiser, 2018, p. 1423). In Sweden, there is one. The answer to the second question was that odds and race outcomes are well calibrated. This really is not all that surprising given that the racetrack is a kind learning environment, whose feedback allows bettors to learn and improve. Thaler (2015) says bets on favorites generally yield returns up to six times higher than bets on longshots. I got 1.3, accounting for a very small market share. This is an interesting finding as the anomaly in question has been used as key evidence for the prevalence and robust- ness of cognitive biases, being one of the classics (see Griffith, 1949). The answer to the third question is multifaceted. Simply put, it boils down to uncertainty. In itself, Big Data does not make managerial heuristics redundant. The real issue is not the data so much as to which degree they improve our prediction models. And we still do not know that for sure. It naturally depends on what to predict, how data are used, and what is meant by Big Data in the first place (more rows, columns, or both?). So time will tell if big nudging ever comes into fruition. For now, the case for the supe- riority of machines over humans actually remains unconvincing for many large world prediction tasks, although some expect that to change.

Impact

Like most doctoral dissertations, the impact of mine will likely be minor. However, I believe it has made revelatory, consolidatory, incremental and replicatory contributions to the JDM literature. Formally, Swedish law requires higher education to contribute to research and education while en- suring public outreach. My work is either published or under review and/or has been presented at the main JDM conferences and before top research centres. It has featured in teaching, and has won some medial recognition. CHAPTER 7 61

Going forward, my work might help to nuance the expectations on nudging, which I believe would be a good thing–in academia, public administration and business alike. Some additional implications follow below. Implications for nudge narratives In Chapter 2, I presented a narrative review of the JDM literature. It covered its normative, descriptive, and prescriptive phases. One reading of that history would place nudging at today’s research frontier. By standing on the shoulders of giants, to quote Isaac Newton. However, as I explained in Chapter 3, there are in fact several ideologies in JDM research. Its mainstream paradigm–bounded rationality–upon which the official history of nudging is based, does not stand unchallenged. Even if the bounded rationality paradigm’s normative, descriptive and prescriptive conditions are not met, there will still be nudging. But its narrative would then have to change. What the empirical research on nudging has produced this far is first and foremost a catalogue of experiments, compiled over the past decade or so. It contains many observations of temporary behaviour changes through nudging, but is not particularly extensive yet. It is probably fair to assume that this body of work has the same limitations in terms of reproducibility as science in general and the social sciences in particular. Nudging goes back as far as ancient Greece (remember the trick at Me- kone), or at least the late 1990s. I do not know how its story ends. Whether the concept is here to stay or will fizzle out in a few years’ time. Either way, it is its yet unknown future, and not its ideological past, which will determine nudging’s eventual place in the history of JDM research. Implications for nudge research Most of the nudges in Article 1 were popular. However, that did not make them immune to controversy and politicization, which here means that their approval differed across socio-demographics and political views among the general public. In a polarized political discourse, that might hin- der the implementation even of generally popular nudges. If nudging’s fea- sibility generally increases with public support, but decreases with controversy and politicization, the last two ought to be researched more. 62 THE GOOD PLACE

Article 2 included three items from a previous survey by Reisch and Sunstein (2016). Its respondents (n= 537) were cautiously positive towards those nudges (M= 4.61, SD= 1.32, 95% CI = 4.50, 4.72).43 Now study Fig- ure 5 below and compare that number to the traditional alternatives with the same policy goals that the nudges were benchmarked against, and it puts the nudges’ public support in a different light. I expect more, similar comparisons and contextualisation in new studies moving forward.

Figure 5. The public support for carrots, sticks, sermons, and nudges with consistent policy goals. Means as data points (n= 537). Error bars are 95 % CIs.

1 Information Subsidies Nudges Taxes Mandates

Articles 1 and 2 illustrate that there are more to research on the public support for nudging than the approval rating studied in isolation. In previous research, much has been made of the fact that most nudges are popular on country-level internationally. But the devil is in the details. The support for a nudge is dependent upon its policy goal and political opinions among the general public. And public opinions may change. Sometimes because there is an entire industry specialized at changing them. The U.S. used to have an inheritance tax affecting the wealthiest 2 % of Americans. It was called the estate tax and had been around since the beginning of the 20th century. Few opposed it. Until some political strategists renamed it and began calling it the death tax instead, which saw it abolished during George W. Bush’s presidency (Graetz & Shapiro, 2005). Likewise,

43 On a 7-points Likert response scale ranging from 1 (“very bad”) to 7 (“very good”). CHAPTER 7 63 even though many nudges are popular as things stand, that does not necessarily prevent them from becoming, or being made, unpopular some time in the future. How robust the public support for nudging is, also relative other policy tools, could be an interesting research question in future work. Implications for nudge theories In Article 3, I investigated a well-established, yet barely reproducible, cognitive bias. In Article 4, I defended the use of heuristics in business; at least until we know for sure that we are replacing them with something better. Cognitive biases and heuristics are both integral to the descriptive theories of how nudges work and why they are needed. For example, defaults work because of the status quo bias; our unwillingness to actively choose (Samuelson & Zeckhauser, 1988). Other cases are more complicated. Thaler and Benartzi (2004) invented a nudge, save more tomorrow, that managed to increase retirement saving by allowing employees to decide–in advance–to put future pay raises into a retirement account. Next consider Thaler and Shefrin’s (1981) original explanation for why: unreliable self-control due to an inner conflict between one’s future- and present-oriented selves. To strengthen their case, they cited Sigmund Freud as someone “deserving special mention” (p. 394). However, psychologists have long known that Freud’s work does not de- serve special mentioning. Effective nudges–like save more tomorrow–must work independently of Freud’s theories, as we know them to be false. This naturally implies the following too. Nudges can work whether or not dual-process theories of cognition are valid; regardless of the (dis)similarity between errors in judgment and visual illusions; and independently of alleged cognitive biases and heuristics (see Gigerenzer, 2015). I have recently been asked to define the psychological nature of the problems that nudges may solve. I would argue that they involve a combination of cognitive, motivational and behavioural factors. There are typologies that classify nudges as System 1 or System 2 (Jung & Mellers, 2016), pro-self or pro-social (Hagman et al., 2015). But typologies require mutually exclusive categories. Psychology is often interactional. Take smoking. You might start because of its image (System 1) but continue because of nicotine addiction (the epinephrine and dopamine sys- 64 THE GOOD PLACE tems). Quit, and it benefits your own health (pro-self). And frees up healthcare for others (pro-social). The bounded rationality paradigm–an ideology based on normative, descriptive and prescriptive assumptions–is nudging’s grand theory. But it is not a prerequisite for nudges to be effective. There could be a case for sep- arating nudging’s grand theory from the method’s measureable effects. Implications for nudge policies The choice architecture vices–unpopular nudging, unnecessary nudging and big nudging–all recently came in for criticism in the U.K. (Almqvist & An- dersson, 2020; Sibony, 2020). With general implications for nudge policy. David Halpern leads the Behavioural Insights Team, also known as the nudge unit. He is also a member of SAGE, the Scientific Advisory Group for Emergencies, which advised the U.K. government during the COVID- 19 crisis. Halpern made nudging an integral part of their strategy. Early on, the government went with nudging instead of coercive measures, waiting for herd immunity (see Fine, Eames & Heymann, 2011) to develop in the meantime. The strategy received heavy criticism; forcing the government to deny it had even existed. Ergo, unpopular nudging. Nudging was then said to counteract a cognitive bias named behavioral fatigue. A group of researchers responded with an open letter to the government, saying behavioral fatigue did not exist (Hahn et al., 2020). In other words, it was an imaginary bias leading to unnecessary nudging. Another move by the government was to launch a contact tracing application. It used smartphone data and would help to isolate cases and communicate risk. It did not take long for the government to receive yet another open letter from researchers. This one raised integrity concerns and warned for mission creep (Albrecht et al., 2020). That is, big nudging. By violating the choice architecture vices, some of the U.K. government’s behavioral insights did not pass their litmus test. This may have consequences also in the long run as “governments can misuse behavioural arguments and tarnish a reputation for sound evidence-based policy- making” (Sibony, 2020). We would not want that. Going forward, it would accordingly be advised for governments to acknowledge the choice architecture vices. Nudging with caution. CHAPTER 7 65

Final remarks

Eutopia

Utopia and Eutopia are two compounds from Greek. They sound the same but have different meanings. The former, as in More’s (1516) book, origi- nates from οὐ (not) and τόπος (place). The latter comes from εὖ (good) and τόπος (place). More never meant for Utopia to be taken seriously, so he named it No Place. Nudge, however, is seriously meant. It is Eutopian. The Good Place. While Utopia had everyone work 6 hour-days and wear identical clothes, choice architects envision a society full of healthy, wealthy and happy people, as promised by Nudge’s subtitle. Thaler’s favourite nudge story comes from Amsterdam’s airport. Its cleaning manager came up with the idea to paint flies in the urinals, to help men take aim. Spilling decreased and cleaning costs went down. But there is still a long way from there to choice architects’ promised land, “saving the world, one randomised controlled test at a time“ (Rutter, 2014). Thalerland The JDM literature has seen large revisions to the earlier portrayal of the human mind as flawed and susceptible to cognitive biases. In his Nobel lecture, Thaler nevertheless maintained that man resembles a cartoon character–Homer Simpson–infamous for his stupidity. His latest book, Misbe- having, says the same. Even though such a negative view is not merited, it supposedly explains and motivates nudging. Therein lies a contradiction. Sunsteinia There is state nudging all over the world. Many nonetheless find Sunstein’s (2016b) proposal of establishing an Council of Psychological Advisers–an independent nudge unit–worrisome. It sides with technocrats over the government and parliament. For example, when I asked a sample (N= 1,032) representative of the Swedish adult population what they thought about his idea, the vast majority of them rejected it, 95 % CI = [.76, .81]. 66 THE GOOD PLACE

Dystopia

There are two dystopian novels in the western canon. One is Huxley’s (1932) Brave New World. The other is Orwell’s (1949) Nineteen Eighty-Four. Huxlian nudging The Huxlian dystopia is a technocratic society whose citizens live conven- iently. It is, for most part, happy. However, not everyone may want to live in a society modelled after a Netflix account. Where money is charged au- tomatically, recommendations bombard you, and choices are made for you unless you actively opt-out. It resembles Nozick’s (1974) pleasure machine. This dissertation began with two quotes. The first one comes from the TV series The Good Place. Season one takes place in a heavenly suburbia that (spoiler alert) turns out to be hell. The second one is from Radiohead’s song Fitter Happier. Its lyrics are read by text-to-voice software (the same one as Stephen Hawking used) to sound as clinical as the life it portrays. I chose the quotes because the thought of happiness by design scares me. Orwellian nudging The Orwellian dystopia is a dictatorship. For all of Thaler’s reassurances (he signs his books “Nudge for good”), it is naïve to assume that nudging can only be done in the spirit of utilitarianism. In Western democracies, it evokes constitutional and ideological questions best left to the voters. In other parts of the world, totalitarian regimes may use nudging to ex- ert control. Hong Kong media cite Thaler as saying “[w]hatever the problem is that China’s government is trying to solve, then thinking about it through this kind of lens [nudging] would help” (Yeung, 2019). Well, the Chinese government already does (Creemers, 2018) and the result is disturbing (The Asan Institute for Policy Studies, 2017; Qiang, 2019). The TV series Black Mirror features an episode in which citizens constantly rate each other on their smartphones. The ratings then determine their social statuses. China has its own version in place: the social credit system. It affects access to capital, employment and welfare services. Per- mittance to take the train or travel abroad. Dating prospects. Under such circumstances, nudging becomes part of the problem, not the solution. CHAPTER 7 67

Epilogue

I close with some personal reflections on the pursuit of helping people to make better decisions. On nudging for health, wealth and happiness.

Health

My first reflection goes back to when I studied psychology and interned as therapist. Some of my patients suffered from mood disorders; others from phobias or neuropsychiatric conditions. Literary every patient I met acknowledged that improved decision making was integral to tackle his or her problems. At the start of therapy, following standard CBT (cognitive behavioral therapy) protocol, I would ask them to list a few things they hoped to get out of therapy. From that, we set treatment goals. We discussed how I could help them reach those goals, agreed on a plan, and got to work. In JDM terms, I did boosting. Rather than having them rely solely on me, we worked together towards strengthening their self-efficacy. JDM and CBT are different in some regards and similar in other. A difference is that JDM (CBT) focuses on the general (pathological) population. A similarity is their mutual interest in erroneous thinking (Leddy, Anderson & Schulkin, 2013). Prescriptive JDM–nudging included–and CBT also share the same end goal: to help people. Going back to my experience as a therapist, I cannot see how adding nudging into the mix could have helped. While CBT uses two-way communication, nudging is one-way. A CBT therapist is transparent; a choice architect typically is not (but see Loewenstein et al., 2015; Schmidt, 2017). CBT assumes that people’s preferences are different. Nudging says everyone wants pretty much the same (Hertwig, 2017; Reijula, Kuorikoski, Ehrig, Katsikopoulos & Sunder, 2018). As CBT clearly is a more inclusive approach, I would have been equally as hesitant to use nudging as therapist as to receive it had I been the patient. That being said, maybe there will be CAT (choice architecture therapy) some day. I see the first session ending something like this: “When you get home, you’ll find your life’s already changed. I got you some job offers and put a list of your new hobbies on the fridge. Don’t worry, you’ll thank me later.” 68 THE GOOD PLACE

Wealth

My second reflection dates back to December 14th 2017, when Thaler visited SSE having just been awarded the Economics Prize. Directly afterwards, he visited parliament to give a presentation on how the Swedish pension system could be reformed. Later did I learn that his recommendation had been to reduce the leverage of the default investment fund (AP7 Såfa). A nudge designed to lower the financial risk of most Swedes’ retirement savings. By April 2018, AP7 Såfa had seen its leverage ratio lowered by 10 percentage-points. Nudges should make us better off, as judged by ourselves. By getting us to do what we really wanted all along or by leading to consequences we like (Sunstein, 2018).44 For the past two and a half years, Thaler’s default has made me better off in three quarters and worse off in seven. In 2019 alone, Swedish retirement savers lost out on tens of billions (SEK) in returns because of the reduced leverage ratio. Then came the COVID-19 pandemic. In a matter of weeks, the bull market became a bear market that morphed into a bat market. Suddenly, Thaler’s advice seemed almost clairvoyant.45 When the stock market bounced back in the next quarter, millions of Swedes again look destined to miss out on what could have been an extra vacation as retirees. Surely we cannot let the evaluation of nudges (or pension system reforms) boil down to timing and chance.

Happiness

My third and final reflection comes from trying to watch television this morning. It takes two machines for me to do so. A TV and a digital TV box. They both use the same nudge, designed to reduce energy consumption: power saving mode. Meaning they both turn themselves off after certain times, unless I actively choose to continue to watch. Their respective opts-out menus are identical but for one thing. Pressing continue means “continue to watch” on one of them, but “continue with the automatic shut-down” on the other. Unfortunately, they cannot be told apart.

44 But see Sugden (2018). 45 Although Thaler would be the first to admit he cannot predict the stock market (no one can). CHAPTER 7 69

The power saving modes on my TV set belong to a category of nudges I call twisted nudges. They are neither good nor bad. Just erratic. Here are some more twisted nudges I have encountered. • Glass recycling. Where I live, coloured glass goes into a container marked with an uncoloured label reading “coloured”. And vice versa. It turns it into a Stroop (1935) task. • Pedometers with the target set to 10,000 steps per day. The number is arbitrary.46 The same health effects can be achieved at much short- er walking distances as long as you increase your heart rate. • The “wellness” e-mails sent by Microsoft Office telling people how to structure their workdays. Stressful and disturbing one’s workflow. • Signals that remind bus passengers in public transportation to wear their seatbelts, which they never do anyway. • The SRRI risk disclosure on investment fund prospectuses. Gives most comparable funds the same exact score. • City marathons involve plenty of tricks to make you run that tiny bit faster. Which you do. But you always pay for it over the full distance. Study the split times of Kenya’s Eliud Kipchoge, the world record holder, and they differ by less than 1 %. • Smartphone applications against overusing one’s smartphone. • The orange envelope. Confuses and guilt-trips retirement savers. • Betting sites that remind me of previous losses, to help me limit my betting. They make me increase my stakes, hoping to break even. • Auto-play of the evening news. Up next will be yesterday’s program. • Most loyalty programs, sign-ups and newsletters. • During the COVID-19 outbreak, the National Board of Health and Welfare sent out flyers on the importance of good hand hygiene, trying to prevent the virus from spreading. They warned against hand contact. One of them ended up in our lunchroom. My colleagues and I passed it around the table, for everyone to read.

46 The first mass market pedometer was launched in Japan in time for the 1964 Olympics in Tokyo. The manufacturer named it manpo-kei (10,000 steps) simply because they liked how it sounded. It stuck.

Part Two

[Pages 73-156 containing Article 1, 2, 3, and 4 are not included in this e-version of the dissertation.]

References

Adolfson, M., Andersson, M. K., Lindé, J., Villani, M., & Vredin, A. (2006). Modern Forecasting Models in Action: Improving Macroeconomic Analyses at Central Banks [Report]. Sveriges Riksbank. Agrawal, A., Gans, J., & Goldfarb, A. (2018). Prediction machines: the simple economics of artificial intelligence. Harvard Business Press. Aikman, D., Galesic, M., Gigerenzer, G., Kapadia, S., Katsikopoulos, K., Kothiyal, A., ... & Neumann, T. (2014). Financial Stability Paper No 28: Taking uncertainty seriously-simplicity versus complexity in financial regulation. [Research paper]. Bank of England. Aischylos (1903). The Prometheus bound. Cambridge University Press. Albrecht, M., et al. (2020, April 29). Joint Statement. [Blog post]. Retrieved from https://drive.google.com/file/d/1uB4LcQHMVP-oLzIIHA9SjKj1uMd3erGu/vie w Ali, M. M. (1977). Probability and utility estimates for racetrack bettors. Journal of political Economy, 85(4), 803-815. Ali, M. M. (1979). Some Evidence of the Efficiency of a Speculative Market. Econometrica, 47(2), 387-392. Ali, M. M. (1998). Probability models on horse-race outcomes. Journal of Applied Statistics, 25(2), 221-229. Allais, M. (1953a). L'extension des théories de l'équilibre économique général et du rendement social au cas du risque. Econometrica, Journal of the Econometric Society, 269-290. Allais, M. (1953b). Le Comportement de l’Homme Rationnel devant le Risque: Critique des Postulats et Axiomes de l’Ecole Americaine.́ Econometrica, 21, 503-546. Allcott, H., & Mullainathan, S. (2010). Behavior and energy policy. Science, 327(5970), 1204-1205. Almqvist, G. (2016). Om risk och osäkerhet. In R. Wahlund (Ed.), Risker och riskhantering i näringsliv och samhälle (pp. 25-47). Stockholm: Studentlitteratur. Almqvist, G. (2018a). Den enda vägens beteendeekonomi – om bakgrunden till utnämnandet av Richard Thaler till Ekonomipristagare. Ekonomisk debatt, 46(3), 79- 83. 158 THE GOOD PLACE

Almqvist, G. (2018b). Uncertainty and Complexity in Predictions from Big Data: Why Managerial Heuristics will Survive Datafication. In P. Andersson, S. Movin, M. Mähring, R. Teigland, & K. Wennberg (Eds.), Managing Digital Transformation (pp. 189-202). Stockholm: Stockholm School of Economics Institute for Research (SIR). Almqvist, G., & Andersson, P. (2020). Carrots, sticks, sermons, and nudges during the pandemic. [Research report]. Retrieved from https://www.hhs.se/contentassets/8d0724b30dc249458eb6ecced784cf26/a06.pdf Alvesson, M., & Sandberg, J. (2011). Generating research questions through problematization. Academy of management review, 36(2), 247-271. Alvesson, M., & Sandberg, J. (2013). Has management studies lost its way? Ideas for more imaginative and innovative research. Journal of Management Studies, 50(1), 128- 152. Andersson, P., & Almqvist, G. (2016). Presentationsformatets betydelse vid riskkommunikation. In R. Wahlund (Ed.), Risker och riskhantering i näringsliv och samhälle (pp. 67-92). Stockholm: Studentlitteratur. Andersson, P., & Nilsson, H. (2015). Do bettors correctly perceive odds? Three studies of how bettors interpret betting odds as probabilistic information. Journal of Behavioral Decision Making, 28(4), 331-346. Andersson, P., & Rakow, T. (2007). Now you see it now you don’t: The effectiveness of the recognition heuristic for selecting stock. Judgment and Decision Making, 2(1), 29- 39. Angner, E., & Loewenstein, G. (2012). Behavioral economics. In U. Mäki (Ed.), Handbook of the Philosophy of Science: Philosophy of Economics (pp. 641-690). Amsterdam: Elsevier. Anscombe, F. J., & Aumann, R. J. (1963). A definition of subjective probability. The annals of mathematical statistics, 34(1), 199-205. Ariely, D. (2008). Predictably irrational. New York: HarperCollins. Ariely, D. (2009). The end of rational economics. Harvard business review, 87(5), 78-84. Arkes, H. R., Gigerenzer, G., & Hertwig, R. (2016). How bad is incoherence?. Decision, 3(1), 20-39. Arkes, H. R., Wortmann, R. L., Saville, P. D., & Harkness, A. R. (1981). Hindsight bias among physicians weighing the likelihood of diagnoses. Journal of Applied Psychology, 66(2), 252-254. Armstrong, J. S., Green, K. C., & Graefe, A. (2015). Golden rule of forecasting: Be conservative. Journal of Business Research, 68(8), 1717-1731. Arrow, K. J. (1992). I know a hawk from a handsaw. In M. Szenberg (Ed.), Eminent Economists. Their Life Philosophies (pp. 42-50). Cambridge University Press. Arrow, K. J. (2004). Is bounded rationality unboundedly rational? Some ruminations. Models of a man: Essays in memory of Herbert A. Simon, 47-55. REFERENCES 159

Artinger, F., Petersen, M., Gigerenzer, G., & Weibler, J. (2015). Heuristics as adaptive decision strategies in management. Journal of Organizational Behavior, 36, 33-52. Asch, P., Malkiel, B. G., & Quandt, R. E. (1982). Racetrack Betting and Informed Behavior. Journal of Financial Economics, 10(2), 187-194. Asch, P., Malkiel, B. G., & Quandt, R. E. (1984). Market efficiency in racetrack betting. Journal of Business, 165-175. Asch, P., & Quandt, R. E. (1987). Efficiency and Profitability in Exotic Bets. Economica, 54(215), 289-298. Asch, P., & Quandt, R. E. (1988). Betting bias in ‘exotic’ bets. Economics Letters, 28(3), 215-219. Ayton, P., & Fischer, I. (2004). The hot hand fallacy and the gambler’s fallacy: Two faces of subjective randomness?. Memory & cognition, 32(8), 1369-1378. Bacon-Shone, J., Lo, V. S. Y., & Busche, K. (1992a). Modelling the winning probabilities [Research report]. The University of Hong Kong. Bacon-Shone, J., Lo, V. S. Y., & Busche, K. (1992b). Logistic analyses for complicated bets [Research report]. The University of Hong Kong. Baron, J. (2004). Normative Models of Judgment and Decision Making. In D. J. Koehler, & N. Harvey (Eds.), Blackwell Handbook of Judgment and Decision Making (pp. 19-36). Malden, MA: Blackwell Publishing. Battaglio Jr, R. P., Belardinelli, P., Bellé, N., & Cantarelli, P. (2019). Behavioral public administration ad fontes: A synthesis of research on bounded rationality, cognitive biases, and nudging in public organizations. Public Administration Review, 79(3), 304- 320. Bell, D. E., Raiffa, H., & Tversky, A. (1988). Descriptive, normative, and prescriptive interactions in decision making. Decision making: Descriptive, normative, and prescriptive interactions, 1, 9-32. Bemelmans-Videc, M. L., Rist, R. C., & Vedung, E. O. (Eds.). (2011). Carrots, sticks, and sermons: Policy instruments and their evaluation. New Jersey: Transaction Publishers. Benartzi, S., Beshears, J., Milkman, K. L., Sunstein, C. R., Thaler, R. H., Shankar, M., ... & Galing, S. (2017). Should governments invest more in nudging?. Psychological science, 28(8), 1041-1055. Benartzi, S., & Thaler, R. H. (2001). Naive diversification strategies in defined contribution saving plans. American economic review, 79-98. Benartzi, S., & Thaler, R. H. (2002). How much is investor autonomy worth?. The Journal of Finance, 57(4), 1593-1616. Benartzi, S., & Thaler, R. H. (2003). Using behavioral economics to improve diversification in 401 (k) plans: Solving the company stock problem. [Working paper]. University of California. Benartzi, S., & Thaler, R. H. (2007). Heuristics and biases in retirement savings behavior. The journal of economic perspectives, 21(3), 81-104. 160 THE GOOD PLACE

Berg, N. (2010). Behavioral economics. 21st century economics: A reference handbook, 2, 861- 872. Berg, L. (2014). Who benefits from behavioural economics?. Economic Analysis and Policy, 44(2), 221-232. Berg, C., & Davidson, S. (2017). Nudging, calculation, and utopia. Journal of Behavioral Economics for Policy, 1, 49-52. Berg, N., & Gigerenzer, G. (2007). Psychology implies paternalism? Bounded rationality may reduce the rationale to regulate risk-taking. Social Choice and Welfare, 28(2), 337- 359. Berg, N., & Gigerenzer, G. (2010). As-if behavioral economics: Neoclassical economics in disguise?. History of economic ideas, 133-165. Bernstein, P. L. (1996). Against the gods: the remarkable story of risk. New York: Wiley. Bernoulli, D. (1954). Exposition of a New Theory on the Measurement of Risk. Econometrica, 22, 23-36. Bertrand, M., Mullainathan, S., & Shafir, E. (2004). A behavioral-economics view of poverty. The American Economic Review, 94(2), 419-423. Bertrand, M., Mullainathan, S., & Shafir, E. (2006). Behavioral economics and marketing in aid of decision making among the poor. Journal of Public Policy & Marketing, 25(1), 8-23. Bjorkman, D., & Bukszar, E. (2003). Efficiency Reigns: Disappearance of the ‘Favorite- Longshot Bias’ in Parimutuel Racehorse Betting. In Western Decision Sciences Annual Conference Proceedings (pp. 197-201). Bohnet, I., Van Geen, A., & Bazerman, M. (2015). When performance trumps gender bias: Joint vs. separate evaluation. Management Science, 62(5), 1225-1234. Borges, B., Goldstein, D. G., Ortmann, A., & Gigerenzer, G. (1999). Can ignorance beat the stock market? In G. Gigerenzer, P. M. Todd, & the ABC Research Group (Eds.), Simple heuristics that make us smart (pp. 59–72). New York: Oxford University Press. Brandstätter, E., Gigerenzer, G., & Hertwig, R. (2006). The priority heuristic: making choices without trade-offs. Psychological review, 113(2), 409-432. Branson, C., Duffy, B., Perry, C., & Wellings, D. (2012). Acceptable behaviour: Public opinion on behaviour change policy. [Report]. Ipsos MORI, London. Brighton, H., & Gigerenzer, G. (2015). The bias bias. Journal of Business Research, 68(8), 1772-1784. Brunswik, E. (1955). Representative design and probabilistic theory in a functional psychology. Psychological review, 62(3), 193-217. Busche, K., & Walls, W. D. (2000). Decision costs and betting market efficiency. Rationality and Society, 12(4), 477-492. Camerer, C. F. (1999). Behavioral economics: Reunifying psychology and economics. Proceedings of the National Academy of Sciences, 96(19), 10575-10577. REFERENCES 161

Camerer C. F. (2000). Prospect theory in the wild: evidence from the field. In D. Kahneman, & A. Tversky (Eds.), Choices, Values, and Frames, (pp. 288-300). Cambridge University Press. Camerer, C. F., Babcock, L., Loewenstein, G., & Thaler, R. H. (1997). Labor supply of New York City cabdrivers: One day at a time. The Quarterly Journal of Economics, 112(2), 407-441. Camerer, C. F., Dreber, A., Forsell, E., Ho, T. H., Huber, J., Johannesson, M., ... & Heikensten, E. (2016). Evaluating replicability of laboratory experiments in economics. Science, 351(6280), 1433-1436. Camerer, C. F., Dreber, A., Holzmeister, F., Ho, T. H., Huber, J., Johannesson, M., ... & Altmejd, A. (2018). Evaluating the replicability of social science experiments in Nature and Science between 2010 and 2015. Nature Human Behaviour, 2(9), 637-644. Camerer, C. F., Issacharoff, S., Loewenstein, G., O'donoghue, T., & Rabin, M. (2003). Regulation for Conservatives: Behavioral Economics and the Case for "Asymmetric Paternalism". University of Pennsylvania law review, 151(3), 1211-1254. Camerer, C. F. & Loewenstein, G. (2003). Behavioral Economics: Past, Present Future. In C. F. Camerer, G. Loewenstein, & M. Rabin (Eds.), Advances in Behavioral Economics (pp. 3-51). Princeton University Press. Canfield, B., Fauman, B. C. and Ziemba, W. T. (1987). Efficient Market Adjustment of Odds Prices to Reflect Track Biases. Management Science, 33, 1428-1439. Canymeres, F. (1946). L’Homme de la belle époque. Paris: Les Editions Universelles. Claverie-Berge, I. (2012, March 13). Solutions Big Data IBM. [Presentation]. Retrieved from https://www-05.ibm.com/fr/events/netezzaDM_2012/Solutions_Big_Data .pdf Coleman, L. (2004). New light on the longshot bias. Applied Economics, 36(4), 315-326. Costa, D. L., & Kahn, M. E. (2013). Energy conservation “nudges” and environmentalist ideology: Evidence from a randomized residential electricity field experiment. Journal of the European Economic Association, 11(3), 680-702. Cosmides, L., & Tooby, J. (1996). Are humans good intuitive statisticians after all? Rethinking some conclusions from the literature on judgment under uncertainty. Cognition, 58(1), 1-73. Creemers, R. (2018, May 22). China's Social Credit System: An Evolving Practice of Control. [Article]. Retrieved from https://ssrn.com/abstract=3175792 Croson, R., & Sundali, J. (2005). The gambler’s fallacy and the hot hand: Empirical data from casinos. Journal of risk and uncertainty, 30(3), 195-209. Damgaard, M. T., & Gravert, C. (2018). The hidden costs of nudging: Experimental evidence from reminders in fundraising. Journal of Public Economics, 157, 15-26. Daston, L. J. (1979). The Reasonable Calculus: Classical Probability Theory, 1650-1840. [Doctoral dissertation]. Harvard University. Daston, L. J. (1980). Probabilistic expectation and rationality in classical probability theory. Historia mathematica, 7(3), 234-260. 162 THE GOOD PLACE

Daston, L. J. (1988). Fitting numbers to the world: The case of probability theory. History and philosophy of modern mathematics, 11, 221-237. De Finetti, B. (1974). Theory of probability. New York: Wiley. De Oliveira, J. M., & Pagano, A. S. (2006). The research article and the science popularization article: a probabilistic functional grammar perspective on direct discourse representation. Discourse studies, 8(5), 627-646. Dhami, M. K., Hertwig, R., & Hoffrage, U. (2004). The role of representative design in an ecological approach to cognition. Psychological bulletin, 130(6), 959-988. Diebold, F. (2012). A Personal Perspective on the Origin (s) and Development of “Big Data”: The Phenomenon, the Term, and the Discipline, Second Version. [Article]. Penn Institute for Economic Research, Department of Economics, University of Pennsylvania. van Dijk, T. A. (1998). Ideology: A multidisciplinary approach. London: Sage. Djupegot, I. L., & Hansen, H. (2020). If it works, I like it: consumer acceptance of food- related nudging. Journal of International Food & Agribusiness Marketing, 32(4), 361-378. Dolbear, F. T. (1993). Is Racetrack Betting on Exactas Efficient? Economica, 60(237), 105-111. Dowd, K., Cotter, J., Humphrey, C., & Woods, M. (2008). How unlucky is 25-sigma? The Journal of Portfolio Management, 34(4), 76-80. Duflo, E., Kremer, M., & Robinson, J. (2011). Nudging farmers to use fertilizer: Theory and experimental evidence from Kenya. American economic review, 101(6), 2350-2390. Dunwoody, P. T. (2009). Introduction to the special issue: Coherence and correspondence in judgment and decision making. Judgment and Decision Making, 4(2), 113-115. Dyson, F. (2004). A meeting with Enrico Fermi. Nature, 427(6972), 297-298. Earl, P. E. (2018). Richard H. Thaler: A Nobel Prize for Behavioural Economics. Review of Political Economy, 30(2), 107-125. Edwards, W. (1954). The theory of decision making. Psychological bulletin, 51(4), 380-417. Edwards, W. (1955). The prediction of decisions among bets. Journal of experimental psychology, 50(3), 201-214. Egan, M. (n.d.). Nudge Database v.1.2. [Blog post]. Retrieved from https://www.stir.ac.uk/media/stirling/services/faculties/social-sciences/research/ documents/Nudge-Database-1.2.pdf Einhorn, H. J., & Hogarth, R. M. (1981). Behavioral decision theory: Processes of judgement and choice. Annual review of psychology, 32(1), 53-88. Einhorn, H. J., & Hogarth, R. M. (1985). Ambiguity and uncertainty in probabilistic inference. Psychological Review, 92, 433-461. Eisenberg, E., & Gale, D. (1959). Consensus of subjective probabilities: The pari-mutuel method. The Annals of Mathematical Statistics, 30(1), 165-168. Ellsberg, D. (1961). Risk, Ambiguity, and the Savage Axioms. Quarterly Journal of Economics, 75, 643-669. REFERENCES 163

Engwall, L. (2009). Mercury meets Minerva: business studies and higher education: the Swedish case. Stockholm: Economic Research Institute, Stockholm School of Economics (EFI). Epstein, D. J. (2019). Range: how generalists triumph in a specialized world. London: Macmillan. Erevelles, S., Fukawa, N., & Swayne, L. (2016). Big Data consumer analytics and the transformation of marketing. Journal of Business Research, 69(2), 897-904. Evans, J. S. B., & Stanovich, K. E. (2013). Dual-process theories of higher cognition: Advancing the debate. Perspectives on psychological science, 8(3), 223-241. Fabricand, B. P. (1965). Horse sense. New York: David McKay. Fan, J., Han, F., & Liu, H. (2014). Challenges of big data analysis. National Science Review, 1(2), 293-314. Fiedler, K., & Juslin, P. (2006). Information sampling and adaptive cognition. Cambridge University Press. Fiedler, K., & von Sydow, M. (2015). Heuristics and biases: beyond Tversky and Kahneman's (1974) judgment under uncertainty. In M. W. Eysenck, & D. Groome (Eds.), Cognitive Psychology: Revisiting the Classical Studies (pp. 146-161). Los Angeles: Sage. Figlewski, S. (1979). Subjective information and market efficiency in a betting market. Journal of Political Economy, 87(1), 75-88. Fine, P., Eames, K., & Heymann, D. L. (2011). “Herd immunity”: a rough guide. Clinical infectious diseases, 52(7), 911-916. Fischhoff, B. (1981). Debiasing. In D. Kahneman, P. Slovic, & A. Tversky (Eds.), Judgmenet under uncertainity. Heuristics and Biases (pp. 422-444). New York: Cambridge University Press. Fischhoff, B., Slovic, P., & Lichtenstein, S. (1977). Knowing with certainty: The appropriateness of extreme confidence. Journal of Experimental Psychology: Human perception and performance, 3(4), 552-564. Flyvbjerg, B. (2009). Survival of the unfittest: why the worst infrastructure gets built— and what we can do about it. Oxford review of economic policy, 25(3), 344-367. Finucane, M. L., Alhakami, A., Slovic, P., & Johnson, S. M. (2000). The affect heuristic in judgments of risks and benefits. Journal of behavioral decision making, 13(1), 1-17. Foa, R. S., & Mounk, Y. (2016). The danger of deconsolidation: The democratic disconnect. Journal of democracy, 27(3), 5-17. Fox, C., Erner, C., & Walters, D. (2014). Decision under risk: From the field to the lab and back. In G. Keren, & G. Wu (Eds.), Blackwell Handbook of Judgment and Decision Making. Blackwell. Fraser, C. (2020, January 3). Target didn’t figure out a teenager was pregnant before her father did, and that one article that said they did was silly and bad [Blog post]. Retrieved from https://medium.com/@colin.fraser/target-didnt-figure-out-a-teen- girl-was-pregnant-before-her-father-did-a6be13b973a5 164 THE GOOD PLACE

Frederick, S., Loewenstein, G., & O'donoghue, T. (2002). Time discounting and time preference: A critical review. Journal of economic literature, 40(2), 351-401. Friedman, M. (1953). The methodology of positive economics. Essays in positive economics, 3(3), 145-178. Friedman, M., & Savage, L. J. (1948). Utility Analysis of Choices Involving Risk. Journal of Political Economy, 56, 279-304. Frischmann, B. M. (2019, September 19). Nudging Humans. [Article]. Retrieved from http://dx.doi.org/10.2139/ssrn.3440791. Gandomi, A., & Haider, M. (2015). Beyond the hype: Big data concepts, methods, and analytics. International Journal of Information Management, 35(2), 137-144. Geman, S., Bienenstock, E., & Doursat, R. (1992). Neural networks and the bias/variance dilemma. Neural computation, 4(1), 1-58. Gigerenzer, G. (1991a). How to make cognitive illusions disappear: Beyond “heuristics and biases”. European review of social psychology, 2(1), 83-115. Gigerenzer, G. (1991b). On cognitive illusions and rationality. Poznan Studies in the Philosophy of the Sciences and the Humanities, 21, 225-249. Gigerenzer, G. (1996). On narrow norms and vague heuristics: A reply to Kahneman and Tversky. Psychological review, 103(3), 592-596. Gigerenzer, G. (2004). Fast and Frugal Heuristics: The Tools of Bounded Rationality. In D. J. Koehler, & N. Harvey (Eds.), Blackwell Handbook of Judgment and Decision Making (pp. 62-88). Padstow, Cornwall: Blackwell Publishing Ltd. Gigerenzer, G. (2006). Bounded and Rational. In R. J. Stainton (Ed.), Contemporary Debates in Cognitive Science (pp. 115-133). Oxford: Blackwell. Gigerenzer, G. (2008). Why Heuristics Work. Perspectives on Psychological Science, 3, 20-29. Gigerenzer, G. (2009). Bounded rationality. In D. Sander, & K. R. Scherer (Eds.), The Oxford companion to emotion and the affective sciences (pp. 79-80). New York: Oxford University Press. Gigerenzer, G. (2015). On the supposed evidence for libertarian paternalism. Review of philosophy and psychology, 6(3), 361-383. Gigerenzer, G. (2016). Taking Heuristics Seriously. In A Samson (Ed.), The Behavioral Economics Guide 2016 (pp. v-xi). Behavioral Science Solutions Ltd. Gigerenzer, G. (2018). The bias bias in behavioral economics. Review of Behavioral Economics, 5(3-4), 303-336. Gigerenzer, G., & Brighton, H. (2009). Homo Heuristicus: Why Biased Minds Make Better Inferences. Topics in Cognitive Science, 1, 107-143. Gigerenzer, G., & Goldstein, D. G. (1996). Reasoning the Fast and Frugal Way: Models of Bounded Rationality. Psychological Review, 103(4), 650-669. Gigerenzer, G., & Hoffrage, U. (1995). How to improve Bayesian reasoning without instruction: frequency formats. Psychological review, 102(4), 684-704. Gigerenzer, G., & Selten, R. (2001). Bounded rationality: The adaptive toolbox. Cambridge, MA: MIT Press. REFERENCES 165

Gigerenzer, G., Hell, W., & Blank, H. (1988). Presentation and content - The use of base rates as a continuous variable. Journal of Experimental Psychology: Human Perception and Performance, 14, 513-525. Gigerenzer, G., Hoffrage, U., & Kleinbölting, H. (1991). Probabilistic Mental Models: A Brunswikian Theory of Confidence. Psychological Review, 98, 506-528. Gigerenzer, G., Todd, P. M., & the ABC Group (1999). Simple heuristics that make us smart. New York: Oxford University Press. Gigerenzer, G., Swijtink, Z., Porter, T, Daston, L., Beatty, J., & Kriiger, L. (1989). The empire of chance: How probability changed science and everyday life. Cambridge University Press. Gilovich, T. D., & Griffin, D. W. (2010). Judgment and decision making. In S. T. Fiske, D. T. Gilbert, & G. Lindzey (Eds.), Handbook of social psychology (pp. 542-588). Hoboken, NJ: John Wiley & Sons. Gilovich, T., Vallone, R., & Tversky, A. (1985). The hot hand in basketball: On the misperception of random sequences. Cognitive psychology, 17(3), 295-314. Goldstein W. M. & Hogarth, R. M. (1997). Judgment and decision research: some historical context. In W. M. Goldstein, & R. M. Hogarth (Eds.), Research on Judgment and Decision Making: Currents, Connections, and Controversies (pp. 3-68). Cambridge University Press. Goldstein, D. G., & Gigerenzer, G. (1999). The recognition heuristic: How ignorance makes us smart. In Simple heuristics that make us smart (pp. 37-58). Oxford University Press. Goldstein, D. G., & Gigerenzer, G. (2002). Models of ecological rationality: the recognition heuristic. Psychological review, 109(1), 75-90. Goldstein, D. G., & Gigerenzer, G. (2009). Fast and frugal forecasting. International journal of forecasting, 25(4), 760-772. Goldstein, D. G., Johnson, E. J., Herrmann, A., & Heitmann, M. (2008). Nudge your customers toward better choices. Harvard Business Review, 86(12), 99-105. Goldstein, W. M., & Hogarth R. M. (1997). Judgement and decision research: Some historical context. In W. M. Goldstein, & R. M. Hogarth (Eds.), Research on judgement and decision making. New York: Cambridge University Press. Graetz, M. J., & Shapiro, I. (2005). Death by a thousand cuts: the fight over taxing inherited wealth. Princeton, N. J.: Princeton University Press. Griffith, R. M. (1949). Odds Adjustments by American Horse-Race Bettors. The American Journal of Psychology, 62(2), 290-294. Griffith, R. M. (1961). A Footnote on Horse Race Betting. Transactions of the Kentucky Academy of Sciences, 22, 78-81. Gulati, A., & Shetty, S. (2007). Empirical evidence of error in pricing of favorites and longshots in greyhound racing. Journal of Economics and Finance, 31(1), 49-58. Gärdenfors, P., Johannesson, M., Molander, P., Sjöström, T., & Strömberg, P. (2017). 2017 års ekonomipris till Richard Thaler. Ekonomisk Debatt, 8, 6-16. 166 THE GOOD PLACE

Hagman, W. (2019). When are nudges acceptable?: influences of beneficiaries, techniques, alternatives and choice architects. [Doctoral dissertation]. Linköpings universitet. Hagman, W., Andersson, D., Västfjäll, D., & Tinghög, G. (2015). Public views on policies involving nudges. Review of Philosophy and Psychology, 6(3), 439-453. Hagman, W., Erlandsson, A., Dickert, S., Tinghög, G., & Västfjäll, D. (2019). The effect of paternalistic alternatives on attitudes toward default nudges. Behavioural Public Policy, 1-24. Hahn, U., et al. (2020, March 16). Open letter to the UK Government regarding COVID-19. [Blog post]. Retrieved from https://sites.google.com/view/covidopenletter/home Hammond, K. R. (1954). Representative vs. systematic design in clinical psychology. Psychological Bulletin, 51(2), 150-159. Hammond, K. R. (1996). Human judgment and social policy: Irreducible uncertainty, inevitable error, unavailable injustice. New York: Oxford University Press. Hansen, P. G. (2016). The definition of nudge and libertarian paternalism: Does the hand fit the glove?. European Journal of Risk Regulation, 7(1), 155-174. Halpern, D. (2015). Inside the nudge unit: how small changes can make a big difference. London: WH Allen. Harville, D. A. (1973). Assigning probabilities to the outcomes of multi-entry competitions. Journal of the American Statistical Association, 68(342), 312-316. Hasan, S., Shamsuddin, S. M., & Lopes, N. (2014). Machine Learning Big Data Framework and Analytics for Big Data Problems. International Journal of Advances in Soft Computing & Its Applications, 6(2), 1-14. Hausch, D. B., Lo, V. S., & Ziemba, W. T. (2008). Efficiency of racetrack betting markets. Singapore: World Scientific Publishing Co. Pte. Ltd.. Hausch, D. B., & Ziemba, W. T. (1990). Arbitrage strategies for cross-track betting on major horse races. Journal of Business, 63(1), 61-78. Hausch, D. B., Ziemba, W. T., & Rubinstein, M. (1981). Efficiency of the market for racetrack betting. Management science, 27(12), 1435-1452. Helbing, D., Frey, B. S., Gigerenzer, G., Hafen, E., Hagner, M., Hofstetter, Y., ... & Zwitter, A. (2017, February 25). Will Democracy Survive Big Data and Artificial Intelligence? [Article]. Retrieved from https://www.scientificamerican.com/article/will-democracy-survive-big-data-and- artificial-intelligence/ Hertwig, R. (2017). When to consider boosting: some rules for policy-makers. Behavioural Public Policy, 1(2), 143-161. Hertwig, R., & Grüne-Yanoff, T. (2017). Nudging and boosting: Steering or empowering good decisions. Perspectives on Psychological Science, 12(6), 973-986. Hesiod (1914). The Homeric Hymns and Homerica with an English Translation by Hugh G. Evelyn-White. Theogony. Harvard University Press. REFERENCES 167

Heukelom, F. (2005, August). Gigerenzer the Decided: A Tale of Difficult Distinctions. [Working paper]. Retrieved from https://www.econstor.eu/bitstream/10419/86455/1/05-113.pdf Hodges, S. D., Tompkins, R., & Ziemba, W. T. (2003, March 20). The favorite/longshot bias in S&P 500 and FTSE 100 index futures options: the return to bets and the cost of insurance. [Conference paper]. Retrieved from http://www.cofar.uni- mainz.de/dgf2003/paper/paper19.pdf Hoffrage, U. (2011). Recognition judgments and the performance of the recognition heuristic depend on the size of the reference class. Judgment and Decision Making, 6(1), 43-57. Hoffrage, U., & Hertwig, R. (2006). Which world should be represented in representative design?. In K. Fiedler, & P. Juslin (Eds.), Information sampling and adaptive cognition (pp. 381-408). Cambridge University Press. Hogarth, R. M. (2001). Educating intuition. University of Chicago Press. Hogarth, R. M., & Karelaia, N. (2007). Heuristic and linear models of judgment: Matching rules and environments. Psychological Review, 114(3), 733-758. Hogarth, R. M., Lejarraga, T., & Soyer, E. (2015). The two settings of kind and wicked learning environments. Current Directions in Psychological Science, 24(5), 379-385. Hogarth, R. M., & Makridakis, S. (1981). Forecasting and planning: An evaluation. Management Science, 27(2), 115-138. Hooghe, L., Marks, G., & Wilson, C. J. (2002). Does left/right structure party positions on European integration?. Comparative political studies, 35(8), 965-989. Hughes, G. (1968). On the mean accuracy of statistical pattern recognizers. IEEE Transactions on Information Theory, 14(1), 55-63. Hummel, D., & Maedche, A. (2019). How effective is nudging? A quantitative review on the effect sizes and limits of empirical nudging studies. Journal of Behavioral and Experimental Economics, 80, 47-58. Hurley, W., & McDonough, L. (1995). A note on the Hayek hypothesis and the favorite-longshot bias in parimutuel betting. The American Economic Review, 949-955. Huxley, A. (1932). Brave new world: a novel. London: Chatto & Windus. Isaacs, R. (1953). Optimal horse race bets. The American Mathematical Monthly, 60(5), 310- 315. Jaderberg, M., Czarnecki, W. M., Dunning, I., Marris, L., Lever, G., Castaneda, A. G., ... & Sonnerat, N. (2019). Human-level performance in 3D multiplayer games with population-based reinforcement learning. Science, 364(6443), 859-865. Javetski, B., & Koller, T. (2018, May). Debiasing the corporation: An interview with Nobel laureate Richard Thaler. [Interview]. Retrieved from https://www.mckinsey.com/business-functions/strategy-and-corporate-finance/o ur-insights/debiasing-the-corporation-an-interview-with-nobel-laureate-richard- thaler Johnson, E. J., & Goldstein, D. (2003). Do defaults save lives?. Science, 302, 1338-1339. 168 THE GOOD PLACE

Johnson, E. J., Shu, S. B., Dellaert, B. G., Fox, C., Goldstein, D. G., Häubl, G., & Wansink, B. (2012). Beyond nudges: Tools of a choice architecture. Marketing Letters, 23(2), 487-504. Jolls, C., Sunstein, C. R., & Thaler, R. (1998). A behavioral approach to law and economics. Stanford law review, 1471-1550. Jullien, B., & Salanié, B. (2000). Estimating Preferences under Risk: The Case of Racetrack Bettors, Journal of Political Economy, 108, 503-530. Jullien, B., & Salanié, B. (2007, December 26). Empirical evidence on the preferences of racetrack bettors. [Article]. Retrieved from http://www.columbia.edu/~bs2237/HandbookJullienSalanie26Dec07.pdf Jung, J. Y., & Mellers, B. A. (2016). American attitudes toward nudges. Judgment and Decision Making, 11(1), 62-74. Juslin, P. (1994). The overconfidence phenomenon as a consequence of informal experimenter-guided selection of almanac items. Organizational Behavior and Human Decision Processes, 57(2), 226-246. Kahan, D. M. (2012). Cultural cognition as a conception of the cultural theory of risk. In S. Roeser (Ed.), Handbook of risk theory: Epistemology, decision theory, ethics, and social implications of risk, (pp. 725-759). Springer Science & Business Media. Kahneman, D. (2003). Maps of bounded rationality: A perspective on intuitive judgment and choice. In T. Frangsmyr (Ed.), Les Prix Nobel 2002. Stockholm: Almquist & Wiksell International. Kahneman, D. (2011). Thinking, fast and slow. New York: Farrar, Straus and Giroux. Kahneman, D. (2012, September 26). A proposal to deal with questions about priming effects. [Letter]. In E. Yong (2012, October 3), Nobel laureate challenges psychologists to clean up their act [Blog post]. Retrieved from https://www.nature.com/news/nobel-laureate-challenges-psychologists-to-clean- up-their-act-1.11535 Kahneman, D., Knetsch, J. L., & Thaler, R. H. (1991). Anomalies: The endowment effect, loss aversion, and status quo bias. Journal of Economic perspectives, 5(1), 193-206. Kahneman, D., Slovic, S. P., Slovic, P., & Tversky, A. (1982). Judgment under uncertainty: Heuristics and biases. Cambridge University Press. Kahneman, D., & Tversky, A. (1972). Subjective probability: A judgment of representativeness. Cognitive psychology, 3(3), 430-454. Kahneman, D., & Tversky, A. (1973). On the psychology of prediction. Psychological review, 80(4), 237-251. Kahneman, D., & Tversky, A. (1979). Prospect Theory: An Analysis of Decisions Under Risk. Econometrica, 47, 263-291. Kahneman, D., & Tversky, A. (1982). On the study of statistical intuitions. Cognition, 11(2), 123-141. Kahneman, D., & Tversky, A. (1984). Choices, values and frames. American Psychologist, 39(4), 341-350. REFERENCES 169

Kahneman, D., Knetsch, J. L., & Thaler, R. H. (1991). Anomalies: The endowment effect, loss aversion, and status quo bias. The journal of economic perspectives, 5(1), 193- 206. Kanto, A., & Rosenqvist, G. (2008). On the efficiency of the market for double (quinella) bets at a Finnish racetrack. In D. B. Hausch, V. S. Lo, & W. T. Ziemba (Eds.), Efficiency of racetrack betting markets (pp. 485-498). Singapore: World Scientific Publishing Co. Pte. Ltd.. Kanto, A. J., Rosenqvist, G., & Suvas, A. (1992). On utility function estimation of racetrack bettors. Journal of Economic Psychology, 13(3), 491-498. Kassin, S. M., Dror, I. E., & Kukucka, J. (2013). The forensic confirmation bias: Problems, perspectives, and proposed solutions. Journal of applied research in memory and cognition, 2(1), 42-52. Katona, G. (1953). Rational behavior and economic behavior. Psychological Review, 60(5), 307-318. Katona, G. (1975). Psychological economics. New York: Elsevier Scientific. Kelly, J. L. (1956). A new interpretation of information rate. Bell System Technical Journal, 35, 917-926. Keren, G. (2013). A tale of two systems: A scientific advance or a theoretical stone soup? Commentary on Evans & Stanovich (2013). Perspectives on Psychological Science, 8(3), 257-262. Keren, G., & Wu, G. (2015). The Wiley Blackwell Handbook of Judgment and Decision Making. New York: John Wiley & Sons. Keynes, J. M. (1921). A Treatise on Probability. London: Mcmillan. Keynes, J. M. (1937). The general theory of employment. The Quarterly Journal of Economics, 51, 209-223. Khuong, A., Gautrais, J., Perna, A., Sbaï, C., Combe, M., Kuntz, P., ... & Theraulaz, G. (2016). Stigmergic construction and topochemical information shape ant nest architecture. Proceedings of the National Academy of Sciences, 113(5), 1303-1308. Knight, F. H. (1921). Risk, uncertainty and profit. Cambridge, MA: Riverside Press. Koehler, D. J., & Harvey, N. (2008). Blackwell handbook of judgment and decision making. New York: John Wiley & Sons. Koivuranta, M., & Korhonen, M. (2018). Misperception explains favorite-longshot bias: evidence from the Finnish and Swedish harness horse race markets. Empirical Economics, 1-12. Kotlikoff, M. I. (2018, 20 September). Statement of Cornell University Provost Michael I. Kotlikoff. Retrieved from https://statements.cornell.edu/2018/20180920- statement-provost-michael-kotlikoff.cfm Krüger, L., Gigerenzer, G. & Morgan, M. (1987). The probabilistic revolution. Cambridge, MA: MIT Press. Kruglanski, A. W., & Gigerenzer, G. (2011). Intuitive and deliberate judgments are based on common principles. Psychological review, 118(1), 97-109. 170 THE GOOD PLACE

Kühberger, A. (1995). The framing of decisions: A new look at old problems. Organizational Behavior and Human Decision Processes, 62(2), 230-240. Kuhn, T. S. (1970). The structure of scientific revolutions. University of Chicago Press. Lakatos, I (1970). Falsification and the methodology of scientific research programmes. In I. Lakatos, & A. Musgrave (Eds), Criticism and the Growth of Knowledge (pp. 91-195). Cambridge University Press. Laney, D. (2001, February 6). 3-D Data Management: Controlling Data Volume, Velocity and Variety. [Research Note]. Retrieved from https://blogs.gartner.com/doug-laney/files/2012/01/ad949-3D-Data-Managemen t-Controlling-Data-Volume-Velocity-and-Variety.pdf LaRiviere, J., McAfee, P., Rao, J., Narayana, V. K., & Sun, W. (2016, May 25). Where predictive analytics is having the biggest impact. [Online article]. Retrieved from https://hbr.org/2016/05/where-predictive-analytics-is-having-the-biggest-impact Lazer, D., Kennedy, R., King, G., & Vespignani, A. (2014). The parable of Google Flu: Traps in big data analysis. Science, 343(6176), 1203-1205. Leddy, M. A., Anderson, B. L., & Schulkin, J. (2013). Cognitive-behavioral therapy and decision science. New Ideas in Psychology, 31(3), 173-183. Lewis, M. (2017). The Undoing Project: a friendship that changed our minds. New York: W. W. Norton. Lin, Y., Osman, M., & Ashcroft, R. (2017). Nudge: concept, effectiveness, and ethics. Basic and Applied Social Psychology, 39(6), 293-306. Lo, V. S., & Bacon-Shone, J. (1994). A comparison between two models for predicting ordering probabilities in multiple-entry competitions. The Statistician, 317-327. Lo, V. S., Bacon-Shone, J., & Busche, K. (1995). The application of ranking probability models to racetrack betting. Management Science, 41(6), 1048-1059. Lodge, M., & Wegrich, K. (2016). The rationality paradox of nudge: Rational tools of government in a world of bounded rationality. Law & policy, 38(3), 250-267. Loewenstein, G., Brennan, T., & Volpp, K. G. (2007). Asymmetric paternalism to improve health behaviors. Jama, 298(20), 2415-2417. Loewenstein, G., Bryce, C., Hagmann, D., & Rajpal, S. (2015). Warning: You Are About to Be Nudged. Behavioral Science & Policy, 1(1), 35-42. Loewenstein, G., & Chater, N. (2017). Putting nudges in perspective. Behavioural Public Policy, 1(1), 26-53. Lodge, M., & Wegrich, K. (2016). The rationality paradox of nudge: Rational tools of government in a world of bounded rationality. Law & policy, 38(3), 250-267. Loibl, C., Sunstein, C. R., Rauber, J., & Reisch, L. A. (2018). Which Europeans like nudges? Approval and controversy in four European countries. Journal of Consumer Affairs, 52(3), 655-688. Losey, R. L., & Talbott, J. C. (1980). Back on the track with the efficient markets hypothesis. The Journal of Finance, 35(4), 1039-1043. Luce, R. D., & Raiffa, H. (1957). Games and decisions. New York: John Wiley & Sons. REFERENCES 171

Ly, K., & Soman, D. (2013, September 3). Nudging around the world. [Research report]. Rotman School of Management, University of Toronto. Madi, M. A. (2020). The Dark Side of Nudges. New York: Routledge. Makridakis, S., & Hibon, M. (2000). The M3-Competition: results, conclusions and implications. International journal of forecasting, 16(4), 451-476. Makridakis, S., Hogarth, R. M., & Gaba, A. (2010). Why forecasts fail. What to do instead. MIT Sloan Management Review, 51(2), 83-90. Makridakis, S., Spiliotis, E., & Assimakopoulos, V. (2018a). Statistical and Machine Learning forecasting methods: Concerns and ways forward. PLoS ONE, 13(3), e0194889. [Article]. Retrieved from https://doi.org/10.1371/journal.pone.0194889 Makridakis, S., Spiliotis, E., & Assimakopoulos, V. (2018b). The M4 Competition: Results, findings, conclusion and way forward. International Journal of Forecasting, 34(4), 802-808. Mayer, P. C. (1989). On ‘Anomalies, Parimutuel Betting Markets: Racetracks and Lotteries,’ by Thaler and Ziemba. The Journal of Economic Perspectives, 3(2), 198. Mayer-Schönberger, V. & Cukier, K. (2013). Big Data: A Revolution That Will Transform How We Live, Work, and Think. London: John Murray Publishers. McAfee, A. & Brynjolfsson, E. (2012, October). Big data: The management revolution. [Article]. Retrieved from https://hbr.org/2012/10/big-data-the-management- revolution McAfee, A., Brynjolfsson, E., & Davenport, T. H. (2012). Big data: The management revolution. Harvard Business Review, 90(10), 60-68. McGlothlin, W. H. (1956). Stability of choices among uncertain alternatives. The American journal of psychology, 69(4), 604-615. Meder, B., Fleischhut, N., & Osman, M. (2018). Beyond the confines of choice architecture: a critical analysis. Journal of Economic Psychology, 68, 36-44. Metzger, M. A. (1985). Biases in betting: An application of laboratory findings. Psychological Reports, 56(3), 883-888. Mill, J. S. (1860). On liberty. London: John W. Parker & Son. Miller, J. B., & Sanjurjo, A. (2018). Surprised by the hot hand fallacy? A truth in the law of small numbers. Econometrica, 86(6), 2019-2047. Millroth, P., Nilsson, H. & Juslin, P. (2019). The decision paradoxes motivating Prospect Theory: The prevalence of the paradoxes increases with numerical ability. Judgment and decision making, 14(4), 513-533. Mols, F., Haslam, S. A., Jetten, J., & Steffens, N. K. (2015). Why a nudge is not enough: A social identity critique of governance by stealth. European Journal of Political Research, 54(1), 81-98. More, T. (1984). Utopia and other essential writings of Thomas More. New York: New American Library. 172 THE GOOD PLACE

Moscati, I. (2016). Retrospectives: how economists came to accept expected utility theory: the case of Samuelson and Savage. Journal of economic perspectives, 30(2), 219- 236. Murphy, A. H. (1993). What is a good forecast? An essay on the nature of goodness in weather forecasting. Weather and Forecasting, 8(2), 281-293. Nelson, L. (2015). Audio recording (m4a) of Leif Nelson's interview of Danny Kahneman. [Interview]. Retrieved from http://www.sjdm.org/archive/KahnemanInterview.m4a von Neumann, J., & Morgenstern, O. (1947). Theory of games and economic behavior. Princeton University Press. Newall, P. W. (2018). Dark nudges in gambling. Addiction Research & Theory, 27(2), 65-67. Newall, P. W., & Love, B. C. (2015). Nudging investors big and small toward better decisions. Decision, 2(4), 319-326. Newell, B. R. (2005). Re-visions of rationality?. Trends in cognitive sciences, 9(1), 11-15. Newell, B. R. (2011). Recognising the recognition heuristic for what it is (and what it's not). Judgment and Decision Making, 6(5), 409-412. Nicholson, J. D., LaPlaca, P., Al-Abdin, A., Breese, R., & Khan, Z. (2018). What do introduction sections tell us about the intent of scholarly work: A contribution on contributions. Industrial Marketing Management, 73, 206-219. Nilsson, A., Erlandsson, A., Västfjäll, D., & Tinghög, G. (2020). Who are the opponents of nudging? Insights from moral foundations theory. Comprehensive Results in Social Psychology, 1-34. Noyes, J., McLaughlin, L., Morgan, K., Walton, P., Curtis, R., Madden, S., ... Stephens, M. (2019). Short-term impact of introducing a soft opt-out organ donation system in Wales: before and after study. BMJ Open, 9(4), e025159. [Article]. Retrieved from https://doi.org/10.1136/bmjopen-2018-025159 Nozick, R. (1974). Anarchy, state, and Utopia. Oxford: Blackwell. Nudel, L., & Wiik, E. (2017). Nudge is all around – but what is around the nudge? [Unpublished M.Sc. thesis]. Stockholm School of Economics. OECD (2017, March 1). Behavioural Insights and Public Policy. Lessons from Around the World. [Report]. Retrieved from http://www.oecd.org/gov/regulatory- policy/behavioural-insights-and-public-policy-9789264270480-en.htm Oliver, A., & Ubel, P. (2014). Nudging the obese: a UK-US consideration. Health Economics, Policy and Law, 9(3), 329-342. Olsson, H. (2014). Measuring overconfidence: Methodological problems and statistical artifacts. Journal of Business Research, 67(8), 1766-1770. Open Science Collaboration. (2015). Estimating the reproducibility of psychological science. Science, 349(6251), 943. Ortmann, A., Gigerenzer, G., Borges, B., & Goldstein, D. G. (2008). The recognition heuristic: a fast and frugal way to investment choice?. Handbook of experimental economics results, 1, 993-1003. REFERENCES 173

Orwell, G. (1949). Nineteen eighty-four: a novel. London: Secker & Warburg. Osman, M. (2004). An evaluation of dual-process theories of reasoning. Psychonomic Bulletin & Review, 11(6), 988-1010. Osman, M., Radford, S., Lin, Y., Gold, N., Nelson, W., & Löfstedt, R. (2018). Learning lessons: how to practice nudging around the world. Journal of Risk Research, 1-9. Oscarsson, H. (2016). Floating voters. Democracy Statistics Report, 21. Oscarsson, H., & Holmberg, S. (2016). Issue voting structured by left–right ideology. In J. Pierre (Ed.), The Oxford Handbook of Swedish Politics (pp. 260-274). London: Oxford University Press. Ottaviani, M., & Sørensen, P. N. (2008). The favorite-longshot bias: An overview of the main explanations. Handbook of Sports and Lottery markets, 83-101. Ottaviani, M., & Sørensen, P. N. (2009). Surprised by the parimutuel odds? The American Economic Review, 99(5), 2129-2134. Ottaviani, M., & Sørensen, P. N. (2010). Noise, information, and the favorite-longshot bias in parimutuel predictions. American Economic Journal: Microeconomics, 2(1), 58-85. Payne, J. W., Bettman, J. R., & Johnson, E. J. (1993). The adaptive decision maker. Cambridge University Press. Pedersen, S. K., Koch, A. K., & Nafziger, J. (2014). Who Wants Paternalism? Economics Bulletin, 66, 147-166. Peters, E., Västfjäll, D., Gärling, T., & Slovic, P. (2006). Affect and decision making: A “hot” topic. Journal of behavioral decision making, 19(2), 79-85. Plott, C. R., Wit, J., & Yang, W. C. (2003). Parimutuel betting markets as information aggregation devices: experimental results. Economic Theory, 22(2), 311-351. Plott, C. R., & Zeiler, K. (2005). The willingness to pay-willingness to accept gap, the “endowment effect,” subject misconceptions, and experimental procedures for eliciting valuations. American Economic Review, 95(3), 530-545. Pohl, R. F. (2011). On the use of recognition in inferential decision making: An overview of the debate. Judgment and Decision Making, 6(5), 423-438. Polk, J., Rovny, J., Bakker, R., Edwards, E., Hooghe, L., Jolly, S., ... & Steenbergen, M. (2017). Explaining the salience of anti-elitism and reducing political corruption for political parties in Europe with the 2014 Chapel Hill Expert Survey data. Research & Politics, 4(1), 1-9. Pollio, V. (1914). Vitruvius: the ten books on architecture. Harvard University Press. Pólya, G. (1945). How to solve it: a new aspect of mathematical method. Princeton University Press. Puaschunder, J. M. (2017). Nudging in the digital big data era. European Journal of Economics, Law and Politics, 4(4), 18-23. Preston, M. G., & Baratta, P. (1948). An experimental study of the auction-value of an uncertain outcome. The American journal of psychology, 61(2), 183-193. Qiang, X. (2019). The Road to Digital Unfreedom: President Xi's Surveillance State. Journal of Democracy, 30(1), 53-67. 174 THE GOOD PLACE

Quandt, R. E. (1986). Betting and equilibrium. The Quarterly Journal of Economics, 101(1), 201-207. Radiohead (1997). Fitter Happier. In N. Godrich, C. Greenwood, J. Greenwood, E. O'Brien, P. Selway, & T. Yorke (producers), OK Computer [album]. London: EMI. Ramsberg, F. (2016). När det rätta blir det lätta – en ESO-rapport om ”nudging”. [Report]. Retrieved from https://eso.expertgrupp.se/wp- content/uploads/2016/02/Hela_ESO_2016-7_web.pdf Ramsey, F. P. (1931). Truth and probability. The foundations of mathematics and other logical essays, 156-198. Rawls, J. (2009). A theory of justice. Harvard University Press. Rebonato, R. (2012). Taking liberties: A critical examination of libertarian paternalism. Basingstoke: Palgrave Macmillan. Rebonato, R. (2014). A critical assessment of libertarian paternalism. Journal of Consumer Policy, 37(3), 357-396. Reijula, S., Kuorikoski, J., Ehrig, T., Katsikopoulos, K., & Sunder, S. (2018). Nudge, boost, or design? Limitations of behaviorally informed policy under social interaction. Journal of behavioral economics for policy, 2(1), 99-105. Reynolds, J. P., Archer, S., Pilling, M., Kenny, M., Hollands, G. J., & Marteau, T. M. (2019). Public acceptability of nudging and taxing to reduce consumption of alcohol, tobacco, and food: a population-based survey experiment. Social Science & Medicine, 236, 112395. Reisch, L. A. (2020). Nudging hell und dunkel: Regeln für digitales Nudging. Wirtschaftsdienst, 100(2), 87-91. Reisch, L. A., & Sunstein, C. R. (2016). Do Europeans like nudges?. Judgment and Decision Making, 11(4), 310-325. Reisch, L. A., Sunstein, C. R., & Gwozdz, W. (2017). Beyond carrots and sticks: Europeans support health nudges. Food Policy, 69, 1-10. Rich, A. S., & Gureckis, T. M. (2019). Lessons for artificial intelligence from the study of natural stupidity. Nat Mach Intell, 1, 174-180. Rossett, R. N. (1965). Gambling and rationality. Journal of Political Economy, 6, 595-607. Rossett, R. N. (1971). Weak experimental verification of the expected utility hypothesis. Review of Economic studies, 38, 481-492. Rutter, T. (2014, June 1). What next for the nudge unit? [Article]. Retrieved from https://www.theguardian.com/public-leaders-network/2014/jun/01/nudge-unit- behavioural-insights-team-conference Samuelson, P. A. (1952). Probability, utility, and the independence axiom. Econometrica: Journal of the Econometric Society, 670-678. Samuelson, W., & Zeckhauser, R. (1988). Status quo bias in decision making. Journal of risk and uncertainty, 1(1), 7-59. Sanders, N. R. (2014). Big data driven supply chain management: A framework for implementing analytics and turning information into intelligence. Upper Saddle River: Pearson Education. REFERENCES 175

Savage, L. J. (1954). The Foundations of Statistics. New York: John Wiley and Sons. Scheiber, N. (2017, April 2). How Uber Uses Psychological Tricks to Push Its Drivers’ Buttons. [News article]. Retrieved from https://www.nytimes.com/interactive/201 7/04/02/technology/uber-drivers-psychological-tricks.html Schimmack, U. (2018, December 29). 2018 Journal Replicability Rankings. [Blog post]. Retrieved from https://replicationindex.com/2018/12/29/2018-replicability-rankings/ Schmidt, A. T. (2017). The power to nudge. American Political Science Review, 111(2), 404- 417. Schoemaker, P. J. (2004). Forecasting and scenario planning: the challenges of uncertainty and complexity. Blackwell Handbook of Judgment and Decision Making, 274- 296. Schroeck, M., Shockley, R., Smart, J., Romero-Morales, D., & Tufano, P. (2012). Analytics: The real-world use of big data. IBM Global Business Services, 1-20. Schumpeter, J. A. (1954). History of economic analysis. London: Allen & Unwin. Schur, M., & Goddard, D. (2016). Pilot. In D. Goddard, D. Miner, M. Sackett, & M. Schur (executive producers), The Good Place [TV series]. New York: NBC. Sent, E. M. (2004). Behavioral economics: how psychology made its (limited) way back into economics. History of political economy, 36(4), 735-760. Serwe, S., & Frings, C. (2006). Who will win Wimbledon? The recognition heuristic in predicting sports events. Journal of Behavioral Decision Making, 19(4), 321-332. Sheldon, M. (2016, February 18). Income Targeting and the Ridesharing Market. [Working paper]. Retrieved from https://static1.squarespace.com/static/56500157e4b0cb706005352d/t/56da1114e 707ebbe8e963ffc/1457131797556/IncomeTargetingFeb16.pdf Sibony, A. L. (2020). The UK COVID-19 Response: A Behavioural Irony?. European Journal of Risk Regulation, 1-11. Simon, H. A. (1955). A behavioral model of rational choice. The Quarterly Journal of Economics, 69, 99-118. Simon, H. A. (1956). Rational Choice and the Structure of the Environment. Psychological Review, 63(2), 129-138. Simon, H. A. (1957). Models of man: social and rational: mathematical essays on rational human behaviour in a social setting. New York: John Wiley & Sons. Simon, H. A. (1958). The role of expectations in an adaptive or behavioristic model. In M. J. Bowman (Ed.), Expectations, uncertainty, and business behavior (pp. 49-58). New York: Social Science Research Council. Simon, H. A. (1962). The architecture of complexity. Proceedings of the American Philosophical Society, 106(6), 467-482. Simon, H. A. (1976). From substantive to procedural rationality. In S. J. Latsis (Ed.), Method and Appraisal in Economics (pp. 129-148). Cambridge University Press. Simon, H. A. (1985). Human nature in politics: The dialogue of psychology with political science. American Political Science Review, 79(2), 293-304. 176 THE GOOD PLACE

Simon, H. A. (1986). Rationality in psychology and economics. Journal of Business, 59(4), 209-224. Simon, H. A. (1989). The scientist as problem solver. Complex information processing: The impact of Herbert A. Simon, 375-398. Simon, H. A. (1990). Invariants of human behavior. Annual review of psychology, 41(1), 1- 20. Simon, H. A. (1997). Administrative Behavior: a Study of Decision-Making Processes in Administrative Organizations. New York: Simon & Schuster. Sivarajah, U., Kamal, M. M., Irani, Z., & Weerakkody, V. (2017). Critical analysis of Big Data challenges and analytical methods. Journal of Business Research, 70, 263-286. Slovic, P., Fischhoff, B., & Lichtenstein, S. (1977). Behavioral decision theory. Annual review of psychology, 28(1), 1-39. Smith, A. (1759). The theory of moral sentiments. New York: Garland. Snowberg, E., & Wolfers, J. (2010). Explaining the Favorite–Long Shot Bias: Is it Risk- Love or Misperceptions?. Journal of Political Economy, 118(4), 723-746. Snyder, W. W. (1978). Horse racing: Testing the efficient markets model. The Journal of finance, 33(4), 1109-1118. Sobel, R. S., & Travis Raines, S. (2003). An examination of the empirical derivatives of the favourite-longshot bias in racetrack betting. Applied Economics, 35(4), 371-385. Soyer, E., & Hogarth, R. M. (2015). The golden rule of forecasting: objections, refinements, and enhancements. Journal of Business Research, 68(8), 1702-1704. Stern, H. (1990). Models for distributions on permutations. Journal of the American Statistical Association, 85(410), 558-564. Stroop, J. R. (1935). Studies of interference in serial verbal reactions. Journal of experimental psychology, 18(6), 643-662. Sugden, R. (2017). Do people really want to be nudged towards healthy lifestyles?. International Review of Economics, 64(2), 113-123. Sugden, R. (2018). ‘Better off, as judged by themselves’: a reply to Cass Sunstein. International Review of Economics, 65(1), 9-13. Sung, M. C., Johnson, J. E. V., & Dror, I. E. (2009). Complexity as a guide to understanding decision bias: A contribution to the favorite-longshot bias debate. Journal of Behavioral Decision Making, 22(3), 318-337. Sunstein, C. R. (1997). Behavioral analysis of law. The University of Chicago law review, 64(4), 1175-1195. Sunstein, C. R. (2013). Simpler: The future of government. New York: Simon & Schuster. Sunstein, C. R. (2014). Nudging: A Very Short Guide. Journal of Consumer Policy, 37(4), 583-588. Sunstein, C. R. (2015a). The ethics of nudging. Yale journal on regulation, 32(2), 413-450. Sunstein, C. R. (2015b). Nudges do not undermine human agency. Journal of Consumer Policy, 38(3), 207-210. REFERENCES 177

Sunstein, C. R. (2015c, June 22). Which nudges do people like? A national survey. [Discussion paper]. Retrieved from http://www.law.harvard.edu/programs/olin_center/papers/pdf/Sunstein_849.pdf Sunstein, C. R. (2016a). The ethics of influence: Government in the age of behavioral science. Cambridge University Press. Sunstein, C. R. (2016b). The Council of Psychological Advisers. Annual Review of Psychology, (67), 713-737. Sunstein C. R. (2017a) People Like Nudges (Mostly). In: Human Agency and Behavioral Economics. Palgrave Advances in Behavioral Economics. Palgrave Macmillan, Cham. Sunstein, C. R. (2017b). Human agency and behavioral economics: nudging fast and slow. New York, NY: Palgrave Macmillan. Sunstein, C. R., Misconceptions About Nudges (2017c, September 6). [Article]. Retrieved from https://ssrn.com/abstract=3033101 Sunstein, C. R. (2017d). Nudges that fail. Behavioural Public Policy, 1(1), 4-25. Sunstein, C. R. (2018). “Better off, as judged by themselves”: a comment on evaluating nudges. International Review of Economics, 65(1), 1-8. Sunstein, C. R., Reisch, L. A., & Kaiser, M. (2019). Trusting nudges? Lessons from an international survey. Journal of European Public Policy, 26(10), 1417-1443. Sunstein, C. R., Reisch, L. A., & Rauber, J. (2018). A worldwide consensus on nudging? Not quite, but almost. Regulation & Governance, 12(1), 3-22. Sumpter, D. (2019). Uträknad: sanningen om algoritmerna som styr världen. Stockholm: Volante. Stephensson, D. B. (2002, October 29). David B. Stephenson’s proposed classification scheme for forecasts. Retrieved from http://www.cawcr.gov.au/projects/verification/forecast-types.pdf Svenson, O. (1981). Are we all less risky and more skillful than our fellow drivers?. Acta psychologica, 47(2), 143-148. Sveriges Riksbank (2017, May 2). Utvärdering av Riksbankens prognoser. [Report]. Retrieved from http://www.riksbank.se/Documents/Rapporter/Riksbanksstudie/2017/rap_riksb anksstudie_170502_sve.pdf Taleb, N. N. (2007). The black swan: The impact of the highly improbable. New York: Random House. Tannenbaum, D., Fox, C. R. & Rogers, T. (2017). On the misplaced politics of behavioural policy interventions. Nature Human Behaviour, 1, 1-7. Thaler, R. H. (1980). Toward a positive theory of consumer choice. Journal of Economic Behavior & Organization, 1(1), 39-60. Thaler, R. H. (1985). Mental accounting and consumer choice. Marketing science, 4(3), 199-214. Thaler, R. H. (1990). Anomalies: Saving, fungibility, and mental accounts. The Journal of Economic Perspectives, 4(1), 193-205. 178 THE GOOD PLACE

Thaler, R. H. (2015). Misbehaving: The making of behavioral economics. New York: WW Norton & Company. Thaler, R. H. (2017, June 2). Much Ado About Nudging. [Blog post]. Retrieved from https://bppblog.com/2017/06/02/much-ado-about-nudging/ Thaler, R. H. (2018). Nudge, not sludge. Science, 361(6401), 431. Thaler, R. H., & Benartzi, S. (2001). Naive diversification strategies in defined contribution saving plans. American Economic Review, 91(1), 79-98. Thaler, R. H., & Benartzi, S. (2004). Save more tomorrow™: Using behavioral economics to increase employee saving. Journal of political Economy, 112(1), 164-187. Thaler, R. H., & Shefrin, H. M. (1981). An economic theory of self-control. The Journal of Political Economy, 89(2), 392-406. Thaler, R. H., & Sunstein, C. R. (2003). Libertarian paternalism. The American Economic Review, 93(2), 175-179. Thaler, R. H., & Sunstein, C. R. (2008). Nudge: Improving decisions about health, wealth and happiness. London: Penguin Books. Thaler, R. H., & Ziemba, W. T. (1988). Anomalies: Parimutuel betting markets: Racetracks and lotteries. The Journal of Economic Perspectives, 2(2), 161-174. The Asan Institute for Policy Studies (2017, February 28). [Blog post]. Orwell’s Nightmare: China’s Social Credit System. Retrieved from http://en.asaninst.org/contents/orwells-nightmare-chinas-social-credit-system/ The Committee for the Prize in Economic Sciences in Memory of Alfred Nobel (2017, October 9). The Prize in Economic Sciences 2017. [Pressrelease]. Retrieved from https://www.nobelprize.org/uploads/2018/06/press-43.pdf The World Bank (2015). World development report 2015: Mind, society, and behavior. [Report]. World Bank Publications. Thunström, L. (2019). Welfare effects of nudges: The emotional tax of calorie menu labeling. Judgment and Decision making, 14(1), 11-25. Todd, P. M., & Brighton, H. (2016). Building the theory of ecological rationality. Minds and Machines, 26(1-2), 9-30. Todd, P. M., & Gigerenzer, G. (2000). Preciś of Simple heuristics that make us smart. Behavioral and Brain Sciences, 23, 727-780. Todd, P. M., & Gigerenzer, G. (2007). Environments that make us smart: Ecological rationality. Current Directions in Psychological Science, 16, 167-171. Todd, P. M., Gigerenzer, G., & the ABC Research Group (2012). Ecological rationality: Intelligence in the world. New York: Oxford University Press. Tooby, J., & Cosmides, L. (1992). The psychological foundations of culture. In J. Barkow, L. Cosmides, & J. Tooby (Eds.), The adapted mind: Evolutionary psychology and the generation of culture (pp. 19-136). New York: Oxford University Press. Tooby, J., & Cosmides, L. (2005). Conceptual Foundations of Evolutionary Psychology. In D. M. Buss (Ed.), Evolutionary psychology: The new science of the mind (pp. 5-67). New York: John Wiley & Sons. REFERENCES 179

Tversky, A., & Kahneman, D. (1971). Belief in the law of small numbers. Psychological bulletin, 76(2), 105-110. Tversky, A., & Kahneman, D. (1973). Availability: A heuristic for judging frequency and probability. Cognitive psychology, 5(2), 207-232. Tversky, A., & Kahneman, D. (1974). Judgment under uncertainty: Heuristics and biases. Science, 185, 1124-1131. Tversky, A., & Kahneman, D. (1981). The Framing of decisions and the psychology of choice. Science, 211, 453-458. Tversky, A., & Kahneman, D. (1983). Extensional Versus Intuitive Reasoning: The Conjunction Fallacy in Probability Judgment. Psychological Review, 90(4), 293-315. Tversky, A., & Kahneman, D. (1986). Rational choice and the framing of decisions. Journal of business, 59(4), 251-278. Tversky, A., & Kahneman, D. (1992). Advances in prospect theory: Cumulative representation of uncertainty. Journal of Risk and uncertainty, 5(4), 297-323. Tversky, A., & Kahneman, D. (1996). On the reality of cognitive illusions. Psychological Review, 103(3), 582-591. Ward, J. S., & Barker, A. (2013, 20 September). Undefined By Data: A Survey of Big Data Definitions. Retrieved from http://www.adambarker.org/papers/undefined_by_data_ ward_barker.pdf Weimer, D. L. (2020). When Are Nudges Desirable? Benefit Validity When Preferences Are Not Consistently Revealed. Public Administration Review, 80(1), 118-126. Weinmann, M., Schneider, C., & vom Brocke, J. (2016). Digital nudging. Business & Information Systems Engineering, 58(6), 433-436. Weitzman, M. (1965). Utility Analysis and Group Behavior: An Empirical Study. The Journal of Political Economy, 73(1), 18-26. Wennberg, K. & Nykvist, B. (2007). The Psychology of Economic Forecasting. Global Business and Economics Review, 9(2), 211-226. Whitehead, M., Jones, R., Howell, R., Lilley, R., & Pykett, J. (2014). Nudging all over the world: Assessing the global impact of the behavioural sciences on public policy. [Research report]. Economic Social & Research Council. Wilkinson, T. M. (2013). Nudging and Manipulation. Political Studies, 61, 341-355. Wilson, D. K., Kaplan, R. M., & Schneiderman, L. J. (1987). Framing of decisions and selections of alternatives in health care. Social Behaviour, 2(1), 51-59. Winkler, R. L., & Murphy, A. H. (1968). “Good” probability assessors. Journal of Applied Meteorology, 7(5), 751-758. Wübben, M., & Wangenheim, F. V. (2008). Instant customer base analysis: Managerial heuristics often “get it right”. Journal of Marketing, 72(3), 82-93. Yeung, K. (2019, March 19). Nobel Prize winner Richard Thaler says China should use “nudge theory” to improve the lives of citizens. [News article]. Retrieved from https://www.scmp.com/economy/china-economy/article/3002362/nobel-prize- winner-richard-thaler-says-china-should-use-nudge 180 THE GOOD PLACE van der Zee, T. (2017, 21 March). The Wansink Dossier: An Overview. [Blog post]. Retrieved from http://www.timvanderzee.com/the-wansink-dossier-an-overview/ Ziemba, W. T., & Hausch, D. B. (1986). Betting at the Racetrack. New York: Norris M. Strauss.

  

End.