Boosting Through Structured Introspection : Exploring Decision-Making in Relation to the COVID-19 Pandemic

(1)

Linköping University | Department of Computer and Information Science Bachelor’s thesis 18 ECTS | Cognitive Science Spring Term 2020 | LIU-IDA/KOGVET-G--20/010--SE

Boosting Through Structured

Introspection

- Exploring Decision-Making in Relation to the

COVID-19 Pandemic

Christoffer Campbell

Tutor: Erkin Asutay

(2)

Copyright

The publishers will keep this document online on the Internet – or its possible replacement – for a period of 25 years starting from the date of publication barring exceptional circumstances.

The online availability of the document implies permanent permission for anyone to read, to download, or to print out single copies for his/her own use and to use it unchanged for non-commercial research and educational purpose. Subsequent transfers of copyright cannot revoke this permission. All other uses of the document are conditional upon the consent of the copyright owner. The publisher has taken technical and administrative measures to assure authenticity, security and accessibility.

According to intellectual property law the author has the right to be mentioned when his/her work is accessed as described above and to be protected against infringement.

For additional information about the Linköping University Electronic Press and its procedures for publication and for assurance of document integrity, please refer to its www home page: http://www.ep.liu.se/.

(3)

Abstract

This thesis explores boosting to improve decision-making in the context of the COVID-19 pandemic using a structured introspection. Structured introspection is an intervention where individuals are prompted with and are asked to estimate the importance of a set of attributes relevant to the decision in order to limit the prevalence of potential cognitive biases. To test the intervention, 281 participants divided into an intervention and control group answered an online survey with a dilemma about COVID-19. The dilemma was whether Sweden should shut down the economy or keep it open during the COVID-19 pandemic. The intervention group was asked to rate how important the attributes “saving lives”, “saving the economy”, “concern for the health of the elderly and risk groups”, and “concern for the quality of life and well-being of all citizens” should be for their decision. The control group was only prompted with the question and asked to think carefully. All participants were asked a set of control variables such as risk perception for self and others and emotions when thinking about COVID-19. The results did not show a significant influence on choice on decisions based on the intervention. They did however show a significant correlation with choice on risk perception as well as a correlation between choice on the dependent variable and the attributes in the intervention group.

The conclusion of the thesis is that structured introspection may not be suitable on a contemporary issue affecting participants directly, as they may already have strong opinions about the issue. Further and broader research needs to be conducted to determine in which circumstances this boost can be effective.

(4)

Acknowledgements

A massive thank you to my supervisor Erkin Asutay for giving me exceptionally good advice on my constant bombardment of small and large questions, and thanks to all my friends and family who have put up with me during these past months.

(5)

3.5 Data Analysis ... 16 3.6 Exclusions ... 16 4 Results ... 17 4.1 Descriptive Results ... 17 4.2 Dependent Variable... 18 4.3 Decision Attributes ... 18 4.4 Secondary Analysis ... 20 5 Discussion ... 21 5.1 Results... 21 5.2 Method ... 22 5.3 Future Research ... 23 6 Conclusion ... 24 References ... 25 Appendix ... 27

(7)

1 Introduction

The most common way to influence behavior has historically been by using hard paternalism such as legislation or economic incentives. An example of this is financing retirement funds. It is common for employers to match the savings of an employee in order to incentivize saving money for their retirement fund. Another example is smoking cigarettes, where the taxes have been increased and some public places have outlawed it in order to make smoking less

convenient and sometimes impossible. Since the discovery of bounded rationality and cognitive

heuristics, other methods have been developed as an alternative, attempting to change behavior

without restricting the freedom of choice. One prominent strategy is by using nudges. Nudges are simple, fast, and cost-effective ways to change the choice architecture, the way a decision is presented, to favor a specific decision (Thaler & Sunstein, 2008).

Recently, another method called boosting have been presented as an alternative to nudges. Critics claim that nudges, while proved to be very effective, are manipulative and lack transparency, as they are often presented in a way that do not notify the decision maker that they are being influenced. Furthermore, nudges require that the designer has the individual's best as their intention, and that they have a realistic view of what that is. Boosting instead has the

intention to improve the person's already present competences, or help them acquire new ones, to make better decisions (Hertwig & Grüne-Yanoff, 2017). Boosting therefore does not lead to pushing the decision maker into a pre-determined decision, but rather what the decision maker considers a good decision.

Nudges and boosts have been compared before, but testing boosts in an experimental setting has been limited. Since there are several arguments against the use of nudges, it is important to study if boosts can be used as an alternative or complement in order to help

individuals and organizations make better decisions while maintaining transparency and avoiding the need for a benevolent designer. This thesis will use a boost in the form of a structured

introspection. A structured introspection is a task where participants are given several attributes

related to the decision to consider before deciding, in order to limit incorrect use of potential heuristics and biases. The aim of this thesis is to discover the potential for structured

introspection-tasks as an alternative to other strategies.

1.1.1 COVID-19

COVID-19, also known as the novel Coronavirus, is an infectious disease caused by the virus SARS-CoV-2. COVID-19 started spreading in China in December 2019, and more widely across the world during the first months of the year 2020. On the 11th of March, the disease was

classified as a pandemic by the World Health Organization (2020). On the 2nd_{of April, the world}

reached 1,000,000 confirmed cases and 60,000 deaths (BBC, 2020). A rapid infection rate and relatively high mortality rate has led to a large variation in the strategies of countries to battle the

(8)

virus, as it has been a threat with several of unknown factors that could influence the health of society in more ways than infections.

It is important to study COVID-19 since the world is battling the threat of a pandemic. With this, governments and experts alike face new decisions where the wrong decision could have extreme consequences. Studying boosts through the lens of COVID-19 can irradiate how people generally think of decisions, especially unclear and emotional dilemmas. The hope is that another piece to the puzzle can be laid for how to consider this type of decisions.

1.2 Purpose

The purpose of this thesis is to explore the concept and mechanisms of boosting through an experiment using structured introspection. This is to increase the understanding of how people make decisions about emotional and unforeseeable dilemmas. By doing this, we can understand how to present information more efficiently, leading to both better insight and decisions.

The aim of the experiment is to discover to what extent boosting in general, and

structured introspection in particular, can be used in order to improve decision-making, and how much these effects are influenced by heuristics, risk perception, emotions, and utilitarianism.

1.2.1 Problem specification

Can boosting through structured introspection be used to aid reasoning and decision-making?

1.2.2 Delimitations

The survey was conducted in Swedish and the questions were specific to the COVID-19 spread in Sweden. Only adults above the age of 18 participated in the study. Data was quantitative and collected through an online survey.

(9)

2 Theoretical Background

2.1 Decision-Making

A traditional model of human decision-making is the one of the economical human, which claims that in order to make decisions, individuals gather all necessary information and make a rational and comprehensive decision based on this data. A person with an infinite amount of time and rationality, devoid of emotions and other circumstances influencing their decisions, might be able to use above-mentioned strategy for decisions. But in recent decades, behavioral science has made a case that this is rarely the case in real life. Simon (1956) famously coined the term

bounded rationality, explaining that in real-life situations, humans are boundedly rational,

making decisions with a finite amount of cognitive resources and time. We therefore use heuristics, mental shortcuts or rules of thumb, to make fast and ‘good enough’ decisions. Many of these heuristics are subconscious and we might not be aware of using them.

Heuristics are present in many areas of cognition. Tversky and Kahneman (1974) modelled a series of cognitive heuristics and biases, and in extension a theory about human rationality where they argued that decision-making-based heuristics in many cases are irrational and exploitable. Based on this work, Kahneman later divided cognitive processes into two systems: System 1, responsible for quick, emotional, and rough estimative thinking, and System

2, which instead does slower, more exact, and logical tasks. For example, simple arithmetic that

one has practiced over and over to calculate next to instantly is done by System 1, while more advanced calculations are done by System 2. Shouting and being enraged at the car in front of us for not using the turn-signal is a product of System 1-thinking while parking in a small space requires more of System 2. The division is, however, not exact, and there is a lot of overlap between the systems. For example, some parts within more advanced calculations might still be carried out by System 1. Many of the heuristics formulated by Kahneman and Tversky are based on this division, with System 1 being affected by heuristics and biases the most (Kahneman, 2011).

Not all rules of thumb can be considered heuristics. Gigerenzer (2004) argues that there are three criteria for what can be considered a cognitive heuristic:

1. Heuristics exploit evolved capacities. They are simple to use and learn based on our biology and past experiences.

2. Heuristics exploit how environments are structured. Heuristics are specific to a domain and designed to solve specific types of problems. Evolved capacities based on our biology make the heuristics simple, but the way the environment is structured is how a heuristic becomes smart.

3. Heuristics are not as-if optimization models. As-if models are cognitive models where people act ‘as if’ they were completely rational.

(10)

There are however different views on the mechanics and rationality of heuristics. Arkes,

Gigerenzer and Hertwig (2016) instead argue that people rely on simple heuristics when making decisions under uncertainty, time pressure and a lack of knowledge, both on an individual level and in organizations. Unlike the heuristics and biases-approach by Kahneman and Tversky, the simple heuristics approach claims that humans are not inherently irrational and error-prone, but rather describes heuristics as an adaptive toolbox and that the use of heuristics in situations with a lack of information, cognitive resources and time rather makes people better at inference and predictive reasoning and are reliable ways to take advantage of the information presented around us. There are many examples of heuristics that have been developed from the different

approaches, many of which have clear positive uses as well as examples of when they backfire. Based on these models for decision-making, new strategies have been developed to change public policies for how to help people improve their decisions.

2.1.1 Focusing illusion

Focusing illusion is a cognitive bias common in decision-making. When people focus on only

one factor for a decision, the weight they put on this factor becomes very large. People often do not have a clear view on their own opinions toward normative factors of well-being, such as happiness or perceived health. Instead, they cling to the most accessible data point and let this be the foundation for their judgement. A famous example is a study where participants were asked how they would rate their happiness and how many dates they had been to in the last month. The researchers found that their answer was influenced by the order of the questions. When the question about happiness came first, it had little, 0.12, correlation with how many dates the person had been to in the last month. But when the question about number of dates was asked first, it had a much higher correlation with how they rated their happiness, 0.66. The focusing illusion can create the illusion that one factor dominates one's well-being as it is difficult to give an appropriate amount of influence to both the focused factor and those in the background (Kahneman, Krueger, Schkade, Schwartz, & Stone, 2006; Schkade & Kahneman, 1998).

2.1.2 Availability heuristic

Availability heuristic is a cognitive bias discovered by Tversky and Kahneman (1974). The

availability heuristic works from an illusion that if something can be easily recalled in our minds, it must be frequent and important. Kahneman and Tversky argues that being able to easily recall a phenomenon makes us neglect the actual probability or effect of that phenomenon. An example used in the article is assessing the risk of heart attacks in middle-aged people. If someone can recall any of their middle-aged acquaintances having a heart attack, they are more likely to overestimate the probability.

(11)

2.1.3 Affect heuristic

Affect heuristic is another prominent heuristic involved in decision-making. We have evolved to

use our emotions as qualitative inference, to assess potential risks and benefits in the world around us based on our emotions toward them. We use feelings as information – If we feel scared or disgusted by a possible outcome, according to our emotional system, it means that something is bad. If we feel good about something, it means it is good (Slovic & Peters, 2006).

Pfister and Böhm (2008) argue that the affective system is integrated in decision-making in four distinct ways. The affective system is good at giving behavioral cues to a person, if an activity or event gives us bad feelings, we are likely to avoid it. The affective system can improve the speed of decisions by making them more intuitive – the affective system gives us cues which are easier to make decisions about than rational reasoning. Additionally, the affective system can help assess personal relevance for something. Our past experiences can lead to positive or negative emotions toward a person, activity or object, which can help a person assess if they should do it again based on the emotion they got in the past, such as satisfaction,

disappointment or regret.

The final aspect of Pfister and Böhm's model is that the affective system enhances our commitments. Love towards our child might make us more likely to try to be more patient or have healthier habits in order to be a role model, while negative feelings, such as being stressed for not writing enough on our thesis, might lead to a commitment toward better focus and allocating more time.

Based on Pfister and Böhm's model, the affective system has both positive and negative aspects. Instead of spending time thinking about rational choices, we trust our emotions to make faster decisions.

But emotional decisions can also be bad because emotions sometimes do not correspond with the actual qualities of something. The thought of exercise might give us negative emotions because it is difficult and makes us tired, while eating candy gives us positive feelings as the high content of sugar rewards our brains. The emotional system can lead to bad things being rewarded in the short run and good things to be penalized.

Affect heuristic can also be used as a measurement for probabilities. We overestimate the likelihood of events that give us strong emotions and underestimate the ones that activates our emotions less. This effect is visible on the whole spectrum of risk but can be seen clearly in highly affective events with low probability, such as winning the lottery or terrorist attacks. We overestimate their probabilities based on the strong emotions we feel. Since many people fear terrorism intensely or can vividly daydream about themselves and their positive feelings after winning the lottery, they tend to overestimate their probabilities. According to Sunstein (2003), people in the aftermath of 9/11 were ready to pay more for a flight insurance covering only terrorism than a flight insurance against all causes, including terrorism. Since the thought of these events creates so strong emotions, people overestimate their likelihood (Johnson, J. Hershey, Meszaros, & Kunreuther, 1993).

(12)

2.1.4 Identifiable victim effect

Identifiable victim effect is a psychological effect claiming that people tend to be more ready to

help fewer identified victims than many statistical victims, even if the latter are more in numbers, severity or easier to help. An identified victim could be a specific victim who we have seen a picture of or that has been described to us, making them more human than a number on a chart or part of a larger population. According to Jenni and Loewenstein (1997), the identifiable victim effect exists in part because identifiable victims are more vivid, we see them as human beings who we can relate to emotionally. Furthermore, helping identifiable victims give a direct and clear feedback – if we help a specific child by paying for their surgery, we will get direct

feedback when they have the surgery. If we instead donate money fund mosquito nets for malaria victims, we do not get a clear feedback when someone is saved. Moreover, helping identifiable victims mean we help a larger portion of the target group. Saving one family for sure rather than contributing to a larger organization helping 500 out of 2000 people makes us react differently.

2.1.5 Heuristics summary

Heuristics are often based on the System 1 way of thinking, as they are quick rules of thumb to avoid rational thinking which can sometimes be inefficient. In many cases, heuristics are necessary for us to function in the world. They can give us a quick and clear path for when the information is lacking or difficult to access. Affect heuristic gives us an intuitive path for decisions by recoding information into emotions while availability heuristic accesses whatever information is the most accessible in our minds and assumes it is the most relevant information, which often is correct. The focusing illusion can serve a purpose as well. When we need to quickly assess the state of a phenomenon, in a fast-paced world with finite time and resources, it is not possible to consider every possible variable responsible for an assessment. If our emotions tell us a piece of information is important (by making us feel strongly about it), it might be all we need to make an assessment. But in a complex world, these heuristics can malfunction. As we have seen in the examples above, these heuristics are often used for the same problems and during similar circumstances but can be used differently depending on several factors.

2.2 Boosts and Nudges

Traditionally, the most common way for policy makers to change behavior on a large scale has been by using hard paternalism, restrictive strategies such as legislation or economic incentives. An example of this is reduction of smoking, where banning of smoking in public places and taxes on cigarettes has made smoking more inconvenient and, in some cases, impossible. In recent decades, based on the discovery of bounded rationality and by understanding how people make decisions in real life and how those are influenced by their environment, strategies

alternative to hard paternalism have been developed. Two of these strategies are called boosts and nudges.

(13)

2.2.1 Boosts

Hertwig and Grüne-Yanoff (2017, p. 977) describe boosts as “interventions that make it easier for people to exercise their own agency by fostering existing competences or instilling new ones”. According to the authors, cognitive and motivational processes are both malleable and worth developing and should therefore be assisted and improved – boosted – instead of circumvented or manipulated.

An example of when a boosting intervention can be used is when a person is trying to decide what to eat. We all have different preferences – eating for health, distribution of macronutrients, weight loss or gain, allergies, or dietary restrictions for ethical or religious reasons. To assist the decision-making, a nutritional label can be used as a boosting intervention. Nutritional labels generally contain information about ingredients and allergy information, as well as content of calories and macronutrients. After learning some basics about what parameters are important for one's goals, by reading a nutritional label one can easily decide how well a certain product fits their dietary goals. They are not forced to comply with it, someone on a strict diet for weight loss might still decide to eat more cookies than planned without repercussions, but it works as a guide and might make it easier for a person to make a decision.

Another example of boosts are flowcharts for medical doctors diagnosing clinical depression. Instead of loosely gathering an overview of a patient's mood, the doctor asks the patient a series of 21 specific questions to determine if the patient could be clinically depressed or if they should look at something else. It is not used as conclusive evidence, but rather to get a fast and clear picture for making a more elaborate assessment. The boost structures information to help the doctor focus on what is important and to remind him or her about signs of depression that may not be obvious (Jenny, Pachur, Williams, Becker, & Margraf, 2013).

Another famous example is helping people make decisions about retirement funds. According to Loewenstein & Prelec (1992), people tend to favor the present above the future, especially regarding themselves. This effect is commonly known as present bias or time inconsistent behavior and has been demonstrated by the fact that many people tend to favor smaller present rewards over larger future rewards. Because of this, people sometimes do not save enough money from their salary to have a comfortable retirement. In a study by Hershfield et al. (2011), the researchers attempted to mitigate the present bias by using a boosting

intervention. People in two different groups were asked to decide how much they wanted to save every month for their retirement fund. The intervention group was shown a computer-generated rendering of their future self with wrinkles and white hair, and were then asked to make their decision, while the control group made their decision without specific instructions or guidance. The idea is that showing a person how they might look in the future connects them with their future self and mitigates the present bias. However, by using this intervention, people are still allowed to keep their pension fund small, or leave it completely empty, without other

repercussions than poverty after retirement.

Some criteria for an intervention to be called a boost are that it is completely transparent, it fosters competences rather than direct behavior, and that the effects of the intervention should

(14)

last even after the intervention is removed (Hertwig & Grüne-Yanoff, 2017). When someone needs to solve a problem or make a decision, there are many variables to take into consideration. Since people are boundedly rational, they tend to use heuristics to guide them toward the right decision. Sometimes, these heuristics lead to a bias in the weighting of a variable, such as the focusing illusion or availability heuristic.

Boosts can therefore be used to help an individual to focus on relevant information, quickly building new competences for the specific task, or assisting already present ones. The main distinction of boosts compared to nudges, according to the proponents, is that if people accept the objectives of a boost, they can commit to it, otherwise they can choose to disregard it. (Hertwig & Grüne-Yanoff, 2017).

2.2.2 Nudges

Another common intervention for changing people's behavior is called a nudge. The term and concept of nudges was popularized by Thaler and Sunstein (2008), and is described as

“nonregulatory and nonmonetary interventions that steer people in a particular direction while preserving their freedom of choice” (Hertwig & Grüne-Yanoff, 2017, p. 973). Sunstein and Thaler explains that public policies generally view humans as perfectly rational, similar to Simon's model of the economical human. Their view is instead that people are boundedly rational, "fallible, inconsistent, ill-informed, unrealistically optimistic, and myopic, and they suffer from inertia and self-control problems" (Thaler & Sunstein, 2008; Hertwig & Grüne-Yanoff, 2017, p. 974).

Typically, nudges are cheap and simple interventions on a choice architecture, the part of a decision problem external to the person making the decision. By exploiting the limits of human rationality according to the concept of heuristics and biases, nudges change the way that

decisions are framed to favor a specific decision. (Hertwig & Grüne-Yanoff, 2017)

An example of a nudge intervention is pension funds. People tend to favor their present self before their future self, an effect called the present bias. In a study by Thaler and Benartzi (2004), the participants were instead instructed to choose between having an extra dollar in the very near future or in the far future. The nudge intervention was made through removing the option of choosing their present, forcing them to choose between two different futures rather than present and the future. In this study, people in the intervention group were acting more favorable toward their far future selves than in the control group.

Critics of nudges, such as Hertwig and Grüne-Yanoff (2017) claim that nudges are too manipulative in their nature, they often lack transparency since they often are designed as, and often are required to be, hidden from the decision maker. Furthermore, they require a benevolent designer of the choice architecture, who not only has the person's best as their intentions, but also has a correct view of what that is.

(15)

2.2.3 Nudges and boosts compared

Boosts and nudges are two interventions for changing behavior without using hard paternalistic methods. Nudges have so far been the most prevalent method of the two, with many academic studies conducted and real-life applications. Some researchers, such as Sunstein (2016) claim that a boost is simply a type of nudge. Hertwig and Grüne-Yanoff (2017, p. 977) instead state that “educative nudges and short-term boosts largely overlap”, but that the main distinction is that while nudges change choice architecture, boosts instead change people's ability to deal with choice architecture.

What boosts and nudges have in common is that these models attempt to improve decision-making without removing the freedom of choice. As mentioned in the paragraphs above, examples for a nudge to improve the retirement savings is changing the choice

architecture from choosing between present self and future self to near future self and far future self. An example for a boost to improve retirement savings is to connect the individual more with their future self by a computer-generated rendering of their potential future self. As a reference-point, a common hard paternalistic approach is for the employer to financially incentivize savings by matching the retirement savings of an employee, leading to the person saving 2 dollars for every 1 dollar they deposit into their savings.

2.3 Structured Introspection

Structured Introspection is a boosting intervention first studied by Mayorga (2019) which has the intention of helping a person make more informed and debiased decisions by giving them a frame for their reasoning. Structured introspection allows a participant to rate how important a set of factors should be when making a certain decision, which might diminish the limiting effects of cognitive heuristics such as focusing illusion, affect heuristic and availability heuristic that can otherwise appear when doing an unstructured deliberation task.

2.3.1 Structured introspection in past studies

In the first experiment by Mayorga (2019), participants in three groups were given information about the humanitarian crisis in Yemen and were asked to donate an optional amount of money between 0 and $100. In the first group, participants were asked to simply make the choice between the two options. In the second group, the participants were asked to deliberate about the decision at hand before making a choice. The last group, the structured introspection group, was given a set of factors about the information they had been given on the humanitarian crisis in Yemen and were asked how much those factors should influence their decision.

In the second experiment by Mayorga (2019), participants were instead asked questions regarding donating blood. All participants were first given a set of information about blood donation, but the structured introspection group were given three attributes about blood donation and asked how much each of them should influence their feelings about donating blood.

(16)

The principle is that by using structured introspection, the participants will be assisted with structuring their deliberation and seeing the important factors, which in extension leads to debiasing and a decreased use of poor heuristics for that decision. By doing this, the focusing illusion should be mitigated since it helps people focus on more than one relevant factor. It may also decrease the effects of the affect heuristic. Since affect heuristic gives emotions a

disproportionate amount of influence over decisions due to decisions being done largely by System 1, it can be mitigated by reframing a problem to give rational factors more influence over the decision. Finally, structured introspection may also be able to mitigate the influence of availability heuristic by making more information more widely available.

2.3.2 Structured introspection as a boost

Structured introspection can be considered a boost since it follows the criteria by Hertwig and Grüne-Yanoff (2017). It is completely transparent, meaning that it does not consist of any components the participant is not aware of, it fosters competences rather than direct behavior as it helps people deliberate about their own priorities and are not told what to do, and the effects should last even after the intervention, as the intention of the introspection is to lead to the person realizing their own motivations and priorities which should carry over to the future.

2.4 Current Study

The current study was created to evaluate the structured introspection-boost through the lens of the COVID-19 pandemic. Structured introspection-experiments have previously only been conducted under circumstances of lower stakes that had no direct effect on the decision maker (Mayorga, 2019). The experiment in this thesis was conducted during the COVID-19 pandemic between 26th of March 2020 and 7th of April 2020.

Unlike Mayorga (2019), the current study used a between-group design consisting of only a structured introspection group and a control group. The study was conducted using the online survey tool Qualtrics. As the dependent variable, participants were asked to consider two potential strategies for Sweden during the COVID-19 pandemic. The first alternative was keeping the economy up and running and thus not slowing down the infection rate at the fastest possible rate, leading to a higher number of lives lost. The other alternative was to shut down the economy, slowing down the infection rate faster but at a larger economic cost for the society.

The boost consisted of four attributes given to the intervention group after the initial dilemma. Each participant was asked to consider how much each attribute should influence their decision. The study also used control variables for both groups after the stimulus to assess any differences between the groups in terms of demographics or pre-existing values.

When the study was conducted, COVID-19 was spreading rapidly in Sweden and the rest of the world. On the 26th of March 2020, the first day of the study, there were 2,996 confirmed cases in Sweden and 78 confirmed dead. On the last day of the study, there were 8,300 cases and 618 deaths from COVID-19. During the data collection, the confirmed cases were increasing by

(17)

approximately 10% per day. Meanwhile, in other parts of Europe, such as Italy, the healthcare system was on the edge of collapse. Italy, which was at the time considered the epicenter of the pandemic, had its peak death rate thus far on the 27th of March. Because of the spread in Sweden and other countries, there were hectic debates about the Swedish strategy. At the time, the

Swedish strategy, guided by the Public Health Agency of Sweden, was to keep the country as open as possible but to avoid overwhelming the healthcare system. There were many proponents of this, but also many people displeased with the Swedish strategy, demanding the government to make stricter restrictions to slow the spread (Kantar Sifo, 2020).

(18)

3 Method

3.1 Participants and Design

The study used a test design partially based on Mayorga (2019) and was a part of a larger research project. The experiment was of a between-group design consisting of an intervention group with 153 participants and a control group with 128 participants, 281 in total. 117

participants were male, 161 female and 3 preferred not to say. Ages ranged between 18 and 88 (M=30,98, Mdn=26, SD=12,344). Participants were recruited using a participant pool belonging to the JEDI-lab at Linköping University and through Facebook. Participants were not offered any compensation. The study ran between 26th_{of March and 7}th_{of April 2020.}

3.2 Materials

Materials used was a survey using the online survey-tool Qualtrics. All materials were either written originally in Swedish or translated to Swedish from English (the questions on the Oxford

Utilitarianism Scale).

All participants were sent a link leading to the start of the survey, using either their own computer, a tablet, or a smartphone. A boosting intervention in the form of a structured

introspection-task based on Mayorga (2019) was used, along with several control variables to assess the validity of the intervention.

3.2.1 Dependent variable

The dependent variable was a dilemma of choosing between two options for what the Swedish strategy to battle the COVID-19 pandemic should be. The dilemma was:

“If it was up to you to decide what the Swedish strategy in dealing with the Coronavirus-situation, which of the following options would you choose:

1. Keeping the Swedish economy running as much as possible during the COVID-19 pandemic, thus potentially leading to the death of an estimated 10,000 people, most of them above the age of 70.

2. Shut down the economy as much as needed, leading to saving more people which could lead to an estimated loss of 300–400 billion kr (Swedish crowns) and future negative consequences for all Swedish citizens.”

3.2.2 Intervention

The intervention group was introduced to an intervention called a structured introspection. The structured introspection-method was invented by Mayorga (2019) and consists of a set of

(19)

mitigate the effect of potential heuristics and biases. The intervention consisted of four attributes and the participants were asked to rate on a scale of 1–7 how important those attributes should be when deciding whether to shut down the economy.

The following attributes were used:

Saving human lives

Saving the Swedish economy from collapsing

Concern for the health of the elderly and risk groups Concern for the quality of life and well-being of all citizens

The attributes were chosen to assess the potential of mitigating potential cognitive biases, such as the focusing illusion, affect heuristic and the identifiable victim effect. They were picked as they were considered as neutral scales that did not promote either option on the dependent variable as objectively better morally. The attributes were also chosen as they are all factors considered to potentially influence choice in the COVID-19 pandemic and to be related to the choice on the dependent variable. The control group was instead asked to make their decision right after reading the information.

3.2.3 Risk perception

All participants were asked to estimate individually the likelihood of themselves and others suffering from serious consequences from COVID-19 on a scale from 1-7, from no risk at all to

very high risk. This was done to evaluate if risk perception influenced decisions, and if risk

perception was different between the groups.

3.2.4 Emotions

The participants were asked to individually rate to what extent they experienced a set of emotions when thinking about COVID-19. The emotions used in the study were anger, worry, sadness, helplessness, optimism, pessimism, anxiety and hope and all scales ranged between 1 – 7, from not at all to very much. The participants were asked these questions to assess any

potential difference in the emotions of people in the intervention group and in the control group, and if the intervention would be enhanced or mitigated by this difference. Additionally, it served to assess a potential relationship between the choice people made and the emotions they

experienced regarding COVID-19.

3.2.5 Control variables

There were several variables used in the study to ensure the validity of the study by comparing group differences in factors that could influence choice. These variables were utilitarianism, whether the participant had been quarantined because of COVID-19, their personal finances, and their satisfaction with the response of the Swedish government so far.

(20)

3.2.6 Oxford Utilitarianism Scale

Oxford Utilitarianism Scale (Kahane, et al., 2018) is a set of dilemmas consisting of two scales

which assesses a person's utilitarianism on two factors. Those are Impartial Beneficence (IB), the extent to which we believe in maximizing the happiness of many people regardless of if they are close to us physically or emotionally or not. The other factor is Instrumental Harm (IH) – the conditions under which people accept causing harm for the greater good. IB consisted of five dilemmas and IH consisted of four dilemmas. Participants were asked to rate each dilemma 1-7 from strongly disagree to strongly agree. The questions on the Oxford Utilitarianism Scale can be found in Appendix.

These dilemmas were presented to find any potential differences in utilitarian aspects of participants’ values between the groups and their potential influence over choices.

3.2.7 Self-quarantine

Three questions were asked whether a participant had been quarantined, if they themselves were or if any of their friends or family members were sick in COVID-19. These were all yes or no-questions asked to assess if personal risk affected decisions and if there was any difference between the groups.

3.2.8 Economy

Participants were asked to assess their personal finances, their finances in relation to others, and how much they expect COVID-19 and the Swedish government's response to it to affect their finances. They were asked to rate these factors on a scale 1–5, from not at all to very much. This was done to check any differences between groups in personal finances and how it affected an individual's choice on the dependent variable.

3.2.9 Satisfaction with the Swedish response

Participants were asked to rate on a scale from 1–5 how satisfied they were with the Swedish government's response to COVID-19 so far. This was done to assess any difference between the groups in satisfaction and if it would influence choice. At the time of conducting the study, many parts of the Swedish society remained open. Schools up to 9th_{grade and restaurants, bars and}

businesses were kept open, contrary to most other countries in Europe at the time.

3.3 Ethics

All participants were informed about the fact that the study gathered anonymous information and that data would be stored and analyzed, and that they could withdraw their consent at any time. Following this information, participants were required to give explicit consent before starting the survey.

(21)

3.4 Procedure

After each participant was recruited, they were linked to the survey on the online survey-tool Qualtrics. There they received some short information about the study and that their participance was voluntary and they could at any time withdraw from the study. After this, they were asked to either agree to the terms or to decline.

After the participant gave their informed consent, they were given a participant ID and were randomly divided into an experiment group or control group. They were sorted randomly by a function from the Qualtrics survey toolkit.

Each group were then posed with a set of several questions. They were first presented with a text about the COVID-19 spread in the world as a whole and in Sweden in particular. The groups were then given the following dilemma (translated from Swedish, see the full experiment in Appendix):

“Imagine that you had to make the decision on how Sweden should respond to the novel Coronavirus and had these two options:

1. Continuing normal activity of the economy, including all schools and businesses, which will lead to the death of 10.000 Swedes (most of whom above the age of 70).

2. Partial to full shut down of the economy, including all schools and businesses, which will lead to a loss of between 300 and 400 billion SEK and future damage to the life-quality of almost all citizens.”

The control group was then directed to make their decision, while experiment group was

redirected to a page containing a set of attributes, asking the participants to consider and declare, on a scale 1–7, how much each of the following considerations should influence the decision they were asked to make:

Saving human lives

Saving the Swedish economy from collapsing

Concern for the health of the elderly and risk groups

Concern for the quality of life and the well-being of all citizens

After answering these questions, the experiment group was then asked to make their decision on the initial dilemma.

After giving their answer to the dilemma, all participants were asked to answer several sets of additional questions, about their perception of their own risk from suffering severe consequences from COVID-19, other people's risk, to evaluate their own feelings on a set of emotions, their satisfaction with the Swedish response to the COVID-19 pandemic, their assessment of their own economy, their own economy compared to others, how much they are expecting COVID-19 and the Swedish government's response to it to affect their own economy, to what extent they perceive the risk of severe consequences for themselves and others due to

(22)

COVID-19, if they have personally been quarantined, if they are personally sick in COVID-19 or if any of their friends or family is sick in COVID-19. Finally, participants answered the

dilemmas on the Oxford Utilitarianism scale.

After responding to all questions, the participants handed in the survey and were thanked for their participation.

3.5 Data Analysis

To assess if groups varied on any demographic data, independent samples t-tests were conducted on age and gender. Independent samples t-tests were conducted on the control variables risk for self, risk for others and quarantine. Independent samples t-tests were also conducted on the control variables IB and IH.

A Mann Whitney U-test was conducted to assess a difference between intervention and control group on the dependent variable, the choice whether to shut down the economy or keep it running.

Four One-way ANCOVAs were conducted to investigate the effect attribute ratings (saving lives, saving the economy, concern for the health of the elderly and risk groups and concern for the quality of life and well-being of all citizens.)

A bivariate logistic regression was conducted on the emotions anger, worry, sadness, helplessness, optimism, pessimism, anxiety and hope as well as if the participant had been sick in COVID-19 themselves, in order to assess their potential influence on choice on the dependent variable, and any difference between control and introspection groups.

All analyses were performed with SPSS 26.

3.6 Exclusions

10 participants were removed from the study, either after indicating that they had already

participated in the study, or who perfectly matched previous participants on both IP-address, age, and gender. These were not counted toward the total number of participants.

(23)

4 Results

This chapter is divided into four subsections. Descriptive results show a demographic comparison between the groups as well as a comparison on some of the control variables.

Dependent variable shows the result on difference between the groups on the dependent variable. Decision attributes show the results on the ANCOVAs on correlation between choice on

dependent variable and the four different attributes in the intervention group. Finally, secondary analysis shows the results of a logistic regression on risk perception, emotions and choice.

4.1 Descriptive Results

Table 1

Frequency Table on dependent variable and control variables.

Control Introspection

Sample Size 128 153

Shut Down Economy 69 (53.9%) 88 (57.5%)

Keep Economy Functioning 59 (46.1%) 65 (42.5%)

Age (Median) 26 26

Age (Mean) 31.65 30.41

Gender M/F/O* 55/71/2 62/90/1

Personally Sick 2/126 (1.6%) 13/140 (8.5%)

Family Sick 28/100 (21.9%) 31/122 (20.3%)

Has Been Quarantined 44 (34.4%) 45 (29.4%)

Instrumental Harm Mean (1–7) 3.64 3.58

Impartial Beneficence Mean (1–7) 3.53 3.62

Note. * M = Male, F = Female, O = No answer/Other.

Data was analyzed to compare the groups regarding sample size, age, and gender distributions and whether participants had been quarantined for or sick in COVID-19 themselves or had a

(24)

friend or family member who had been sick in COVID-19. Group sizes differed marginally as participants were sorted into groups randomly and the introspection group was by chance bigger.

Independent samples t-tests were conducted to check if there was a significant difference in gender or age distributions between the groups. No differences were found.

Independent samples t-tests were also done on the Oxford Utilitarianism Scale-variables Instrumental Harm and Impartial Beneficence individually to assess if there was a difference between groups. No significant difference was found. Additionally, independent samples t-tests were done on the control variables self-quarantine and family-has been sick and were both insignificant.

An independent samples t-test was conducted on whether the participants were personally sick in COVID-19. There was a significant difference in the scores for being personally sick in COVID-19 in the control group (M=1.98, SD=.125) and the introspection group (M=1.92,

SD=.280); t (279)=2.60, p = .01. Only 1.6% (n=2) in the control group were sick in COVID-19

while 8.5% (n=13) were sick in the introspection group. It did however not show a significant result on the dependent variable when analyzed as a part of a logistic regression (See Table 2).

4.2 Dependent Variable

A Mann-Whitney test indicated no difference between choice on the dependent variable between the control group (M=138.24) and the intervention group (M=143.31). (U = 9438, p = .545) (See Table 1).

4.3 Decision Attributes

Four one-way ANCOVAs were conducted on the intervention group for the individual attributes (saving lives, saving the economy, concern for the health of the elderly and risk groups, and concern for the quality of life and the well-being of all citizens). All test analyses used age, gender, own finances, own finances compared to others, the expected effect of COVID-19 and the Swedish government’s response to it on their own finances, impartial beneficence,

instrumental harm, and satisfaction with the Swedish government’s response to COVID-19 as covariates. These tests were done to assess if any difference between people choosing one answer on the dependent variable correlated with their choices on the attributes, even after accounting for potential influence of other variables. Figure 1 shows the difference in importance of attributes for the different choices.

The first ANCOVA was conducted on the attribute saving lives, and showed a significant effect on choice based on how high a person rated the importance of saving lives (1–7) after accounting for the covariates, where a higher estimation correlated with higher likelihood of choosing to shut down the economy. F(1, 142) = 22.68, p < .001. Comparing the estimated marginal means showed that participants choosing to shut down the economy rated the importance of saving lives higher (M=6.44) compared to those choosing to keep the economy functioning (M=5.57).

(25)

The second ANCOVA was conducted on the attribute saving the economy and showed an effect on choice after checking the effect of the covariates age, gender, own economy, own economy compared to others and expected effect on economy from COVID-19 and the Swedish response to it. F(1, 142) = 11.52, p < .001. Comparing the estimated marginal means showed that participants choosing to keep the economy functioning rated the importance of saving the

economy higher (M=5.78) compared to those choosing to shut down the economy (M=4.78). The third ANCOVA was conducted on the attribute concern for the health of the elderly and risk groups and showed an effect on choice after checking the effect of the covariates age, gender, own economy, own economy compared to others and expected effect on economy from COVID-19 and the Swedish response to it. F(1, 142) = 19.43, p < .001 Comparing the estimated marginal means showed that participants choosing to shut down the economy rated the

importance of concern for the sick higher (M=6.07) compared to those choosing to keep the economy functioning (M=5.19).

Finally, the fourth and last ANCOVA was conducted on the attribute concern for the quality of life and well-being of all citizens. There was no effect on choice after checking the effect of the covariates age, gender, own economy, own economy compared to others and expected effect on economy from COVID-19 and the Swedish response to it. F(1, 142) = .115, p = .735. Comparing the estimated marginal means showed that participants choosing to shut down the economy rated the importance of the quality of life and well-being of all citizens as

marginally lower (M=5.09) compared to those choosing to keep the economy functioning (M=5.20).

Figure 1. Mean answers on individual attributes in the intervention group.

0 1 2 3 4 5 6 7

Saving Lives Saving Economy Concern for Sick Concern Well-Being All Citizens Keep Economy Functioning Shut Down Economy

(26)

4.4 Secondary Analysis

A logistic regression analysis was performed on the participants’ emotions, perception of risk, and whether they were sick in COVID-19. The logistic regression was significant on the

variables perceived risk for self and risk for others. The logistic regression table (Table 2) shows that risk for self correlated with choosing to keep the economy functioning while risk for others correlated with choosing to shut down the economy. The other variables did not show a

significant result.

Risk for self and others accounted for 9% of the variance on the dependent variable, and correctly classified 60.5% of cases.

Table 2

Logistic regression table for emotions, risk perception and sick in COVID-19, effect on choice.

Variable B _SE _p Constant -1.28 1.31 .329 Risk Self -.20 .1 .035 Risk Others .26 .10 .011 Anger -.05 .08 .506 Worry .22 .12 .065 Sadness -.02 .09 .855 Helplessness -.15 .09 .095 Optimism .061 .11 .560 Pessimism .044 .10 .651 Anxiety .042 .09 .645 Hope .48 .10 .630 Sick Self .076 .60 .897 R2 .09 N 281

(27)

5 Discussion

5.1 Results

The aim of the study was to find out if a structured introspection-boost can influence decisions about a contemporary dilemma. The results show that the intervention was not successful at changing decisions in a significant way. The study showed three main results of interest.

The first result showed a correlation between what people in the intervention group chose on the dependent variable and how high they rated the importance of the attributes saving lives, saving the economy, and concern for the health of the elderly and risk groups. There was no correlation between choice and the attribute concern for quality of life and well-being of everyone in the society. All correlations were still significant after accounting for the influence of the other attributes. This means that three out of four attributes successfully showed difference of values between the choice alternatives. This means that while there was no significant

difference in choice between groups, some participants in the intervention group made their decisions based on their opinions on these attributes rather than heuristics.

Second, the results showed that the control variables perceived risk for self and others correlated with specific decisions on the dependent variable. Perceived risk for self correlated with choosing to keep the economy open while risk for others correlated with choosing to shut down the economy. This, together with the first findings, indicates that people were guided in their decisions by something else than heuristics such as the above-mentioned focusing illusion, availability heuristic, identifiable victim effect and affect heuristic. Risk for self had positive correlation with the choice of keeping the economy open. This could potentially mean that participants had a broad interpretation of what personal risk meant, or that those considering themselves at high risk were more willing to distance themselves for the good for society at large.

Last, the results from the logistic regression show that emotions did not have an influence on decisions in either group, which could be an additional reason why there was no effect

between the groups. This may be because participants were influenced by risk perception or had predetermined opinions already, which could mitigate the effect.

The result was expected, as boosting in general, and more specifically structured introspection has not been widely studied and never for a phenomenon such as the COVID-19 pandemic. The first part of the previous structured introspection-study had a focus on the humanitarian crisis in Yemen, something that according to Mayorga (2019) is often forgotten and unheard of. The second part of the previous structured introspection-study focused on blood donation. Both studies had an answer considered a moral high ground, where the attributes acted to distinct the objectively unimportant factors from the more important ones. Additionally, those studies excluded participants who had a strong opinion from before (in the blood donation study, participants who reported giving blood often were excluded). COVID-19, however, was at the

(28)

time of the study the most reported news topic in media. Perhaps this could have led to some participants having a strong stance on COVID-19 and opinions about what the Swedish strategy should be. This could mitigate the focusing illusion, as the focusing illusion is used when people lack a clear view of abstract concepts such as how happy they are. But regarding the COVID-19 pandemic, people might have already had a clear view of how scared they were and what they prioritized (saving risk groups vs. saving the economy).

Additionally, the current study, unlike the experiments in previous structured

introspection-interventions, used a stimulus which lacked a preferred answer, as both alternatives on the dependent variable were intended to be presented as equally correct, it may be difficult to find an effect. This is because participants may be influenced in both directions, either toward shutting down the economy or toward keeping it open. The results indicate that this may be the case, as the results of the ANCOVAs suggest that choice correlate strongly with three out of four attributes.

The identifiable victim effect may have been present in the study in the way that risk groups are at more direct risk and by shutting down the society, there is a clearer feedback of lives saved. It might be that the structured introspection did not lead to a clearer insight. It could also be that the information given before the structured introspection which both groups read included enough information for the participants to make a debiased answer, one which was not influenced by the structured introspection-task.

The results imply that the affect heuristic was not a determining factor in the participants’ decisions, as their emotions did not predict their answers. This could either mean that people were making more rational choices, or that their affective response was correlated with other factors, such as the perceived risk for others or oneself. Finally, the perceived risk for self and others showed to be a predicting factor (by up to 9%) for choice. This could perhaps mean that those feelings overruled potential influence from the intervention. Additionally, looking at Pfister and Böhm’s (2008) model, the affective system can lead to feelings based on our experiences. This could also mean that personal risk such as being in a risk group or having family members in risk groups can lead to higher negative emotions, which may therefore influence one’s decision.

5.2 Method

The structured introspection-design might have been problematic as the stimulus lacked a

preferred answer. In the context of this study, it means participants could be boosted both toward shutting down society and keeping the economy open. There are some indications that this was the case, as choice correlated strongly with three out of four of the attributes. Changing the structured introspection to be more biased toward a certain choice could however compromise the integrity of the boost as an intervention with the intention of increasing an individual’s competences rather than changing the choice architecture.

(29)

group or their perception of the potential risk for their friends and family. As the results show that risk perception played a role in decisions, perhaps controlling for these variables could have showed whether these participants were less likely to be influenced by the structured

introspection-boost.

Furthermore, the study could have introduced a variable where the participant had to self-assess how strong their opinions were regarding COVID-19 or how well-informed they

considered themselves to be. As mentioned in the theoretical background, boosting strategies intend to mitigate the use of cognitive heuristics, mainly those based on the decision maker having limited time or knowledge on the subject. Perhaps it could have been of interest to assess if participants with stronger opinions or more knowledge were easier to be helped by the SI-intervention.

Additionally, it could have been assessed more thoroughly how the stimulus, the

informative text, was interpreted. It is possible that terms such as life-quality can mean different things to different people. Losing 10,000 lives in a disease is a clear and vivid outcome while the loss of 300–400 billion kr has more abstract potential outcomes, such as unemployment,

disruptions in the welfare system or lower quality of education. These are second order effects which may have been remembered by some and not by others. This is suggested by the fact that one attribute, the quality of life and well-being of all citizens, did not correlate with answers on the dependent variable.

It is also possible that the nature of the decisions in the current study and Mayorga (2019) was different. In the dissertation by Mayorga, the stimuli were on charitable giving in the context of the humanitarian crisis in Yemen and blood donations. The dependent variable in this study may have been a different type of decision, as the studies conducted by Mayorga are decisions about being a small contributor to a larger cause. In the current study, the participant was required to make a larger decision, influencing more people directly. This could mean that individuals reason and decide differently about these types of decisions, or that more information is required to influence their opinion.

In the informative part of the dependent variable, the number 3–4% mortality rate, 10,000 lives lost, or 300–400 billion kr lost was claimed. There were very different opinions about this, and different governmental agencies and media outlets were claiming different things as these factors were unknown at the time. Additional work could have been made to make sure that estimates for death rates and economic cost were more grounded in the current predictions at the time.

5.3 Future Research

Structured introspection should be studied in a broader sense, and more work should be done on looking at how different types of decisions can be influenced by a boost. Additionally, future research should control more for how well-informed participants are on the topic already. Future studies could also focus on assessing if decisions which people already have strong opinions about can be boosted in a significant and meaningful way.

(30)

6 Conclusion

The intervention did not influence decisions in a significant way. However, it was concluded that participants in both groups made decisions about COVID-19 in relation to their perception of their own and others risk related to it. Additionally, it was assessed that emotions did not have a significant role in decisions, which indicate that emotions were used less for the decision on COVID-19. An indication for this is that choice correlated with three of the four attributes, meaning that decisions in the introspection group were made dependent on their opinions on these factors. This could be because people have personal involvement in COVID-19 or are informed to a larger extent compared to previous structured introspection-interventions, reducing the biasing effect of heuristics. The study indicates that structured introspection-tasks may be difficult to use as a successful boost on participants with personal involvement or with a previously formed opinion. The experiment shows that risk perception both for self and others correlates with decisions, and the results indicate a strong overlap with three of the attributes in the structured introspection.

The hope is that the work made in this thesis has clarified which parts of structured introspection needs development and further investigation and has led to a better understanding of boosts. Perhaps it can lay another brick in the foundation for how boosting interventions should be constructed to assist people in making more informed and well-rounded decisions in the future.

(31)

References

Arkes, H. R., Gigerenzer, G., & Hertwig, R. (2016). How bad is incoherence? Decision, 3(1), 20-39.

BBC. (2020, 04 02). Coronavirus: Confirmed global cases pass one million. Retrieved from BBC News: https://www.bbc.com/news/world-52144390

Gigerenzer, G. (2004). Fast and Frugal Heuristics: The Tools of Bounded Rationality. In D. J. Koehler, & N. Harvey, Blackwell Handbook of Judgment and Decision Making (pp. 62-88). Malden, MA, USA: Blackwell Publishing Ltd.

Hershfield, H. E., Goldstein, D. G., Sharpe, W. F., Fox, J., Yeykelis, L., Carstensen, L. L., & Bailenson, J. N. (2011). Increasing Saving Behavior Through Age-Progressed

Renderings of The Future Self. Journal of Marketing Research, XLVIII (Special Issue 2011), S27-S37.

Hertwig, R., & Grüne-Yanoff, T. (2017). Nudging and Boosting: Steering or Empowering Good Decisions. Perspectives on Psychological Science, 12(6), 973-986.

Jenni, K., & Loewenstein, G. (1997). Explaining the "Identifiable Victim Effect". Journal of Risk

and Uncertainty, 14(3), 235-257.

Jenny, M. A., Pachur, T., Williams, L. S., Becker, E., & Margraf, J. (2013). Simple rules for detecting depression. Journal of Applied Research in Memory and Cognition, 2(3), 149-157.

Johnson, E., J. Hershey, J., Meszaros, J., & Kunreuther, H. (1993). Framing, Probability Distortions, and Insurance Decisions. Journal of Risk and Uncertainty, 7, 35-41.

Kahane, G., Everett, J. A., Earp, B. D., Caviola, L., Faber, N. S., Crockett, M. J., & Savulescu, J. (2018). Beyond Sacrificial Harm : A Two-Dimensional Model of Utilitarian Psychology.

Psychological Review, 125(2), 131-164.

Kahneman, D. (2011). Thinking, Fast and Slow. New York, NY, USA: Farrar, Straus and Giroux.

Kahneman, D., Krueger, A. B., Schkade, D., Schwarz, N., & Stone, A. A. (2006). Would You Be Happier If You Were Richer? A Focusing Illusion. Science, 312(5782), 1908-1911. Kantar Sifo. (2020, 03 18). Frågor gällande Coronaviruset. Retrieved from Kantar Sifo:

https://www.kantarsifo.se/sites/default/files/reports/documents/kantar_sifo_fragor_gallan de_coronaviruset.pdf

(32)

Loewenstein, G., & Prelec, D. (1992). Anomalies in Intertemporal Choice : Evidence and an Interpretation. The Quarterly Journal of Economics, 107(2), 573-597.

Mayorga, M. W. (2019). An Introspection Intervention for Perceived Inefficacy in Charitable Giving. University of Oregon.

World Health Organization (2020, 03 12). WHO announces COVID-19 outbreak a pandemic. Retrieved from World Health Organization: http://www.euro.who.int/en/health- topics/health-emergencies/coronavirus-covid-19/news/news/2020/3/who-announces-covid-19-outbreak-a-pandemic

Pfister, H.-R., & Böhm, G. (2008). The multiplicity of emotions: A framework of emotional functions in decision making. Judgment and Decision Making, 3(1), 5-17.

Schkade, D. A., & Kahneman, D. (1998). Does Living in California Make People Happy? A Focusing Illusion in Judgments of Life Satisfaction. American Psychologal Society, 9(5), 340-346.

Simon, H. (1956). Rational Choice and the Structure of the Environment. Psychological Review, 63(2), 129-138.

Slovic, P., & Peters, E. (2006). Risk Perception and Affect. Current Directions in Psychological

Science, 15(6), 322-325.

Sunstein, C. (2003). Terrorism and Probability Neglect. Journal of Risk and Uncertainty, 26(2-3), 121-136.

Sunstein, C. R. (2016). The Ethics of Influence: Government in the Age of Behavioral Science. Cambridge, England: Cambridge University Press.

Thaler, R. H., & Benartzi, S. (2004). Save More Tomorrow(TM): Using Behavioral Economics to Increase Employee Saving. Journal of Political Economy, 112(1), S164-S187. Thaler, R., & Sunstein, C. (2008). Nudge: Improving Decisions About Health, Wealth and

Happiness. New Haven, CT: Yale University Press.

Tversky, A., & Kahneman, D. (1974). Judgment under Uncertainty: Heuristics and Biases.

(33)

Appendix INTRO

Den här studien handlar om det nya Coronaviruset (COVID-19). Det nya Coronaviruset började spridas i Kina i slutet av 2019, och har de senaste veckorna spridit sig snabbt i många andra länder. Viruset leder till influensaliknande symptom, men är betydligt farligare än influensa. Enligt de senaste beräkningarna väntas mellan 3 och 4 procent av de smittade att dö i sjukdomen (främst äldre och personer med underliggande sjukdomar), och omkring 20 procent av de

smittade kommer behöva sjukvård. Med hänsyn till de risker som viruset medför har många länder beslutat att ta omfattande beslut för att stoppa smittspridningen.

Vänligen läs noga igenom texten på nästa sida och besvara därefter de efterföljande frågorna. INFORMATION

Det har på sistone debatterats mycket kring hur spridningen av Coronaviruset ska besvaras för att begränsa dess inverkan.

Vissa förespråkar strategin att stänga ner så stora delar av samhällsekonomin som möjligt. De beräknar att 10 000 svenskar, de flesta av dem över 70, kommer dö av Coronaviruset om företag och affärer inte stängs ner. Andra anser att dessa åtgärder är överdrivna och att en nedstängning av samhällsekonomin kommer att kosta mellan 300–400 miljarder kronor och påverka

privatekonomin samt möjligtvis den mentala hälsan hos en majoritet av befolkningen. Det här dilemmat är inte unikt för något land, och olika länder i världen har agerat olika för att hantera hotet från COVID-19.

Föreställ dig att du var ansvarig för att fatta beslutet om hur Sverige ska agera angående det nya Coronaviruset och fick välja mellan dessa två alternativ:

1. Upprätthålla normal aktivitet i samhällsekonomin inklusive alla skolor och företag, vilket kommer leda till att ungefär 10 000 svenskar dör, de flesta av dem över 70 år.

2. Delvis till fullständig nedstängning av samhällsekonomin inklusive alla skolor och företag, vilket kommer leda till en förlust av mellan 300 och 400 miljarder kronor, och i framtiden negativa konsekvenser för livskvaliteten hos samtliga medborgare.

MANIPULATION (between subjects)

Control group: Vilken av de två strategierna hade du valt?

Introspection group: Innan du fattar ett beslut, fundera på hur mycket följande faktorer bör

påverka beslutet gällande vilken strategi som ska användas:

(Uppskattat betydelse på en skala 1–7: Inte alls – Väldigt mycket) Rädda människoliv

Boosting Through Structured Introspection : Exploring Decision-Making in Relation to the COVID-19 Pandemic