_____________________________________________
Nudging & Brands
The effect of becoming aware of system 1 nudges on
attitude towards the brand
___________________________________________
Prepared by: Piet Degen (11934212) Under supervision of: Dr. Alfred Zerres
21st of June 2018
Final Master’s thesis
MSc. in Business Administration, specialization Marketing. Amsterdam Business School, UVA
Statement of Originality
This document is written by Student Piet Adriaan Degen who declares to take full responsibility for the contents of this document. I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the text and its references have been used in creating it. The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.
Table of Contents
Acknowledgements ... 5
Abstract ... 6
Introduction ... 7
Literature review ... 10
Paternalism & libertarian paternalism ...10
Nudges ...11
Social proof & default nudges ...13
Perceived restriction of autonomy ...14
Transparency and awareness of nudging ...16
Attitude towards nudges and the brand ...17
Conceptual framework ... 21
Data and method ... 22
Preliminary study ...22 Main study ...24 Sample ...25 Measures ...26 Results ... 29 Descriptive statistics ...29
Social proof & default nudge ...30
Attitude towards the brand ...30
Perceived restriction of autonomy ...31
Conceptual model of hypotheses tests ...35
Discussion ... 36
Social proof & default nudges in healthy consumption behaviour ...36
Attitude towards the brand, health consciousness, and perceived threat of autonomy ...36
Desire for control ...38
Practical implications ...38
Limitations ...39
Directions for further research ...40
References ... 41
Appendices ... 45
Appendix 1: Flight selection, decoy question ...45
Appendix 2: Comfort class selection, decoy question ...45
Appendix 3: Conditions ...46
Acknowledgements
This study could not have been completed without the essential help of several people. First of all, a special thanks to all the participants who partook in this research. I have requested quite some help from friends and family, and I am greatly thankful for their dedication and continued assistance in completing my surveys and providing feedback. Particularly thanks to Heleen, Luuk, Paul, Doris, thank you all!
Second, thank you Linda for showing support during this entire process. You were always ready to help me vent during the writing of my thesis –but also helped through the different discussions we had regarding my ideas throughout this project.
Lastly, my supervisor Dr. Alfred Zerres provided me with critical feedback and food for thought, but especially encouragement during times that I was a bit stuck in the process. Openly discussing all parts of my thesis with him has contributed largely to not only the outcome of this project, but also to my personal development.
With gratitude,
Piet Degen
Abstract
To guard consumers from the perception of manipulation through nudges, policy makers and choice architects strive to make their intentions of these policies transparent. Specifically, social proof and default nudges have become an increasingly popular method to move consumers towards healthy eating behaviour. This study aims to shed light on the effect of becoming aware of social proof and default nudges on the companies implementing them. While a social proof and default nudge did lead to enhanced healthy choice in a flight booking setting, consumers did not rate the company significantly different when they became aware of the nudge. Perceived restriction of autonomy was not found as an influential factor in the attitude formation towards a brand after becoming aware of the nudge, nor did desire for control influence the attitude
towards the brand during this study. These results suggest that using these nudges lead to consumers making a healthier choice, and that more research is necessary regarding the perceived restriction of autonomy after receiving system 1 nudges.
Introduction
15September 2015 is described as a historic moment by Cass Sunstein, Director of the U.S. Program on Behavioural Economics and Public Policy, and the co-author of the book “Nudge:
improving decisions about health, wealth, and happiness”. It is the day that U.S. President Barack
Obama signs an executive order for federal organizations to use behavioural science insights to design more effective policies to steer citizens towards making better choices (White House, 2015). One of these insights is the frequently discussed concept of nudges: which are adjustments in an environment to gently push people towards desirable behaviour either for themselves or for society.
Nudging can have a tremendous effect on people’s willingness to donate organs (Johnson & Goldstein, 2003), saving habits for pensions (Thaler & Sunstein, 2008), and food consumption behaviour (Salmon et al., 2015; Schwartz, Riis, Elbel, & Ariely, 2012). Unlike junk-food bans or other forms of coercion, nudges preserve freedom of choice but create an environment in which you are encouraged to commit to socially or politically desirable behaviour. This line of thought is widely embraced by governmental institutions, because of their cost efficiency and effectiveness. Not only U.S president Barack Obama, but also former UK prime minister David Cameron has set up “nudging units” to encourage pro-self and pro-societal behaviour from their citizens (Thaler & Sunstein, 2008).
More often than not, nudges work by making use of the boundaries of people’s cognitions and decision-making abilities. A promising heuristic to encourage healthy food consumption is the social proof heuristic: which is the principle that individuals look at the behaviour of others to determine their own actions (Salmon et al., 2015; Sinha, 2016). Although nudges encourage beneficial behaviour, ongoing work in current literature has addressed a possible issue. Because nudges often work on a subconscious level –they are found to disrespect the individual’s ability to
make autonomous decisions. This may restrict people’s perceived autonomy making the so-called
libertarian paternalism not so libertarian at all.
Furthermore, as nudging policies are on the rise companies are becoming increasingly exposed to implementation of interventions that encourage healthy food consumption. Many governments strive to be open when making use of nudges, in order to offer transparency of their policies and motives (Sunstein, 2014; Thaler & Sunstein, 2003). However, consumers becoming aware of the fact that they are nudged may result into perceived restriction of their autonomy, even when the nudge is for their own benefit. Current empirical research is scarce, that investigates the extent to which consumers perceive their autonomy to be restricted once they become aware that they are nudged through exploitation of the social proof heuristic by using a default option. This study aims to shed light on this topic, and to gain insight in the consumer’s attitude of brands that implement such nudges. Desire for control and perceived threat of autonomy were expected to play a role within the attitude formation of the brand after awareness of the nudge. Through this study, I aim to demonstrate if there are any potential consequences of disclosure of the nudge for the companies implementing them.
Although there is relevant research regarding the effectiveness of becoming aware of a nudge, there is a lack of in-depth research regarding the consequences awareness of nudges have for the attitude towards a brand. The paper contributes to current literature in several different ways. First, the results of this study provide additional evidence to studies of social cues with regards to eating behavior (Kroese, Marchiori, & De Ridder, 2016; Salmon et al., 2015;
Schwartz et al., 2012; Sinha, 2016). When cued with a social proof and default nudge, consumers in this research chose a healthy option more often than when they did not receive such a cue. This study shows that these effects exist even in a private and non-immediate consumption
setting. Second, this study extends current literature regarding attitude formation through the use of nudges. While most papers measure attitude formation by describing several types of nudges (Felsen, Castelo, & Reiner, 2013; Hansen & Jespersen, 2013; Jung & Mellers, 2016; Junghans, Cheung, & De Ridder, 2015; Reisch & Sunstein, 2016), this study exposes participants to a nudge first-hand to measure the effects after becoming aware of this manipulation. Although more research is necessary to draw conclusive results regarding this effect, first steps have been taken to establish the consequences of awareness towards the brands implementing them.
This study has relevant practical implications; managers of companies that offer food offerings with a non-immediate consumption moment can use these findings to move its consumers to a healthier choice. While companies often restrict unhealthy items, it might be more effective, more cost-efficient, and less restrictive to use the results of this study to
encourage healthy eating behavior. Furthermore, transparency of nudging is an important factor in respecting the individual’s ability to make autonomous decisions. In turn, this may lead to different judgements of the brand implementing these nudges based on the individual’s goals and beliefs. This research shows some insight into this topic, although additional research is needed to establish direct effects.
Participants in this study were exposed to a flight booking situation, where they were asked to choose a meal plan for future consumption during a flight. They were then nudged by placing a default option on the healthiest meal plan, and attitude towards brand was measured. The results of this research are beneficial towards the ongoing discussion regarding the ethics of nudging, and how choice architects can be transparent regarding their efforts to move people towards a healthy choice.
Literature review
Paternalism & libertarian paternalism
Paternalism originates from public policy and was first theoretically defined by Dworkin (1972), who states that paternalism is interfering with one’s liberty to address the individual’s own good. One example of such a policy is the law that require motorcyclists to wear a helmet while driving. The driver’s safety is protected, but in return he/she needs to give up a degree of freedom for his/her own benefit. It is argued that these steps are necessary, due to the decision making processes of humans. People’s rationales are transformed by heuristics and biases, which lead to irrational and sometimes unfavourable decisions and actions. The costs of potential (immediate or non-immediate) risks are often miscalculated, or the potential benefits are often over- or undervalued leading people to make poor decisions such as not fastening their seatbelts or consistently making poor food choices (Kahneman, Knetsch, & Thaler, 1991; Thaler & Sunstein, 2003). An obvious implication of paternalistic policies is the degree to which people feel restricted by such interventions, which may deprive self-determination and autonomy (Dworkin, 1972).
Dworkin (1972) however did not address the idea that paternalism can possibly exist without coercion, or strict limitations of one’s freedom. The idea of a less coercive form of paternalism was addressed much later in the literature, and addressed by the authors Thaler and Sunstein (2003) with the term Libertarian Paternalism. The authors argue that the term paternalism is based on two false assumptions. The first assumption being that paternalism is avoidable, or can be replaced by alternatives: organizations or agents must make decisions for others which by definition is paternalistic. Restricting one’s freedom in choice is therefore unavoidable, and in the direct power of the choice architect (i.e the agent responsible for setting an environment in which individuals make choices)(Thaler & Sunstein, 2008). The second and
most important misconception is that paternalism always requires coercion. For example, inhabitants of a country can by default be registered as an organ donor, but can choose to opt-out at any time. Although the context of the choice changes, the same options exist in both an opt-in or opt-out default. This approach has been proven extremely successful in European countries with regards to how much of the general population subscribes as a donor. Countries that are similar in culture, can yield extremely different results across these two different systems. For example, Austria boasts 99.8% of their population as donors through their opt-out system, while Denmark has a mere 4.25% of their population as donor through their opt-in system (Johnson & Goldstein, 2003).
Nudges
A popular form of libertarian paternalism are nudges, which was popularized by Thaler and Sunstein (2008) in their book “Nudge: Improving choices about Health, Wealth, and Happiness”. Nudges are used to move people towards socially and/or politically favourable behaviour, or behaviour that is line with consumer’s own goals without restricting their choice. Often, nudges work by using heuristics, routines, and/or habits (Hansen, 2015; Thaler and Sunstein 2008). Barton & Yanoff (2015) give a clear illustration of nudges by placing it in a continuum. On the one side, there are interventions that use rational persuasion such as informing (the most liberal and non-controlling extreme) and on the other side, interventions are based on coercion (the most forceful extreme). In the middle of this continuum lies nudging –which is neither controlling nor based on pure rational persuasion only.
Two different types of nudges can be categorized according to Kahneman's (2003) dual-process theory (Hansen & Jespersen, 2013). According to this perspective, human thoughts can
arise in two different ways: through either system 1 or system 2. The first system is intuitive, automatic, and unconscious while the second system is reflective, rational, and conscious (Kahneman, 2003). Nudges that work through system 2 are more informative in nature and aim at making certain aspects in the environment more salient so that the individual becomes aware and can make a better decision (Hansen & Jespersen, 2013). One example of such a nudge is the look
right campaign in the United Kingdom. In contrast to the home countries of most tourists that visit
the country, traffic in the UK drives on the left side rather than the right side. As a result, many traffic accidents occur due to tourists looking the wrong way when crossing the street. To minimize these types of accidents, the UK government marked streets of tourist concentrated areas with the words “look right”. This simple nudge encouraged people to look the right way to avoid accidents (Thaler & Sunstein, 2008).
A system 1 nudge aims to either block or trigger certain heuristics to encourage desired behaviour (Barton & Grüne-Yanoff, 2015). One example of a heuristic-triggering nudge is shown in a study of Kroese, Marchiori, & De Ridder (2016). They show that merely placing healthy products next to the cash register almost doubles the sales of these foods as opposed to when the healthy products are in a less salient place in the store. Neither the healthy or unhealthy products were removed in either conditions, but rather the location in the store changed. People generally overestimate their level of self-control and are subject to impulse purchases. While these impulse-oriented locations are often saturated with unhealthy products, it can also be used in order to trigger this heuristic and encourage healthy consumption behaviour. In an additional condition of this study, consumers were informed that they were “being helped to make a healthy choice” when products were put in a prominent location. This however did not significantly influence the sales of the healthy products. In other words, awareness of the nudge did not influence the effect.
An illustrative example of a heuristic blocking nudge was shown through a field experiment by Schwartz et al. (2012). In the study servers of a Chinese fast-food restaurant asked customers whether they would like to “downsize” their meal. They found that a significant amount of the consumers (14%-33%, across three similar experiments) choose to downsize, without even receiving a discount. The authors of this study argue that customers mindlessly overconsume and choose large portion sizes, based on inertia: they previously ordered such meal sizes and use this decision as a reference for future decisions. Asking consumers to downsize removes this inertia and makes them more aware of the portion size, leading to a healthier choice in this example.
Social proof & default nudges
Social context has proven to be an effective fundament when designing nudges, especially within food consumption behaviour (Salmon et al., 2015; Schwartz et al., 2012; Sinha, 2016; Thaler & Sunstein, 2008). One particular behavioural aspect of the social context is the social proof heuristic, which is often used as a base for a heuristic-triggering nudge. As humans look for appropriate behaviour, they tend to look at how others are behaving and use this as a reference for their own actions (Cialdini & Trost, 1998). For example, Sinha (2015) vividly showed that eating in public (vs. private) contexts influenced the amount and type of calorie intake of participants. In the study, overweight individuals consumed more healthy calories in a public setting to repair their self-identity while comparing themselves to their peers. These same individuals consumed even more unhealthy calories when in a private setting to repair their affect. Although this example shows how social context can have negative consequences for the individual’s eating behaviour, there are also more beneficial applications (Kroese et al., 2016; Salmon et al., 2015). For example Salmon et al. (2015) used a social context to design a heuristic-triggering nudge. A simple sign stating that
several healthy products were “most sold in this supermarket” lead to higher purchase from consumers during a low state of self-control.
An even more effective form of nudging are manipulations that use default options. The before mentioned opt-in donor example is considered one of the most controversial yet effective examples of these types of nudges. They exploit the status quo bias, which is the tendency to stick towards an option which is already familiar. As people are loss averse, they tend to overestimate the disadvantages of an unknown option and overestimate the advantages of a familiar option. This phenomenon drives the force behind the default nudges, by offering a “safe” option (Kahneman et al., 1991). Targeting these two nudge manipulations towards a healthy choice is therefore likely to improve the consumption behaviour of consumers, therefore:
H1: Consumers exposed to a social proof and default nudge will choose a healthy choice
option more often than when they receive no nudge.
Perceived restriction of autonomy
Although system 1 nudges can serve as highly effective measures, the sentiment in the literature regarding this topic is somewhat bilateral in terms of ethics. While proponents of nudging see it as completely libertarian, opponents tend to take the term libertarian less narrow. As nudges often shape choices of consumers without their awareness, they can be considered to not be so libertarian at all. The intended use of nudges is targeted on some level of control of people’s choices in the form of exploiting limitations in decision making processes. This may be problematic for two reasons. The first being that people cannot guard themselves when they are nudged towards a decision that is not inherently theirs (Hausman & Welch, 2010). The second issue is that exposure
to a nudge implies that an individual needs assistance in making decisions, which disrespects of the individual’s ability to make autonomous decisions (Felsen et al., 2013). These issues are especially apparent when the nudge moves people towards a direction that they do not agree with. People feel that nudges should intend to “do the right thing”. In other words, if the nudge is congruent with the participant’s goals and morals (s)he is more likely to accept the nudge and react positively to the company nudging them (Jung & Mellers, 2016).
Consumers’ attitude towards nudges change if they feel that these issues come to play. In a study by Reisch & Sunstein (2016), consumers’ attitudes towards nudges was measured in six European countries. They found that consumers are generally positive about nudges, but only when they meet several conditions. The first condition is that nudges are designed to have legitimate goals that fit in the interests of people. Second, a nudge should not be aimed at influencing people for which they cannot guard themselves, in this case they are often considered manipulative and are not morally approved. Being able to guard oneself towards external manipulations targeted on subconscious levels requires awareness of the manipulation and it’s intended effect, often not present by system 1 nudges. Not knowing that a certain environment is steered towards influencing your choice can have consequences on the person’s ability to guard themselves from the nudge.
These findings are extended by a study of Jung & Mellers (2016), who found that system 1 nudges were rated as less favourable than system 2 nudges across the American public. Consumers feel that system 1 nudges threaten their autonomy, which also led to lower ratings of attractiveness for companies that implement such practices. The authors found that the person’s desire for control was an important moderator in the extent to which consumers perceived nudges to be autonomy threatening. Groups who scored high on this personal trait were more sensitive to feeling restricted in their autonomy and are more likely to show resistance towards the nudge.
Both studies however are limited in the sense that subjects are not exposed to a nudge manipulation themselves, but rather rate different nudges from a third-person perspective. When consumers become aware of being nudged themselves, this may have a more representative effect on the extent to which they feel manipulated. This is likely to influence the attitude towards the company or brand that is implementing such nudges.
Transparency and awareness of nudging
To counter the potential ethical concerns of (system 1) nudging, proponents encourage governments to be transparent and open about implementing such policies (Sunstein, 2014). Thaler and Sunstein (2008) argue that if a government adopts a policy that is not defendable in public because of lack of transparency, it risks forfeit of the policy even when it is highly effective. They argue that nudges should be visible and when the nudge is not visible itself, it should be disclosed by choice architects.
Although several scholars argue that nudges often work best when “in the dark” (Bovens, 2009; Hansen & Jespersen, 2013), many recent empirical studies seem to argue otherwise (Bruns et al., 2016; Loewenstein, Bryce, Hagmann, & Rajpal, 2014). Various nudges seem to be effective, even when they are disclosed and made explicit by institutions exposing people to such interventions. For example, in a laboratory study by Bruns et al. (2016), participants were given 10 euro’s which they could use to contribute to the “climate protection fund’. They were allowed to choose their own contribution and the rest of the money would be their private payoff. Between two different conditions (plus an additional control group with no default), both had a default option of 8 euro’s. The second condition however informed participants that the default option was intended to encourage the participant to contribute more of their amount. Compared to the control
group, the default option resulted into a higher contribution. However, when informing participants of the purpose of the default, the contribution to the fund did not significantly change compared to when they were not informed. In other words, even when informing participants of the intended effect of the default option, they still contributed more compared to the control group with no default option. This seems to also be the case in the realm of healthy eating behaviour. Kroese et al. (2015) placed items in a more salient place in a train station shop to manipulate people’s choices and to encourage healthy eating behaviour. Disclosure of this nudge did not have any influence on this effect.
The ongoing discussion regarding ethical acceptability of nudges has motivated much research in studying the effect of disclosure of the nudge and its effectiveness. There is growing evidence that awareness of the nudge does not in fact decrease its effectiveness. If this is the case, organizations can design open and disclosed nudges without necessarily compensating on effectiveness. However, a missing part in this line of research are the consequences disclosure has for the brand. Surprisingly, this has been hardly subjected within the current body of research but is essential for the practical acceptance of ethical nudging.
Attitude towards nudges and the brand
In general, people feel less comfortable with system 1 nudges than system 2 nudges (Felsen et al., 2013; Jung & Mellers, 2016; Reisch & Sunstein, 2016). It is these types of manipulations that make consumers feel as if the company implementing them believes that he/she cannot make autonomous decisions. Felsen et al. (2013) found that covert nudges which worked on a subconscious level were deemed less favourable than overt, conscious-level nudges. These results were found in several contexts, including eating behaviour. Although there is always a certain
receives from these nudges, system 1 nudges are generally regarded as somewhat manipulative (Felsen et al., 2013; Jung & Mellers, 2016). The authors argue however that when the nudge is in line with the individual’s conscious decisions, allows rational thoughts and reflections before a decision, and if it is free of non-transparent and covert factors, the individual is more likely to accept the nudge as a base for his/her decision making. Inconsistencies with those values however are likely to lead to resistance and negative attitude formation towards choice architects providing the nudge. System 1 nudges are generally not disclosed, and it is these types of nudges that aim at exploiting cognitive limitations of consumers’ decision making processes. This rejects two important criteria for acceptance of the nudge; (1) the fact that people cannot guard themselves from the manipulation and therefore (2) they do not allow the individual to make autonomous decisions (Felsen et al., 2013). Therefore I expect that disclosure of these nudges will have a negative influence on the consumer’s attitude towards the brand.
H2: Awareness of the social proof & default nudge will lead to decreased attitude towards the
brand compared to when consumers are not aware of the nudge.
It is important to consider that different types of individual ideologies react differently to nudges. In a study of Costa & Kahn (2013), they showed that a high desire for control lead to reactance, which is a psychological state of resistance when people perceive a threat in their autonomy. In order to encourage people to consume less energy, participants where shown energy consumption of similar households. While participants that are liberals with low desire for control consumed less energy after being exposed to the consumption of other households, conservatives with a high desire for control responded much differently. These participants had less confidence
that this program was effective, disliked it more than liberals, and made them actually consumer
more energy. While liberals decreased their consumption by almost 6% on average, conservatives
increased their consumption by 1%. While consumers generally have complete freedom of choice with system 2 nudges, system 1 nudges are more difficult to counter. Nudges that exploit intuitive and unconscious decision making processes, lead to consumers feeling that they have limited control over these responses (Hansen & Jespersen, 2013), especially when they have a high desire for control (Costa & Kahn, 2013; Jung & Mellers, 2016). Therefore, it is expected that perceived threat of autonomy mediates the attitude formation process and that desire for control moderates this relationship.
H3: Perceived restriction of autonomy mediates the effect between awareness (vs no
awareness) of a social proof and default nudge on attitude towards the brand.
H4: Desire for control moderates the extent to which consumers perceive a restriction of their
autonomy after becoming aware of a social proof and default nudge compared to when they are unaware.
As stated in the paper of Felsen et al (2013), a nudge should be in line with the individual’s goals and conscious decisions. This is an important criterion for the acceptance of the nudge, due to the fact that potential benefits of the nudge can outweigh the perceived restrictions. In the case of healthy eating, the individual’s eating goals is therefore likely to influence the hypothesized effects. As this study looks into nudging in the field of healthy eating, health consciousness is expected to influence that effect:
H5: Health consciousness moderates the extent to which awareness (vs no awareness) of a
Conceptual framework
Figure 1 Conceptual model 1a
Figure 2 Conceptual model 1b
Awareness of Nudge Attitude towards brand Perceived threat of autonomy Desire for control H2 Health Consciousness Nudge exposure Healthy Choice H1 H3 H4 H5
Data and method
Preliminary study
A pre-test was conducted in order to measure the effect of a nudge manipulation through setting a social proof and default option. An online experiment was conducted through a survey. 94 participants were acquired through Amazon’s Mechanical Turk, and were paid 0.40 euro per completed survey. Participants were exposed to a hypothetical airline named “Sky-High”, and were briefed to go through the process of booking a flight for a business trip. A flight booking was chosen because it offers choice of a meal for future consumption. This eliminates any potential hedonic effects of an immediate consumption environment. After the introduction, participants were asked to answer an attention check question to check whether they were paying attention and could actually speak English. They were requested to indicate what instructions they received: either to “Write a short essay”, “Book a business flight” or “Pick a product in the supermarket”. Participants whom answered another question than to book a business flight could not further participate in the study. After exclusion of these participants (13), 81 participants remained (Mage=29.9, 74.1% male). Most participants are single (49.4%), employed fulltime (60.5%), and from either the USA (42%) or India (51,9%). The respondents were generally highly educated, with 69.1% having a Bachelor’s Degree.
Participants were then exposed to several decoy questions in order to refrain them from guessing the hypothesis. First they needed to select one of two flights that differed in price, departure/arrival time, transfers, and flight time (appendix 1). Second, participants were exposed to another decoy question and asked to pick a comfort level ranging from standard economy to business class for an extra fee (appendix 2). After these two decoy questions, participants were exposed to the target manipulation of the survey. Participants were randomly assigned to one of
two conditions. In the first (control) condition, they were asked to select a meal plan from three choices: a “Healthy meal set (300kcal)”, “International meal set (580kcal)”, or an “Oriental meal
set (498kcal)”. The calories per meal are based on the average airline meal calories (Washington,
2017). In the second (nudge manipulation) condition respondents were able to select the same three meal sets. However in this condition, the “Healthy meal set (300kcal)” was pre-selected as a default option. An additional text was added stating “a default has already been selected based
on the most chosen option”. This text is a social proof nudge and is adopted from a study of Salmon
et al. (2015). Please refer to appendix 3 for an overview of these conditions and how the nudge was displayed. In all conditions and options, participants had the possibility of selecting a vegetarian option to rule out any vegetarian preferences.
In the control condition participants chose the healthy option 17 times, participants who received the social proof and default nudge chose the healthy option 26 times. Figure 3 summarizes these results. A crosstab with Pearson Chi-Square analysis was conducted. The difference was significant χ(1)= 4.34 p= 0.04, meaning that participants successfully chose the healthy option more often when exposed to a social proof and default nudge. The intended manipulation is therefore deemed successful and is used for the main study of this research.
0 5 10 15 20 25 30 Nudge Control
Healthy choice
Figure 3 Pre-testMain study
The same manipulation was used as in the pre-test, which was the process of booking a flight for a business trip and nudging consumers towards a healthy meal plan through a default setting and social proof text. The same decoy questions and targeted manipulations were used, however participants were now exposed to one of three conditions. The first condition nudged consumers towards a healthy choice by showing a default option with a text stating that the option was defaulted based on the most chosen option. The second condition had the same manipulation, however the participant was made aware of the nudge after the manipulation. The awareness making text was adopted from Bruns et al., (2016), which stated the following:
“When you selected a meal plan, the above default was already selected for you. A statement next to the choice “Healthy meal plan” noted that this option was most chosen by other passengers. Please consider that the preselected default, and the statement “most chosen by other passengers” might have had an influence on your decision.
The default and text are part of Sky High's efforts to encourage healthier eating behaviour from you as a passenger.”
Appendix 3 shows the different conditions, and the awareness making text used in the survey.
The third and final condition was the control condition, were participants were asked to choose a meal plan without a nudge or awareness making manipulation. This non-factorial condition was added in order to measure the effect of the nudge compared to a control condition, specifically for the first hypothesis. This isolated the effect of the nudge to determine whether the participants
have successfully been steered into a healthy direction. Participants were randomly assigned to one of the three conditions.
After the manipulation, all respondents from all conditions were informed that they had successfully booked a flight at Sky-High airlines. They were then asked the same questions regarding the dependent variables of this study. They were asked to state the extent to which they agree/disagree with several statements regarding their attitude towards the company, perceived restriction of autonomy, desire for control, health-consciousness, and they were asked several demographic questions. Please refer to appendix 4 for all item scales and measures. When the survey was completed, participants were debriefed. This stated a text with the actual purpose of the study, and that they were exposed to a nudge and decoy questions. They were then asked if they would like to receive a copy of the results of this study.
Sample
The anticipated sample for this research are all consumers whom purchase products or meals that they consume themselves in the Netherlands. Dutch consumers above the age of 18 were therefore targeted as sample for this study. Participants were collected using a non-probability sampling strategy, through convenience sampling. They were acquired through the use of social media, email, and word-of-mouth. Respondents were collected over a period of one and a half weeks, in which the online experiment opened on the the 25th of April 2018 and closed on the 4th of May 2018.
An a-priori power analysis was conducted in order to anticipate the desired number of participants to find a significant effect. The anticipated effect size was based on the study of Salmon et al. (2015) (Cohen’s d=0,8) which also used a social proof based nudge in the context
of healthy eating. The effect size was decreased due to the difference in the nature of this study. In this research, the differences between groups are slightly smaller which may lead to a slightly more conservative effect size. With the anticipated effect size at 0,65 and the intended probability level at 0,05 –the minimum sample size per group amounted to 39 participants. In total across the three different conditions, a sample size of 117 would theoretically suffice to find an effect. 134 participants (Mage= 28,8 54,5% female) completed the online survey. The age varied between 19 and 68, with a standard deviation of 11.13. Most participants are between 21-26 (63,5%), are full-time students (46,3%), and acquired a Bachelor’s Degree (60,5%). In general, participants were quite conscious about their health (M=4.84, 7-point likert scale, SD= .85) and had a relatively high desire for control (5.81, 7-point likert scale, SD=.83).
Measures
All scale items can be found in appendix 4.
Perceived threat of autonomy.
Perceived threat of autonomy was measured using a 4-item scale developed by Roubroeks, Ham, & Midden (2011). Participants were asked to rate the extent to which they agreed to several statements (e.g. “The airline’s booking process tried to manipulate me”) on a 7-point likert scale (1=Strongly disagree, 7=Strongly agree). This scale proved to have sufficient reliability, with a Cronbach’s Alpha of .75. A principal axis factor analysis indicated one component with an eigenvalue of > 1 (KMO = .68, χ2 (6) = 137.781, p < 0.001), consistent with literature (Roubroeks
Attitude towards brand.
A 2-item scale developed by Reimer & Benkenstein (2018) was adopted for this study. Two slider scales were presented to participants, and they were asked to rate the extent to which they found the airline “Sky High” to be either good (vs bad), and positive (vs negative). A reliability analysis showed a Cronbach’s Alpha of .88.
Health consciousness
To measure the extent to which the participant is already conscious about their health, an 11-item scale of Hong (2009) was used. A reliability analysis showed that one item (“I only worry about
my health when I get sick”) did not properly correlate to the total score of the scale, with a corrected
item-total correlation of -.27. This is likely explained due to the specific population of this study, most participants are relatively young and may have less worries about their health than an older population. After deletion of this item, the Cronbach’s Alpha was .87 (vs .82 with all 11 items). A principal axis factoring analyses was conducted on the scale. The Kaiser-Mayer-Olkin measure showed adequate sampling (KMO = 0.848). Bartlett’s test of sphericity χ2 (45) = 659.936, p < 0.001 indicated that correlations between items were large enough for a principal axis factor analysis. Three components had eigenvalues higher than one with a combined explained variance of 72.52%, and the scree plot shows levelling off after three factors. Therefore three factors were retained and due to the low correlation between factors, rotated with a varimax rotation. These findings are consistent with the original literature regarding this scale, which intends the scale to load to three distinct factors: self-health awareness (SHA), personal responsibility for health (PR), and health motivation (HM). Table 1 shows the factor loadings after rotation.
Table 1 Factor loadings of health consciousness scale
Item Rotated factor loadings
SHA PR HM
I’m very self-conscious about my health. 0.765 -0.005 -0.371 I’m generally attentive to my inner feelings about my health. 0.749 0.09 -0.459
I reflect about my health a lot. 0.719 0.301 -0.16
I’m concerned about my health all the time 0.319 0.856 0.211 I notice how I feel physically as I go through the day 0.465 0.206 0.174 I take responsibility for the state of my health. 0.739 -0.279 -0.076 Good health takes active participation on my part 0.741 -0.273 0.198 Living life without disease and illness is very important to
me.
0.572 -0.056 0.368
My health depends on how well I take care of myself. 0.634 -0.195 0.33 Living life in the best possible health is very important to me. 0.801 -0.105 0.105
Note: Factor loadings over .3 are bolded. Desire for control
An adjusted 3-point scale by Peluso, Bonezzi, De Angelis, & Rucker (2017) which was adopted from Burger & Cooper's (1979) original 20-point scale was used to measure this variable. Participants were asked to rate the extent to which a statement applies to them (e.g “I enjoy having
control over my destiny”) on a 7-point likert scale (1=does not apply to me at all, 7= always applies to me. A reliability analysis shows that the Cronbach’s Alpha is .79. A principal axis factor
analysis indicated one component with an eigenvalue of > 1 (KMO = .65, χ2 (3) = 138.205, p < 0.001), consistent with literature.
Results
Descriptive statistics
The standard deviations, means, correlations, and Cronbach alphas are presented in Table 2. There is a positive and significant correlation between participants that received a nudge manipulation and healthy choice (r=.18, p<0.05). Health consciousness was negatively correlated to age (r=-.20, p<0.05), and desire for control was positively correlated to attitude towards brand (r=.23 p<0.05).
Table 2 Correlations Among and Descriptive Statistics For Key Study Variables
Variables M(SD) 1 2 3 4 5 6 7 8 9 1. Age 28.8 (11.13) - 2. Sex 1.54 (.50) -.19* - 3. Education 6.50 (1.85) -.04 .09 - 4. PTOA 3.74 (1.14) -.00 .07 .04 (.75) 5. Health consciousness 4.84 (.85) .20* -.00 .09 .02 (.87)
6. Attitude towards brand 25.22 (10.6) .08 .09 .04 .00 -.14 (.88) 7. Desire for control 5.81 (.83) -.04 -.03 .14 .10 -.16 .23* (.79)
8. Healthy choice .51(.50) -.02 .15 -.01 .04 .03 -.05 -.03 - 9. Condition 1.01(.83) .18* .02 -.06 -.09 -.12 .09 -.12 .13 -
Social proof & default nudge
The first hypothesis of this study was to test whether exposure to a social proof and default nudge would lead to participants choosing the healthier option more often. Figure 4 shows a graphic representation of the results. In both the nudge and nudge + awareness condition, consumers chose more healthy products (26 & 25) compared to the control condition (17). The data was analysed using a crosstab with Pearson’s Chi Square. The analysis showed
that there was a significant association between the nudge condition and whether participants would choose a healthy choice χ(1)= 4.56 p = 0.03. Hypothesis one is therefore accepted.
Attitude towards the brand
An independent t-test was conducted in order to analyse the effect of awareness of the nudge towards the attitude towards a brand. On average, participants that are aware of the nudge had a slightly higher attitude towards the brand (M=26.51 SE=1.78) than participants who were not aware (M=24.75 SE=1.17). This difference however, -2.06, CI [-6.108, 1.987], was not significant t(113)=-1.08 p=0.32. The effect size was calculated and resulted in a Cohen’s d of 0.19. When selecting participants that had a health consciousness score below the mean score (<4.84), participants had a slightly lower attitude towards the brand when made aware of the nudge (M=21.33 SE=2.66) compared to those whom were unaware (M=25.11 SE=1.77). This difference,
17 26 25 0 5 10 15 20 25 30
Control Nudge condition Nudge + awareness
condition
Healthy choice
3.78 CI [-2.508, 10.064], however is not significant t(52)=1.21 p=.23. From this analysis, hypothesis two is rejected.
Perceived restriction of autonomy
Although there was no direct effect found between awareness of the nudge towards attitude towards the brand, it may still be statistically relevant to discover whether an indirect effect exists across the independent, mediating, and dependent variable. It is possible that a negative effect exists between independent variable X and mediating variable M, while a positive effect exists between M and dependent variable Y. In sum this results in no visible direct effect, however a deeper analysis could show that there is mediation between the variables (Hayes, 2009). Therefore, a regression analysis through Andrew Hayes’ process tool model 4 (Hayes, 2012) was used in order to analyse whether there is an indirect effect of perceived threat of autonomy between awareness of the brand and attitude towards the brand. Table 3 and 4 summarize the results of this analysis. There was a positive yet insignificant effect between nudge awareness and attitude towards brand (c1=2.06 p=.32), meaning that awareness increased the rating of the brand. The effect between perceived threat of autonomy and awareness was negative (b1=-.04 p=.90) yet insignificant. A bias-corrected bootstrap interval for the indirect effect (a1b1=-.00) based on 5,000 bootstrap samples had an interval around zero (-.50 and .47) indicating likeliness of no significant effect. Hypothesis three is therefore rejected. A similar analysis was conducted for participants with a high desire for control (M>5.84) but yielded no significant results, please refer to table 5 and 6 for a summary of these results.
Table 3 Mediation analysis
Consequent
PTOA (M) ATB (Y)
Antecedent Coeff. SE p Coeff SE p
Awareness (X) a1 .02 .22 .94 c1 2.06 2.04 .32
PTOA (M) - - - b1 -.04 .90 .96
constant i1 3.78 .13 .00 i2 24.61 3.62 .00
F(1,113) = .01, p=.94 F(2,112) = 5.05, p=.61
Table 4 direct, indirect, and total effect.
Effect SE p
Direct effect c1 2.06 2.04 .31
Total effect c1 2.06 2.05 .32
Boot SE Boot LLCI Boot ULCI
Indirect effect a1b1 -.000 .23 -.50 .47
Table 5 Mediation analysis with high desire for control
Consequent
PTOA (M) ATB (Y)
Antecedent Coeff. SE p Coeff SE p
Awareness (X) a1 -.12 .26 .65 c1 3.04 3.10 .33
PTOA (M) - - - b1 .01 1.79 .99
constant i1 4.06 .17 .65 i2 21.33 7.56 .00
F(1,44) = .20, p=.65 F(2,43) = .48, p=.62
Table 6 direct, indirect, and total effect with high desire for control
Effect SE p
Direct effect c1 3.04 3.10 .315
Total effect c1 2.06 2.05 .317
Boot SE Boot LLCI Boot ULCI
Desire for control
To analyse whether desire for control moderates the relationship between awareness of a nudge and perceived threat of autonomy a regression analysis was used through Andrew Hayes’ process tool, model 1 (Hayes, 2012). Table 7 summarizes these findings. There was no significant interaction between nudge awareness and desire for control on perceived threat of autonomy (c3=.05 p=.84), rejecting hypothesis four.
Table 7 Moderation analysis
Coefficient SE t p Intercept i1 3.72 .12 .20 .84 Awareness (X) c1 .05 .24 .21 .83 DFC (M) c2 -.13 .16 -.81 .42 Awareness * DFC (XM) c3 .05 .25 .20 .84 R2=.01 p=.78 F(3,130) = .36 Health consciousness
Table 8 shows the output of the moderation analysis regarding the moderating effect of health consciousness, through Andre Hayes’ process tool model 1 (Hayes, 2012). The analysis shows that there is no significant main effect between becoming aware of a nudge and attitude towards the brand (c1=-17.00 p=.17) nor between health consciousness and attitude towards the brand (c2=.48 p-.73). The analysis also shows that there is no significant interaction effect between health consciousness and awareness of the nudge on attitude towards the brand (c3=-3.17 p=.22). In other words, no significant evidence was found that a change in attitude towards the brand through becoming aware of a nudge from that brand is influenced by the consumers’ health consciousness, thus rejecting hypothesis five.
Table 8 Moderation analysis Coefficient SE t p Intercept i1 24.97 1.00 25.00 .00 Awareness (X) c1 -17.00 12.29 -1.38 .17 Health Consciousness (M) c2 .48 1.40 .34 .73 Awareness * HC(XM) c3 3.78 2.47 1.53 .13 R2=.05 p=.14 F(3,111) = 1.84
Conceptual model of hypotheses tests
Figure 5 Results of conceptual model 1a
Figure 6 Results of Conceptual model 1b
Awareness of Nudge Attitude towards brand Perceived threat of autonomy Desire for control 0.19 Health Consciousness Nudge exposure Healthy Choice 4.56* -.00 .05 3.17
Discussion
Social proof & default nudges in healthy consumption behaviour
As nudging becomes a more prominent way of moving people towards healthy behaviour, policymakers strive to make these methods transparent to consumers being nudged. This study investigated the effect of becoming aware of a nudge and its implications towards the brands implementing them. Participants of this study were exposed to a social proof and default nudge first-hand, which lead them to a healthier choice in a hypothetical flight booking situation. This extends the current body of literature (Salmon et al., 2015; Schwartz et al., 2012; Sinha, 2016) by providing additional evidence that social manipulations are successful in consumption choices, even in a private and non-immediate consumption setting. A mere default setting of the healthy option and a text stating that the option has been pre-selected based on the most chosen option lead to a 47% and 53% increase of healthy food options across two different conditions. This same manipulation yielded similar results with a different population in the pre-test. People look at the actions of others to determine their own behaviour, and move to default options that are safe and familiar (Cialdini & Trost, 1998). Exploiting this phenomenon can be used to encourage healthy eating behaviour without restricting the individual’s choice.
Attitude towards the brand, health consciousness, and perceived threat of autonomy
It was expected that in line with several studies (Felsen et al., 2013; Jung & Mellers, 2016; Reisch & Sunstein, 2016), attitude towards the brand would decrease when participants were made aware that they are nudged due to consumers feeling restricted in their autonomy. In this study however, attitude towards the brand did not significantly differ across conditions and participants did not feel that their autonomy was restricted by these types of nudges. The current consensus in literature
is that system 1 nudges generally lead to negative attitudes due to their autonomy-restricting nature (Felsen et al., 2013; Jung & Mellers, 2016; Junghans et al., 2015; Thaler & Sunstein, 2003). The results of this study however do not show this effect. More specifically, the direction of the relationship between becoming aware of the nudge and attitude towards the brand was actually opposite than hypothesized. In other words, awareness of the nudge did not decrease the attitude towards the brand but slightly increased it. Although this effect was not significant, the reason of this unintended effect may be explained due to the specific dynamics of this study. As stated in the paper of Jung & Mellers (2016), the consumer’s attitude of a nudge is dependent to the extent that they feel the nudge “does the right thing”. As the participants of this study were generally quite health conscious, it may be that they found that the company was actually helping them with their own intended behaviour. When analysing the data of participants with a lower health consciousness however, participants rated the attitude towards brand lower when they were made aware of the nudge –which is the hypothesized direction of the effect. Although these findings were not statistically significant, they do possibly shed some light on the formation of attitudes towards brands that implement nudges.
Although covert and sub-conscious level nudges are generally seen as autonomy restrictive (Jung & Mellers, 2016; Junghans et al., 2015; Thaler & Sunstein, 2003), there is some evidence that in specific cases these types of nudges actually increase autonomy (Felsen et al., 2013; Trout, 2005). Sometimes a lower-order (subconscious) desire is influenced through a nudge in order to be in line with a higher-order (conscious) desire. For example, when the person’s perception of hunger is altered through a nudge which is in line with the person’s goal to eat less. The individual’s goal with regards to health is therefore a likely important factor for the attitude formation. As stated by Felsen et al., (2013), potential benefits of the nudge can outweigh the
restrictions people feel in their autonomy. It is likely that in this specific study, consumers felt that the added benefit of the nudge contributed to their wellbeing.
Desire for control
Desire for control was expected to play a part in the attitude towards the brand after becoming aware of a nudge. Individuals with a high desire for control are generally more sensitive to restriction of their autonomy (Hausman & Welch, 2010). Making people with a high desire for control aware that they have been nudged subconsciously was therefore expected to result into negative associations with that brand. The results of this study however did not yield any significant insights into this phenomenon. The explanation for this lack of effect is likely due to the variance of the population sample. Generally, participants that contributed to this study had a remarkably high desire for control, and the low standard deviation in this measurement showed some homogeneity in the population.
Practical implications
The results of this study provide findings that are relevant for practitioners. First and foremost, social proof and default nudges can be used together to move consumers towards healthy behaviour. This study has shown that these manipulations also work in private and non-immediate consumption situations. Managers of airlines, rail transport companies, or any other organization that includes a meal plan for future consumption in their offering can use these insights to contribute to a healthier life for its customer. This specific method is interesting for practitioners as it does not require any elimination of choice for the customer, but rather a setting where they are encouraged to make a healthy choice.
Limitations
The results of this research need to be interpreted with some caution. First, the study illustrated a hypothetical situation that was conducted through an online survey. Participants were asked to imagine a situation where they book a business flight, but the research did not measure actual behaviour. This has some implications for the external validity, as the study may yield different results when performed in a real-life setting. The situation illustrated was quite specific, focused on a flight booking situation only. While this specificity was necessary to isolate the potential effects, it also has some limitations for the external validity.
Second, the specific sample of this study has given some limitations to this study. The participants were generally young, business and economics university students from the Netherlands. After analysis of this group, there was some homogeneity in both demographics and personal traits. For example, the standard deviations of the respondent’s desire for control and health consciousness were quite low, showing a low amount of variance. Respondents scored high on both of these traits which is could be a slightly skewed representation of the general population.
A third limitation is the amount of respondents that partook in this study. An a-priori analysis was conducted; however the effect size was significantly smaller than anticipated (0.19 vs the anticipated 0.65). The difference between conditions was smaller than expected, making the effect size more conservative than the theoretical benchmark study of Salmon et al. (2015).
Directions for further research
Several directions can be explored to enhance the current body of literature regarding nudges and its consequences towards brands. The results of this study indicated that, although insignificant, health consciousness played a role in the attitude formation of the brand. In this research, the underlying health goals influenced the outcome of the participant’s attitude towards the brand. The consumer’s perceived restriction of autonomy is dependent on whether the individual’s higher order desires align with lower-order mechanisms that these nudges address (Felsen et al., 2013). Congruency between the individual’s goal and the intention of the nudge is likely an important factor in consumers’ attitude formation that needs to be further addressed in future research.
Furthermore, similar research needs to be conducted using other populations than a student sample. A more heterogeneous sample with regards to both demographics and personal traits (such as desire for control and health consciousness) will contribute to a more solid conceptual model. Doing so in a real-life setting will provide a better framework of actual behaviour as opposed to a hypothetical situation.
Lastly, other system 1 nudges need to be researched in order to discover the differences amongst different types of nudges that work on a subconscious level. This research should be focused on finding when and why customers feel restricted in their autonomy, and how underlying personal traits moderate this effect. Doing so will enhance the body of knowledge regarding the ethics behind nudging, and will contribute to overt and transparent nudging. This study used a heuristic-triggering nudge, by exploitation of the status quo and social proof heuristic. Conducting studies that use nudges that work by either triggering or blocking other heuristics and biases than the before mentioned will extend the robustness of this theoretical framework.
References
Barton, A., & Grüne-Yanoff, T. (2015). From Libertarian Paternalism to Nudging—and Beyond.
Review of Philosophy and Psychology, 6(3), 341–359.
https://doi.org/10.1007/s13164-015-0268-x
Bovens, L. (2009). The Ethics of Nudge. In Preference Change. https://doi.org/10.1007/978-90-481-2593-7_10
Bruns, H., Kantorowicz-Reznichenko, E., Klement, K., Luistro Jonsson, M., & Rahali, B. (2016). Can Nudges Be Transparent and Yet Effective? SSRN Electronic Journal. https://doi.org/10.2139/ssrn.2816227
Burger, J. M., & Cooper, H. M. (1979). The desirability of control. Motivation and Emotion,
3(4), 381–393.
Cialdini, R. B., & Trost, M. R. (1998). Social influence: social norms, conformity, and compliance. In The handbook of social psychology (pp. 151–192).
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. https://doi.org/10.1111/jeea.12011
Dworkin, G. (1972). Paternalism. The Monist, 56(1), 64–84. https://doi.org/10.2307/27902250 Felsen, G., Castelo, N., & Reiner, P. B. (2013). Decisional enhancement and autonomy : public
attitudes towards overt and covert nudges. Judgment and Decision Making. https://doi.org/10.1177/2332858416674007
Hansen, P. G., & Jespersen, A. M. (2013). Nudge and the Manipulation of Choice. European
Journal of Risk Regulation, 4(01), 3–28. https://doi.org/10.1017/S1867299X00002762
Philosophy, 18(1), 123–136. https://doi.org/10.1111/j.1467-9760.2009.00351.x
Hayes, A. F. (2009). Beyond Baron and Kenny: Statistical mediation analysis in the new millennium. Communication Monographs. https://doi.org/10.1080/03637750903310360 Hayes, A. F. (2012). PROCESS: A versatile computational tool for observed variable mediation,
moderation, and conditional process modeling. White Paper. https://doi.org/978-1-60918-230-4
Hong, H. (2009). Scale Development for Measuring Health Consciousness:
Re-conceptualization. 12th Annual International Public Relations Research Conference. https://doi.org/10.1016/j.pubrev.2009.01.003
Johnson, E. J., & Goldstein, D. (2003). Do Defaults Save Lives? Science. https://doi.org/10.1126/science.1091721
Jung, J. Y., & Mellers, B. A. (2016). American attitudes toward nudges. Judgment and Decision
Making, 11(1), 62–74. Retrieved from
http://ideas.repec.org/a/jdm/journl/v11y2016i1p62-74.html
Junghans, A. F., Cheung, T. T., & De Ridder, D. D. (2015). Under consumers’ scrutiny - An investigation into consumers’ attitudes and concerns about nudging in the realm of health behavior Health policies, systems and management. BMC Public Health, 15(1).
https://doi.org/10.1186/s12889-015-1691-8
Kahneman, D. (2003). Maps of bounded rationality: Psychology for behavioral economics.
American Economic Review. https://doi.org/10.1257/000282803322655392
Kahneman, D., Knetsch, J. L., & Thaler, R. H. (1991). Anomalies: The Endowment Effect, Loss Aversion, and Status Quo Bias. Journal of Economic Perspectives.
Kroese, F. M., Marchiori, D. R., & De Ridder, D. T. D. (2016). Nudging healthy food choices: A field experiment at the train station. Journal of Public Health (United Kingdom), 38(2), e133–e137. https://doi.org/10.1093/pubmed/fdv096
Loewenstein, G., Bryce, C., Hagmann, D., & Rajpal, S. (2014). Warning: You are About to Be Nudged. SSRN Electronic Journal. https://doi.org/10.2139/ssrn.2417383
Peluso, A. M., Bonezzi, A., De Angelis, M., & Rucker, D. D. (2017). Compensatory word of mouth: Advice as a device to restore control. International Journal of Research in
Marketing, 34(2), 499–515. https://doi.org/10.1016/j.ijresmar.2016.10.003
Reimer, T., & Benkenstein, M. (2018). Not just for the recommender: How eWOM incentives influence the recommendation audience. Journal of Business Research, 86, 11–21. https://doi.org/10.1016/j.jbusres.2018.01.041
Reisch, L. A., & Sunstein, C. R. (2016). Do Europeans Like Nudges? SSRN Electronic Journal. https://doi.org/10.2139/ssrn.2739118
Roubroeks, M., Ham, J., & Midden, C. (2011). When artificial social agents try to persuade people: The role of social agency on the occurrence of psychological reactance.
International Journal of Social Robotics, 3(2), 155–165.
https://doi.org/10.1007/s12369-010-0088-1
Salmon, S. J., De Vet, E., Adriaanse, M. A., Fennis, B. M., Veltkamp, M., & De Ridder, D. T. D. (2015). Social proof in the supermarket: Promoting healthy choices under low self-control conditions. Food Quality and Preference, 45, 113–120.
https://doi.org/10.1016/j.foodqual.2015.06.004
Schwartz, J., Riis, J., Elbel, B., & Ariely, D. (2012). Inviting consumers to downsize fast-food portions significantly reduces calorie consumption. Health Affairs, 31(2), 399–407.
https://doi.org/10.1377/hlthaff.2011.0224
Sinha, J. (2016). We are where we eat: How consumption contexts induce (un)healthful eating for stigmatized overweight consumers. Journal of Consumer Psychology, 26(2), 289–297. https://doi.org/10.1016/j.jcps.2015.06.015
Sunstein, C. R. (2014). Nudging: A Very Short Guide. Journal of Consumer Policy, 37(4), 583– 588. https://doi.org/10.1007/s10603-014-9273-1
Thaler, & Sunstein. (2003). Thaler, Sunstein (2003)_Libertarian Paternalism. The American
Economic Review, 67(4), 420. https://doi.org/10.1080/01944360108976250
Thaler, & Sunstein. (2008a). Nudge, Improving Decisions on Health, Wealth and Happiness. New Haven CT: Yale University Press.
Trout, J. D. (2005). Paternalism and cognitive bias. Law and Philosophy. https://doi.org/10.1007/s10982-004-8197-3
Washington, P. (2017). Airline average calories per meal. Retrieved from
https://www.washingtonpost.com/lifestyle/travel/new-study-ranks-healthy-airline-food-
options-hint-bring-your-own/2017/12/27/9e90a16a-e4d8-11e7-ab50-621fe0588340_story.html?utm_term=.2dce3cd94fb5
White House. (2015). Executive Order -- Using Behavioral Science Insights to Better Serve the American People. White House Office Secretary. https://obamawhitehouse.archives.gov/the- press-office/2015/09/15/executive-order-using-behavioral-science-insights-better-serve-american
Appendices
Appendix 1: Flight selection, decoy question
Appendix 3: Conditions Nudge
Appendix 4: Scales & items
Perceived threat of autonomy Cronbach’s alpha = .75
Please state the extent to which you agree or disagree with the following statements:
1 The airline’s booking process restricted me in my autonomy to choose how I wanted to book my flight
2 The airline’s booking process tried to manipulate me
3 The airline’s booking process tried to make a decision for me 4 The airline’s booking process tried to pressure me
7-point Likert-scale. 1 = strongly disagree 7 = strongly agree. Adopted from Roubroeks, Ham, & Midden, (2011)
Attitude towards brand Cronbach’s alpha = .88
Please state your attitude towards the airline Sky High according to the following statements The airline Sky High is…..
1 Good/Bad
2 Negative/Positive
Health Consciousness Cronbach’s alpha = .87
Please rate the extent to which you agree with the following statements
1 I’m very self-conscious about my health.
2 I’m generally attentive to my inner feelings about my health. 3 I reflect about my health a lot.
4 I’m concerned about my health all the time
5 I notice how I feel physically as I go through the day 6 I take responsibility for the state of my health. 7 Good health takes active participation on my part 8 I only worry about my health when I get sick*
9 Living life without disease and illness is very important to me. 10 My health depends on how well I take care of myself.
11 Living life in the best possible health is very important to me.
7-point Likert-scale. 1 = strongly disagree 7 = strongly agree. Health consciousness scale: adopted from Hong (2009). * Item was deleted due to low reliability score. See methods.
Desire for control Cronbach’s alpha = .79
Please indicate the extent to which you agree with the following statements:
1 I enjoy having control over my own destiny 2 I enjoy making my own decisions
3 I prefer a job where I have a lot of control over what I do and when I do it
7-point Likert-scale. 1 = does not apply to me at all 7 = always applies to me. Desire for control adopted from (Burger & Cooper, 1979; Peluso et al., 2017).
Demographic questions
1 What is your age? 2 What is your gender?
3 What is your highest level of education? 4 Where do you currently live?
