Put a stop to license to sin : using subliminal priming to help individuals make healthier decisions

(1)

Put a stop to license to sin

Using subliminal priming to help individuals make healthier

decisions

Master’s Thesis MSc Business Administration – Marketing Track Amsterdam Business School – University of Amsterdam

Martine Daniëlle van Rijssel 10876642 Final version

Supervisor: MSc Tina Dudenhöffer

(2)

Statement of originality

This document is written by Martine Daniëlle van Rijssel, 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 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.

(3)

Abstract

This study explores the use of subliminal priming as a marketing tool to help individuals choose a healthier option when confronted with a situation in which they usually are inclined to sin and threat themselves. The research is based on existing theories about subliminal priming and consumer decision making, in particular self-licensing as a justification mechanism. Justification theory states that individuals who are reminded of refrain are more likely to choose the hedonistic option. Subliminal priming theory has discovered the effectiveness of a goal-relevant cognition when individuals are motived to reach this goal. The aim of the current study is to investigate if subliminal priming is effective in reminding individuals of their motivation to make health food decisions when they are placed in a hedonistic food decision situation.

A bilateral experiment was conducted measuring participants’ ‘health consciousness’ and asses if individuals who are considered to be health conscious are more likely to opt for the healthy alternative when primed with the words “choose healthy” when they are placed in a hedonistic situation compared to individuals who are not primed, are placed a non-hedonistic setting and who are not to be considered health conscious. Results indicate that subliminal priming does have a positive effect on making healthier food choices. An association between health consciousness and subliminal priming could not be established. An unexpected result occurred when the hedonistic condition was added to the equation. The results indicate that respondents in a hedonistic situation who are subliminally primed are less likely to choose the healthy food option compared to respondents in the same situation who are not primed.

(4)

Acknowledgements

The experiment conducted in this master thesis would not have been possible without the permission and support of Frits Nieuwenhuizen, CEO and owner of NH bioscopen and Dirk Groenendijk, manager of CineMeerse Hoofddorp. Thank you both for given me the opportunity to

conduct my experiment at their cinema and use their facilities.

A big thank you to my lovely family, friends and colleagues, without your support, encouragement and trust I would not have been able pull through. Thank you for believing in me when I did not believe in myself. To my love, without your unconditional faith in my abilities and unconditional love and support I would not have been able to finish what I started. Last, but definitely not least I owe a huge thank you to my supervisor Tina, who believed in this research from the start and who has kept me on the right path when I was wondering off. And off course dr. Alfred Zerres, who’s critical eye has kept me on my toes the entire process.

(5)

List of Figures

Figure 1 ………13 Figure 2 ………17 Figure 3 ………21

List of Tables

Table 1 ……….28 Table 2 ……….59 Table 3 ……….29 Table 4 ……….30 Table 5 ……….62 Table 6 ……….63 Table 7 ……….33 Table 8 ……….34 Table 9 .………34

(7)

Introduction

The number of overweight citizens has extremely increased from one in three in 1981 to one in two in 2015. Currently 50.3 percent of Dutch citizens, 20 years or older, are overweight, of which 36.6 percent is moderately overweight and 13.7 percent is suffering from obesity. Even though the extreme increase has flattened over the last couple of years, in 2015 12 percent of Dutch children, between 14 and 20, were overweight, with 3 percent being obese (Overgewicht > Cijfers & Context > Huidige situatie, 2015). The primary cause of death for Dutch citizens in 2015 were cancer and cardiovascular diseases (Statline - Overledenen; belangrijke doodsoorzaken (korte lijst), leeftijd, geslacht, 2016). These diseases are also known as nutritional diseases, as the majority of the diseases is caused by an unhealthy lifestyle containing eating too much (unhealthy) food, drinking (too much) alcohol and too little exercise.

The Dutch government has founded several initiatives to help citizens adopt a healthy lifestyle. An example is “30 minuten bewegen”, an initiative that stimulates citizens to participate in 30 minutes of moderate intensive exercise a day for at least five days a week. That could include chasing the dog or riding a bicycle to work (Nederlands Instituut voor Sport & Bewegen, sd). Another example is the update of the “Schijf van Vijf”, a diagram including the five different food groups that citizens should eat to create a diverse and healthy menu, in the second half of 2016. The new “Schijf van Vijf” includes more vegetable consumption per day and a decrease in the consumption of meat (replacements). In Appendix A an illustration of the current “Schijf van Vijf” can be found. The recommended daily intake of fruit and vegetables is currently two pieces of fruit and 250 grams of vegetables (Voedingscentrum, sd).

Furthermore, existing legislation has been altered to prevent citizens purchasing and using products that are unhealthy. As of January 1st_{, 2014 citizens have to be 18 or older to purchase and} consume alcohol and tobacco products, instead of 16 years. Youngsters under 18 are not allowed to carry alcohol in public (Alcohol, n.d.) The packaging of tobacco products has to display a warning

(8)

image, showing the consequences of consuming the products as of May 20th_{, 2016. An example of} such packaging can be found in Appendix B. It is also prohibited to advertise e-cigarettes and refill containers (Wet & Beleid, n.d.).

Additionally, new legislation has been created especially to protect children and help parents make healthier choices. A ban for food advertising towards children under the age of 13 - unless the product meets specific nutritional requirements – exists since January 2015. There is a complete ban for food advertising towards children under the age of 7. However, these bans do not exist for the packaging of the products ('Verbod op reclame voor kinderen werkt al goed', 2015). In December 2016 the Dutch State Secretary of Public Health Van Rijn announced that during 2017 images of movie- and comic characters will be banned from being featured on food packaging. The popular characters are mostly featured on food products that are unhealthy and contain too much sugar, salt or saturated fat. According to Van Rijn, it is important that children and parents are not confronted with tempting advertising on unhealthy food products while shopping in grocery stores (Tooms, 2016). In Appendix C, an example of such packaging can be found.

The need for a healthier lifestyle is also apparent from the consumer’s point of view. Fitgirls (young women who eat healthy and exercise extensively aspire to inspire peers by posting their meals, exercise regime and progress via social media such as Instagram) are becoming a new subculture among youngsters (Corstius, 2015). Famous chefs such as Jamie Oliver are increasingly adopting superfoods: “a food (such as salmon, broccoli, or blueberries) that is rich in compounds (such as antioxidants, fibre, or fatty acids) considered beneficial to a person’s health” (Merrian Webster Dictionary, n.d.), and teaching home cooks how to use these foods to create healthy meals for their family. New companies such as ‘Marley Spoon’, ‘Hello Fresh’ and large grocery store chains provide meal boxes that include all the necessary fresh ingredients that consumers need to prepare several fresh and healthy dinners per week using provided recipes. The meal boxes combine today’s need for convenience (consumers preselect what type of dinner they want, for how many people, for how many days of the week, and deliver it at home) with the need to eat healthy.

(9)

The importance of convenience is made clear when looking at eating out and ordering in. 29 percent of Dutch Millennials (individuals born between 1981 and 2000) state that at least one night a week they go out to have dinner and almost one in three orders take out dinners once a month. It is assumed that this development is a result of the increasing number of one person households, and two income households. In practice, the two income households with kids function as two separate households as children have to attend sport practices or other activities and the parents are working late. Every household member eats at a different time when it suits them best, therefore ordering in is more convenient. Additionally, the perspective on going out to dinner has changed between generations, Millennials do not only go out at a special occasion. As they are less materialistic than their previous generation, Millennials have money to spend on restaurants. Finally, women emancipation could also be an underlying reason as women today are not expected to have dinner on the table when their partner comes home. Not only dinner is consumed outside of the home, breakfast and lunch are most often consumed out of the home by Millennials than any other age group. Additionally, no other group purchases as much (snack) food and drinks at train and petrol stations as the Millennials (Berg, 2017).

There seems to be a discrepancy between what is good for the Dutch citizens and their current lifestyle. They might know what is good for them, but their actions speak louder than words and the numbers do not lie. It seems that one of the most common means of the Dutch government to help citizens improve their lifestyle and to decrease the number of overweight citizens is to ban the marketing of unhealthy foods. What about using marketing and advertising to encourage citizens to adopt a healthier lifestyle and eat healthier foods? What if there was a way to encourage citizens to choose healthier options but without being too obtrusive?

Subliminal advertising could be just that: “a technique of exposing consumers to product pictures, brand names, or other marketing stimuli without the consumers having conscious awareness” (Trappey, 1996, p. 571). Subliminal advertising is the marketing application of subliminal priming. Subliminal priming is the presentation of stimuli at such a speed or visual level

(10)

that it is below the threshold of consciousness and has some effect on individuals’ attitudes and/or behaviours (Smarandescu & Shimp, 2015).

Subliminal priming has been a topic of academic interest since the early 1900’s, however it did not became public knowledge until James Vicary’s hoax research during the 1950’s. He claimed to have projected the words ‘Eat Popcorn’ and ‘Drink Coca-Cola’ for three milliseconds during the frames of a movie in a movie theatre. This would have increased the sales of Coca-Cola with 18 percent and of the popcorn sales with more than 50 percent. In the early 1960’s Vicary had to admit that his experiment had been a hoax and that he had made up the results. Nonetheless, his experiment had made citizens aware of subliminal priming and it had sparked researchers’ interest in reproducing his research. It was not until the early 1990’s that researchers were able to find proof for the effectiveness of subliminal priming by providing reproducible scientific evidence (Pratkanis, 1992).

Murphy and Zajonc (1993) provided evidence that subliminal priming can influence individuals’ attitude towards an unfamiliar object and Bargh, Chen and Burrows (1996) confirmed that even behaviour can be affected by subliminal priming, but attempts to influence individuals’ level of persuasion had failed up until then. Just after the turn of the century Strahan, Spencer & Zanna (2002) found that in order for subliminal priming to enhance persuasion it has to meet two criteria: (1) The subliminal prime has to be a goal-relevant cognition and (2) The one being primed has to be motivated to pursue the goal in order for the prime to direct behaviour. Bargh stresses “the importance of importance” (Bargh, 2016, p. 50), if the goal of the prime is important to the person that is being primed, the size of the priming effect increases noticeably.

Based on the findings of Strahan et al. (2002) and Bargh (2016) the goal of this research is to see whether subliminal priming could help individuals who are motivated to live a healthy lifestyle and eat healthy make the right choice in situations in which the temptation to indulge in unhealthy food/snacks is high. Those situations are referred to as hedonistic situations. In this study these situations can be going out to dinner, ordering food, or other outings such as going to the

(11)

cinema, going to a music festival etc. These particular situations are tempting due to the fact that for the average person these situations do not occur on a very frequent basis and are associated with being a treat, therefore when they do occur individuals might feel that they can indulge. Especially if they have restrained from indulging at other occasions and therefor they can justify the indulgent choice (Mukhopadhyay & Johar, 2009). As mentioned previously as one of the underlining arguments for an increase in going out to dinner is the decrease in materialism in the younger generations, therefor the frequency in which hedonistic situations are encountered increases. However, if the association of these situations with indulgence and treat remains it will have negative effects on a healthy lifestyle and eating healthy.

The current study has an experimental research design, with an experiment which takes place at a local cinema. This location has been chosen as the researcher is an employee of the cinema and the cinema is seen as a hedonistic situation, as people do not visit the cinema on a frequent basis. In 2016 the average cinema visit for Dutch citizens was two times a year (Holland Film Nieuws, 2017). Respondents were divided into two groups and asked to watch a movie. One group has seen a subliminal prime displayed as an intermission clip including the words “choose healthy”. The other group functioned as the control group. During the intermission the respondents were allowed to purchase drinks and snacks. Respondents were then asked to fill out a survey. The research sample consisted of Dutch citizens above the age of 18, as subliminal priming is illegal in The Netherlands and is only allowed in a research setting for which individuals have to agree to participate.

The study will continue with the literature review which will discuss subliminal priming and decision making factors in more detail as well as the study’s hypotheses. The review will be followed by the data and method chapter, discussing the participants as well as the study’s measurements. The results section of the study will closely follow thereafter and the final chapter of the study entails the contribution of the study to the current knowledge, its limitations and suggestions for future research.

(12)

Literature review

Subliminal priming

The term subliminal priming consist of two components, subliminal and priming. Subliminal can be described as processing a stimulus below the threshold of perception (Hawkins, 1970). Priming comes from prime, which is defined by Oxford Dictionary as “to prepare somebody for a situation so that they know what to do, especially by giving them special information” (Oxford Advanced Learner's Dictionary, 7th edition). Smarandescu and Shimp explain that “subliminal priming involves presenting people with single words or images at a speed that is below the conscious threshold. These words/images activate people’s knowledge, beliefs, stereotypes, or cognitions. These cognitions are capable of influencing individual’s attitudes, choices, and behaviours…” (2015, p. 717).

Probably the most well-known example of subliminal priming is Vicary’s 1950’s coke and popcorn experiment, in which he claimed to have increased the sales of popcorn and Coca-Cola by projecting subliminal primes in a movie. When the news broke of this experiment the public opinion was one of concern: if a machine was capable of influencing people’s decisions to purchase refreshments what else could such a machine do? (Pratkanis, 1992) Ever since, the use of subliminal priming in advertising has been legally banned in several countries such as the U.S., Britain and Australia (Karremans, Stroebe, & Claus, 2006) as well as in The Netherlands (Stichting Reclame Code, n.d.).

Even though Vicary had to admit to fabricating the experiment results as he had never run the experiment, academia’s interest in the topic was sparked. It was not until several decades later that researchers were able to find reproducible analytical evidence for the effectiveness of subliminal priming (Pratkanis, 1992).

As mentioned previously it was first established that subliminal priming could influence an individual’s attitude (Murphy & Zajonc, 1993), followed by the discovery that it could affect behaviour (Bargh, Chen, & Burrows, 1996) it were Strahan, Spencer & Zanna (2002) however, who

(13)

discovered that subliminal priming could enhance persuasion, but only if two criteria are met. The two criteria being that (1) The prime should be a goal-relevant cognition and (2) The person being primed needs to have the motivation to pursue the goal. Emphasis is placed upon importance, if the goal is important to the individual being primed, the effect size increases (Bargh, 2016).

Based up on the two criteria from Strahan et all. (2002) and Bargh’s (2016) findings the following hypothesis is drafted: H1: Individuals who are subliminally primed with words associated

with health will have a higher tendency to make a healthy food choice compared to individuals that are not primed. Figure 1 represents a graphical representation of the hypothesis.

Figure 1: Hypotheses 1: Individuals who are subliminally primed with words associated with health will have a higher tendency to make a healthy food choice compared to individuals that are not primed.

Factors influencing the effectiveness of subliminal priming

Now that it was established that subliminal priming could indeed have an effect on persuasion behaviour, Karremans, Stroebe & Claus (2006) researched the effect of subliminal priming on brand choice. They reaffirmed that when the prime is highly relevant or applicable to a person’s motivation, the effect of the prime affecting a person’s actions is more likely to occur. In addition they have discovered that priming the means that could reach that goal will have a positive influence on the actual choice for those means. The Karremans et al. (2006) research was limited in terms of only measuring the participants’ intentions, not the actual consumption.

In order to find support for Karremans et al. (2006) findings, another type of motivation was researched in an attempt to reproduce the previously mentioned research. Additionally Bermeitinger, Goelz, Johr, Neumann, Ecker & Doerr (2009) tackled the limitations of Karremans et al. (2006) by measuring actual consumption instead of intention. The research reaffirmed the two

Subliminal priming words associated

with health

Tendency to make a healthy food choice H1 (+)

(14)

criteria needed for subliminal priming to be effective in terms of persuasion and having an effect on actual behaviour (in this case consumption). Moreover, by using made-up product brands it was concluded that a subliminal cue does not have to be familiar in order for it to be effective in an advertising context (Bermeitinger, et al., 2009). Translating these findings to the current study implies that the researcher does not have to use brands or branded products in the subliminal prime for the prime to be effective.

Verwijmeren, Karremans, Stroebe & Wigboldus (2011) wondered whether habit would influence the effect of subliminal priming. Existing habits act as a moderator for subliminal priming. Priming the habitual consumed brand does not influence brand-choice, even though it complies with the first criteria of effect subliminal priming. Surprisingly, if the primed brand is not the habitual brand, the influence of brand-priming is stronger than habit. The primed brand is more likely to be chosen over the habitual brand. Based on these results the researcher can assume that in the present study the food choices made by the primed participants should not be attributed to the habits of the participants.

Now that multiple studies have confirmed that subliminal priming has an effect on consumer behaviour, the question arises for how long the prime holds. Muscarella, Brintazzoli, Gordts, Soetens & Van den Bussche (2013) found several studies which indicated that the effect of subliminal priming on behaviour is not long lasting. The effect decreases rapidly and just after a few hundreds of milliseconds the significant effect is lost. They further mention that other studies state that these findings imply that the primed information should be applied almost immediately after exposure. This poses difficulties when implementing subliminal priming in everyday life. However, they also found a handful of studies that concluded that the effects of subliminal priming could be long-lasting. A limitation of those studies is that the priming was paired with visible positive rewards. In their own research Muscarella et al. (2013), tested if there is a long-term effect of subliminal priming without this reward. The primes were either above or near the threshold of consciousness and the time between the prime and the behaviour which it should have an effect on

(15)

were manipulated to be either 334ms, 1000ms or 5000ms to test the short- and long-term effects. Ultimately they were able to conclude that subliminal information can have an effect for at least five seconds after priming. Additionally they found an apparent crucial factor for the long-term effectiveness: the nature of the unconscious information. In this research the pictures of brands that were familiar to the respondents were the most powerful primes compared to non-brand pictures. In concluding the study it is mentioned that for subliminal priming to compete effectively with present stimuli (e.g. traditional forms of advertising) over a longer time interval the prime itself should be very powerful in familiarity for example by depicting highly familiar brand logos. Applying these results to the current study, the researcher has to make sure that the time between the prime and the consumption purchase is as short as possible. This time could be a little extended by using brands that are familiar to the participants, however the main suppliers of the cinema do not allow brands that do not belong to their house of brands to be sold in the cinema. The only option therefor is to use a non-branded prime and to place the prime as close to the intermission as possible.

Smarandescu & Shimp (2015) felt it was time to test subliminal priming in a more realistic purchase situation. They conducted three different studies. Study 1 reaffirmed that purchase intentions are significantly influenced by goal relevance. Study 2 placed the first study in a simulated store situation and found confirmation for their previous findings. In study 3 the second study was repeated, but with a time delay of fifteen minutes between the prime and the actual behaviour instead of the five minute interval used in study 1 and 2. The results of the third study indicated that there was no effect of the priming on the consumer choice decision. The researchers concluded that “subliminal priming can affect behavioural intentions and brand choice in situations where familiar brands are involved and a familiar brand is pitted against a market leader—but only when the dependent measures are presented in temporal proximity to the priming procedure (studies 1 and 2), but not after a 15 minutes delay (study 3)” (Smarandescu & Shimp, 2015, p. 725). Smarandescu & Shimp find it too premature to conclude that subliminal priming is not able to

(16)

influence consumer choice behaviour in real purchases situations and call for continuation of research (2015). The present study will create a real-life purchase decision in comparison to a simulated purchase situation as participants will have to purchase the products they desire. Additionally, the time interval between the prime and the purchase will be kept as small as possible, therefor the prime does not need to contain familiar brands as discovered by Bermeitinger et al. (2009).

Health consciousness and health related studies

Health consciousness has several definitions as well as operationalisations. The adopted definition for the present study is as follows: “a concept reflecting a person’s readiness to do something to his/her own health. Health conscious individuals are aware of and concerned about their wellness. They are motivated to improve and maintain their quality of life by engaging in healthy behaviors and being self-conscious about health” (Kaynak & Eksi, 2014, p. 773).

According to Papies, Potjes, Keesman, Schwinghammer and Van Koningsbruggen (2014) attractive food cues are one of the reasons why people tend to overeat, since they trigger hedonistic and reward processes in the brain. This is especially pronounced for people who are overweight or obese. Papies et al. (2014) state that there is a need for effective environmental interventions when it comes to such hedonistic cues, in order to protect influential people. One of these environments that is especially tempting is the grocery store. The food decisions there made influences several individuals for several days. The goal of the study is to see whether a simple supraliminal prime could decreases the purchase of energy-dense snacks among overweight people in a grocery store. The study focuses on energy-dense snacks as they “play a role in the increasing prevalence of being overweight, and because overweight and obese people consume more snacks than normal-weight consumers” (2014, p. 597). Respondents were approached in front of the grocery store and asked if they were willing to participate in a research. Participants received a recipe flyer with either a health and diet prime or a neutral supraliminal prime. After the purchase of the groceries the participants were giving a questionnaire to learn their height and weight, their attention to and awareness of the

(17)

primed flyer during their shopping trip. The purchases of the participants were analysed using the receipts. Overweight and obese participants who received the health and diet prime purchased 75 percent fewer snacks compared to overweight and obese participants in the control group. The prime was found to be effective for participants who were overweight or obese and only if they had paid attention to the primed flyer. However, participants did not have to be continuously aware of the prime during the shopping for it to be effective (Papies et al., 2004). These results show promise for a positive outcome of the present study as participants will see the prime only on the big screen and buy the consumption in the foyer of the cinema, they will not be continuously in the presence of the prime.

Taking the findings of health consciousness and health related studies into account, the following two hypotheses for the present study were drafted:

H2a: The effect of subliminal priming towards healthy food choices is more pronounced for individuals who are health conscious when exposed to healthy choice related primes.

H2b: The effect of subliminal priming towards healthy food choices is less pronounced for individuals who are not health conscious when exposed to healthy food choice related primes.

Figure 2 visually depicts these hypotheses in the research model.

Figure 2: The effect of subliminal priming towards healthy food choices is positively moderated by the level of health consciousness. Subliminal priming words associated with health Subliminal priming words associated with health Tendency to make a healthy food choice

Level of health consciousness

H1 (+)

(18)

Consumer choice decisions and hedonism

Research on consumer purchase decisions has primarily focused on one time decisions and on the quantity of the purchase, both in situations where a purchase was intended. Mukhopadhyay and Johar (2009) state that, in contrary to popular belief, decisions made at one point in time may have an influence on later decisions. They argue this statement by investigating the effects of withstanding the temptation to purchase or consume a tempting product on a subsequent tempting opportunity. They hypothesize that consumers who withstand the temptation of an impulse purchase at one point, are more likely to indulge at a later point in time. The effect only occurs when the previous shopping restraint is salient. The researchers propose that remembering the refrain of the impulse purchase allows consumers to justify the current indulgence. The hypothesis is based on prior research stating the importance of justification: consumers will make a decision more easily when they can justify that decision. Additionally if consumers have made an effort, they can justify their decision for choosing indulgence over necessity more easily. Using a series of four experiments Mukhopadhyay and Johar confirmed their hypothesis that “shopping restraint can serve as a justification for subsequent indulgence as long as the restraint is salient” (2009, p. 342). This condition was tested in study 2; respondents were asked to recall a situation in which they had seen a product on sale, which they had not intended to purchase before, but where tempted to purchase now and whether they did or did not purchase it in the end. The respondents in the high salience group where asked immediately after if they preferred chocolate cake or a fresh fruit salad in a direct comparison. The respondents who had refrained from purchasing the item on sale were more likely to choose chocolate cake over a fruit salad than respondents who had purchased the product on sale. Additionally, they found that increasing the self-esteem increases indulgence. This moderator of self-licensing was found in their third study. During this study respondents were put in a situation in which they could choose to use some of their experiment compensation for a bar of “organic European chocolate that was new to the market”. After the purchase decision, half of the respondents were told that the revenue of the chocolate bar was donated to charity and the other

(19)

half was not told anything. All respondents faced a second decision, to choose chocolate cake or a fresh fruit salad. Respondents who had purchased the chocolate bar when the earnings were donated to charity were more likely to choose the chocolate cake than respondents who had purchased a chocolate bar without donation (Mukhopadhyay & Johar, 2009).

De Witt Huberts, Evers and de Ridder (2012) build on the justification literature by looking into self-licensing. They describe self-licensing as “the tendency to rely on reasons and arguments to justify subsequent gratification” (2012, p. 490). If the context in which a decision is made allows justification of that decision, people are more likely to choose hedonistic goods. De Witt Huberts et al. found that previous researches investigating self-licensing did not represent real dilemmas of choosing instant gratification over negative future consequences of that decision (i.e. many researches offered a choice between a hedonistic option and a functional option of equal value). In a pilot study the researchers investigated whether depletion of self-control resources is the cause of self-licensing. Participants were asked to do a task once (control group) or twice (experimental group) in order to manipulate their perceived effort of doing the task, although the duration of performing either the one or two tasks was the same (1 x 10 minutes or 2 x 5 minutes). The Stroop task was performed by participants as a measurement for self-control, from the task could be concluded that participants who perceived to make more effort did not deplete self-control resources compared to the control group. Therefore, if an increase in hedonistic intake is measured after the effort is made, it cannot be accounted to depletion of self-control resources. For the main study participants were asked to participate in two studies, one dyslexia study and one consumer test for a large supermarket chain. The dyslexia study was, same as the pilot study, divided into two parts, the experimental group was told to have been selected to perform the study twice for validation of the dyslexia screener. The control group just had a break in between two parts of one study. This study manipulated the level of perceived effort. In the subsequent task participants were asked to taste two brands of four different snacks. It was hypothesized that the participants who perceived to have made more effort by performing two studies were more likely to feel licensed to indulge by

(20)

eating more unhealthy snacks. The results of the study confirmed the hypothesis that participants who believed to have completed two studies consumed an average of 26 grams more snacks than the control group. In other words, they took in 130 calories more in 10 minutes than participants who performed the same task but believed that they finished only one study (de Witt Huberts, Evers, & de Ridder, 2012).

The studies by Mukhopadhyay and Johar (2009) and De Witt Huberts et all. (2012) illustrate that justification and self-licensing are important processes when it comes to decision making. These mechanisms allow people to choose the hedonistic option even if it is not necessarily the best option for them. Based on these studies it should be logical to state that when placed in a hedonistic setting people are more likely to choose for the unhealthy option. However, in the current study it is proposed that placing people in such a setting but reminding them of their goal: being, living and choosing healthy i.e. being health conscious, by subliminal priming will result in choosing the healthy option. This all depends on the fact that the person exposed to the prime is goal relevant. Therefor the last hypotheses are:

H3a: The effect of subliminal priming toward healthy food choices is more pronounced for individuals who are health conscious when exposed to healthy choice related primes in a hedonistic setting.

H3b: The effect of subliminal priming toward healthy food choices is less pronounced for individuals who are less health conscious when exposed to healthy choice related primes in a hedonistic setting.

(21)

Figure 3: The complete research model

Subliminal priming words associated

with health

Tendency to make a healthy food choice

Level of health consciousness H1 (+) H2 (+/-) Hedonistic condition H3 (+/-)

(22)

Design

The present study was set up to have an independent measures experimental design. An experiment as previously described was conducted at a local cinema. Unfortunately, the number of participants was not sufficient to run statistical analyses, only 48 participants were involved in the experiment. In order to obtain the relevant data necessary to statistically test the hypotheses a different research design was used. An experimental vignette methodology study presented as an online survey was created. In order to test the hypotheses in a hedonistic situation, a restaurant setting was chosen to function as the hedonistic research setting. Aguinis and Bradley (2014) discuss that one of the criticisms of experimental vignette studies is the lack of realism. The use of a video in the vignette study increases the level of immersion of the participants and therefore the realism. To increase the level of realism even further, the data from the survey were then substantiated by the actual healthy food choices made by the participants in the cinema experiment.

The dependent variable of this study is the healthy food choice, defined as the participant’s intention to choose or purchase healthy food compared to unhealthy food. The independent variables are participant’s level of health consciousness, if the participants were placed in a hedonistic situation and if the participants were exposed to a subliminal prime.

Procedure

The original set up of the study consisted of an independent measures experiment. At a local cinema two movie theatres were made available to exclusively execute the experiment. One theatre was used for the experimental group and the other for the control group. Participants were randomly assigned to either group. There was a twenty minute interval between the different groups in order to control for simultaneous intermissions. Participants were asked to refrain from purchasing consumptions prior to the start of the experiment. At the intermission a five second clip announcing the intermission and an advertisement for a new movie was shown. Within the clip the words “choose healthy” were flashed for 40 milliseconds. Appendix D provides a screenshot of the prime.

(23)

After the clip the participants were asked to leave the theatre. They were given the opportunity to purchase consumptions. In addition to the regular offer of drinks and snacks three healthy items were added to the assortment: freshly squished orange juice, a fresh fruit salad and a granola bar. After participants, who had wanted to, had ordered their consumptions a survey was handed to them to fill out. All surveys were collected and participants were asked to return to the theatre to watch the remainder of the movie. After the movie the participants were debriefed and thanked for their participation.

The adapted research design entailed an experimental vignette methodology study presented as an online survey. The survey was made up of five sections. Prior to filling out the survey respondents were informed that they would participate in a survey studying the relationship between a healthy lifestyle and food choice. In Section I respondents were asked questions relating to their demographics. Section II consisted of a question related to the respondents’ food choice motives. Steptoe, Pollard and Wardle’s (1995) Food Choice Questionnaire was used as a measurement for the food choice motives. Section III was designed to measure the respondents’ lifestyle characteristics; Reeves and Rafferty’s (2005) Healthy Lifestyle Characteristics were used as a measurement. Section IV represents the hedonistic and experimental condition: respondents are randomly placed in either the hedonistic condition or the non-hedonistic condition. In the hedonistic condition the respondents were asked to imagine a situation in which they had accomplished something meaningful to them, for example a job promotion, graduating etc. Respondents were told that they would take their partner, family and friends to dinner to celebrate. Continuing they would be shown a video featuring the restaurant that they would go to. The respondents in the non-hedonistic condition were only informed that they would be asked to watch a video (the same video as in the hedonistic condition). Respondents were randomly selected to either the experimental group or the control group. The experimental group were shown the restaurant video with the subliminal prime “choose healthy”, the control group were shown the video without the subliminal prime. The prime was displayed for 20 milliseconds. All respondents were then presented with the

(24)

restaurant’s two menu options, one being a healthy menu, the other being an unhealthy menu. The respondents were asked to make a decision between the two menus. The final question of this section consisted of a subliminality check. Participants were asked, in an open-end question, if they had noticed anything unusual in the video. Section V consisted of the closing question, if respondents wanted to have a chance at winning a €20, - gift card of a product/store of their preference they were asked to fill in their e-mail address. Appendix E consist of screenshots of the online survey. In Appendix F a screenshot of the subliminal prime used in the online survey is included.

Participants

Due to the large population, obtaining a sampling frame was impossible. A non-probability sampling technique called volunteer sampling was used to reach and attract participants.

Participants for the cinema experiment were recruited via flyers that were put up throughout the cinema and via different social media channels. The participants were offered watching a movie in the cinema for free in return for their participation in the experiment. Participants were asked to send an e-mail to the researcher to sign up for the experiment, upon signing up the participants received an information e-mail explaining what they could expect from the experiment. The day before the experiment the participants were contacted via e-mail to inform them of the group in which they were selected, they were either placed in the group starting at 17.00 hours or in the group starting at 17.20 hours. The flyer, both information letters as well as the consent form are included in Appendix (G). Appendix H consists of the survey for the cinema experiment. The initial sample size of the experiment consisted of 48 participants. One participant’s survey was answered careless and therefore excluded from the analysis. The final sample was made up of N=47 participants. As this sample is too low to run statistical analysis (Field, 2013) an experimental vignette study executed as an online survey was conducted.

(25)

The participants for the online experimental vignette survey were attracted via e-mail and social media. In return for their participation respondents were offered the opportunity to fill in their e-mail address in the survey to have a chance at winning a €20, - gift card of their preference. In total 324 responses were recorded by ‘Qualtrics’, from these respondents 44 failed to complete the survey and were therefore excluded from analysis. From the remaining 279 respondents 40 answered the question whether they had seen anything unusual in the video. Of those 40 respondents, five respondents indicated that the video did not work, and three respondents reported actually seeing the prime. These respondents were excluded from analysis and therefore the final sample was made up of N=272 respondents. Of the final sample 134 respondents were randomly assigned by ‘Qualtrics’ to the experimental condition and 138 respondents were randomly assigned to the control condition. 146 respondents were allocated to the non-hedonistic condition, the remaining 126 were assigned to the hedonistic condition. The vast majority of the respondents were female (199) compared to male (72), and one respondent indicated to not identify with either gender. The majority of respondents are grouped in age group 18 – 24 (158 respondents).

Measurements

Health consciousness is a multifaceted; there are several definitions and operationalisations of this concept. Therefore there are multiple ways to measure health consciousness. Traditionally health consciousness is measured by behaviour, Hong (2009) re-conceptualized the measurement scale by measuring underlying psychological traits of health consciousness. As this scale measurement has not been validated since Hong’s primary version, the decision was made to use the more traditional measurement of actual behaviour to measure health consciousness as this has been a valid measurement scale for numerous years. The difficulty lies in measuring the actual day-to-day food consumption of respondents, therefore Steptoe et al. (1995)’s Food Choice Questionnaire (FCQ) is partially used as a measurement of the motives that underlie the selection of food on a typical day. Two factors: health and weight control were measured using 6 items which were scaled on a 5-point Liker scale (1 = strongly disagree to 5 = strongly agree). One of the variables measuring weight

(26)

control consisted of a counter-indicative item or otherwise known as a negatively-keyed item, this item was recoded. The health scale has high reliability with Cronbach’s Alpha = .851. The corrected item-total correlations indicate that all times have good correlation with the total score of the scale (all above .30) Also, none of the items would substantially affect reliability if they were deleted. The weight control scale has a low reliability with Cronbach’s Alpha = .591, however if one item is to be deleted the score goes up to α = .729, which slightly exceeds Field’s (2013) cut off point of .7. The remaining items have good correlation with the total score of the scale, as all are above .30. To complement the previous measurement, the Healthy Lifestyle Characteristics (HLC) by Reeves and Rafferty (2005) are used to identify other factors that were taken into account in determining the respondents’ health consciousness. In their research Reeves and Rafferty (2005) identified four HLC’s, namely non-smoking, healthy weight (measured as a person’s BMI), consuming five or more fruits and vegetables per day, and regular physical activities (≥ 30 minutes for ≥ 5 times per week). As these characteristics applied to American citizens, they were compared to Dutch standards and only the recommended daily quantity of fruit and vegetable intake is phrased differently in The Netherlands: 250 grams of vegetables and two pieces of fruit (Voedingscentrum, sd). The researcher included the respondents’ alcohol consumption as Raad voor Volksgezondheid en Zorg (2011) mentioned in their report about preventing nutritional diseases that excessive consumption of alcohol is one of the probable causes of several nutritional diseases such as various types of cancer. The decision was made not to include the respondents’ BMI into the analyses as the survey did not assess whether the respondents might have a disease that could affect their weight and/or if they are in the process of losing weight (with a survey being conducted in February that might be a real possibility). Therefore a new variable was created based on the respondents’ score on the above mentioned HLCs. Respondents were classified to be health conscious if they did not smoke, as well as if they answered ‘always’ or ‘most of the time’ in the question regarding fruit and vegetable consumption and weekly exercise and if their alcohol usage in the past 12 months was

(27)

Motives variables combined. 119 out of the sample of 272 respondents classified as health conscious, which comes down to 43.8%, 153 respondents were classified as not health conscious, which comes down to 56, 3% of the final sample.

Results

Assumptions categorical data analysis

The data used in the hypotheses testing analysis are categorical, therefore two assumptions have to be met before the analysis can take place: (1) independence and (2) expected frequencies (Field, 2013). The independence assumption is met in the present research as an independent measures design is used. The expected frequencies assumption is dependent on the number of categorical variables that are being tested. “With two categorical variables both with two categories (2 x 2 contingency tables) no expected values should be less than five” (Field, 2013, p. 735). In association testing between more than three categorical variables all expected counts should be greater than one and no more than 20 percent of expected counts should be less than five. Evidence of meeting the second assumption will be provided per analysis.

Hypothesis testing

The first hypothesis (H1) proposed that there is a positive relationship between the subliminal priming of words associated with health and making a healthy food choice. A Pearson Chi-square test was used to examine whether there is an association between the menu choice of respondents and the exposure to the subliminal prime. A Pearson Chi-square test has the assumption that no expected values should be less than five. In this analysis 0 cells (0, 0. percent) have an expected count less than 5. The minimum expected count is 47, 29. The analysis showed a significant association between exposure to a subliminal prime and the choice of a menu, with χ2_{(1) = 4,882,} p (.27) < .05. Based on the contingency table, displayed in Table 1, the odds ratio was calculated.

(28)

Count Percentage Experimental group Respondent choose the healthy menu 56 41,8%

Respondent choose the unhealthy menu 78 58,2%

Total 134 100%

Control group Respondent choose the healthy menu 40 29,0% Respondent choose the unhealthy menu 98 71,0%

Total 138 100%

Table 1 Contingency table condition*menu choice

The calculation of the odds ratio showed that for respondents who were subliminally primed the odds of choosing the healthy menu were 1.759 times higher than for respondents who were not subliminally primed.

The second hypotheses (H2 a and b) proposed that the positive relationship between subliminally priming words associated with health and making healthy food choices is moderated by the level of health consciousness of the respondents. At a glance, the data seem to support such an interaction. Calculations of the odds ratios of the healthy menu choice of primed respondents who are health consciousness (1.148) and for non-primed health consciousness respondents (.525) result in an odds ratio of 2.187. Health conscious respondents who are subliminally primed with words associated with health should be 2.187 times more likely to choose the healthy menu over the unhealthy menu compared to health conscious respondent who were not subliminally primed.

The moderating effect of health consciousness as well as the possible association of the hedonistic research setting are examined using a loglinear model analysis using backward elimination. The variables included in this analysis are MenuChoice, HealthConsciousness,

Condition and HedonisticCondition The cross tab analysis of the four categorical variables

indicated that all expected counts were greater than one and 0,0 percent of excepted counts were less than five. Table 2, included in Appendix (I) provides a graphical representation of the cross tab analysis.

(29)

The SPSS output of the analysis indicated that there were no missing values and the Goodness-of-Fit Test indicated that the significance level was ‘.’, which means that it cannot be measured. According to Fields (2013, p. 751) that should be the case as “at this stage the model predicts the data perfectly”. The K-Way and Higher-Order Effects table displays which components of the model can be removed without significantly changing the fit of the model. Table 3 displays the K-way and Higher-Order Effects table.

Likelihood Ratio Pearson

K Df Chi-Square Sig. Chi-Square Sig. Number of Iterations

K-way and Higher Order Effectsa 1 15 66,506 ,000 68,588 ,000 0 2 11 36,833 ,000 39,355 ,000 2 3 5 19,438 ,002 19,174 ,002 4 4 1 ,005 ,942 ,005 ,942 3 K-way Effectsb 1 4 29,673 ,000 29,233 ,000 0 2 6 17,394 ,008 20,181 ,003 0 3 4 19,433 ,001 19,169 ,001 0 4 1 ,005 ,942 ,005 ,942 0

Table 3 K-way and Higher-Order Effect log linear analysis a. Tests that k-way and higher order effects are zero. b. Tests that k-way effects are zero.

From the table it can be concluded that looking at the significance level of the Likelihood Ratio and the Pearson (p = ,942; p ≥ ,05) that the four way interaction

(Condition*MenuChoice*HealthConsciousness*HedonisticCondition) can be excluded from the

model without significantly changing the fit of the model. The combined effect of the variables and the interactions is not significant. Continuing on, the Backward Elimination Statistics, which are displayed in Table 4 show the step-wise backwards elimination of components of the model.

(30)

Stepa _Effects _Chi-Squarec _df _Sig. _{Number of}

Iterations 0 Generating Classb _{Condition*MenuChoice*Healthconsc}

iousness*HedonisticCondition

,000 0 .

Deleted Effect 1 Condition*MenuChoice*HealthCons ciousness*HedonisticCondition

,005 1 ,942 3

1 Generating Classb _{Condition*MenuChoice*HealthCons}

ciousness, Condition*MenuChoice*HedonisticC ondition, Condition*HealthConsciousness*He donisticCondition, MenuChoice*HealthConsciousness* HedonisticCondition ,005 1 ,942 Deleted Effect 1 Condition*MenuChoice*HealthCons ciousness ,091 1 ,763 3 2 Condition*MenuChoice*HedonisticC ondition 16,020 1 ,000 3 3 Condition*HealthConsciousness*He donisticCondition ,832 1 ,362 3 4 MenuChoice*HealthConsciousness* HedonisticCondition ,521 1 ,470 3

2 Generating Classb _{Condition*MenuChoice*HedonisticC}

ondition, Condition*HealthConsciousness*He donisticCondition, MenuChoice*HealthConsciousness* HedonisticCondition ,096 2 ,953

Deleted Effect 1 Condition*MenuChoice*HedonisticC ondition

(31)

2 Condition*HealthConciousness*Hed onisticCondition ,835 1 ,361 3 3 MenuChoice*HealthConsciousness* HedonisticCondition ,459 1 ,498 3

ondition,

Condition*HealthConsciousness*He donisticCondition,

MenuChoice*HealthConsciousness

,556 3 ,907

Deleted Effect 1 Condition*MenuCHoice*Hedonistic Condition 16,790 1 ,000 3 2 Condition*HealthConsciousness*He donisticCondition ,723 1 ,395 3 3 MenuChoice*HealthConsciousness 4,740 1 ,029 2

ondition, MenuChoice*HealthConsciousness, Condition*HealthConsciousness, HealthConsciousness*HedonisticCon dition 1,279 4 ,865 Deleted Effect 1 Condition*MenuChoice*HedonisticC ondition 18,160 1 ,000 3 2 MenuChoice*HealthConsciousness 6,138 1 ,013 2 3 Condition*HealthConsciousness ,291 1 ,589 3 4 HealthConsciousness*HedonisticCon dition 4,420 1 ,036 3

ondition,

MenuChoice*HealthConsciousness,

(32)

HealthConsciousness*HedonisticCon dition Deleted Effect 1 Condition*MenuChoice*HedonisticC ondition 18,144 1 ,000 3 2 MenuChoice*HealthConsciousness 5,883 1 ,015 2 3 HealthConsciousness*HedonisticCon dition 4,385 1 ,036 2 6 Deleted Effect 1 Condition*MenuChoice*HedonisticC ondition, MenuChoice*HealthConsciousness, HealthConsciousness*HedonisticCon dition 1,570 5 ,905

Table 4 Backwards Elimination Statistics log linear analysis

a. At each step, the effect with the largest significance level for the Likelihood Ratio Change is deleted, provided the significance level is larger than ,050

b. Statistics are displayed for the best model at each step after step 0

c. For ‘Deleted Effect’, this is the change in the Chi-Square after the effect is deleted from the model

In step 0 the previously mentioned four-way interaction was deleted and it did not significantly change the fit of the model. However, when looking at step 1, if the three-way interaction of

Condition*MenuChoice*HedonisticCondition were to be deleted it would significantly change the

model fit (p=.000) < .05. In Appendix (J) the Partial Associations, displayed in Table 5 and in Appendix (K) displayed in Table 6 Parameter Estimates are included, both tables support the statement that the combined effect of the variables and interactions of the three way interaction is significant. The final Goodness-of-Fit Tests, represented in Table 7 displays how well the final model fits. “A non-significant test statistic indicates that the expected values generated by the model are not significantly different from the observed data (put another way, the model is a good fit of the data)” (Field, 2013, p. 753). With a value of p = .905 for the Likelihood Ratio, the model is very non-significant, subsequently it can be concluded that the model is a good fit of the data.

(33)

Chi-Square df Sig. Likelihood Ratio 1,570 5 ,905

Pearson 1,552 5 ,907

Table 7 Goodness-of-Fit test final model log linear analysis

A follow up Pearson’s Chi-Square test was conducted. The SPSS data file was split based on the Condition variable and the variables MenuChoice and HedonisticCondition were used to run the analysis. Table 8 provides the output of the analysis for the respondents assigned to the experimental group, and Table 9 provides the output for the respondents assigned the control group. For the respondents assigned to the experimental group there was a significant association between the additional condition to which the respondents were assigned to (hedonistic vs. non-hedonistic) and the type of menu that the respondents choose χ2_{(1) = 13,619, p < .001. Based on} the odds ratio, the odds of respondents in the experimental group assigned to the hedonistic condition choosing the healthy menu were .26 times higher than respondents assigned to the non- hedonistic situation.

The control group also showed a significant association between the additional condition to which the respondents were assigned to (hedonistic vs. non-hedonistic) and the type of menu that the respondents choose χ2_{(1) = 5,36, p < .05. Based on the odds ratio, the odds of respondents in the} control group assigned to hedonistic condition choosing the healthy menu were 2.42 times higher than respondents assigned to the non- hedonistic situation. Therefore, the analysis seems to reveal an unexpected difference: respondents who are subliminally primed are less likely to choose the healthy food option in a hedonistic setting than respondents who are not primed.

(34)

Value df Asymptotic Significance (2-sided) Exact. Sig. (2-sided) Exact Sig. (1-sided) Point Probability Pearson Chi-Square 13,619b ₁ _,000 _,000 _,000 Continuity Correctionc 12,352 1 ,000 Likelihood Ratio 13,998 1 ,000 ,000 ,000

Fisher’s Exact Test ,000 ,000

Linear-by-Linear Association

13,517d ₁ _,000 _,000 _,000 _,000

N of Valid Cases 134

Table 8 Pearson Chi-square test output experimental group

a. Condition to which he respondent is assigned to by Qualtrics = Experimental group b. 0 cells (0,0%) have expected count less than 5. The minimum expected count is 25,49 c. Computed only for a 2x2 table

d. The standardized statistic is 3,677

Value df Asymptotic Significance (2-sided) Exact. Sig. (2-sided) Exact Sig. (1-sided) Point Probability Pearson Chi-Square 5,360b ₁ _,021 _,025 _,017 Continuity Correctionc 4,525 1 ,033 Likelihood Ratio 5,388 1 ,020 ,025 ,017

Fisher’s Exact Test ,025 ,017

Linear-by-Linear Association

5,322d ₁ _,021 _,025 _,017 _,011

N of Valid Cases 138

Table 9 Pearson Chi-square test output control group

a. Condition to which the respondent is assigned to by Qualtrics = Control group b. 0 cells (0,0%) have expected count less than 5. The minimum expected count is 18,84 c. Computed only for a 2x2 table

(35)

Substantiate the data

The data collected in the cinema experiment substantiated the findings from testing hypothesis H1. The participants who were assigned to the experimental group are 1.32 times more likely to purchase healthy/healthier consumptions than participants who were assigned to the control condition.

The data from the cinema expert also confirm the assumption of a higher odds ratio of participants purchasing healthy/healthier consumptions if they are health conscious compared to

participants who are not health conscious (odds ratio is 5.60).

The cinema data can not substantiate the data used for the final hypotheses. Based on the definition used in this study for a hedonistic situation the cinema is categorized as one. Consequently, there is not a non-hedonistic setting in the experiment

(36)

Conclusions, discussion, limitations and future research

The following section addresses the conclusions of the present study and discusses the implications of the conclusions. Afterwards, some limitations of the study and some suggestions for future research will be discussed.

Conclusions

The present study set out to examine if subliminal priming could help individuals make better food choices; especially when these individuals care about their health and are faced with a situation that tempts them to make a hedonistic choice. After the initial research method of an experiment proved to have too few participants, an experimental vignette methodology study presented as an online survey was set up. The final sample of N=272 respondents were attracted using a convenience sampling method.

A Pearson Chi-square test was able to statistically confirm that there is a significant association between subliminally priming individuals with words associated with health and the food choice that they make after priming, χ2_{(1) = 4,882, p (.27) < .05. The calculation of the odds} ratio showed that for respondents who were subliminally primed the odds of choosing the healthy menu were 1.759 times higher than for respondents who were not subliminally primed. Based on these findings H1 Individuals who are subliminally primed with words associated with health will

have a higher tendency to make a healthy food choice compared to individuals that are not primed

could be accepted. The data retrieved from the cinema expert endorse these findings. At a glance the collected data did seem to support the hypotheses H2ab that the positive association found in H1 is moderated by the level of health consciousness of the respondents. However, when a loglinear analysis including the variables MenuChoice, HealthConsciousness,

Condition and HedonisticCondition was performed, the analysis resulted in concluding that the

four-way interaction could be deleted without significantly changing the fit of the model, (p = ,942;

(37)

if it were to be deleted, (p=.000) < .05. Based on these finding the hypotheses H2a: The effect of

subliminal priming towards healthy food choices is more pronounced for individuals who are health conscious when exposed to healthy choice related primes and H2b: The effect of subliminal priming towards healthy food choices is less pronounced for individuals who are not health conscious when exposed to healthy food choice related primes could not be accepted.

Consequently, the hypotheses H3a: The effect of subliminal priming toward healthy food

choices is more pronounced for individuals who are health conscious when exposed to healthy choice related primes in a hedonistic setting and H3b: The effect of subliminal priming toward healthy food choices is less pronounced for individuals who are less health conscious when exposed to healthy choice related primes in a hedonistic setting could not be accepted as they build on the

assumption that the level of health consciousness has a moderating influence on the positive association between subliminally priming words associated with health and the food choices individuals make after being primed. Nonetheless, the results of the follow up Pearson Chi-square tests of the loglinear analysis did turn out to be quite remarkable. The SPSS data was split based on the Condition variable and Pearson Chi-square tests were performed for the variables MenuChoice and HedonisticCondition. Both tests found a significant association between the menu which the respondents choose and if they were assigned to a hedonistic setting or not; for the experimental group: χ2_{(1) = 13,619, p < .001 and for the control group: χ}2_{(1) = 5,36, p < .05. The odds ratios} indicated that the odds of choosing the healthy menu was .26 times higher for respondents in the experimental group assigned to the hedonistic condition when compared to the respondents in the non-hedonistic test condition, but 2.42 times higher for respondents in the control group assigned to the hedonistic condition when compared to the respondents in the non-hedonistic test condition. Therefore, the analysis seems to reveal an unexpected difference: respondents who are subliminally primed are less likely to choose the healthy food option in a hedonistic setting than respondents

who are not primed.

(38)

affect individuals’ choice behaviour. The study was not able to find a significant moderating effect of health consciousness. Surprisingly the study did find that respondents who were assigned to a hedonistic condition and were being subliminally primed are less likely to choose the healthy food option when compared to respondents in the same condition but who are not subliminally primed.

Discussion

Although the study did not significantly identify a moderating effect of health consciousness, it should not be discarded as a factor that does not have an influence. Viewing the data at glance gave the impression that there might be something going on. As mentioned before, there are several definitions of health consciousness and multiple ways of measuring it. In the present study health consciousness is measured by the traditional measurement scale of actual behaviour. A clear notion of what types of behaviour should be included in the measurement scale does not exist. The fact that in the present study weight/BMI are not included in the measurement scale and that alcohol consumption is included might have affected the outcome. Perhaps a different approach such as Hong (2009) mentioned in his research, approaching health consciousness as a personal attribute might provide different outcomes. In that case the underlying psychological attributes are measured, which might increase the predictive power of health consciousness on certain health behaviours.

The remarkable finding that subliminally primed respondents placed in a hedonistic setting are less likely to choose the healthy menu when compared to respondents in a hedonistic setting that are not primed might be traced back to the justification literature. Mukhopadhyay and Johar (2009) found that individuals who are reminded of previous restraint are more likely to indulgence in a subsequent indulgence situation. The remarkable finding might be a result of just that; an unintended effect of the subliminal prime functioning as a reminder of the restraint individuals who are health consciousness have when making food choices. Because they are reminded of their restrain, they are more likely to justify their decision to indulge. This unintended effect might be caused by the order of the online survey: by asking the question about food decisions on a regular day before the menu choice the respondents might have been reminded of their healthy choices

Put a stop to license to sin : using subliminal priming to help individuals make healthier decisions