Healthy Food at Football Clubs

(1)

Healthy Food at Football Clubs

The Influence of Promotion Posters on Healthy Food Choice

Among Children

Floor Zanderink (s1960075) University of Groningen

(2)

HEALTHY FOOD AT FOOTBALL CLUBS

The Influence of Promotion Posters on Healthy Food

Choice among Children

Master thesis, MSc Marketing Management University of Groningen

Faculty of Economics and Business

[28 – 06 – 2017] FLOOR ZANDERINK Stationsstraat 1 9711 AR Groningen Tel.: +31 (0)612241484 Email: f.zanderink@student.rug.nl Student number: S1960075

Supervisor: Prof. dr. ir. K. van Ittersum Second supervisor: M.T. van der Heide

(3)

MANAGEMENT SUMMARY

Obesity is a growing problem worldwide, and children in the Netherlands are no exception to this problem. The food industry is pointed out as one of the main contributors to the obesity problem, as marketing to children is primarily focused on promoting unhealthy products, such as savory snacks and soft drinks. Techniques such as using fun elements and celebrity endorsers on packaging and advertisements are a popular manner of attracting children to buy specific products. The aim of this study is to examine if these techniques are also efficient in promoting healthier food options instead of unhealthy products.

Sports canteens in the Netherlands are still offering a wide arrange of unhealthy products. Nevertheless, they might be the perfect venue to improve healthy consumption behavior of children and adolescents in more ways than only physical activity. Therefore, this study will conduct a field experiment at a football canteen in the province of Groningen, The Netherlands. Four different experimental conditions are tested. First of all, children can choose between two food options, an unhealthy one and a healthier one, which are both shown on two advertising posters. Thereafter, certain fun elements, a sports celebrity endorser (Dutch football player Arjen Robben), and both are subsequently added on the poster for the healthier food option, and it is investigated whether these fun elements increase the preferences for the healthier food option. In the different experimental conditions, the advertising poster for the unhealthy food option is kept constant. A binary logistic regression is performed to see whether the different experimental conditions significantly affect the option children chose. Furthermore, a survey is conducted among 184 children, aged 7 to 14, to measure the influence of age and winning a sports game on the choice children make.

With respect to previous research, it was expected that using fun elements, a sports celebrity endorser and/or both will increase the preferences for the healthier food option. However, results show that solely using a sports celebrity on a promotion poster for a healthy food option significantly increases the preferences for the healthier option. Furthermore, results show that age negatively influences the relationship between fun elements and the healthy choice children make. In other words, using fun elements will become less effective when children become older.

To conclude, using a sports celebrity endorser on a promotional poster for a healthier food option is a potentially successful, and easy, way for sport canteens to increase healthy consumption behavior of children.

Keywords: obesity, marketing to children, sports canteens, fun elements in marketing, sports

(4)

PREFACE

This paper has been written in order to complete the master Marketing Management at the University of Groningen. After finishing my thesis for the master International Financial Management in June 2016, I was interested in writing a thesis with a more practical background, rather than the more theoretical and mathematical thesis I have written before. As I always played football when I was younger and as I am still very interested in the sport, the opportunity to conduct an experiment in a football canteen caught my attention. During the process, from February to June, there were ups and downs in writing my master thesis, however I am looking back with a satisfied feeling and I am very glad that I have been given the opportunity to see what it is like to conduct a field experiment.

(5)

1. INTRODUCTION

1.1. Background

Childhood obesity is a growing problem worldwide, and, according to the World Health Organization (2016), “one of the most serious public health challenges of the 21st_{century”. The health of an entire} generation is at risk (Vilsack 2011) and there are indeed alarming facts and numbers to support these allegations. In the United States, the percentage of children who are obese has tripled since the 1970s (Ogden, Carroll, and Lawman 2016), recent research in Australia indicates that a quarter of school-aged children are overweight or obese (Kelly et al. 2008a), and research by Statistics Netherlands (CBS) and the Dutch National Institute for Health and Environment (RIVM) shows that 13.6% of children between the age of 4 and 17 were overweight in 2016.

These are alarming facts, as childhood obesity results in long-term consequences on physical, social, and emotional health. Heart diseases, strokes, circulatory problems, diabetes, cancer, and osteoporosis are only some of the many diseases which obese children might face in their adult life (McGinnis, Gootman, and Kraak 2005). Choices children and teenagers make today, determine the diets of tomorrow’s adults and as these dietary habits are extremely resistant to change once they are adopted, it makes it even more important to target these habits in the years that they are developed (Lien, Lytle, and Klepp 2001; Shepherd and Dennison 1996).

The food industry is identified as being one of the main contributors to the childhood obesity problem (Elliott 2007). Together with parents, dietitians and health care organizations, they significantly influence the eating behavior of children and youth (McGinnis, Gootman, and Kraak 2005). These groups all face difficulties in creating a healthy environment for the future generation. Children nowadays have more money to spend and they have become increasingly involved in purchasing decisions made by their parents. Previous research shows that the promotion of food is greatly observed by children and also influences their (parents’) shopping behavior (Carruth et al. 2000; French et al. 2001; Goldberg 1990).

1.2. Marketing to Children

(8)

great opportunity for firms and marketers to promote the food which is recommended by health advisors and dietitians to encourage children to eat more healthily, one might think. However, actual practice shows the opposite. Children’s food promotion is dominated by ‘the big four’ of unhealthy products: pre-sugared breakfast cereals, soft-drinks, confectionery, savory snacks, and additionally fast food (Alexander et al. 1998; Hastings et al. 2003). In the Netherlands, a total of 94% of the commercials on television aimed at children in 2011 were promoting products which do not fit in a healthy lifestyle, which leaves only 6% to products which do fit in promoting a healthy lifestyle (Consumentenbond 2011).

(9)

increase since 2010. Sport clubs are getting bigger, and they can help to increase the health of children and adolescents in more ways than only physical activity (Geidne, Quennerstedt, and Eriksson 2013). Research by the GfK Netherlands (2015) shows that 96% of children who visit their sports canteen, also buys something to eat or drink, however only a small percentage actually buys something healthy. On the other side, repeated research by the Mulier Institute (2014) and the GfK also show that the demand for more healthy products in sport canteens is increasing. This is contradictory, and reasons for this might include the current supply of healthy products, habits, or social pressure (GfK 2015), e.g. ‘if my friend buys fries, I can buy fries as well’.

This paper will try to find underlying reasons on when and why children will choose the healthy option, instead of the unhealthy option in sport canteens at the point of purchase. By means of a field study and survey, children are exposed to different posters promoting healthy products. By using a brightly colored poster, decorated with balloons and garlands, healthy products will be promoted as fun and will try to persuade children in buying these products. Furthermore, an additional condition will be added, which includes a celebrity endorser promoting the healthy product. The unhealthy product will not be promoted by means of celebrities or fun aspects. The goal is to attract children to buy healthy products, whereas they do not directly realize that they are actually consuming more healthy products.

1.3. Research Questions

The overarching question based on the above considerations is: ‘how can sport canteens stimulate healthy food choice among children?’ Several research questions can be made: ‘does including fun elements in promoting healthy products lead to increases in preferences for these healthy products, compared to not promoting these healthy products?’; ‘Does including a sports celebrity endorser in promoting healthy products lead to increases in preferences for these healthy products, compared to not promoting these healthy products?’; ‘Does including both fun aspects and a sports celebrity endorser in promoting healthy products lead to increases in preferences for these healthy products, compared to only using one of these two techniques?’ Next to these main effects, the moderating role of age is considered in influencing the relationship between using fun aspects and healthy food choice, which leads to the following research question: ‘Does age moderate the effect of the relationship between using fun aspects in promoting healthy products and preferences for these healthy products?’. Additionally, it is tested whether winning a football game influences this relationship: ‘does winning a football game influence the preference of (fun) healthy products?’.

1.4. Contribution

(10)

healthy foods in sports canteens. Previous research shows that especially unhealthy products are marketed in a fun way to children (Elliot 2008; Hastings et al. 2003), but few studies consider that this can also influence children’s healthy eating behavior. This study may be a starting point by attracting children to make more healthier choices. Existing research on this topic mainly focuses on different settings, such as schools or changing the packaging of food. This research will be conducted in a sports canteen as these are known for offering unhealthy products, whereas sports clubs themselves are promoters of a healthy lifestyle by encouraging children to involve in sports activities.

1.5. Structure

The remainder of this article is structured as follows: first of all, prior research is consulted and underlying theories on children’s food preferences, advertising to children, and the influence of emotions on product choice will be discussed. Then, the conceptual framework and hypotheses are presented, and the methodology is provided. Thereafter, results of the field study will be shown. This paper concludes by summarizing and discussing the results, evaluating limitations of the study, and outlining implications and recommendations for further research.

2. THEORETICAL BACKGROUND

This section will introduce the theoretical background relevant for this research. First of all, children’s preferences for unhealthy and healthy food will be discussed. Secondly, marketing techniques which are mainly used in children’s food marketing are considered. Thereafter, the effects of winning a football game on emotions and choices are explained.

2.1. Unhealthy versus Healthy Choices

(11)

without parental permission, they tend to choose candy, snacks, soft drinks, or fast food. Also, consumption of products with added sugars has increased consistently among children in the past two decades (Guthrie and Morton 2000). Which factors are responsible for children’s preference for sweet and unhealthy food? Ventura and Mennella (2011) give two explanations for these preferences. First of all, it might be due to biological preferences (congenital, evolutionary preferences), which humans have developed. Secondly, learned preferences, as a consequence of repeated exposure through advertising, might also increase children’s preference for unhealthy food. The following subsections will elaborate on these factors.

2.1.1. Biological Preferences.

The mechanisms which are responsible for recognizing and responding to different tastes are already well-shaped during pregnancy and continue to develop and mature after birth (Ganchrow and Mennella 2003). The complex processes which are involved in these mechanisms are beyond the scope of this paper, however some examples will be explained here. Research on infants’ responses to sweet-flavored beverages or food, show that these responses are unlearned and can be explained as some kind of a reflex (Ventura and Mennella 2011). Crook (1978) investigated newly-born infants’ taste preference for sucrose solutions and water and found that, as expected, newborn infants can distinguish between these substances and prefer the sucrose solution above plain water. Also, Steiner et al. (2001) study facial expressions in response to the intake of a sweet solution. They find that infants’ faces tend to relax when given a sweet solution, which then is followed by a smile, indicating happiness. Among others, these studies show that preferences for sugary and unhealthy food and beverages is partly congenital and this preference will evolve during childhood, whereas during adulthood this might decrease while other taste preferences, such as bitterness, are developing.

2.1.2. Learned Preferences.

(12)

clown mascot of fast-food restaurant McDonald’s, probably knows what is good and healthy for children (Choice 1990). Again, this shows the vulnerability of children to food promotion.

Hitchings and Moynihan (1998) found proof that food advertising on television affects the food choice children make. In a research involving interviews with children from nine- to eleven-years-old, a positive and significant relationship was found between recalled food advertisements and actual food consumed, which was strongest for soft drinks and crisps and savory snacks. Additionally, more recent research by Halford et al. (2008) also shows that 11- to 13-year-old children choose more food items after exposure to advertisements of these items. Especially obese and overweight children, in comparison with normal weight children, showed a significant correlation between the recall of food advertisements and the food chosen.

All in all, children tend to make an unhealthy product choice when they buy food without parental permission due to both biological and learned preferences. These learned preferences are the result of the domination of unhealthy food advertisements and research that shows that these advertisements influence product choice.

2.2. Marketing to Children

Marketing to children is often done by trying to attract attention by, among other things, using cartoon-characters, bright colors, spokes-characters (Kelly et al. 2008b), and celebrities (Martin and Bush 2000). These marketing techniques are associated with positive attitudes towards the advertised product and can persuade children to buy certain products (Brand 2007). According to Hoyer et al. (2012), children are cognitively limited and therefore easier to persuade than adolescents or adults. This paper will mainly focus on the use of fun elements in marketing to children, and the following subsections will explain how this is done, what it entails, and how fun can influence the food choice that children make. Additionally, the effect of using celebrities will be discussed.

2.2.1. Determinants of fun food.

(13)

products had to satisfy at least two out of the five following criteria: “direct claims or allusions to ‘fun’/play on the package; cartoon iconography pointedly directed to children; tie-ins with children’s television programs, merchandise or films; the foregrounding of strange shapes, unusual colors or unconventional tastes; puzzle or games targeted at children” (Elliot 2007). Findings show that around 84% of the products which sufficed the criteria selected by Elliot (2007) made use of a font style which was cartoonish and resembled a child’s handwriting, and over 75% included a cartoon character (e.g. an animal, a cartoonish boy or girl). Furthermore, slogans such as “Fun to eat!” or “Fun fried potatoes”, are popular techniques by marketers to attract children to notice their products and eventually buy them.

Additionally to font styles, characters, and slogans, the use of color seems to be an important element in food marketing to children. Elliott (2007) found that four colors are especially dominant in the children’s food market, namely the primary colors (red, blue, and yellow) and subsequently green. In her research from 2007 and 2008, she found that blue was the dominant color in 27.5% of the 367 products, yellow was the dominant color in 18.5% of the products, red was dominant in 14.4% and green in 10.4% of the products. Furthermore, if these colors were not the dominant color, they were often used as the secondary color. Additionally to Elliot (2007), research by Walsh et al. (2012) shows that red, green, orange, and yellow are popular colors in food marketing to children. All in all, the most important determinants of classifying products or marketing actions as fun is the use of funny fonts, cartoon characters, fun slogans, and bright colors. Next, the influence of these marketing methods on children’s food choice will be discussed.

2.2.2. The Effect on Purchase Behavior.

As shown in the previous subsection, the use of fun in marketing to children is a popular method used by companies. Two logical follow-up questions are then: how does this influence children and does it actually influence their purchase behavior? This subsection will look into previous research and elaborate on these questions.

(14)

Livingstone and Helsper (2006) offer a review of empirical findings in the domain of advertising and the food choice made by children. They argue that the different routes to persuasion and the processes which belong to these routes are at work at different ages. Additionally, previous research suggest that younger children (until the age of 11) are more persuaded by peripheral cues such as colors, entertaining images, or celebrity endorsers (Carruth et al. 2000; Valkenburg and Cantor 2001). Therefore, Livingstone and Helsper (2006) conclude that peripheral route processing is more typical for younger children, and as they become teenagers they tend to rely more on central route processing. Gunter and McAleer (1997) explain this by arguing that older children become more skeptical to advertising, but are also able to better express themselves and form a more stable judgment about the advertisement while relying on the arguments given. Nevertheless, it is still argued by several scholars that even children from the age of 8 until 16 do not possess the ample cognitive mechanisms to detect persuasion techniques (Dias and Agante 2011).

In the field of color perception, previous research shows that colors provoke emotions, affect consumer behavior and can act as a cue to differentiate from the competition (Aslam 2006). Also, older research by e.g. Alpert and Alpert (1991) and Kotler (1973) found that, among other things, the use of colors can positively influence purchase intentions. As mentioned previously, especially the colors red, blue, green, orange and yellow are popular among marketers. Reasons for preferences for these colors could be that they are all bright colors and these colors seem to arouse positive reactions and happy feelings from children (Boyatzis and Varghese 1993; Cimbalo et al. 1978), which is an important factor in food marketing to children. Also, research by Walsh et al. (2012) showed that, for children from both 5-years-old and 9-years-old, a significant effect exists between the preference for food choice and the use of the colors red, green, orange, and yellow.

(15)

use of cartoon characters is indeed effective and can help the brand in creating awareness, recognition and ultimately increase purchase intentions, why is it not used to promote healthy products?

Previous research in this area is limited. Pires and Agante (2011) studied the influence of food packaging on children’s purchase decisions in making healthier choices. They used the fun packaging which is generally used on the packaging of unhealthy products and used it for the packaging of healthier products. 408 Portuguese children between the age of 10 and 14 years took part in the study, and a fun package design for apple slices was created, including bright colors and a ‘cartoon apple’. Results of this study showed that both willingness to buy and willingness to consume significantly increased in comparison to normal packaging. Pires and Agante (2011) suggest that these results are important given that ‘children are driven by fun’ and that their intuitive to show resistance against healthy products might be reduced by increasing the fun element in promoting healthy products.

All in all, following these results and the possibility to use fun elements and approaches to increase healthy behavior of children, the following hypothesis is developed:

H1: A promotion poster for a healthy food option including fun elements will lead to higher preferences for the healthy food option in comparison to a poster without fun elements.

2.2.3. The Moderating Effect of Age

As already explained before, Livingstone and Helsper (2006) argue that children of different ages process information via different channels. Younger children tend to rely more on simple cues and heuristics (e.g. colors/characters), whereas teenagers become more skeptical and rely more on arguments and information given. As fun marketing is explained as using funny fonts, cartoon characters, fun slogans, and bright colors, it can be expected that younger children are more influenced by fun marketing, in comparison with older children. It is not expected that these older children are not at all influenced by this fun marketing, however they might be more skeptical and persuaded less easily by these fun aspects.

H2: The age of children weakens the positive relationship between fun marketing and

healthy food choice. Specifically: if age is increasing, fun marketing will become less effective.

2.2.4. Celebrity Endorsement

(16)

voice of cigarette brand Chesterfield and promoted this brand with the words: “My cigarette is the MILD cigarette… that’s why Chesterfield is my favorite.” Celebrity endorsement thus has a long history and have been very successful in achieving various marketing and communication objectives (Pradhan, Duraipandian and Sethi 2014). It can lead to increases in e.g. brand awareness, purchase intention, profit and sales (Elberse and Verleun 2012; Erdogan 1999; Hung, Chan and Tse 2011; Kaikati 1987).

As this study is conducted at a sports canteen, the main focus here will be on sports celebrities, as it is assumed that children who perform a certain sport, are also interested in this sport. For children and adolescents, sports celebrities can be very important in developing their own identity. Martin and Bush (2000) suggest that when children and adolescents develop their own set of self-image and consumption patterns, they are affected by a wide range of outside interests including the role sports celebrities play. They see sports celebrities as their ‘ideal self’ and want to develop certain personality traits which are similar to those of their role model (Boon and Lomore 2001).

Chan, Ng and Luk (2013) performed a research among 76 young Chinese consumers and found that the majority of the interviewees agreed that using a celebrity endorser increases brand awareness, purchase confidence and that it encourages trial. For example, one of the interviewees in the study said that he only started noticing a specific sort of candy, because a local male celebrity endorsed the brand in a new commercial. Beforehand, he had never paid attention to the brand. Bush, Martin and Bush (2004) examined the influence sports role models have on purchase intentions and behavior of 218 adolescents. They found that these sports celebrities are quite important to adolescents and influence favorable word-of-mouth and also enhance brand loyalty. Furthermore, Dix, Phau and Pougnet (2010) researched whether athlete role models influence the purchase behavior and behavioral intentions of students in Australia. Findings show that they indeed have a positive impact on various consumer behavior attributes, including switching behavior, favorable word-of-mouth and purchase intentions. Same results were found for, among others, young Brazilian consumers (Grohmann, Battistella and Radons 2013) and teenagers in Botswana (Makgosa 2010). Based on these findings, it is expected that a sports celebrity endorser has a positive effect on the purchase intentions of young consumers. Therefore, the following hypothesis is derived:

H3: A promotion poster for a healthier food option including a sports celebrity endorser will lead to higher preferences for the healthier option in comparison to a poster without the sports celebrity endorser.

(17)

the healthy option, it can be expected that using both these techniques increases preferences even more. Therefore, the following hypothesis is derived:

H4: A promotion poster for a healthy product including both fun elements and a sports celebrity endorser will lead to higher preferences for the healthy option in comparison to a poster without fun elements and/or a sports celebrity endorser.

2.3. Emotions After Winning a Football Game

Experiencing a positive or negative event results in different emotional responses. Baker-Ward, Eaton and Banks (2009) examined the emotional responses of forty-eight 10-year-old children after playing a final game in a football tournament. They found that children who experienced a negative outcome (i.e. lost the final game) reported very different emotional states than those children who experienced a positive outcome (i.e. won the final game). As expected, children who won the game experienced more positive emotions. This section will explain why and how different outcomes of a football game can have different influences on the purchase behavior of healthy products. Furthermore, the effect of winning a football game on the relationship between promotion including fun elements and purchase behavior of healthy products is examined.

2.3.1. Emotions and Purchase Behavior

Experiencing certain emotions can have an effect on eating behavior (Macht 2008), and thus purchase behavior. Emotions such as anger, sadness, happiness and guilt can change, among other things, food choice (Gibson 2006) and the motivation to eat (Macht and Simons 2000). Whereas for one group of people, emotions may actually increase food intake or increase the intake of unhealthy food, for a different group of people, it might actually decrease (unhealthy) food intake. Previous research shows that for restrained eaters, as measured by a questionnaire, food intake increases more than for unrestrained eaters when experiencing negative emotions (Heatherton, Striepe and Wittenberg 1998; Macht 2008; Rotenberg and Flood 1999). The same holds for the consumption of high caloric foods: restrained eaters are more likely to overeat than unrestrained eaters. As it shows that experiencing emotions can go either way (increasing and decreasing unhealthy food intake), and it is assumed that children experience more positive emotions after winning a football game than children who do not win a football game, the following hypotheses are designed:

H5a: When children win a football game, and subsequently can choose between an unhealthy and a healthy food option, their preferences for the unhealthy option will increase in comparison to children who do not win a football game.

(18)

unhealthy and a healthy food option, their preferences for the healthy option will increase in comparison with children who do not win a football game.

2.3.2. Emotions and Persuasion Techniques

Bless et al. (1990) researched the effect of persuasion when being in a happy or sad mood. They predict that individuals who are in a good mood, are more easily persuaded by an advertising message than individuals who are in a bad mood. This is explained by three reasons: first of all, when evaluating the advertising message, the criteria which are used to evaluate might be affected by the emotional state of the individual. Bless et al. (1990) expect that individuals who are in a bad mood, may use harsher criteria when evaluating the message. Secondly, it is possible that individuals use their emotional status as a peripheral cue. If this is true, attitudes toward the advertised message could be more favorable when the individual is in a ‘favorable’ (good) mood. Finally, Bower (1981) and Bless et al. (1990) also refer to mood-congruent memory, and explain that consumers in a happy mood may provoke more positive feelings with the advertised message and are therefore more easily persuaded. Additionally, Schwarz et al. (1991) argue that consumers in a positive mood see their current environment as a secure place, and therefore they might take more risks and as a consequence use simple cues to process information.

Rosenbaum (1980) found that for individuals who are opposing a negative or bad mood, are more concentrated on other activities which can distract them from thinking about negative thoughts. Therefore, they might more elaborate on the content of a message and consider all options presented. However, Bless et al. (1990) also point at the fact that bad moods can go hand in hand with a decrease in an individuals’ motivation, and consequently, he or she might rely more on simple heuristics.

Previous research does not seems to focus on children when studying the effects of mood and persuasion of an advertising message, but mainly focus on adults. Therefore, this paper will try to find if these effects also apply to children. As already explained above, children who win a football game are expected to experience more positive emotions. Also, fun elements which will be included in the advertising of healthy products during this study, are expected to act as heuristic cues. Together with the expectation that individuals who are in a good mood are more easily persuaded by an advertising message and rely on simple cues, the following hypothesis is developed:

(19)

2.4. Conceptual Model

The aforementioned hypotheses combined lead to the proposed conceptual model, which is shown in figure 1.

Figure 1. Conceptual Model

3. METHODOLOGY

A field study is conducted to test the hypotheses which are introduced in the previous part and to collect relevant data. In this section, the research method, data collection and plan of analysis are explained.

3.1. Research Method

3.1.1. Field Experiment.

The aim of this study is to see whether including fun elements and/or a sports celebrity endorser on a promotion poster of a healthier food option increases the food choice of this healthier option at the point of purchase. To examine this, a field study is conducted at the canteen of football club S.V. Lycurgus in the city of Groningen. For children below the age of 15, S.V. Lycurgus has 34 teams and is one of the biggest amateur football clubs in Groningen. The field study is conducted during two Saturday mornings (the 6th_{of May and the 13}th_{of May), between 09.00 and 13.00, when most youth} teams have to play a match.

(20)

extraneous variables. However, field experiments are necessary to see whether certain theories and effects will work in the real world (Johnson and Christensen 2012) . Because of its natural setting, there is better external validity and also less probability of demand characteristics which might influence the results (Roe and Just 2009). Field experiments are a correct way to see the relative influence of different theories to provide a better and more complete rationale for certain behaviors (Sudhir, Roy and Cherian 2016).

3.2.2. Study Design.

A table is set up in the football canteen including two posters, where four conditions will be tested. These conditions will differ in the way the promotion posters of the healthier food option will look like. The poster with the unhealthy food option will remain the same in each condition. The unhealthy food option includes Coca-Cola and Croky Natural Crisps. The healthier food option includes Jimmy’s Salt Popcorn and Dubbel Frisss Light Apple and Peach. Table 1 and table 2 of Appendix A shows a comparison of the nutrition facts of these products to indicate the overall healthiness of the products. The following four conditions will be tested:

1. The control condition. There will be one poster that displays the unhealthy food option, and one poster that displays the healthier food option. None of the posters will depict fun elements or a celebrity.

2. The fun condition. There will be one poster that displays the unhealthy food option, which will not include fun elements and there will be one poster that displays the healthier food option, including fun elements. None of the posters will depict a celebrity.

3. The celebrity condition. There will be one poster that displays the unhealthy food option, which will not include a celebrity and there will be one poster that displays the healthier food option, including a celebrity. None of the posters will depict fun elements.

4. The fun-celebrity condition. There will be one poster that displays the unhealthy food option, which will not include fun elements or a celebrity and there will be one poster that displays the healthier food option, including both fun elements and a celebrity.

(21)

Table 1. Experimental conditions, time frame, and number of respondents

Day Time Frame Condition Number of Respondents 6th_{of May, 2017} _{09.00 – 10.00} _Control ₃₈ 10.00 – 11.00 Celebrity 35 11.00 – 12.00 Fun 42 12.00 – 12.30 Fun-Celebrity 15 13th_{of May, 2017} _{09.30 – 11.00} _{Fun-Celebrity} ₃₂ 11.00 – 11.30 Control 11 11.30 – 12.00 Celebrity 11 3.2.3. Survey Design

Survey researchers and academic research institutes are realizing more and more that there is a high need to obtain survey data directly from children, rather than depending on proxy data, such as data collected from parents (Bell 2007). However, there has been a lot of debate about the minimum age at which children are able to independently fill in a questionnaire, understand the questions asked and also provide answers that are meaningful. Piaget (1929) developed a theory of cognitive growth, which is still used extensively in research involving children. He recognized four cognitive stages or periods in the development of children. This paper will not go into much detail about the scientific explanations of these stages, however children between age 7 to 11 enter into the third phase. During this stage, thoughts become more coherent (Piaget and Inhelder 1972). According to Borgers, de Leeuw, and Hox (2000) children in this phase are in a stage of ‘development of concrete operations’ and can be successfully surveyed. They also recognize that children from the age of 4 and onwards can, basically, also be interviewed. However, this is not easy: children from 4 until 7 are in a stage of intuitive thought and are still developing language use, whereas this limits comprehension of questions and can bias answers. Therefore, the survey will be conducted among children from 7 to 14-years-old who visit the sports canteen at the two Saturdays.

(22)

questions. Borgers, de Leeuw and Hox (2000) found that responses will be more reliable, if there are more words used in the introductory text. In addition, Pires and Agante (2011) argue that, to keep children motivated, the questions should be understandable by making them ‘clear and simple’. Therefore, instead of using a normal 5-point Likert scale, the ‘smileyometer’ is used (Read, MacFarlane and Casey 2001).The smileyometer uses smileys instead of text, where a very sad smiley represents the most negative outcome and a very happy smiley represents the most positive outcome. Zaman et al. (2013) explain that children will be more motived to answer questions by seeing the smiley faces, rather than seeing text. Furthermore, to motivate the children even more, it is said in the introduction that the survey will not take too long and that there are no right or wrong answers. Also, it is stressed that they can always ask a question if they do not understand a question. To make sure the children have noticed the posters, children are made aware of the posters by handing them the survey and additionally, the first question refers to the posters. Malhotra (2010) claims that it might be good to ask respondents for their opinions in the first questions. Therefore, the first question refers to the choice of food the respondent likes most. When children finish filling in the questionnaire, it is asked if they succeeded and it is also checked whether they filled the survey in completely. Thereafter the children receive the option of their choice for free and they also get a stamp on their hand, to make sure they will not come back for a second time. Notes are made where necessary.

The final survey can be found in Appendix C, which includes both the original Dutch version (figure 1) and the translated English version (figure 2) of the survey.

3.2. Variables

3.2.1. Independent variables

The independent variables in this study are the four different treatment conditions, which are the manipulated posters. In all four conditions the poster depicting the unhealthy food option will not be changed and will not include fun elements and/or a celebrity endorser. The poster depicting the healthier food option will be manipulated, which will result in the conditions explained before.

3.2.2. Dependent variable

(23)

Furthermore, the survey includes two questions which present six options of respectively drinks and food. Children are asked to tick the ones they do not like. These options include the products used in the experiment. This way, it is known that children who do not like the drinks or food at all, have not based their choice on the posters, but simply because of distaste.

3.2.3. Moderator Variables

The moderator age is measured in the form of an open question, by simple asking: ‘How old are you?’. To measure the influence of winning a football game on the choice children subsequently make, it is first asked whether the respondent has played a football match today. If yes, a follow-up question is asked whether the result of the match was a win, a loss, or a draw.

3.2.4. Control Variables

Gender, degree of hunger, importance of eating healthy, and likeness of Arjen Robben are used as control variables in this study. To account for differences in healthy choice between boys and girls, gender is added as control variable. It is measured by simply asking whether the respondent is a boy or a girl. Degree of hunger and importance of eating healthy are both measured on a 5-point Likert smiley scale ranging from definitely not to definitely yes. Siep et al. (2009) claim that when a person is hungry, the perceived attractiveness of unhealthy food increases. Additionally, when an individual finds it very important to care about their healthy eating habits, it can be assumed that they are more likely to choose a healthier food option instead of an unhealthy food option. Finally, as one poster includes the sports celebrity Arjen Robben, the likeness of Arjen Robben might influence the decision to choose the healthier food option (or not). Therefore, a 5-point Likert smiley scale is added where the respondent can indicate how they feel about Arjen Robben, ranging from very stupid to very fun.

3.3. Plan of Analysis

(24)

typically the case in statistics (Malhotra, 2010).

The general estimation for a binary logit analysis can be modeled as follows (Malhotra, 2010): 𝑙𝑜𝑔𝑒(

𝑝

1−𝑝) = 𝑎0+ 𝑎1𝑥1+ 𝑎2𝑥2+ ⋯ + 𝑎𝑘𝑥𝑘 , [1] which leads to:

𝑝 = exp ( ∑𝑘𝑖=0𝑎𝑖𝑥𝑖 ) 1+exp ( ∑𝑘𝑖=0𝑎𝑖𝑥𝑖 )

, [2]

where:

𝑝 = probability that an event occurs 𝑥𝑖 = independent variable 𝑖

𝑎𝑖 = parameter to be estimated

The next chapter will provide descriptive statistics and results of the analyses performed.

4. ANALYSES AND RESULTS

This chapter will show the results of the study. First of all, the data is prepared to obtain a valid sample. Next, some descriptive statistics are given and thereafter the regression results will be shown and discussed.

4.1. Data preparation

(25)

4.2. Sample

184 children have filled in the survey, of which 36 (=19.6%) were girls and 148 (=80.4%) were boys. Whereas the experiment was targeted at children between 7 to 14 years, there are a few

participants beyond this age range. However, as this percentage is very small (=5.4%), they will be included for further analyses. Moreover, there are 58 children who indicated that they did not like one (or more) of the products which were offered. As it is unlikely that these children chose an option which included a product they do not like, the analyses will be performed excluding these children. Therefore, the final sample consists of 126 respondents, of which 26 (=20.6%) are girls and 100 (=79.4%) are boys. The average age is 10.45 with a minimum age of 5 and a maximum of 15 years. Table 2 shows descriptive statistics and age information per condition and figure 2 shows the age distribution graph.

As can be seen in table 2, participants per condition lie between 27 and 35 participants. About two-third made an unhealthy choice and approximately one-two-third made a healthy choice. 57 children won a football game on the day they filled in the survey, whereas 38 children lost, 10 children played a draw, and 21 children did not (yet) play a football game on that day.

Table 2. Descriptive Statistics

Frequency Percentage(%) Mean Age Minimum Age Maximum Age Condition N=126 100 Control 27 21.4 10.59 5 14 Fun 31 24.6 11.00 8 13 Celebrity 35 27.8 11.06 5 15 Fun + Celebrity 33 26.2 9.67 7 13 Choice N=126 100 Healthy 42 33.3 Unhealthy 84 66.7 Gender N=126 100 Male 100 79.4 Female 26 20.6 Result N=126 100 Win 57 45.2% Loss 38 30.2% Draw 10 7.9%

(26)

Figure 2. Age Distribution Graph

Figure 2shows that for the target group (children between 7-14 years old), both the age 7 and 14 are not highly represented (4 and 7 children, respectively). This can make the sample less

generalizable. Nevertheless, children between the age of 8 and 13 are highly represented. Therefore, the sample is seen as externally valid and results are expected to be generalizable beyond the experiment to some extent.

It should also be kept in mind that the experiment is conducted at a football club in Groningen, which might be different than conducting the experiment in other parts of the Netherlands. However, it is expected that there will not be great differences between children from different regions in this area, and therefore external validity will not be affected.

4.3. Binary Logistic Regression

As explained before, the hypotheses will be tested by means of a binary logistic regression. Below, the main effects of hypotheses 1, 3 and both hypotheses 4a and 4b will be tested, and thereafter the moderator variables and control variables will be added to complete the model. However, first it is of importance to see whether the control variables proposed in this paper have a significant influence on the dependent variable and will be taken into further analyses. Subsequently, the assumptions of binary logistic regression will be tested.

4.3.1. Control Variables

Control variables which have a significant influence on the dependent variable, and therefore improve the model, are included into further analyses as covariates. To see which control variables

(27)

add to the predictive power of the model, a binary logistic regression is performed where the control variables are the predictors and the healthy choice children make is the dependent variable.

The following model is estimated: 𝑙𝑜𝑔𝑒(

𝑝_ℎ

1−𝑝_ℎ) = 𝑎0+ 𝑎1𝑥𝐺𝐸𝑁𝐷𝐸𝑅+ 𝑎2𝑥𝐻𝑈𝑁𝐺𝐸𝑅+ 𝑎3𝑥𝐻𝐸𝐴𝐿𝑇𝐻 + 𝑎4𝑥𝑅𝑂𝐵𝐵𝐸𝑁 , [3]

where 𝑝ℎ is the probability that a case belongs to the healthier food choice group. 𝑎𝑖 is the parameter to be estimated. 𝑥𝐺𝐸𝑁𝐷𝐸𝑅 is a dummy variable which equals zero if the respondent is female and one if the respondent is male. 𝑥𝐻𝑈𝑁𝐺𝐸𝑅 represents the degree of hunger, 𝑥𝐻𝐸𝐴𝐿𝑇𝐻 represents the importance of eating healthy, and 𝑥𝑅𝑂𝐵𝐵𝐸𝑁 represents the respondents’ liking of football player Arjen Robben. Table 3 shows the results of the regression.

Table 3. Binary logistic regression of control variables on the choice children make

Variable B S.E. Odds Ratio

Constant .059 1.597 1.061

Gender -1.153** .490 .316

Degree of hunger -2.34 .230 .791

Importance of eating healthy -.020 .274 .980

Liking of Arjen Robben .265 .270 1.304

Nagelkerke Pseudo R2

Chi-Square

.085 7.996* *p < .1 ** p < .05 *** p < .01

The regression results in table 3 show that only the control variable gender significantly improves the healthy choice children make. Therefore, gender will be included into subsequent analyses, whereas degree of hunger, importance of eating healthy, and likeness of Arjen Robben will be excluded from any further analysis.

4.3.2. Assumptions

(28)

• Linearity. The dependent variable should be binary and there should be a linear relationship between the logit of the dependent variable and any continuous independent variables. The dependent variable in this study is the choice children make, which can take two values: healthy and unhealthy. Thus, the dependent variable is indeed binary. To test the linearity of the logit, Hosmer and Lemeshow (1989) claim that a logistic regression should be run including all predictor variables and including interaction terms between the continuous predictors and the log of these predictors. When these interactions are significant, the linearity assumption is violated. As age is the only continuous variable, only the interaction term age times the log of age is added. Appendix D shows the results. As the interaction term is not significant at the 5% level, the assumption of linearity of the logit is met.

• Independence of errors. Each case should be unrelated with every other case in the data, i.e. ‘the categories must me mutually exclusive and exhaustive’ (Burns and Burns 2008). As this is the case for the sample, this assumption is also met.

• Multicollinearity. The predictors of the model should not be too highly correlated. One simple way of seeing whether there might be collinearity, is by creating a correlation matrix. When coefficients between variables are, approximately, above .8, there might be multicollinearity problems (Field 2009). Table 1 of Appendix E shows the correlation matrix. The correlation coefficient between the variable fun and the interaction term fun x age is very high (=.980). It is quite common that there exists high correlation between the interaction term and the original variables (Henseler and Chin 2010). To overcome this multicollinearity problem, residual centering, introduced by Lance (1988), will be used. Hereby, the interaction effect is fully represented and first-order effect variance is filtered out (Little et al. 2006). The product term fun x age is regressed onto the original variables

fun and age and the residuals obtained are then used in further analyses. To see whether

(29)

As the assumptions are satisfied, the following paragraphs will show the results of the binary logistic regression. The obtained centered residuals of the interaction term fun x age will be used to overcome the multicollinearity problem.

4.3.3. Main Effects

By means of two dummy variables (one for the fun condition, and one for the celebrity condition) and their interaction (the control condition and the fun-celebrity condition) it is tested whether there are any significant differences in the choice children make when fun elements, a celebrity, or both are present at the poster in comparison with no fun elements, no celebrity endorser, or neither. The following model is estimated:

𝑙𝑜𝑔𝑒( 𝑝ℎ

1−𝑝ℎ) = 𝑎0+ 𝑎1𝑥𝐹𝑈𝑁+ 𝑎2𝑥𝐶𝐸𝐿𝐸𝐵+ 𝑎3𝑥𝐹𝑈𝑁∗𝐶𝐸𝐿𝐸𝐵, [4]

where 𝑝ℎ is the probability that a case belongs to the healthy choice group. 𝑎𝑖 is the parameter to be estimated. 𝑥𝐹𝑈𝑁 is a dummy variable which value equals 1 if fun elements are present (the fun condition), and 0 when fun elements are not present. 𝑥𝐶𝐸𝐿𝐸𝐵 is a dummy variable which value equals 1 if a celebrity endorser is present (the celebrity condition) and 0 if a celebrity endorser is not present. 𝑥𝐹𝑈𝑁∗𝐶𝐸𝐿𝐸𝐵 is the interaction term between 𝑥𝐹𝑈𝑁 and 𝑥𝐶𝐸𝐿𝐸𝐵, which value equals 1 when both fun elements and a celebrity endorser are present (the fun-celebrity condition) and 0 when no fun elements and no celebrity endorser is present (the control condition). Table 4 and figure 3 show the result of this regression.

Table 4. Logistic regression of the experimental conditions on the healthy

choice children make

Variable B S.E. Odds Ratio

Constant -.693* .408 .500 Fun -.539 .593 1.500 Celebrity .405 .532 1.143 Fun * Celebrity .134 .777 1.143 Nagelkerke pseudo R2 _.034 Chi-square 3.098, df =3 *p < .1 ** p < .05 *** p < .01

(30)

39% 28% 33,30% 28,10% 38,20% 33,30% 0% 40%

Fun Elements Celebrity Fun Elements and

Celebrity % o f h e lath y c h o ic e Experimental conditions Not Present Present

present (38.7%). When both fun elements and a celebrity endorser are present, 33.33% of the children chose the healthy option, whereas when they are both not present also 33.33% chose the healthy option. However, as can be seen in table 4 these differences between the conditions are not significant and no support can be found for the main effects proposed in this paper. Next, moderator variables and control variables will be added to complete the model.

Figure 3. Graphical display of the relationship between the experimental conditions and the percentage of

respondents who made a healthy choice.

4.4.4. Final Model

The final model includes the dependent variable, independent variable, moderator variables, and the control variable. Again, a binary logistic regression analysis is performed to predict the healthy choice children make. The model is estimated as follows:

𝑙𝑜𝑔𝑒( 𝑝ℎ

1−𝑝ℎ) = 𝑎0+ 𝑎1𝑥𝐹𝑈𝑁+ 𝑎2𝑥𝐶𝐸𝐿𝐸𝐵+ 𝑎3𝑥𝐹𝑈𝑁∗𝐶𝐸𝐿𝐸𝐵+ 𝑎4𝑥𝐴𝐺𝐸 [5] + 𝑎5𝑥𝐹𝑈𝑁∗𝐴𝐺𝐸+ 𝑎6𝑥𝑊𝐼𝑁+ 𝑎7𝑥𝐹𝑈𝑁∗𝑊𝐼𝑁+ 𝑎8𝑥𝐺𝐸𝑁𝐷𝐸𝑅,

where:

(31)

which equals 1 if the football game is won and 0 if there is a loss, draw, or no game played. 𝑥𝐹𝑈𝑁∗𝑊𝐼𝑁 indicates the interaction effect between fun elements and winning a football game. 𝑥𝐺𝐸𝑁𝐷𝐸𝑅 is the control variable which equals 1 if the respondent is a boy, and 0 if the respondent is a girl. Table 5 shows the regression results.

Table 5. Logistic regression of the final model on the healthy choice children make

Variable B S.E. Odds Ratio P-value

Constant 2.252* 1.329 9.509 .090 Fun -.307 .723 .736 .671 Celebrity 1.233* .651 3.432 .058 Fun*Celebrity -.787 .942 .455 .403 Age -.162 .111 .850 .144 Fun*Age (residual-centered) .390 .261 1.477 .134 Win -1.025 .627 .359 .102 Fun*Win .132 .980 1.141 .893 Gender -1.494*** .523 .224 .004 Nagelkerke pseudo R2 _.199 Chi-square 19.500*, df =8 *p < .1 ** p < .05 *** p < .01

(32)

Table 5 also shows that when both fun elements and a celebrity endorser are present at a promotional poster for a healthier food option, healthy consumption behavior of children decreases (b = -.787, odds ratio = .455). However, these results are not significant (p=.403) and therefore hypothesis 4 cannot be accepted.

The significance level of the effect of age on the choice children make is .144 (2-sided tested). Age has a negative coefficient (b=-.162), and the odds ratio is .850. Decreasing age by 1 year, makes it 1.176 times more likely that children will chose the healthier food option. For example, children from the age of 8 are 1.176 times more likely to choose the healthier option than children from the age of 9. Figure 4 displays this effect. The linear trend line shows a negative effect of age on the choice children make.

Figure 4. The effect of age on the healthy choice children make.

The interaction term fun*age has a significance level of .134 (2-sided tested), with a positive coefficient (b=.390, odds ratio = 1.477). If age increases, the effect of fun increases. As fun has a negative effect, it would mean that increasing the age of a child with one, increases the negative effect of fun with 1.477. For example, a child with the age of 11 who is exposed to fun elements is 1.477 times more likely to choose the unhealthy option than a child with the age of 10 who is exposed to fun elements. Figure 5 also shows this effect. For children above the age of 10, the decline in percentage of healthier choice is much steeper than for children below the age of 10. Hypothesis 2 can be partly accepted, as it was expected that when age is increasing, fun marketing will become less effective. However, hypothesis 2 also expected a positive relationship between fun marketing and food choice, whereas this relationship seems to have a (insignificant) negative effect.

(33)

Figure 5. The percentage of children who make a healthy choice, with the interaction effect age*fun

The significance level of the effect of winning a game on the choice children make is slightly above the 10 percent level (p=.102, 2-sided tested). Children who win a football game are more likely to choose the unhealthy food option, in comparison with children who lose, play a draw or not play a football game. Graph 6 visualizes this result. On average, 28.10% of the children who win a football game make a healthier food choice, whereas 37.70% of children who did not win a football game make a healthier food choice. Table 5 shows that children who did not win a football game, before they have to choose between two food options, are 2.786 times more likely to choose the healthier food option than children who have won a football game. Hypothesis 5a can be accepted, whereas hypothesis 5b cannot be accepted. The interaction effect fun times win is insignificant (p=.893), and therefore hypothesis 6 cannot be accepted.

Figure 6. The effect of winning a football game on the healthy choice children make.

(34)

The control variable gender is significant at the 1% level (p=.004). As shown in figure 7 on average, 53.80% of females choose the healthier option, and only 28% of males. To be more specific, females are 4.4641_{times more likely to choose the healthier option than males.}

Figure 7. The effect of gender on the healthy choice children make

5. DISCUSSION

The main aim of this study was to see whether children in a football canteen can be influenced to unconsciously choose a healthier food option, in comparison with a more unhealthy option. This was done by showing two advertising posters: one with a healthier food choice and one with an unhealthy food choice. The poster with the healthier food option was manipulated by including fun elements, a celebrity endorser, and both. Next, children (aged 7-14) where asked to make a choice between the two options and it was analyzed whether this manipulation of the poster with the healthier option led to an increase in preferences for the healthier option. This part will discuss the results of the study.

5.1. Fun Elements

(35)

children between 7 and 14, they might not even focus on the fun elements, but focus mainly on the products which were displayed on the poster and this might explain that there are no significant differences found.

Another reason why fun elements might not influence the option children choose, is that the unhealthy option included Coca-Cola. Achenreiner and John (2003) say that brand names can be a very important perceptual cue for children, which identifies a certain product as being familiar or have a certain meaning or prestige. Coca-Cola has the highest brand value, in the soft drinks category, worldwide (Statista 2017). Kantrowitz and Wingert (1999) found that children between the age of 8 and 14 have a desperate need to belong and are very brand conscious. Owning the right brands is their quickest way to acceptance by peers. As Coca-Cola is a widely accepted brand which children recognize, this might wear off the effect of using fun elements and children might choose the Coca-Cola option anyway, regardless of the promotion poster used.

5.1.1. Fun and Age

Regarding hypothesis 2, it was expected that if age is increasing, fun marketing will become less effective. Findings show that this is indeed the case, however a positive relationship between fun marketing and food choice was expected, whereas results show a (insignificant) negative effect. Thus, the age of children strengthens the negative relationship between fun marketing and food choice. Specifically: if age is increasing, fun marketing will become less effective. This is in line with earlier research of Livingstone and Helsper (2006), who found that younger children are more relying on simple cues (such as colors and characters), and teenagers rely more on the information given, in this case the products itself. However, the results here show that even for younger children (with the age of 8), fun marketing is not effective. It could be that the children participating in this study, were not young enough to be influenced by the fun aspects. As already explained above, from an age of 8 and onwards children develop a certain skepticism towards marketing (Gunter and McAleer 1997). Whereas this study involved children between the age of 7 and 14, this might be the reason why children are not attracted by the fun poster. The results found here seem to indicate that the older the child, the more it turns against the fun poster and choose the other poster instead.

5.2. Sports Celebrity Endorser

(36)

healthier foods might indeed be effective in changing the way children eat. In addition, de Droog et al. (2011) find that using famous characters can push children into the right direction and improve their healthy eating behavior. This paper proves evidence in line with these views. Using, in this case, Arjen Robben as a promoter of a healthier food option, significantly increases the percentage of children who actually choose the healthier food option. Reasons for this might originate from the theory of wishful identification (Feilitzen and Linné 1975; Hoffner 1996). When children see an idol or role model, they have the desire to act the same as this character. In this case, Arjen Robben is proven to be a role model for many of the respondents as many of them play football. Additionally, a question asked about the liking of Arjen Robben on a 5-point Likert scale, ranging from very stupid to very nice, resulted in an average score of 4.1, indicating that on average, children like Arjen Robben. Elaborating some more on the wishful identification theory, children see sports celebrities as their ideal self, and they want to have certain characteristics which are equal to that of the role model (Boon and Lomore 2001). Also, de Droog et al. (2011) suggest that famous characters who promote certain products, can elicit a certain positive affective response where children like the products more which are promoted by the famous person. Celebrity endorsing is widely used and there is a wealth of research which proves that it actually works. However, research which investigates if celebrity endorsement can be used to improve dietary habits of children is very limited. McGale et al. (2016) find prove that displaying brand equity characters on the packaging of certain products, including healthier products, increases children’s preferences for this product. This paper adds to the existing literature by proving that using a sports celebrity endorser on a poster promoting a healthier food option increases the preference and choice for this option.

5.2.1. Fun and Celebrity

With regards to hypothesis 4, it was expected that using both fun elements and a celebrity endorser on a promotional poster for a healthier food option would lead to a higher preference for the healthier food option, than using only one of these techniques or using none. Results presented in this paper have shown the opposite: when both fun elements and a celebrity endorser are present, this leads to a decrease in the healthy consumption behavior of children. However, these results were not significant. Whereas using a celebrity endorser solely gives significant positive results, adding fun elements results in insignificant negative results.

5.3. Winning a Football Game

(37)

(Baker-Ward, Eaton and Banks 2009), this would mean that experiencing positive emotions by means of winning a football game is a trigger for unhealthy food intake. Dingemans et al. (2009) explain this effect through a view that positive emotions leads people to enjoy hedonic foods, rather than focusing on long-term goals, such as eating healthy or watching a diet. This adds to the literature about risk behaviors, which indicates that when people experience high amounts of positive emotions, they are found to have a higher intolerance for alcohol consumption, binge eating, and drug intake (Martin et al. 2002). By means of three studies, Evers et al. (2013) found that unhealthy snack intake results mainly from experiencing positive emotions, as opposed to negative emotions. Results presented in this paper are in line with Evers et al. (2013) and provide evidence that winning a football game can increase unhealthy snack intake. This is especially important for football canteens, who, for example, might be able to reward winning teams with a healthy snack, instead of letting them choose themselves.

5.3.1. Fun and Winning a Football Game

The effect of winning a football game on healthy food choice did not significantly differ for respondents exposed to a poster with fun elements and respondents who were not exposed to the poster with fun elements. This is not in line with hypothesis 6 and the expectations that individuals who won a football game are more easily persuaded by an advertising message and rely on simple cues (Bless et al. 1990), such as the fun elements. This might be explained by the fact that previous research and theory on this subject mainly focused on adults, and not on children. Also, there are no significant results found for believing that including fun elements on a poster influences the choice children make.

5.4. Gender

(38)

5.5. Managerial Implications

Results obtained by this study can have managerial implications for managers of football canteens, or other sports canteens. It was found that using a celebrity endorser on a promotional poster for a healthier food option, increases the preference for this healthier food option. Sports canteens can use this to their benefit, and start using promotion posters with various sports celebrities to, unconsciously, let children change their food habits. However, using fun elements should be taken with care. Results have shown a negative insignificant effect, and if this effect is only due to sample size issues, using fun elements could actually decrease preferences for the healthier option. Therefore, it is recommended to managers of sports canteens, to use solely a sports celebrity endorser and no fun elements to stimulate healthy consumption behavior of children. Furthermore, previous research mainly focused on the packaging of products (de Droog et al. 2011; Pires and Agante). Results might be even more convincing when sports canteens make the packaging of the products which are promoted equal to the promotion posters. For example, sports canteens can use a bag or box to pack the products and depict the sports celebrity endorser on this bag or box. De Droog et al. (2011) found that using a character on the packaging of fruit, makes children more attracted to the fruit and increases intentions to purchase the fruit. This can also work for sports canteens, and gives children a chance to collect the bags or boxes with their favorite sports celebrities on it. Therefore, it is recommended to not only use promotional posters to stimulate children to purchase more healthier food options, but also consider to use other techniques.

Furthermore, children often rely on their parents when they buy something in a sports canteen. Therefore, it is also important to make parents aware of healthier food options in the sports canteen and make them aware of the nutritional facts of certain products. Whereas showing nutritional facts has no significant influence on children’s decisions (Hastings et al. 2003), this might convince parents to push their children in the right, healthier, direction (Cranage et al. 2004; Din, Zahari and Shariff 2012) . This can be done by, for example, sending an email to the members of the sports club or using social media.

Healthy Food at Football Clubs

Healthy Food at Football Clubs

The Influence of Promotion Posters on Healthy Food Choice

Among Children

HEALTHY FOOD AT FOOTBALL CLUBS

The Influence of Promotion Posters on Healthy Food

Choice among Children

MANAGEMENT SUMMARY

PREFACE

TABLE OF CONTENTS

1. INTRODUCTION

2. THEORETICAL BACKGROUND

3. METHODOLOGY

4. ANALYSES AND RESULTS

5. DISCUSSION