The effects of front-of-pack labelling in the form of exercise labels on consumer behaviour

(1)

The effects of front-of-pack labelling in the form of

exercise labels on consumer behaviour

27-minute run

Calories: 350 kcal

Ana Maria Botez S3490548 MSc Marketing

(2)

The effects of front-of-pack labelling in the form of exercise

labels on consumer behaviour

University of Groningen

Faculty of Economics and Business

MSc Marketing Management January 14th, 2019

Master Thesis

First Supervisor: Martine van der Heide Second Supervisor: Dr. Jenny van Doorn

Ana Maria Botez

Metaallaan 133, 9743 BR Groningen 0658707578

(3)

Acknowledgements

Firstly, I would like to express my gratitude to my first supervisor Martine van der Heide, for taking the time to guide me through the whole process and constantly offering her support and helpful feedback. I am incredibly grateful for receiving the opportunity to conduct a study focused on such an important subject. As the obesity rates are drastically increasing, the need for intervention has never been more vital. By writing this study and contributing to the body of knowledge on consumer behaviour, I hope to help consumers in their journey towards a healthy living.

Second, I would like to thank my family and friends who have always supported and encouraged me. During this whole experience, with its ups and downs, they have always listened to me and gave the strength to continue.

Groningen, January 14th 2019

(4)

Abstract

Worldwide, obesity rates have seen a rapid climb, with close to half of the population being overweight. In order to combat this epidemic, many efforts have been made with the sole purpose of nudging people to a healthier lifestyle. Among these efforts employed by policymakers and marketers, food labelling has been a hot topic of discussion and debate. The simple concept of food labelling is to provide simple, clear and accurate information to consumers in order to guide their food choices. Built on previous labelling theories, the present study evaluates the effectiveness of the novel nutritional information format: the physical activity calorie equivalent (PACE) label, also known as the exercise label. Participants in the experiment were randomly assigned to one of the two conditions (normal caloric label vs. exercise label). Following expectations, there were significant differences in caloric consumption between the exercise label and more commonly used caloric label. The participants assigned to the exercise label condition have made healthier food decisions, with a lower calorie count compared to the participants assigned to the control condition. This study encloses key implications for marketers and for policymakers by confirming previous findings, which present the exercise label as an effective tool in guiding consumers towards a healthier choice. Moreover, this study shows that the relatively new front-of-pack label does not induce a negative view on sporting, as some of its critics have feared. In conclusion, the results of this study can be seen as a green light for the introduction of exercise labels. Furthermore, these results call for further research, especially for longitudinal studies, to see whether this front-of-pack label truly has the power to decrease the number of calories purchased by consumers in a shopping trip.

Keywords: front-of-pack labelling, labelling, exercises label, PACE, nutritional information,

(5)

List of figures

Figure 1 Estimated Marginal Means of Calorie Consumption between label Conditions and with-in Type of Food

33

Figure 2 Estimated Marginal Means of Number of Hours dedicated to Future Sporting Intentions in the upcoming

Week ... 34

List of tables

Table 1 Results of ANOVA and Person's Chi-squared test ... 28

Table 2 Normality tests for the dependent variables and for the moderator ... 29

Table 3 Median calorie rank of conditions ... 31

Table 4 Main effect of the food type ... 32

Table 5 Normality test of the independent variable Level of hunger ... 35

Table 6 Normality test of the dependent variable Diet ... 36

(8)

Introduction

According to the World Health Organization (WHO), globally in 2014, 39% men and 40% women aged 18 or over were overweight. In Europe, obesity and weight-related problems (such as diabetes, heart disease, stroke and some cancers) are increasing at an alarming rate in most of the EU state members, with estimates of 51.6% of the EU’s population being overweight in 2018. This percentage places Europe in second place, behind the United States of America, for the highest number of overweight or obese people in 2018 (Overweight and obesity - BMI statistics, 2018). This rapid climb in obesity rates can be attributed to a number of factors such as: financial and geographical barriers in acquiring healthy products, the large availability and intense marketing of energy-dense foods, lack of knowledge and education surrounding nutritional information (Roberto and Khandpur, 2014; Wardle, Parmenter and Waller, 2000), an increase in the intake of energy-dense foods and an increase in physical inactivity (Variyam, 2005). The prevalence of obesity and overweight presents itself as a great burden to the health system and a great challenge to the life course of people across the world.

Numerous steps have been taken by policymakers, health organizations and manufacturers to better inform the consumers about their energy intake, for example, the 1994 Nutrition Label and Education Act or the pledge of manufacturers to implement energy (kilojoules, kJ; kilocalories, kcal) labels on food and non-alcoholic beverages in out of home settings (Health, 2013). The use of these labels varies across subgroups (Campos, Doxey and Hammond, 2011) and calorie labelling presents itself as a promising method in preventing obesity, as in its absence consumers find it very difficult to understand nutritional information. (Roberto, Agnew and Brownell, 2009) However study reviews focusing on fast food menus, show that there are existing concerns regarding its effectiveness (Kiszko, Martinez, Abrams and Elbel, 2014).

(9)

when possessing these skills consumers are still influenced in their purchase by a number of factors, such as taste, cravings (Grunert, Wills, and Fernández-Celemín, 2010,) halo effect(social-psychological phenomenon that causes people to be biased in their judgments by transferring their feelings about one attribute of something to other, unrelated, attributes), quantity anchors or the limited amount of time they look at food before buying it. These factors can lessen the effect of nutritional information on consumer behaviour (Chandon and Wansink, 2007; Van Ittersum and Wansink, 2012; Hamlin, McNeill and Moore, 2015). A review article published in 2011 stated that commonness of self-reported use of these labels is high among the general population (with 82% in New Zealand, 52% in Canada, 47% in the EU and 75% in the USA) (Campos, Doxey and Hammond, 2011). Although, past research has shown that over 70% of consumers believe that new nutritional labels on packaged foods should be easier to understand (Kristal, Levy, Patterson, Li and White, 1998). These findings, coupled with the alarming rates of global obesity, indicate that pressing against overconsumption and healthy food choice via nutritional content and calorie labels alone is insufficient. Different ways of presenting nutritional information in a more understandable manner have been put forward with front-of-package food labels as a useful tool to battle obesity.

(10)

labels could have different effects on consumer behaviour, based on the type of food they are used on. However, research is yet to determine the in-depth effects of the exercise labels.

Aim

The aim of this study is to determine if exercise labels are effective at improving consumer’s understanding of calorie content and the healthiness of their food intake. This experiment will try to fill the gap in the literature by conducting an experiment, which will compare the presence of the exercise labels with its absence in aiding consumers to make healthier food choices. Additionally, this experiment will focus on the labels effect on sporting attitude and sporting intention. The results will gain further insight into the fundamental relationship between nutritional labels and subsequent consumption and product decisions.

Literature review

The economic growth, industrialization, urbanization, a decrease in physical activity, and a transition to processed foods and high calorie/energy diet over the past decades, has made the rates of obesity in many countries double (Hruby and Hu, 2014). In most countries, steps have been taken in an effort of guiding consumers through their decision-making process, with information such as calorie or nutritional content (Montford, Peloza and Goldsmith, 2017). These steps have the final goal of changing the consumer’s eating behaviour.

People’s eating behaviour consists of a combination of what they eat, how much they eat when they eat, where they eat and why they eat. The factors that drive people’s food choices are biopsychosocial factors, that is, it involves biological, psychological and social factors (Cox et

al., 1998). These determining factors of consumer food choice are of interest not only to

(11)

attitudes, knowledge and beliefs about food (The determinants of food choice, 2006). Triggering a dietary change is a major challenge to the health professionals and to the public themselves, this is why different strategies for different groups with different priorities are required. One such strategy implemented across the world is the use of nutritional labels.

The nutritional labels on food have as the main attribute informing the consumer about the exact nutritional content of a food or beverage, and in this way helping him make a healthy decision. In a more scientific view, the nutritional labels represent a cost-effective vehicle for aiding consumers identify healthy food choices (Kelly et al., 2008). However, Wandel (1997) states that when dealing with nutritional labels, consumers find them time-consuming and difficult to understand. Further studies have confirmed that interpreting nutritional information is indeed difficult for many consumers because it requires a set of skills such as high literacy and high numeracy (Krukowski et al., 2006; Roberto et al., 2009). Even when possessing these skills consumers are affected by biases (e.g. halo effect) which can reduce the effectiveness of nutritional information (Chandon and Wansink, 2007; Van Ittersum and Wansink, 2012). Analogously, Sacks in his article entitled ‘States should stand up to the food industry on traffic light labelling’ states that “only a small percentage of the population read the current back-of-pack nutrition information panels and even fewer understand it” (p.1). Going even further into the factors that influence the use of nutritional labels in the consumers food choice behaviour, a review on people’s understanding of nutrition labelling has identified besides difficulty in interpreting information, lack of time and lack of trust in the accuracy of the information provided as defining factors (Cowburn and Stockley, 2005). It has been found that the extent of information provided by the nutritional labels mostly leads to consumers feeling confused and not being able to convert the information into an actual food choice (EUFIC, 2005).

(12)

content (van Kleef et al., 2008). In this manner, consumers can easily understand which products are overall healthier for them and make an informed choice. The simple execution of the FOP labels coupled with their grouping of foods into understandable ‘healthy’ or ‘unhealthy’ tiers is what ensures their success since many consumers limit their information search to the front of the package (Levy and Fein, 1998; Roe, Levy and Derby, 1999). Food information programs have excluded calorie count as a relevant piece of information to the consumer because the calorie count is not a good indicator of foods healthiness (van Kleef et al., 2008). On the other hand, if the caloric information of food is used comparative with the recommended daily amount of energy it could provide consumers with meaningful information (Bussell, 2005). This could be in the detriment of the consumer since caloric information is the most commonly used nutritional indicator and amongst the European consumers is the most recognized as well (European Food Information Council (EUFIC), 2005). A challenge for policy makers, health specialist and marketers is to develop a nutrition label that is accurate, yet simple, with a representation of the caloric information in a format that is appealing, easy to process and easy to understand for consumers healthiness (van Kleef et al, 2008).

Front-of-pack labels

Daily Intake System (%DI)

(13)

amount of nutrients while the DI refers simply to the guideline amounts for a normal eating adult

(

What are Guideline Daily Amounts (GDAs)?, n.d.)

Third Party Label

Consumers are demanding higher quality in the food products they buy from stores, and food manufacturers use a variety of claims to target consumers. Because claims sell products and manufacturers are inclined to exaggerate their product’s quality consumers no longer know which of these claims to trust.

In order to compensate for the official’s lack of action, third-party certifications were created by private voluntary agents to fill in as informative health indicators for consumers (Hartt, 2015). An example of this third-party certification is The Heart Foundation Tick, which is used in retail food settings, such as supermarkets. The aim of the Tick is to allow consumers to identify the healthier choice within the food category they shop in (Heart Foundation Tick, 2012). The third party labels help people make quick and easy healthy food choices they also encourage food manufacturers to develop or adapt food products that support the labels very strict rules and costly restrictions (Hartt, 2015).

There is some controversy surrounding the third-party label, with critics naming the label a money-making scheme. For example, in an article by Burton-Bradley (2012) the Heart Tick is being accused of appearing on foods that do not belong in the healthy category. The label has been seen appearing on foods high in sugar content or other ingredients deemed unhealthy, including burgers, pizzas, fruit bars, cereals, and even some McDonald’s products. Therefore, the main problem that consumers and health authorities have with third-party labels is that they could be misleading in the pursuit of making money.

Health Star Rating

(14)

resistance from manufacturers of having their product ranked against their competitors (Chappman and Watson, 2013).

Traffic Light Labels

The traffic light system uses the colours: green, amber and red to signal to the consumers, in a rapid manner, whether the product they are considering is high, medium or low in fat, saturated fat, sugar, salt or overall energy (Australian Medical Association, 2011). This labelling system makes it easy for the consumer to identify the healthier food choices, by directing him to a product with amber or green colour rather than red (Wartella, 2012). Several studies have shown that the traffic light labels are more advantageous and beneficial than other labelling approaches in nudging consumers toward healthy options. Research conducted in Germany and the UK and afterwards replicated in Australia has revealed that this labelling system enhances the ability of consumers to identify healthier options on packaged foods (Borgmeier and Westenhoefer, 2009; Kelly et al., 2008). Consumers were able to identify the nutrient content of the food based on the colour of the label (Kelly et al, 2008). Based on the findings of these studies it appears that the use of traffic labels leads to reduced cognitive workload for the consumer, making the identification of healthier food options easier (Australian Medical Association, 2011). In their study, Kelly et al. (2008) also discovered that consumers are able to compare products faster, at a first glance, than when using the daily intake system.

The most important benefit of traffic labelling system is the fact that it has high levels of understanding and acceptance across ethnic and income groups (Gorton et al., 2007). Therefore, as it will benefit the entire population and it has relatively low implementation costs, traffic labelling is regarded as one of the top policies in interventions against obesity.

(15)

Exercise Labels

In response to the challenge of effectively conveying nutritional information to the consumers, nutritional experts have discussed exercise labels as a prospective method to inform the public about calorie value. Exercise labels are defined as the amount of physical activity necessary to burn the calories ingested from food (Platkin, Yeh, Hirsch, Wiewel, Lin, Tung and Castellanos, 2014). For example, burning off a hamburger with a calorie count of 300, after expending the calories needed for daily subsistence, requires 75 minutes of walking. When calculating an individual's energy expenditure needed to burn a certain amount of calories, numerous factors play an important role in the calculation (factors such as body weight, age, gender, and basal metabolic rate). On the other hand, exercise labels must be universally relevant and therefore average energy expenditure must be calculated. Blair et al. (2011) calculated the energy expenditure for a 72 kilograms adult at an assumed walking rate of 30 minutes/kilometer, which was converted into a calorie burning rate of approximately 3 kcal per minute. To obtain the necessary number of minutes needed to burn of an item, the number of kcal in that item is divided by the 3 kcal/minute burning rate. Through such calculations, different labels are created featuring either kilometers to walk or miles or minutes. Formulating the information in this set-up makes exercise labels a tool with the potential of simplifying food and restaurant nutrition labels, decrease the misunderstanding of the term calorie, and ensure a decrease in energy intake as well (Bleich et al., 2012).

Several studies have shown that, when presented with the exercise label, consumers have, in more than one occasion, cut down on the calorie they consumed. Recent research by Masic et

al. (2017), presented images of five different food snacks (breakfast bar, café muffin, biscuits,

(16)

success of exercise labels, a real-life experiment was done by Bleich Herring, Flagg, and Gary-Webb (2012), involving African American teenagers from an impoverished neighborhood. The demographic chosen by Bleich et al. is known for having the highest levels of obesity and the lowest levels of health literacy among consumers and a success for the exercise label in this condition could mean high effectiveness across all demographic conditions. The aim of the study was to see whether the exercise labels would influence the consumption of sugary drinks. The participants observed by the researcher were exposed to either: caloric labelling, daily intake labelling or the time required to jog to burn off the calorie ingested (Bleich Herring, Flagg, and Gary-Webb, 2012). The conclusion the study arrived at was that the exercise labels had the most powerful intervention, with the researchers stating that most consumers in the USA would be shocked by the amount of time need to burn off a nutritionally worthless beverage as sugary drinks. These findings are consistent with other studies involving menu labelling, which shows that menus with exercise labels included have stimulated patrons to choose less calorie dense meals.

For example in a web-based study rolled out by Dowray et al. (2013) involving menus with no nutritional information, with caloric information, with calorie information and minutes to walk to burn those calories and with calorie information and miles to walk to burn those calories, there has been found a significant difference in the mean number of calories ordered depending on the condition. When questioned about the exercise labels 82% of the participants reported a preference for a combination of calories and exercise equivalents on menu labels. This study combined with previous ones shows the positive impact of exercise labels in delivering a proper understanding of the term calorie and encouraging consumers to lower their calorie intake.

(17)

illustrations increased understanding in 98 percent of the studies. The explanation for this is that pictures facilitate learning by providing clear examples, extra-lingual information, contexts for interpretation, and redundancy which helps recall. Therefore, the images added to the label, images that portray a certain physical activity help the consumer to understand, recall and eventually learn nutritional information. The third and possibly the strongest reason why exercise labels could trigger a change in calories purchased is based on behavior economics theory. According to this theory, people will by default use mental shortcuts for the majority of decisions made daily simply because the human mind is limited when it comes to processing information (Williams, 2016). That is why the simple design of the exercise label, which appeals to the intuitive system, is more likely to yield results.

(18)

Vice vs. Virtue Food Choice

Consumers might find this label discouraging because people, in general, find food as a source of pleasure (Muñoz-Vilches et al., 2018). It is a direct contributor to their physical well-being but at the same time, it can be a source of worry and stress (Rozin, Fischler, Imada, Sarubin, and Wrzesniewski, 1999). People’s opinion on food as a pleasure and a worry at the same time is generally due to its categorization based on a good/bad dichotomy of virtue and vice (Rozin, Ashmore and Markwith, 1996). Elaborating on previous research conducted by Wertenbroch (1998), virtue decisions are decisions consistent with one’s long-term self-control goals (e.g. losing weight) but do not necessarily present immediate gratification to the consumer. As opposed to vices, defined as decisions consistent with short-term goals of immediate gratification (e.g. eating a burger) and inconsistent with long-term goals (e.g. losing weight). To better illustrate these notions, vegetables and fruits are regarded as virtue foods and are inherently considered healthy, while chocolate and fries are considered vice foods and inherently unhealthy (Hui, Bradlow and Fader, 2007).

(19)

due to the high number of calories in “vice foods” transforming into a high amount of sporting required, bringing forth the delayed consequences of eating “vice foods” might have a more striking effect on the public. Diving deeper into the “vice food” and “virtue food” conflict, studies have shown that when people perceive food as being healthy, they also perceive it as being low in calories (Chandon and Wansink 2007; Raghunathan, Naylor, and Hoyer 2006). This perception combined with the results of van Kleef et al. (2007) on exercise labels and how they draw attention on the health perspective of food it is expected that when applied to “virtue foods” these labels might lead to a sub-estimation of calories and therefore an increased consumption. Also, when consumers are asked to combine healthy and unhealthy food choices they look at them from a qualitative point of view. The health benefit from one food choice compensates the high calories form an unhealthy food choice and surprising actually leads to a lower count in calories.

Building backwards from the delayed gratification theory (Mischel et al., 1989), which refers to the process that one subject might undergo to fight the temptation of immediate reward and on well-documented studies that people tend to compensate effort with hedonic rewards (Kivetz and Simonson, 2002; Kivetz and Zheng, 2006; Mukhopadhyay and Johar, 2009). Could a past hedonic consumption transform future physical effort into a punishment? Food consumption also leads to consumers having mixed feelings with hedonic products leading to greater feelings of guilt (Kuijer and Boyce, 2014). In the case of “vice foods”, this guilt could deepen the view of physical activity as punishment.

Unintended effects on sporting

Even though exercise labels have the potential to be more convincing than the normal calorie label because they deliver the information in terms regarded as familiar, they could also trigger unwanted views or responses from consumers. As previously mentioned the exercise label brings forth the consequences of having an unhealthy diet and experts fear that the label could backfire and lead to negative views on physical activity.

(20)

and patronizing with some participants experiencing a feeling of guilt. The main reasons for disliking the labels were the fact that the label drew attention too much on health and not enough on enjoying the experience of eating and the fact that the exercise amount seemed unrealistic. Although the label received a negative review from some participants, younger participants preferred it as it is easy to read and it motivates them to exercise. Other studies such as Swartz et

al. (2013) have reached the same conflicting conclusions with participants in the focus groups

finding the exercise labels discouraging and others preferring them because they seem more personal. Among the reasons given by participants in both of the studies, the main and recurring one is the guilt, they experience followed by their perceived inability to carry out the label's indication. Even though these previous studies showed whether the participants liked or disliked the label and their in-depth opinion on it, the question of whether the exercise label had any influence on the sporting activity was not answered. In order to answer this question, Antonelli and Viera (2015) conducted a study on fast food ordering with 823 participants. The study revealed that 64% of the participants reported that the exercise label was ‘somewhat likely’ or ‘very likely’ to influence their physical activity level.

(21)

Thesis outline

Hypotheses:

Summarizing the findings in the literature review regarding front-of-pack labels, it seems that ease of understanding and comparison, standardized presentation, reliability and salience to the customer are the most important attributes of the label. Also, the label needs to be compelling enough to steer the consumer towards the healthy choice, and as mentioned beforehand the exercise label has shown a positive effect on healthy food choices by decreasing the number of calories consumed. This is why the first hypothesis is as follows:

H1: Presenting the front-of-pack exercise label (as opposed to the regular caloric

label) will lead to a reduction in calorie consumption.

As previously mentioned applying the exercises label could lead to consumers perceiving delayed consequences as more salient and different types of food could elicit different response due to the way they are perceived. Therefore, hypothesis two, three and four are as follows:

H2: Presenting the front-of-pack exercise label (as opposed to the regular

caloric label) will lead to a greater (smaller) reduction in calorie

consumption in the case of vice food (virtue foods).

(22)

H4: Presenting the front-of-pack exercise label (as opposed to the regular

caloric label) will lead to higher sporting intention.

Methodology

This research includes four variables Label Format (Normal Caloric Format, Exercise Label Format) as the independent variable, Caloric Consumption (which measures the energy intake), Attitude towards Running (which shows the way running was perceived; fun or requirement) and Future Sporting Intention (which shows the desire to sport in the upcoming day/days) as the dependent variables. Type of food acts as a moderator. The study design is a 2 by 2 mixed design, with participants being allocated to one of the 2 conditions (Normal Caloric Label or Exercise Label) but all of them will be exposed to both types of food (Healthy Food and Junk Food).

The experiment

In order to test the hypothesis stated above, primary data was gathered through an online survey, using the platform of Qualtrics (www.qualtrics.com). The Internet was selected as the appropriate delivery method due to its easy and fast distribution, high respondent reach and low costs. Furthermore, in the online medium the bias of social norms is smaller, the researcher can easily manipulate the questionnaire (which is important since different respondents will be exposed to different conditions) and the data is available in digital form, resulting in a more efficient analysis (Malhotra, 2010, p.222; Blumberg, Cooper and Schindler, 2014). The participants received the survey on Facebook and after clicking on the link they were sent to the Qualtrics platform. Since the survey was distributed in the online environment, the respondents should have access to and knowledge of the Internet.

(23)

includes questions regarding demographics, product selection, sporting attitude, sporting, and health knowledge.

Study Procedure

The online survey included an online grocery-shopping environment in the first section and self-report measures in the second part of the questionnaire (health knowledge, sporting attitude etc). The survey was structured in the following manner. Firstly, a short introduction pointing out the framework of the survey and declaring that the study is conducted within the scope of the University of Groningen was presented. Afterward, participants had to declare consent to their participation as well as to confirm the fact that they have shopped for groceries in the previous 6 months, as a precondition for the survey. In addition, the participants were guaranteed anonymity and they were informed that if by any matter they decide so, they can retract their answers whenever they want without any explanations needed.

(24)

therefore according to the paper 12.8 kcal are burned per minute. To determine the number of minutes required to burn one product’s calories, the calories per serving of that product were divided by the 12.8 kcal/min rate. From the categories presented, the participant could only choose one product per category. After choosing one product, they were allowed to move to the next category. The order of the categories, as well as the order of the items inside the category, were randomized in order to eliminate any order preference.

In the second part of the experiment, the participants were asked a series of questions regarding their shopping trip experience, their attitude towards sporting, their intention to sport in the upcoming days, and their nutritional knowledge. At the end of the questionnaire, a couple of demographic questions such as age, sex, self-reported height, and weight were asked. Across the survey, the participants were not allowed to return to the previous questions and all questions were mandatory. At the end of the survey, they received a short debriefing.

Measurements

In the following chapter, the methodology behind the evaluation of the shopping cart as well as the relevant scales used for this study will be presented and explained.

Caloric consumption

(25)

Attitude towards running

To be able to understand if the exercise label has had any influence on the participant’s view about running, the participant’s enjoyment of sporting was measured using the Physical Activity Enjoyment Scale (PACES) by Mullen et al. (2011) adapted to running. The original version contains 18-items but for this study, only 10 items will be used. This decision was taken for feasibility reasons because the dropout rate and the non-response bias caused by measurement administration are some of the most important factors in measurement bias (Strickland, 1996a). Fagarasanu and Kumar (2002) mention the benefits of keeping the self-reported instruments to a small amount in a study due to a generally poor response rate. They also recommend including only items that are below the physical and cognitive capabilities of the target population to reduce the dropout rate. Therefore, the decision was made on reducing the scale items to not produce inconvenience and time demand. Participants were asked to rate on a 7-point bipolar scale (1=” Strongly disagree”; 7=”Strongly agree”) how they felt at the moment about physical activity. Higher scores reflected higher levels of enjoyment by the participants. For this study, the main question was adapted to running due to the usage of this sport in the exercise label. Therefore, the main question was “How do you feel about running?” The ten items chosen for this scale were displayed in a randomized order: I find it pleasurable; I enjoy it; It's very pleasant; It's very invigorating; It's very gratifying; It's very exhilarating; It's very stimulating; It's very refreshing; I like it; I feel as though I would rather be doing something else.

Future sporting intention

(26)

Nutrition Knowledge

In order to measure the participant’s subjective nutrition knowledge, the scale developed by Brucks (1985) and adapted by Moorman et al. (2003). This scale contains two statements where their self-expressed level of nutrition knowledge is compared to that of the average consumer and last statement is refereeing to their confidence in applying that knowledge in their day to day lives. All the statements are rated in a 7-point Likert scale with 1 being ‘Very low’ and 7 being ‘Very high’. The scale contains the following statements: Rate your knowledge of

nutrition information compared to the average consumer; Rate your confidence in using nutrition information compared to the average consumer; I feel confident about my ability to comprehend nutrition information on product labels.

Exercise label check

To test whether the participants in the exercise label condition have observed the caloric information they were exposed to, they will receive a question with the two labels presented and they will have to choose which label they saw in the previous questions.

Demographic information

At the end of the survey, participants received questions regarding their gender, age, height, and weight. The last two are required to asses each participants BMI in order to have a clear description of the sample at hand and in order to check if their answers were in any way influenced by this variable.

Empirical study

Data editing and cleaning

(27)

total, 208 respondents participated in the survey however, 52 did not fill the survey until the end. Therefore, there are 156 valid responses with 78 of them in the control condition (normal caloric label) and the remaining 78 respondents in the manipulated condition (exercise label condition). Moreover, outlier analyses were conducted for the dependent variable calorie consumption (healthiness of the shopping basket), for the moderator type of food and for the other two dependent variables namely attitude towards running and future sporting intention. The box plots detected no outliers for the dependent variable calorie consumption but for the moderator, in the case of virtue foods, an outlier was detected as for the two other dependent variables (Appendix A). To ensure the analysis is appropriate the outliers were fixed by using the process of winsorizing. Winsorization is an effective way to deal with the problem of outliers as it improves statistical efficiency and increases the robustness of statistical inferences (Dixon, 1960). This process consists of modifying the top and/or bottom 10% to the value of the nearest extreme. In the case of the moderator, the outlier can be a legitimate value as respondents could choose what different products with different caloric values they would like to consume and this could lead to a variety of values. After manually checking the outliers for the dependent variable attitude towards running, it was observed that the values were in normal range (between 1 and 10) therefore no action was taken towards this variable either. With regards, to the dependent variable future sporting intentions, three of the extreme outliers (values 16, 30 and 20) where modified to the nearest extreme value (15).

Sample characteristics

(28)

Random assignment

This study is a mixed-designed one, with two in-between conditions, namely normal caloric condition (control condition) and exercise label condition, and two within conditions, type of food (vice and virtue food). In order to deliver a reliable analysis, the study requires at least 50 participants per condition, i.e. at least 100 respondents (Simmons, Nelson and Simonsohn, 2013). This study has 156 valid response, therefore, the sample is large enough for the following analysis to be reliable. Exactly half of the participants (n=78) were assigned to the exercises label condition and the other half to the control condition.

To ensure that the randomly assigned participants are not systematically different across the two conditions with regards to the demographic variables, an ANOVA and a Person’s Chi-squared test were performed. The ANOVA was performed for the metric demographic variables such as age, weight, and BMI and for the remaining non-metric variable, gender, the Person’s Chi-squared test was performed. Additionally, the variable nutrition knowledge was tested as well in order to check that the groups did not differ significantly on this measurement either. The null hypothesis for these tests, there is no significant difference between groups, cannot be rejected (p > .05) for the variables age (p=.262), weight (p=.662), BMI (.364), gender (p=.151) and nutrition knowledge (p=.966). Therefore, it can be concluded that there is no significant difference between the participant’s characteristics in the study’s two conditions (Table 1).

ANOVA F Sig. Age 1.265 .262 Weight .192 .662 BMI .831 .364 Nutrition Knowledge .002 .966

Person’s Chi-squared test

Value Sig.

Gender 2.601 .151

(29)

Normality test

In order to gain a better insight into the distribution of data, normality tests were conducted for the dependent variables (caloric consumption, attitude towards running and future sporting intentions) and the continuous moderator (the type of food). To assess the normality of these variables two tests have been used, the Kolmogorov-Smirnov test and the Shapiro-Wilk test. In order to move forward with the main analysis, the assumption of normality needs to be met (Friedman, 1939). The test has yielded significant results (p<.05) for all the tested variables. Thus, it can be concluded that the null hypothesis for a normal distribution can be rejected.

Kolmogrorov-Smirnov Shapiro-Wilk

Statistic Sig. Statistic Sig

Caloric .134 .000 .930 .000 consumption Attitude towards .129 .000 .912 .000 running Future sporting .157 .000 .892 .000 intentions Vice food .141 .000 .901 .000 Virtue food .060 .200 .972 .003

Table 2 Normality tests for the dependent variables and for the moderator

One of the final tests before diving in the main analysis is the test of homogeneity of variances to the dependent variables. The test applied was Levene’s test of homogeneity of variances which lead to a significant result for the depended variable Caloric consumption (p=.006) and non-significant results for the other two dependent variables, Future porting intention (p=.820) and Attitude towards running (p=.291). Hence, the null hypothesis can only be rejected for the last two dependent variables.

Factor analysis and reliability test

(30)

internal consistency for this scale. Following this analysis, a similar one was conducted to determine if the three items of the nutrition knowledge scale load on one factor as well. A second factor analysis and a second reliability analysis were performed. The KMO is 0.715, which means that the factor analysis is appropriate. The value of Barlett’s test of sphericity is 0.000 and one component was recommended based on the criteria of the total variance explained and the inspection of the scree plot (Malhotra, 2008). The Cronbach’s alpha is larger than 0.6 (p=.881) indicating a high level of internal consistency for this scale. The final scores for both variables were determined by calculating the means of the respective items.

Hypothesis testing

In the following section, different analyses were conducted in order to investigate the hypothesis established in the literature section. Before diving into the analysis, it is worth to give a brief description of the main variables.

During the period the survey was active a total of 156 participants completed the entire experiment. The average caloric consumption for both conditions was 2630 (SD 507) calories and for each condition, separately the average caloric consumptions were 1960 (SD 451) calories for vice foods and 725 (SD 126) calories for virtue foods. For the remaining independent variables, attitude towards running future sporting intention the average scores were 5 (SD 1) and respectively 4 (SD 3).

The effect of exercise labels on the healthiness of the shopping basket

(31)

Condition N Median rank Control condition 78 90.42 Exercise label 78 66.58 condition Total 156

Table 3 Median calorie rank of conditions

The conclusion that can be drawn from this analysis is that exercise labels reduce the number of calories purchased by participants when comparing it to the normal caloric label present in supermarkets. Hence, hypothesis 1, presenting the consumer with an exercise label leads to a reduction in calories purchased, can be confirmed.1

The effect of exercise labels on the healthiness of the shopping basket depending on food type

Even though the exercise labels had a significant impact on the healthiness of the shopping basket in comparison to the caloric label, there is interest in testing whether this effect differs between the two types of food (vice versus virtue) across the two label conditions (caloric label versus exercise label). Hence, the next step is testing the second hypothesis, which states that exercise labels have a greater impact on the healthiness of the shopping basket when presented on vice foods compared to virtue foods.

As already stated in the previous chapters the dependent variables and the continuous moderator were tested for a normal distribution across the independent variable. In order to test the second hypothesis, a mixed repeated-measures ANOVA was selected as the appropriate test. Even though the distribution is not normal, the ANOVA test is robust to moderate departures from the normality when the sample sizes are reasonably large and equal (Winner, Brown and Michels, 1991) and moderate departures from normality are not a reason for concern if the population studied is homogenous in form (Kirk, 1968). As formerly, presented in the Random

1_{A one way ANOVA was executed with the purpose of delivering a further investigation on the matter.}_{An analysis}

(32)

assignment chapter, the sample size is large and equally divided between the two conditions and the populations in each condition do not differ significantly. Prior to analyzing the results of the mixed design repeated measures ANOVA test, the last assumption, sphericity, was verified. As the data violated the assumption of sphericity the analysis will be conducted with a Greenhouse-Geisser correction. There was a significant main effect of the food type, F (1,154) =1263, 76, p<.005, Ƞp2= .90. This significant effect means that if the condition in which the participant is assigned to is ignored, one type of food leads to a significantly different calorie consumption than the other one.

Source Df SS MS F p Type of food 1 118944086,2 118944086,2 1263,37 .000 (Vice vs Virtue)

Table 4 Main effect of the food type

Post hoc tests using the Bonferroni correction revealed that the type of food elicited a reduction in calorie consumption (1960.27 ± 451.06 calories vs 725.3974±126.68924, respectively), which was found statistically significant (p <.05). Therefore, we can conclude that a different type of food elicits a statistically significant reduction in calorie consumption. This does not come as a surprise since vice foods are higher in calories than virtue foods.

The main effect of the label condition (Caloric label vs Exercise label) is listed separately. Before being able to check if the effect is significant it is important to check the assumption for homogeneity of variance for each variable of the repeated measures variables. The Levene’s test indicated that variance is homogenous for only one level of the repeated measures ANOVA variables, therefore, the following results should be interpreted with caution Levene’s test showed that the variance for vice foods was not equal F(1,153)=5,49, p=.02. (view Table X ). There was a significant main effect of the label condition, F (1,154) = 4, 22, p=.041, Ƞp2=. 027.

(33)

Estimated Mariginal Means of Calorie Consumption between label Conditions and with-in Type of Food

2500 2000 1500 1000 500 0

Vice food Virtue food

Caloric label condition Excericse label condition

Figure 1 Estimated Marginal Means of Calorie Consumption between label Conditions and with-in Type of Food

The effect of the exercise label on participant’s future sporting intentions

In order to analyze the third hypothesis, which states that the exercise label will negatively affect the participant’s attitude towards running, a one-way ANOVA was performed with label condition as independent variable and attitude towards running as a dependent variable. Even though the dependent variable is not normally distributed between the conditions of the independent variable, as previously stated ANOVA is fairly robust to moderate departures from normality when the sample sizes are reasonably large, equal and the populations are homogenous in form (Winner, Brown and Michels, 1991; Kirk, 1968). Homogeneity of variance was previously tested and Levene’s test showed that the variances for attitude towards running were equal, F(1, 142) = ,409, p=.950.

(34)

Estimated Marginal Means of Number of Hours dedicated to Future Sporting Intentions in the upcoming Week

6 5 4 3 2 1

Caloric label condition Exercise label condition Figure 2 Estimated Marginal Means of Number of Hours dedicated to Future

Sporting Intentions in the upcoming Week

The effect of the exercise label on participant’s attitude towards running

In order to analyze the forth hypothesis, which states that the exercise label will positively affect the participant’s future sporting intention, a Mann-Whitney U test was performed since the dependent variable, attitude towards sporting is not normally distributed nor did it pass the homogeneity of variance test. Homogeneity of variance was previously tested and Levene’s test showed that the variances for attitude towards sporting were not equal, F (31,106) = 3. 965, p=.000.

The test indicates that attitude towards sporting in the control condition (Mdn=82,40) was not significantly different than for the exercise label condition (Mdn=74, 60), U=2737, 5,

p=.280.

Further analysis

(35)

people to purchase products higher in calories. The variable Level of hunger was created with three different options (Very hungry, Hungry and Full). This was done by dividing the continuous variable Hunger into three equal levels based on range picked by the participant (for example between 0% and 30% the level of hunger is Very hungry). Before any test was run, the normal distribution of data was checked and the dependent variable was not normally distributed between the conditions of the independent variable Level of hunger.

Levels of hunger Kolmogorv-Smirnov Shapiro-Wilk

Statistic Sig Statistic Sig.

Very hungry .129 .200 .920 .059

Hungry .127 .001 .939 .000

Full .140 .065 .900 .003

Table 5 Normality test of the independent variable Level of hunger

(36)

Another important variable that could have influenced participants on their shopping trip is whether they were on diet or not. As before the normal distribution of data and homogeneity of variance were tested before conducting any future tests. The dependent variable, Caloric consumption, does not present a normal distribution or homogeneity of variance across the two conditions of the dependent variable Diet (F (1,154) =4,941, p=.028).

Diet Kolmogorv-Smirnov Shapiro-Wilk

Statistic Sig Statistic Sig.

Following a diet .171 .126 .915 .079 Not following a .135 .000 .932 .000 diet

Table 6 Normality test of the dependent variable Diet

A Mann-Whitney U test showed that there was no statistically significant difference in calories purchased between the people who were following a diet and the people who were not following a diet ( U=1220, p=.458).

Discussion

The effect of exercise labels has been proven to have a positive influence on the healthiness of the consumer’s choices (Dowray et al., 2013) but whether its effect on consumer choice differs based on the type of product it is attributed to has not received much interest. The main goal of this empirical study was to examine how consumer shopping behavior is affected by nutritional information received in the form of the novel exercise label and if the effect differs based on the type of food. The last chapter of this study will discuss the outcome of the analysis as well as some managerial implications.

General discussion

(37)

lowering the number of calories purchased. Furthermore, it was of great interest whether the exercise labels yielded different effects on the basket healthiness base on the type of food (vice versus virtue) it was attributed to. This study was also concerned with the possible effect the label might have on the consumer’s attitude towards running (the sport depicted on the label) and any effect it might have consumer’s future sporting intention.

While the analysis on the exercise label’s effect on basket healthiness, as hypothesized, was significant, its impact of attitude towards running and future sporting intention could not be proven significant. Moreover, the interaction between the exercise label and type of food was not proven significant. The following table provides an overview of the tested hypothesis.

Results of the hypothesis testing

Hypothesis Sig. Results

H1 Exercise labels have positive effect on the basket’s healthiness .001 Supported

(compared to normal caloric label)

H2 Exercise labels have different effects on calorie consumption .254 Not based on the type of food they are applied to Supported

H3 Exercise labels have a negative effect on consumer’s attitude .554 Not

towards running Supported

H4 Exercise labels have a positive effect on consumer’s future .502 Not

sporting intention Supported

Table 7 Results of the hypothesis analysis

(38)

and Volpp, 2010). Therefore, the exercise label with its understandable format has helped participants make individual healthy choices.

The fact that the exercise label had no effect on calorie consumption based on the type of food it was applied could be because people believe that virtue food is generally less fulfilling than vice food (Chandon and Wansink, 2007). Further research has confirmed the negative relationship between perceived healthiness of food and perceived calorie count (Finkelstein and Fishback, 2010). Consequently, if the consumer does not necessarily pay close attention to the nutrition label and perceives an assortment of food as healthy he/she can choose a high-calorie product without being aware. Similar to this line of thinking, if the food presented to the consumer triggers fitness cues or is associated in any matter in the consumer’s mind with physical activity than he/she will consume the product in greater volume (Koenigstorfer and Baumgartner, 2016). In the case of exercise label where the information is presented in a more striking format (as minutes required running in order to burn calories), if they view the label as a recommendation rather than a food selection guide (Fitch, 2009) consumers will still make a choice based on their perception of the product.

Lastly, this study reveals valuable information regarding the effect of the exercise label on consumers’ sporting intention and attitude towards running. There were no significant differences on these two matters between participants in the control conditions and participants in the exercise label condition. This finding might be explained by the previously mentioned statement where consumers do not view the exercise label as more than a recommendation (Fitch, 2009). Moreover, consistent with the literature review, consumers might see the exercise label more as a scare tactic than as a helpful nudge and therefore go against it. Other findings that could have downplayed the effectiveness of the label is that some consumers believe the PACE label is not generalizable to everybody and the activity portrayed is more than they habitually do (Fitch, 2009; Swartz et al, 2013).

(39)

consumers to shop healthy and it does not affect their view on the particular sport portrayed on the label.

Contribution to theory

The study at hand contributes to the body of existing research on exercise labels and consumer well-being. Previous studies have revealed weak and inconsistent proof that normal calorie labels lead consumers to make healthier, lower-calorie food choices (Swartz et al., 2013; Harnack and French, 2008). Considering these findings, different approaches of swaying consumers in the right direction have been deployed. One such approach is the new front-of-pack labelling format, the exercise label. The current research, however, fills a knowledge gap at the product level, since the previous studies have been generally conducted in a restaurant, fast-food restaurants or cafeterias. The findings of this study focus on fast-moving consumer goods that can be bought in any regular supermarket. The exercise label format had a positive effect on the healthiness of the consumer’s shopping basket without affecting the attitude he or she might have about the sport portrayed on the label. These results are reassuring since one of the main fears experts have about the exercise label is that it might negatively affect consumer’s view on sporting and their sporting levels.

Managerial implications

The current study provides marketers and policymakers who are seeking to combat the growing worldwide obesity problem with new insights on the new labelling approach, exercise label or otherwise known as physical activity based nutrition label.

(40)

Limitations and future research

An important limitation that needs to be taken into consideration in this study is the fact that it has been conducted in form of an online survey which can lead to a couple of shortcomings. Even taking into account that participants were asked to imagine they were experiencing an actual shopping trip, the fact that they were behind a computer could have an effect on their purchasing behaviour (Sheeran, 2005). Furthermore, the participants received a limited 14 categories assortment and a restrictive, but as close as possible, amount of in-shop information. This is why the first proposal for further research is an experiment conducted in a real-life environment in order to increase the external validity of the research (Kneale et

al.,2018). Another limitation pertaining to the shopping experience was the fact that participants

did not receive a no-choice option. No assumptions can be made whether participants selected an option because they could not opt-out or they selected it because they actually choose it.

The experiment part of the study ran only for one week, leading to a time limitation. It remains up to investigation whether the exercise label’s effect will evolve over time. Further research should, across a longitudinal study, check whether the effect of the label becomes more salient with time or will consumers become accustom with it and begin to ignore it.

(41)

Conclusion

(42)

Reference

• About Health Star Ratings. (2018, August 29). Retrieved from

http://healthstarrating.gov.au/internet/healthstarrating/publishing.nsf/content/about-health-stars

04/10/2018

• Antonelli, R., & Viera, A. J. (2015). Potential Effect of Physical Activity Calorie Equivalent (PACE) Labeling on Adult Fast Food Ordering and Exercise. Plos One,10(7). doi:10.1371/journal.pone.0134289

• Austin, E. J., Deary, I. J., Gibson, G. J., Mcgregor, M. J., & Dent, J. (1998). Individual response spread in self-report scales: Personality correlations and consequences. Personality and

Individual Differences,24(3), 421-438. doi:10.1016/s0191-8869(97)00175-x

• Australian Medical Association. (2011). Traffic Light Food Labelling The Evidence - Logo. Retrieved from https://ama.com.au/system/tdf/documents/Traffic Light Food Labelling - The Evidence.pdf?file=1&type=node&id=36221 10/10/2018

• Bassett, M. T., Dumanovsky, T., Huang, C., Silver, L. D., Young, C., Nonas, C., . . . Frieden, T. R. (2008). Purchasing Behavior and Calorie Information at Fast-Food Chains in New York City, 2007. American Journal of Public Health,98(8), 1457-1459. doi:10.2105/ajph.2008.135020 • Blair, S. N., Marcus, B., Jaret, P. E., Carpenter, R. A., & Dunn, A. L. (2011). Active living every

day. Champaign, IL: Human Kinetics.

• Bleich, S. (2011). If '250 Calorie' Label Doesn't Stop You, '50 Minute Jog' Label Might. Fox News. 16th December 2011. Retrieved from http://www.foxnews.com/health/2011/12/16/if-250-calorie-label-doesnt-stop-50-minute-jog-label-might/#ixzz2OmNwPcXv on the 02/10/18

• Bleich, S. N., Herring, B. J., Flagg, D. D., & Gary-Webb, T. L. (2012). Reduction in Purchases of Sugar-Sweetened Beverages Among Low-Income Black Adolescents After Exposure to Caloric Information. American Journal of Public Health,102(2), 329-335. doi:10.2105/ajph.2011.300350 • Bleich, S. N., Herring, B. J., Flagg, D. D., & Gary-Webb, T. L. (2012). Reduction in Purchases of

Sugar-Sweetened Beverages Among Low-Income Black Adolescents After Exposure to Caloric Information. American Journal of Public Health,102(2), 329-335. doi:10.2105/ajph.2011.300350 • Bleich, S. N., Herring, B. J., Flagg, D. D., & Gary-Webb, T. L. (2012). Reduction in Purchases of

Sugar-Sweetened Beverages Among Low-Income Black Adolescents After Exposure to Caloric Information. American Journal of Public Health,102(2), 329-335. doi:10.2105/ajph.2011.300350 • Blumberg, B., Cooper, D., & Schindler, P. (2014). Business Research Methods (4th ed.). London:

(43)

• Blumenthal, K., & Volpp, K. G. (2010). Enhancing the Effectiveness of Food Labeling in Restaurants. Jama,303(6), 553. doi:10.1001/jama.2010.85

• Borgmeier I. and Westenhoefer J.( 2009). ‘Impact of different food label formats on healthiness evaluation and food choices of consumers: A randomised controlled study’. BMC Public Health. Vol 9(1), pp.184

• Brucks, M. (1985). The Effects of Product Class Knowledge on Information Search Behavior.

Journal of Consumer Research,12(1), 1. doi:10.1086/209031

• Burton-Bradley, R. (2012). Sugar trap: Critics say Healthy Tick deal too sweet for Heart Foundation to ignore. News.com. 29th January 2012. Retrieved from

http://www.news.com.au/business/sugar-trap-heart-foundation-critics-say-tick-deals-with-nestle-uncle-tobys-mcdonalds-aldi-too-sweet-to-ignore/story on the 04/10/2018

• Bussell, G. (2005). Nutritional profiling vs guideline daily amounts as a means of helping consumers make appropriate food choices. Nutrition & Food Science,35(5), 337-343. doi:10.1108/00346650510625548

• Campos, S., Doxey, J., & Hammond, D. (2011). Nutrition labels on pre-packaged foods: A systematic review. Public Health Nutrition,14(08), 1496-1506. doi:10.1017/s1368980010003290 • Chance, Z., Gorlin, M., & Dhar, R. (2014). Why Choosing Healthy Foods is Hard, and How to

Help: Presenting the 4Ps Framework for Behavior Change. Customer Needs and Solutions,1(4), 253-262. doi:10.1007/s40547-014-0025-9

• Chandon, P. and Wansink, B. (2007), “Is obesity caused by calorie underestimation? A psychophycal model of meal size estimation”, Journal of Marketing Research, Vol. 44 No. 1, pp 84-99

• Chappman, K. and Watson, W. (2013). Out with traffic lights, in with stars – next steps for food labelling by Kathy Chapman. Cancer Society. Retrieved from

http://www.cancercouncil.com.au/65841/news-media/blog/healthy-living-cancer-prevention on the 04/10/2018

• Choice. (2011). Daily intake percentage labels discredited. 28th of November 2011. Retrieved from

http://www.choice.com.au/consumer-action/food-labelling/better-food-labelling-campaign/percentage-daily-intake-guides-discredited.aspx on the 04/10/2018

• Cowburn, G., & Stockley, L. (2005). Consumer understanding and use of nutrition labelling: A systematic review. Public Health Nutrition,8(01). doi:10.1079/phn2004666

• Cox, D. N., Anderson, A. S., Reynolds, J., Mckellar, S., Lean, M. E., & Mela, D. J. (1998). Take Five, a nutrition education intervention to increase fruit and vegetable intakes: Impact on consumer choice and nutrient intakes. British Journal of Nutrition,80(02), 123. doi:10.1017/s0007114598001020

• Dixon, W. J. (1960). Simplified Estimation from Censored Normal Samples. The Annals of

(44)

• Downs, J. S., Loewenstein, G., & Wisdom, J. (2009). Strategies for Promoting Healthier Food Choices. American Economic Review,99(2), 159-164. doi:10.1257/aer.99.2.159

• Dowray, S., Swartz, J. J., Braxton, D., & Viera, A. J. (2013). Potential effect of physical activity based menu labels on the calorie content of selected fast food meals. Appetite,62, 173-181. doi:10.1016/j.appet.2012.11.013

• European Food Information Council (EUFIC) (2005). Nutrition Information and Food Labelling. EUFIC Forum No. 2. Brussels: EUFIC

• Fagarasanu, M., & Kumar, S. (2002). Measurement instruments and data collection: A consideration of constructs and biases in ergonomics research. International Journal of Industrial

Ergonomics,30(6), 355-369. doi:10.1016/s0169-8141(02)00101-4

• Finkelstein, S. R., & Fishbach, A. (2008). When Healthy Food Makes You Hungry. PsycEXTRA

Dataset. doi:10.1037/e722352011-097

• Fitch, R. C., Harnack, L. J., Neumark-Sztainer, D. R., Story, M. T., French, S. A., Oakes, J. M., & Rydell, S. A. (2009). Providing Calorie Information on Fast-Food Restaurant Menu Boards: Consumer Views. American Journal of Health Promotion,24(2), 129-132. doi:10.4278/ajhp.08031426

• Friedman, M. (1939). A Correction: The Use of Ranks to Avoid the Assumption of Normality Implicit in the Analysis of Variance. Journal of the American Statistical Association,34(205), 109. doi:10.2307/2279169

• Gorton, D., Mhurchu, C. N., Chen, M., & Dixon, R. (2008). Nutrition labels: A survey of use, understanding and preferences among ethnically diverse shoppers in New Zealand. Public Health

Nutrition,12(09), 1359. doi:10.1017/s1368980008004059

• Grunert, K. G., Wills, J. M., & Fernández-Celemín, L. (2010). Nutrition knowledge, and use and understanding of nutrition information on food labels among consumers in the UK.

Appetite,55(2), 177-189. doi:10.1016/j.appet.2010.05.045

• Hamlin, R. P., Mcneill, L. S., & Moore, V. (2015). The impact of front-of-pack nutrition labels on consumer product evaluation and choice: An experimental study. Public Health

Nutrition,18(12), 2126-2134. doi:10.1017/s1368980014002997

• Harnack, L. J., & French, S. A. (2008). Effect of point-of-purchase calorie labeling on restaurant and cafeteria food choices: A review of the literature. International Journal of Behavioral

Nutrition and Physical Activity,5(1), 51. doi:10.1186/1479-5868-5-51

• Hartt, A. (2015, April 21). What Is Third-Party Certification? Retrieved from

https://www.naturalproductsinsider.com/labeling/what-third-party-certification on 04/10/2018 • Hayes, A. F. (2013). Introduction to mediation, moderation, and conditional process analysis:

(45)

• Health, D. (2013a). In D. o Health (Ed.), Public health responsibility deal: F1. Out of home energy (kJ/Kcal) labelling.

• Heart Foundation Tick. (2012). Retrieved from http://www.heartfoundation.org.nz/healthy-living/healthy-eating/heart-foundation-tick on the 04/10/2018

• Hill, J.O., Wyatt, H.R., and Peters, J.C. (2012). Energy balance and obesity. Circulation, 126, 126-132

• Hruby, A., & Hu, F. B. (2014). The Epidemiology of Obesity: A Big Picture.

PharmacoEconomics,33(7), 673-689. doi:10.1007/s40273-014-0243-x

• Hui, S. K., Bradlow, E., & Fader, P. (2007). An Integrated Model of Grocery Store

Shopping Path and Purchase Behavior. SSRN Electronic Journal. doi:10.2139/ssrn.960960 • Iles, I. A., Nan, X., & Verrill, L. (2017). Nutrient Content Claims: How They Impact Perceived

Healthfulness of Fortified Snack Foods and the Moderating Effects of Nutrition Facts Labels.

Health Communication,33(10), 1308-1316. doi:10.1080/10410236.2017.1351277

• Kelly B., Hughes C., Chapman K., Dixon L.J. and King L. (2008). Front-of-Pack Food Labelling: Traffic Light Labelling gets the green light. Cancer Council, Sydney. Available from:

http://www.cancercouncil.com.au/html/prevention/healthyeating/downloads/foodlabelling_front ofpack_surveyreport.pdf

• Kirk, R. E. (1968). Experimental Design:Procedures for the Behavioral Science a battle line. Belmont, CA.: Brook/Cole Publishing.

• Kiszko, K. M., Martinez, O. D., Abrams, C., & Elbel, B. (2014). The Influence of Calorie Labeling on Food Orders and Consumption: A Review of the Literature. Journal of Community

Health,39(6), 1248-1269. doi:10.1007/s10900-014-9876-0

• Kivetz, R., & Simonson, I. (2002). Earning the Right to Indulge: Effort as a Determinant of Customer Preferences toward Frequency Program Rewards. Journal of Marketing

Research,39(2), 155-170. doi:10.1509/jmkr.39.2.155.19084

• Kivetz, R., Urminsky, O., & Zheng, Y. (2006). The Goal-Gradient Hypothesis Resurrected: Purchase Acceleration, Illusionary Goal Progress, and Customer Retention. Journal of Marketing

Research,43(1), 39-58. doi:10.1509/jmkr.43.1.39

• Kneale, D., Thomas, J., Omara-Eves, A., & Wiggins, R. (2018). How can additional secondary data analysis of observational data enhance the generalisability of meta-analytic evidence for local public health decision making? Research Synthesis Methods. doi:10.1002/jrsm.1320.

• Koenigstorfer, J., & Baumgartner, H. (2016). The Effect of Fitness Branding on Restrained Eaters’ Food Consumption and Postconsumption Physical Activity. Journal of Marketing

Research,53(1), 124-138. doi:10.1509/jmr.12.0429

• Kristal, A.R., Levy, L., Patterson, R.E., Li, S.S. and White, E. (1998) Trends in food label use associated with new nutrition labelling regulations. American Journal of Public Health, 88(8),