The effect of taxing unhealthy food items and healthy eating calls in the form of a descriptive norm on the

EFFECTIVELY PROMOTING HEALTHY FOOD CHOICES

The effect of taxing unhealthy food items and healthy eating calls in the form of a descriptive norm on the

healthiness of food purchases

Romy Madelon Elshof

June 16, 2019

CALLS IN THE FORM OF A DESCRIPTIVE NORM ON THE HEALTHINESS OF FOOD PURCHASES

Master Thesis, MSc Marketing Management

University of Groningen, Faculty of Economics and Business

June 16, 2019

Author:

Romy Madelon Elshof

Student number: 2768623

Oude Ebbingestraat 82a

9712 HM Groningen

r.m.elshof@student.rug.nl

Supervisor:

M. van der Heide, PhD

Second evaluator:

Prof. dr. L.M. Sloot

RUNNING HEAD: EFFECTIVELY PROMOTING HEALTHY FOOD CHOICES II

ABSTRACT

With the substantial increases in the prevalence of overweight and obesity and its severe consequences for individuals and society, it is crucial to direct consumers towards interchanging unhealthier options for healthier alternatives. This study looks at the effect of taxing unhealthy food items and healthy eating calls in the form of a descriptive norm on the healthiness of the purchases in the target categories and the overall shopping basket. To examine the predictions, participants were recruited via Amazon Mechanical Turk (N = 269) and were asked to do online groceries in 25 categories. Specifically, I expected that taxing unhealthy food items increases the healthiness of the purchases in the target categories and the overall shopping basket. Contrary to the predictions, the results revealed no support. Moreover, I proposed and demonstrated that healthy eating calls in the form of a descriptive norm increase the healthiness of the purchases in the target categories and the overall shopping basket. Lastly, I proposed that the effect of taxing unhealthy food items on the healthiness of the purchases in the target categories and the overall shopping basket is more pronounced in the presence of healthy eating calls in the form of a descriptive norm. Contrary to the predictions, the results revealed no support. This research offers some important theoretical and practical implications, and future research directions.

Keywords: overweight and obesity rates, food pricing strategies, nudging, healthy food

choices, taxing unhealthy food, healthy eating calls, descriptive norms, self-regulatory focus

theory.

PREFACE

When my grandparents visited North America in the 1980s, it was the first time that they were confronted with obese people. In 2018, shockingly, in the Netherlands, 43.9% of the population suffered from overweight and obesity (Centraal Bureau voor de Statistiek, 2018b).

The consequences of overweight and obesity are detrimental. Individuals who are overweight or obese are at higher risk for cardiovascular diseases, type 2 diabetes, certain types of cancer, osteoarthritis, asthma, and other complications (Finer, 2005; Hruby et al., 2016; Pi-Sunyer, 1991; Visscher & Seidell, 2001; World Health Organization, 2018b). During the courses Marketing & Consumer Well-being and Consumer Psychology, I was inspired by how we can use marketing to help consumers to make better food decisions. Although food marketing is often seen as one of the main causes of the obesity epidemic, food marketing has the potential to help consumers to eat healthier (Chandon & Wansink, 2012). For this reason, I decided to write my thesis about effectively promoting healthy food choices to combat overweight and obesity rates.

First of all, I would like to thank my supervisor Martine van der Heide for her enthusiasm and great supervision during the development of my thesis. As this thesis also means the end of my study, I would like to thank my family for always having my back and providing me with great support. Moreover, I appreciated my friends who were a welcome distraction from sometimes stressful times and thanks to them for the joyful times during my study.

I sincerely hope you will enjoy reading my thesis.

Romy Elshof

EFFECTIVELY PROMOTING HEALTHY FOOD CHOICES

TABLE OF CONTENTS

Abstract... II Preface ... III

Introduction ... 1

Theoretical Background ... 4

The Healthiness of Food Choices ... 4

Food Pricing Strategies ... 4

Healthy Eating Calls in the Form of a Descriptive Norm ... 6

The Moderating effect of Healthy Eating Calls in the form of a Descriptive Norm ... 7

Method ... 9

Participants ... 9

Study Design and Intervention ... 9

Procedure ... 9

Measures... 10

Results ... 12

Preparation for Data Analysis ... 12

Descriptive Statistics and Correlations ... 14

Test for Differences between the Conditions ... 15

Hypotheses Testing ... 16

Follow-up Analysis - Covariates ... 18

Follow-up Analysis - The Moderating Role of Self-Regulatory Focus ... 19

Follow-up Analysis - The Moderating Role of Health Consciousness ... 23

Discussion... 25

Summary ... 25

Theoretical Implications ... 26

Limitations and Future Research ... 27

Practical Implications ... 29

Conclusion ... 30

References ... 31

Notes ... 40

Table of Contents - Appendices ... 41

Appendices ... 42

INTRODUCTION

Globally, in 2016, more than 1.9 billion adults were overweight, from which 650 million adults were obese (World Health Organization, WHO, 2018b). According to the Body Mass Index (BMI; person’s weight in kilograms divided by the square of his height in meters), someone is overweight with a BMI higher or equal to 25, and someone is obese with a BMI higher than or equal to 30 (WHO, 2018b). Considering the Netherlands, in 2018, 43.9% of the population suffered from overweight and obesity, which is a major increase in comparison to a rate of 27.4% in 1981 (Centraal Bureau voor de Statistiek, CBS, 2018b). On a global level, researchers even expect overweight and obesity rates to increase in the next decades and there seems to be an obesity epidemic (Finkelstein et al., 2012; Ng et al., 2014).

Individuals who are overweight or obese are at higher risk for cardiovascular diseases, type 2 diabetes, certain types of cancer, osteoarthritis, asthma, and other complications (Finer, 2005; Hruby et al., 2016; Pi-Sunyer, 1991; Visscher & Seidell, 2001; WHO, 2018). These health issues negatively affect the life expectancy of these individuals (Hruby et al., 2016;

Olshansky et al., 2005). Besides the health-related issues, obese individuals are more likely to experience discrimination in education, work, health care, and social relationships (Finer, 2015). For example, obese women are likely to experience discrimination in educational environments since they are about 65% less likely to be accepted by their first-choice college than thinner students (Finer, 2015). In addition to personal well-being, overweight and obesity issues result in major medical costs for society, with an estimation of 9% of annual medical expenditures, leading to an expense of $147 billion per year (Finkelstein, Trogdon, Cohen, &

Dietz, 2009).

These overweight and obesity issues are primarily caused by the overconsumption of calories (Downs, Loewenstein, & Wisdom, 2009; Rolls, 2003; Wallinga, 2010; WHO, 2018;

Wright & Aronne, 2012). The United States Department of Agriculture (USDA) showed that

the average daily calorie intake of American consumers increased with 600 calories from 1970

to 2007 (Wallinga, 2010). According to researchers of the USDA, this increase in calorie intake

is the main cause behind the increasing obesity rates (Wallinga, 2010). Therefore, to decrease

calorie intake, it is crucial “to direct consumers towards interchanging unhealthier options for

healthier alternatives” (Waterlander, Steenhuis, de Boer, Schuit, & Seidell, 2013: 66). For

example, consumers can be directed towards healthier options by food pricing strategies and

packaging (Asfaw, 2011; Chandon & Wansink, 2012; Waterlander et al., 2013). As globally

most food purchases are made in the supermarket, developing interventions that positively

EFFECTIVELY PROMOTING HEALTHY FOOD CHOICES 2

influence the healthiness of food purchases in the supermarket may have a significant impact on public health (Glanz, Bader, & Iyer, 2012; Payne, Niculescu, Just & Kelly, 2014; Regmi &

Gehlhar, 2005). Hence, the purpose of this study is to investigate how consumers can be directed towards interchanging unhealthier options for healthier alternatives in the setting of a supermarket.

Specifically, this study considers both the effect of taxing unhealthy food items and healthy eating calls in the form of a descriptive norm on the healthiness of the purchases in the target categories and the overall shopping basket (see Figure 1). First of all, I propose that taxing unhealthy food items increases the healthiness of the purchases in the target categories (i.e., categories with taxes) and the overall shopping basket. This is because of positive cross-price elasticity; although people are still able to spend the money on the unhealthy food items, they are likely to willfully choose not to buy these foods as they think it is not worth the high price (Epstein, Dearing, Rob, & Finkelstein, 2010; Nederkoorn, Havermans, Giesen, & Jansen, 2011). Moreover, I propose that healthy eating calls in the form of a descriptive norm increase the healthiness of the purchases in the target categories (i.e., categories with healthy eating calls) and the overall shopping basket. Consumers are likely to choose the option with the healthy eating call in the form of a descriptive norm since these calls are used as a social proof heuristic and simplify the decision-making process (Salmon, de Vet, Adriaanse, Fennis, Veltkamp, & de Ridder, 2015). Lastly, I propose that synergy effects may exist when taxing unhealthy food items and healthy eating calls in the form of a descriptive norm are introduced together. This is because the healthy options are likely to become more salient and these interventions are together aiming at consumers with promotion-focused self-regulation (i.e., more focused on behaving towards positive outcomes) and prevention-focused self-regulation (i.e., more focused on avoiding negative outcomes; Higgens, 1997; Higgens, 1998).

By testing these predictions, my research strives to make several contributions. First, this study considers both a pricing and a nudging strategy, in line with the call for researchers to “directly compare nudges and economic incentives and see if they can complement each other” (Cadario & Chandon, 2018: 36). Relatedly, according to Ni Mhurchu, Blakely, Jiang, Eyles and Rodgers (2010: 746), “further work is needed to determine how to augment the effect of pricing strategies on healthier food purchases to achieve a corresponding positive effect on nutrient purchases.” Moreover, this is in line with prior research of Giesen, Payne, Havermans, and Jansen (2011) and Nederkoorn et al. (2011) who argue that it is crucial to investigate taxes on unhealthy food items in combination with other strategies to increase healthy food choices.

Furthermore, although research showed the potential of healthy eating calls in the setting of a

self-service restaurant (van Kleef, van den Broek, & van Trijp, 2015), cafeteria-type restaurant (Policastro, Palm, Schwartz, & Chapman 2017), and in workplace restaurants (Thomas et al., 2017), little research has been carried out to investigate the potential of healthy eating calls in supermarkets. Lastly, this research considers healthy eating calls in the form of a descriptive norm which is in line with the request for researchers to further test the potential of the social norm approach to increase healthy food choices (Thomas et al., 2017). Practically, my research may help organizations, specifically supermarkets, to use implications of this study to stimulate consumers to make healthier food choices which have a meaningful impact on consumer’s well- being.

Next, I will discuss an overview of underlying theories, which leads to the formulation of the hypotheses. Subsequently, the participants, study design and intervention, procedure, and measures are described in the methodology section. This is followed by the results section in which the results of the data analyses are discussed. After that, a summary, theoretical and practical implications, and limitations and future research are provided in the discussion section. Lastly, I will describe the conclusion of this paper.

FIGURE 1

Conceptual Model

EFFECTIVELY PROMOTING HEALTHY FOOD CHOICES 4

THEORETICAL BACKGROUND The Healthiness of Food Choices

Although people often have good intentions to make healthy food choices, “good intentions are thwarted by non-conscious buying patterns and factors such as familiarity, taste expediency, price and how options are presented” (Lassen et al., 2016: 124). For example, food choices are influenced by unhealthy eating habits that are often based on a simple consideration of emotional cues and difficult to change (Kidwell, Hasford, & Hardesty, 2015). Moreover, the socialization process of the family is an essential determinant for food choices since parents primarily influence the diet pattern of their children (Moore, Wilkie, & Desrochers, 2017).

Lastly, consumers’ ability to integrate, remember and understand information is limited and influences the subsequent decisions that are made (Ariely, 2000). Therefore, showing nutritional information does not show the desired effect on consumers’ choices as it is difficult for consumers to interpret the information correctly (Block & Peracchio, 2006; Grunert, Wills,

& Fernandez-Celemin, 2010; Kisko, Martinez, Abrams, & Elbel, 2014). As a consequence, research has focused on investigating interventions to effectively promote healthy food choices.

Interventions like food pricing strategies and easy-to-interpret nutritional information may have the potential to help consumers to interchange unhealthier options for healthier alternatives.

Food Pricing Strategies

In general, healthy food items which are nutrient-rich and low energy-dense (e.g., fresh vegetables and fruit) are relatively more expensive than high energy-dense unhealthy food items with added sugars and added fats (Drewnowski & Darmon, 2005; Drewnowski, 2010;

Finkelstein, Ruhm, & Kosas, 2005; Waterlander et al., 2010). For example, in the Netherlands, the price of healthy food items increased with 22% relative to 13% of unhealthy food items in the past ten years (CBS, 2018a). Similarly, Finkelstein et al. (2005) argued that the price increases in unhealthy food items are substantially smaller than those of healthy food items.

Hence, consumers are prone to choose for the unhealthy alternative based on price and are likely to have an unhealthier diet (Drewnowski, 2004; Drewnowski, 2010; Drewnowski & Darmon, 2005).

Looking at food pricing strategies to stimulate healthy food choices, supermarkets can

choose for taxing or subsidizing strategies or can apply a combination of both (Waterlander,

Steenhuis, de Boer, Schuit, & Scheidell, 2012a). Taxes can be applied to unhealthy food items

to increase the price, and subsidies can be applied to healthy food items to decrease the price

(Waterlander et al., 2012a). These strategies are shown to influence purchase decisions. A key

reason for this is that consumer demand for food is elastic, meaning that changing the price affect purchases of food (Andreyeva, Long, & Brownell, 2010).

First of all, based on price elasticity of demand, subsidizing the prices of healthy food positively affects the purchases of healthy food (French, Jeffery, Story, Breitlow, Baxter, Hannan, MStat, & Snyder, 2001; Ni Mhurchu et al, 2010; Waterlander, Steenhuis, de Boer, Schuit, & Seidell, 2012b; Waterlander et al., 2013). For example, Waterlander et al. (2012a) showed that in the condition with 50% discount on healthy items, participants purchased 821 grams more vegetables and 420 grams more fruits for their household compared to no discount.

Importantly, however, we should take into account that subsidizing healthy food items may also have adverse side effects. Specifically, Waterlander et al. (2012a) found that although participants bought a higher amount of healthy food items when this is subsidized, the proportion of healthy products was not higher; the discounts also lead to a higher total number of items purchased. Additionally, subsidizing healthy food was related to a higher number of calories purchased as well. Only when the price of unhealthy food was increased with 25%, there was no higher food and calorie purchase overall (Waterlander et al., 2012a). Hence, price discounts on healthy food do not discourage the purchase of unhealthy food because individuals may spend money that is saved on healthier food to buy additional unhealthier alternatives, resulting in more caloric intake (Epstein et al., 2010; Waterlander et al., 2013).

Although subsidizing healthy food items does not have the desired result, previous research has shown that taxing unhealthy food items have the potential to reduce the amount of unhealthy food and calories purchased (Epstein et al., 2010; Giesen et al., 2011; Nederkoorn et al., 2011). According to Epstein et al. (2010), taxing High-Calorie-For-Nutrient (HCFN) foods have the dual benefit to reduce purchases of HCFN foods and to increase purchases of Low- Calorie-for-Nutrient (LCFN) foods. This may be explained by positive cross-price elasticity;

although people are still able to spend the money on the HCFN foods, they are likely to willfully choose not to buy these foods as they think it is not worth the high price (Epstein et al., 2010;

Nederkoorn et al., 2011). Hence, they shift to buying healthier food items (Epstein et al., 2010;

Nederkoorn et al., 2011). Similarly, Finkelstein, Zhen, Bilger, Nonnemaker, Farooqui, and Todd (2013) found that sugar-sweetened beverage (SSB) tax of 20% results in a decrease in store-bought energy of 24.3 kcal per day per person across the included food and beverage categories. Additionally, they did not find evidence of substitution to sugary foods. This finding is consistent with results of SSB taxes in France and Mexico, where purchases of taxed beverages decreased, and purchases of untaxed beverages increased (Berardi, Sevestre, Tepaut,

& Vigneron, 2016; Colchero, Popkin, Rivera, & Ng, 2016). For example, in Mexico, the

EFFECTIVELY PROMOTING HEALTHY FOOD CHOICES 6

purchases of taxed beverages decreased with 12%, and purchases of untaxed beverages increased with 4%, mainly related to bottled water (Colchero et al., 2016).

In conclusion, based on the above reasoning, I suggest that taxing unhealthy food items is likely to increase the healthiness of the purchases in the target categories and the overall shopping basket. Therefore, I posit:

Hypothesis 1. Taxing unhealthy food items increases the healthiness of the purchases in the target categories.

Hypothesis 2. Taxing unhealthy food items increases the healthiness of the purchases in the overall shopping basket.

Healthy Eating Calls in the Form of a Descriptive Norm

Besides economic incentives, there are other strategies which can be pursued to stimulate healthy food decisions. One known strategy is called nudging, which can be seen as

“a variety of simple, inexpensive, and freedom-preserving modifications to the choice environment” (Cadario & Chandon, 2018: 3). Thaler and Sunstein (2008: 6) have defined nudges as “any aspect of the choice architecture that alters people’s behavior in a predictable way without forbidding any options or significantly changing their economic incentives.

Putting fruit at eye level counts as a nudge; banning junk food does not.” Prior research has identified a wide variety of behaviors that could be seen as a nudge, like point-of-sale graphic warning labels (Donnely, Zatz, Svirsky, & John, 2018), manipulation in the “choice architecture” where healthy items are more accessible (Levy, Riis, Sonnenberg, Barraclough,

& Thorndike, 2012), and hedonic enhancements with attractive displays or descriptions for healthy options (Cadario & Chandon, 2018).

Specifically, this research is focused on healthy eating calls (i.e., a written or an oral injunction to choose healthier options) to nudge consumers towards making healthier choices (Cadario & Chandon, 2018: 10). These calls aim to influence how consumers feel by, for example, placing signs in a restaurant with “Keep your meal balanced” (Cadario & Chandon, 2018). Previous research has shown that healthy eating calls positively affects healthy food choices. For example, verbal prompts involving healthy side dishes increased the choice for healthy side dishes in a self-service restaurant during breakfast time (van Kleef et al., 2015).

Moreover, Policastro et al. (2017) found that college students were likely to substitute sugar- sweetened beverage by water through posters displaying calorie information.

In this research, healthy eating calls in the form of a descriptive norm are used (e.g.,

“most consumers choose this as a healthy choice” or “most preferred healthy choice”).

Descriptive norms inform consumers about what is normal and typically is done, and

individuals are likely to rely on the norm because “if everyone is doing it, it must be a sensible thing to do” (Cialdini, Reno, & Kallgren, 1990: 1015). These descriptive norms are likely to be effective in the shopping environment since often self-regulation (i.e., the capacity for altering its inner states and behavior) is low, and food choices are often made mindlessly and automatic, cued by environmental stimuli (Baumeister & Vohs, 2007; Jacobsen, Mortensen, & Cialdini, 2011; Marteau et al., 2010; Salmon et al., 2015). For example, Salmon et al. (2015) showed that in a state of low self-control, participants were more likely to buy low-fat cheese when it is associated with a social proof heuristic (i.e., “most sold in this supermarket”), compared to when it is not associated with a social proof heuristic. Moreover, the study of Thomas et al.

(2017) showed that the purchase of vegetables in a workplace restaurant increased when using posters with a descriptive social norm about the vegetable purchase of other individuals. Hence, healthy eating calls in the form of a descriptive norm may be used as a social proof heuristic by adopting the option preferred by others and simplify the decision-making process (Burger &

Shelton, 2011; Cialdini, 1990; Mollen, Rimal, Ruiter, & Kok, 2013; Salmon et al., 2015).

In conclusion, the above reasoning suggests that healthy eating calls in the form of a descriptive norm are likely to increase the healthiness of the purchases in the target categories and the overall shopping basket. Therefore, I posit:

Hypothesis 3. Healthy eating calls in the form of a descriptive norm increase the healthiness of the purchases in the target categories.

Hypothesis 4. Healthy eating calls in the form of a descriptive norm increase the healthiness of the purchases in the overall shopping basket.

The Moderating effect of Healthy Eating Calls in the form of a Descriptive Norm

The impact of food pricing strategies could be augmented by other strategies like nudging to increase healthy food choices (Cadario & Chandon, 2018; Ni Mhurchu et al., 2010).

Synergy effects may exist, meaning that introducing the interventions together have a higher impact on the healthiness of food purchase, than when the interventions are introduced in isolation (Shine, Park, & Wyer Jr, 2007). Specifically, this research looks at whether the effect of taxing unhealthy food items on the healthiness of the food purchases is more pronounced in the presence of healthy eating calls in the form of a descriptive norm.

First of all, as mentioned before, by increasing the price of unhealthy food items,

consumers are likely to willfully choose not to buy these foods, as they think it is not worth the

high price (Epstein et al., 2010; Nederkoorn et al., 2011). Therefore, choosing a healthy option

instead of an unhealthy option becomes more attractive. When combining taxing unhealthy

food items with healthy eating calls in the form of a descriptive norm on healthy food items,

EFFECTIVELY PROMOTING HEALTHY FOOD CHOICES 8

the healthy options are likely to become even more salient (Cialdini et al., 1990; Dotch-Klerk

& Jansen, 2008; Fennis & Stroebe, 2015; Wansink & Chandon, 2006). When a healthy option is salient, it means that this option is noticeably different from its environment (Fennis &

Stroebe, 2015; Kardes, 2002). As consumers’ food decisions are often mindlessly and automatic, they are likely to go for the salient options (Cialdini et al., 1990; Fennis & Stroebe, 2015).

Secondly, based on the self-regulatory focus theory, the interventions are aiming at consumers with promotion-focused self-regulation (i.e., healthy eating calls in the form of a descriptive norm) and prevention-focused self-regulation (i.e., taxing unhealthy food items).

According to the self-regulatory focus theory, “people are motivated to approach pleasure and avoid pain” (Higgens, 1997: 1280). Individuals are predominantly focused on promotion self- regulation (i.e., approach pleasure) or prevention self-regulation (i.e., avoid pain), and this has an essential impact on how they feel, think and act (Higgens, 1997; Higgens, 1998). Consumers who tend to be promotion-focused rely primarily on approach strategies which motivate people to behave towards positive outcomes (Aaker & Lee, 2001; Higgens, 1998; Higgens, 2002;

Kramer & Yoon, 2007; Pham & Avnet, 2004; Pham & Higgens, 2005). In contrast, consumers who tend to be prevention-focused rely primarily on avoidance strategies which motivate people to engage in behaviors that avoid negative outcomes (Aaker & Lee, 2001; Higgens, 1998; Higgens, 2002; Kramer & Yoon, 2007; Pham & Avnet, 2004; Pham & Higgens, 2005).

Hence, consumers who tend to be promotion-focused are more influenced by messages with positive cues, and consumers who tend to be prevention-focused are more influenced by messages with negative cues (Aaker & Lee, 2001; Pham & Avnet, 2004).

The healthy eating calls in the form of a descriptive norm are likely to target and persuade consumers who tend to be promotion-focused. This is because the healthy eating calls in the form of a descriptive norm are likely to be in line with an approach motivation as the message is framed positively and more focused on the gains (i.e., making a healthy choice which is most preferred by other consumers). In contrast, taxing unhealthy food items are likely to target and persuade consumers who tend to be prevention-focused, because the intervention is in line with an avoidance approach with a negative cue merely focusing on the losses (i.e., an increase in price).

Hence, when combining taxing unhealthy food items with healthy eating calls in the form of a descriptive norm, the effectivity of both interventions is likely to be reinforced.

Synergy effects are likely to be expected as healthy options are likely to become more salient,

and these interventions target and persuade both promotion- and prevention-focused consumers. Therefore, I posit:

Hypothesis 5. The effect of taxing unhealthy food items on the healthiness of the purchases in the target categories is more pronounced in the presence of healthy eating calls in the form of a descriptive norm.

Hypothesis 6. The effect of taxing unhealthy food items on the healthiness of the purchases in the overall shopping basket is more pronounced in the presence of healthy eating calls in the form of a descriptive norm.

METHOD Participants

The participants for this study were recruited via Amazon Mechanical Turk in May 2019. Amazon Mechanical Turk is a crowdsourcing marketplace which enables survey participation (Amazon Mechanical Turk, 2019). This platform provided the opportunity to recruit a diverse group of participants in a limited amount of time. All the questions in the survey were formulated in English.

Study Design and Intervention

To test the hypotheses, this study used a randomized between-subjects design with two levels of taxes on relatively unhealthy food items (25% versus no) x two levels of healthy eating calls in the form of a descriptive norm on relatively healthy food items (yes versus no). The participants were randomized into one of the four conditions and were asked to purchase groceries in 25 different food categories. The sequence of the product categories and the groceries within these product categories were randomized to control for possible effects.

The tax of 25% on relatively unhealthy food items was applied in the categories mayonnaise, cola, chips, pasta sauce, and spam classic meat. When considering nutrient criteria, the products that contained a tax were identified as the relatively unhealthiest product within their product category (Roodenburg, Popkin, & Seidell, 2011). The healthy eating calls in the form of a descriptive norm on relatively healthy food items were applied in the categories bread, yogurt, and dressing. When considering nutrient criteria, the products that contained a healthy eating call were identified as the relatively healthiest product within their product category (Roodenburg et al., 2011).

Procedure

At the beginning of the study, participants were informed that they are asked to do

groceries and that they have to answer some questions afterward. They were informed about

the purpose of the study and were told to get $0.20 compensation after finishing the complete

EFFECTIVELY PROMOTING HEALTHY FOOD CHOICES 10

study. Participants were told that for each product category, they will be presented with four different options and that they could only choose one item per category. It was emphasized that they could do the online-groceries like they would do groceries in real-life. After their shopping trip, they were asked to answer questions about their health consciousness, hunger level, dietary restrictions, self-regulatory focus, and demographics. Finally, participants were thanked for their participation and were asked to fill out their worker ID for the reimbursement. Please refer to Appendix A for the complete questionnaire.

Measures

Taxing unhealthy food items. The price of the relatively unhealthiest product within the categories mayonnaise, cola, chips, pasta sauce, and spam classic meet were increased by 25%. The price increase of 25% was clearly communicated with a red line through the old price, and the new price was shown in bold (see Appendix A).

Healthy eating calls in the form of a descriptive norm. The healthy eating calls in the form of a descriptive norm were inspired by the research of Thomas et al. (2017) who used the sentence “most people here choose to eat vegetables with their lunch” in the setting of a restaurant to increase the purchases of vegetables. In the questionnaire within the categories bread, yogurt, and dressing, on the relatively healthiest product was stated: “most consumers choose this as a healthy choice”. Next to the sentence, a green check mark was stated for visualization (see Appendix A).

The healthiness of the purchases the target categories. For each category with an intervention (i.e., mayonnaise, cola, chips, pasta sauce, spam classic meat, bread, yogurt and dressing), the healthiness of the purchases was assessed by a “0” or “1”, where “0’ represents a relatively healthy choice and “1” a relatively unhealthy choice. Within each target category, the two products with the lowest number of calories per 100 grams/ milliliters were categorized as a relatively healthy choice, and the two products with the highest number of calories per 100 grams/ milliliters were categorized as a relatively unhealthy choice. In this way, the healthiness of purchases in the target categories was calculated on a scale of 0 – 1.

The healthiness of the purchases of the overall shopping basket. The healthiness of the overall shopping basket was measured by the sum of the calories in the overall shopping basket. The number of calories of each product was measured by the calories per 100 grams or 100 milliliters. The higher the number of calories of the overall shopping basket, the lower the healthiness of the overall shopping basket.

Control variables. To ensure that no other variables account for the relationships

between the variables in my conceptual model, several control variables were included. To

begin, I controlled for the demographic variables age, gender, and education as these variables could influence the food choices in the supermarket (Boek, Bianco-Simeral, Chan, & Goto, 2012; Drewnowski, 2004; Drewnowski & Darmon, 2005). Moreover, height and weight data were collected to calculate participants’ body mass index (BMI) as research showed that BMI may impact food preferences and choices (Bartoshuk, Duffy, Hayes, Moskowitz, & Snyder, 2006). Furthermore, I controlled for hunger level as consumers are more likely to go for the unhealthy option in a state of hunger (Zepeda & Deal, 2008). Additionally, I controlled for dietary restrictions as it may have an impact on food choices. For example, people may pursue a vegetarian’s lifestyle and do not eat meat and seafood (Dinu, Abbate, Gensini, Casini, & Sofi, 2016; Leitzmann, 2014; Kwok, Umar, Myint, Mamas, & Loke, 2014).

Moreover, the variable health consciousness was included, which is a motivational component that stimulates consumers to take health actions to achieve a healthy lifestyle and to avoid or reduce illness (Kraft & Goodell, 1993; Plank & Gould, 1990). Individuals who are health conscious tend to be aware of and concerned with nutrition (Kraft & Goodell, 1993).

Health consciousness was measured by eight items developed by Roininen, Lähteenmäki, and Tuorila (1999). Responses were given on a seven-point scale (1 = strongly disagree, 7 = strongly agree). Example items are: “It is important for me that my daily diet contains a lot of vitamins and minerals,” “I always follow a healthy and balanced diet,” and “The healthiness of snacks makes no difference to me.” Some items were reversed so that higher scores indicated higher health consciousness. After deleting the first item (“The healthiness of food has little impact on my food choices”), Cronbach’s alpha was .63.

Lastly, consistent with the theory section, self-regulatory focus was included as a

control variable as this may have an impact on how consumers react to a tax of 25% and healthy

eating calls in the form of a descriptive norm. Self-regulatory was measured by fourteen items

of Lockwood, Jordan, and Kunda (2002) on a nine-point scale (1 = Not at all true of me, 9 =

Very true of me). From the original eighteen items, four items were left out as these were mainly

related to academic occasions. Example items are: “I frequently imagine how I will achieve my

hopes and aspirations,” “I am more oriented toward preventing losses than I am toward

achieving gains,” and “I often imagine myself experiencing good things that I hope will happen

to me.” Some items were reversed so that higher scores indicated more promotion-focused

individuals and lower scores indicated more prevention-focused individuals. Cronbach’s alpha

was .63.

EFFECTIVELY PROMOTING HEALTHY FOOD CHOICES 12

RESULTS Preparation for Data Analysis

Sample. Overall, 295 respondents completed the survey on Amazon Mechanical Turk.

All participants were at least 18 years old to make sure that they could make independent food choices. At the end of the survey, participants had to fill in their Amazon ID to receive compensation and to check that they would only fill it in once. Thirteen participants filled in the survey two times, and they did not provide the same answers in both surveys. As this causes reliability issues, both surveys of these participants were deleted from further analysis. All statistical analyses were done in SPSS (Statistical Package for the Social Sciences; Babbie, Wagner, & Zaino, 2018).

The final set of participants consisted of 269 participants. Participants in the final sample had an average age of 33 years (SD = 10 years), and 59,1% was male, 40,2% was female, and 0,7% preferred not to say. Participants were mainly from the United States with a percentage of 59,85%. The education level, current or highest, of most participants, was a bachelor’s degree (57,2%) and master’s degree (19,0%).

Conditions. Participants were randomized into one of the four conditions. In the final

sample, the participants were allocated as follows, 73 participants in the no intervention

condition, 60 participants in the taxing condition, 70 participants in the healthy eating call

condition, and 66 participants in the taxing and healthy eating call condition. In Table I and

Table II, the descriptive statistics for the dependent variables; the healthiness of the purchases

in the target categories (i.e., categories with taxes and healthy eating calls) and the overall

shopping basket is shown. In Figure II and Figure III, the estimated marginal means of the

healthiness of the purchases in the target categories (i.e., categories with taxes and healthy

eating calls) and the overall shopping basket is depicted.

N Minimum Maximum Mean Std. Deviation

No Intervention 73 5228 7895 6918.89 553.07

Taxing 60 5757 8012 6960.48 457.65

Healthy Eating Call 70 5580 7740 6767.26 392.06

Taxing and Healthy Eating Call 66 5592 7931 6752.85 464.18

TABLE II Healthiness of the Purchases in the Overall Shopping Basket (in calories)

N Minimum Maximum Mean Std. Deviation

No Intervention 73 0 1 .60 .23

Taxing 60 .13 1 .60 .21

Healthy Eating Call 70 0 1 .48 .20

Taxing and Healthy Eating Call 66 0 1 .47 .24

TABLE I

Healthiness of the Purchases in the Target Categories (0: healthy choice; 1: unhealthy choice)

EFFECTIVELY PROMOTING HEALTHY FOOD CHOICES 14

Descriptive Statistics and Correlations

In Table III, the means, standard deviations, and bivariate correlations for the variables measured in this study are presented. When looking at this table, it is clear that some interesting and significant correlations exist. I found that dietary restrictions were negatively correlated with the number of calories of the overall shopping basket (r = -.19, p < .01). This indicates that people with dietary restrictions (e.g., vegetarian, vegan, lactose intolerance) purchase a lower number of calories during their shopping trip. As expected, health consciousness was negatively correlated with the number of calories of the overall shopping basket (r = -.31, p < .01).

Moreover, self-regulatory focus was positively correlated with the number of calories of the overall shopping basket (r = .13, p < .05). This indicates that people who are more health conscious and more prevention-focused, purchase a lower number of calories.

Interestingly, BMI was negatively correlated with hunger level (r = -.14, p < .05), indicating that people with a higher BMI had a lower hunger level while filling in the survey.

Moreover, BMI was positively correlated with age (r = .20, p < .01) and negatively correlated with education (r = -.19, p < .01). This outcome indicates that the higher someone’s BMI, the higher someone’s age and the lower someone’s education level. Health consciousness was positively correlated with age (r = .21, p < .01) and education (r = .17, p < .01), indicating that the higher someone’s health consciousness, the higher someone’s age and education as well.

Lastly, self-regulatory focus was positively correlated with age (r = .19, p < .01) and health consciousness (r = .19, p < .01), and negatively correlated with dietary restrictions (r = -.26, p

< .01). This outcome indicates that the higher someone’s promotion-focus, the higher someone’s age and health consciousness, and the lower someone’s dietary restrictions.

Variables Mean SD 1 2 3 4 5 6 7 8

1. Age 33.07 10.24

2. Gender (Male = 1, Female = 2, Other/I prefer not to

say = 3) 1.42 .51 .07

3. Education 4.86 1.09 -.21

.08

4. BMI 25.58 5.83 .20

.10 -.19

5. Hunger level 3.41 1.24 -.28

-.16

.17

-.14

6. Dietary restrictions (No = 0, Yes = 1) .55 .50 -.37

-.03 .30

-.18

.42

7. Health consciousness 4.35 .87 .21** .09 .17** -.00 -.10 .04

8. Self-regulatory focus 5.34 .80 .19** .12 -.09 .09 -.22** -.26** .19**

9. Number of calories of the overall shopping basket 6847.97 477.76 .12 -.02 -.09 .06 -.11 -.19

-.31** .13*

Note. N = 256 – 269 participants.

* p < .05; ** p < .01

Table III

Means, Standard Deviations, and Pearson Correlation Coefficients

Test for Differences between the Conditions

To test whether the dependent variables (i.e., the healthiness of the purchases in the target categories and the healthiness of the purchases of the overall shopping basket) had different means in the no intervention condition and the intervention conditions, I conducted an independent samples t-test. Before conducting the independent samples t-test, I tested whether the dependent variables were normally distributed for each group of the independent variable (see notes, page 40). For hypothesis 3, the Mann-Whitney U-test was used as the condition for normal distribution for an independent sample t-test was not met.

To conduct the independent samples t-tests (hypothesis 1, 2, 4,) and the Mann-Whitney U-test (hypothesis 3), several variables were created. For hypothesis 1 and 2, I used the dummy variable DumTax (0 = no intervention, 1 = tax) and to measure the dependent variables I used DV_Health_TAXcategory (i.e., the healthiness of the purchases in the categories with taxes) and Total_CAL_DV (i.e., the healthiness of the purchases of the overall shopping basket). For hypothesis 3 and 4, I used the dummy variable DumHec (0 = no intervention, 1 = healthy eating call) and to measure the dependent variables I used DV_Health_HECcategory (i.e., the healthiness of the purchases in the categories with healthy eating calls) and Total_CAL_DV (i.e., the healthiness of the purchases of the overall shopping basket).

Hypothesis 1. Contrary to the expectations, the independent samples t-test was not significant (t (131) = .10, p = .92). Hence, the healthiness of the purchases in in the target categories in the taxing condition (M = .56, SD = .24) did not significantly differ with the healthiness of the purchases in the target categories in the no intervention condition (M = .56, SD = .27).

Hypothesis 2. Contrary to the expectations, the independent samples t-test was not significant (t (131) = -.47, p = .64). Hence, the healthiness of the purchases in the overall shopping basket in the taxing condition (M = 6960.48, SD = 457.65) did not significantly differ with the healthiness of the purchases in the overall shopping basket in the no intervention condition (M = 6918.89, SD = 553.07).

Hypothesis 3. In line with the expectations, the results of the Mann-Whitney U-test are in support of hypothesis 3. The null hypothesis of ‘the distribution of healthiness of purchases HEC category is the same across categories of ‘DumHec’ could be rejected (p = .00). Hence, participants in the healthy eating call condition purchased healthier products in the target categories (M = .41, SD = .32) than participants in the no intervention condition (M = .65, SD

= .31).

EFFECTIVELY PROMOTING HEALTHY FOOD CHOICES 16

Hypothesis 4. In line with the expectations, the independent samples t-test was marginally significant (t (130) = 1.90, p = .06). The healthiness of the purchases in the overall shopping basket in the healthy eating call in the form of a descriptive norm condition (M = 6767.26, SD = 392.06) did marginally significantly differ with the healthiness of the purchases in the overall shopping basket in the no intervention condition (M = 6918.89, SD = 553.07).

Hence, participants in the healthy eating call condition purchased healthier products in the overall shopping basket than participants in the no intervention condition.

Hypotheses Testing

In order to analyze the influence of taxing unhealthy food items and healthy eating calls in the form of a descriptive norm on the healthiness of the purchases in the target categories and the overall shopping basket, I conducted a factorial analysis of variance (ANOVA). The dummy independent variables DumTax (-1 = no tax, 1 = tax) and DumHec (-1 = no hec, 1 = hec), and the dependent variables DV_Health_TAXcategory (i.e., the healthiness of the purchases in the categories with taxes), DV_Health_HECcategory (i.e., the healthiness of the purchases in the categories with healthy eating calls), DV_Health_TargetCategory (i.e., the healthiness of the purchases in the categories with taxes and healthy eating calls), and Total_CAL_DV (i.e., the healthiness of the purchases in the overall shopping basket) were included. For the results, please refer to Table IV.

The results of the factorial ANOVA for hypotheses 1, 2, 3, 4 are in line with the results of the independent samples t-tests (hypothesis 1, 2, 4) and the Mann-Whitney U-test (hypothesis 3), supporting the robustness of my findings. In addition, based on the recommendations of Becker (2005), the analyses were repeated controlling for gender, education, BMI, hunger level, dietary restrictions (yes versus no), health consciousness, and self-regulatory focus but the results were essentially identical (see Appendix C.2).

Hypothesis 1. Contrary to what was predicted in hypothesis 1, the results indicate that the main effect of taxing unhealthy food items on the healthiness of the purchases in the target categories is not significant (F (1, 265) = .04, p = .85). This indicates that taxing unhealthy food items did not significantly influence the healthiness of the purchases in the target categories (i.e., categories with taxes).

Hypothesis 2. Contrary to what was predicted in hypothesis 2, the results indicate that

the main effect of taxing unhealthy food items on the healthiness of the purchases in the overall

shopping basket is not significant (F (1, 265) = .06, p = .81). This indicates that taxing unhealthy

food items did not significantly influence the healthiness of the purchases in the overall

shopping basket.

Hypothesis 3. In line with what was predicted in hypothesis 3, the results indicate that the main effect of healthy eating calls in the form of a descriptive norm on the healthiness of the purchases in the target category is significant (F (1, 265) = 44.35, p = .00). This indicates that healthy eating calls in the form of a descriptive norm did significantly influence the healthiness of the purchases in the target categories (i.e., categories with healthy eating calls).

Hypothesis 4. In line with what was predicted in hypothesis 4, the results indicate that the main effect of healthy eating calls in the form of a descriptive norm on the healthiness of the purchases in the overalls shopping basket is significant (F (1, 265) = 9.70, p = .00). This indicates that healthy eating calls in the form of a descriptive norm did significantly influence the healthiness of the purchases in the overall shopping basket.

Hypothesis 5. Contrary to what was predicted in hypothesis 5, the results indicate that the interaction effect of taxing unhealthy food items and healthy eating calls in the form of a descriptive norm on the healthiness of the purchases in the target categories is not significant (F (1, 265) = .09, p = .76). This indicates that this research did not find a moderating effect of healthy eating calls in the form of a descriptive norm on the relationship between taxing unhealthy food items and the healthiness of the purchases in the target categories (i.e., categories with taxes and healthy eating calls).

Hypothesis 6. Contrary to what was predicted in hypothesis 6, the results indicate that the interaction effect between taxing unhealthy food items and healthy eating calls in the form of a descriptive norm on the healthiness of the purchases in the overall shopping basket is not significant (F (1, 265) = .24, p = .63). This indicates that this research did not find a moderating effect of healthy eating calls in the form of a descriptive norm on the relationship between taxing unhealthy food items and the healthiness of the purchases of the overall shopping basket.

Predictors

F -value p- value F -value p- value

Taxing unhealthy food items .04 .85 .06 .81

Healthy eating calls in the form of a descriptive

norm 44.35 .00** 9.70 .00**

Taxing unhealthy food items × healthy eating calls

in the form of a descriptive norm .09 .76 .24 .63

* p < .05; ** p < .01 df = 1

Healthiness of the purchases in the target categories

Healthiness of the purchases in the overall shopping basket Dependent variables

Factorial ANOVA for the influence of taxing unhealthy food items and healthy eating calls in the form of a descriptive norm on the healthiness of the purchases in the target categories and the overall shopping basket

TABLE IV

EFFECTIVELY PROMOTING HEALTHY FOOD CHOICES 18

Follow-up Analysis - Covariates

A follow-up analysis was conducted to analyze the effect of the independent variables on the healthiness of purchases in the target categories (i.e., categories with taxes and healthy eating calls) and the overall shopping basket. Regression analysis were conducted for the following independent variables: taxing unhealthy food items, healthy eating calls in the form of a descriptive norm, age, gender, education, BMI, hunger level, dietary restrictions, health consciousness, and self-regulatory focus. For the results, please refer to Table V.

The healthiness of the purchases in the target categories.

The regression was checked for the variance inflation factor (VIF) to ensure that no multicollinearity between the predictor variables existed (Grewal, Cote, & Baumgartner, 2004;

Malhotra, 2009). As the VIF-scores were below 2, multicollinearity was not an issue. The regression was found to be significant (R

= .19, F (10, 245) = 5.63, p = .00).

The significant findings are as follows. In line with previous results, healthy eating calls in the form of a descriptive norm positively influenced the healthiness of the purchases in the target categories (B = -.05, t (255) = -3.64, p = .00). Furthermore, health consciousness positively influenced the healthiness of the purchases in the target categories (B = -.09, t (255)

= -5.10, p = .00), and self-regulatory focus negatively influenced the healthiness of the purchases in the target categories (B = .06, t (255) = 3.28, p = .00). This indicates that people who were more health conscious and more prevention-focused purchased on average healthier food items in the target categories.

The healthiness of the purchases in the overall shopping basket.

The regression was checked for the variance inflation factor (VIF) to ensure that no multicollinearity between the predictor variables existed (Grewal et al., 2004; Malhotra, 2009).

As the VIF-scores were below 2, multicollinearity was not an issue. The regression was found to be significant (R

= .18, F (10, 245) = 5.42, p = .00).

The significant findings are as follows. In line with previous results, healthy eating calls

in the form of a descriptive norm positively influenced the healthiness of the purchases in the

overall shopping basket (B = -58.17, t (255) = -2.04, p =.04). Moreover, age influenced the

healthiness of the purchases in the overall shopping basket negatively (B = 6.69, t (255) = 2.18,

p = .03). When someone’s age increased with one unit, the number of calories in the overall

shopping basket increased with 6.69 calories. Furthermore, health consciousness influenced the

healthiness of the purchases in the overall shopping basket positively (B = -202.57, t (255) = -

5.68, p = .00). When someone’s health consciousness increased with one unit, the number of

calories in the overall shopping basket decreased with 202.57 calories. Lastly, self-regulatory

focus influenced the healthiness of the purchases in the overall shopping basket negatively (B

= 97.91, t (255) = 2.65, p = .01). When someone’s self-regulatory focus increased with one unit (i.e., someone is more promotion-focus), the number of calories in the overall shopping basket increased with 97.91 calories.

Follow-up Analysis - The Moderating Role of Self-Regulatory Focus

As mentioned in the theory section, I expected that the combination of taxing unhealthy food items with healthy eating calls in the form of a descriptive norm target and persuade both promotion- and prevention-focused consumers. To analyze the interaction effect with self- regulatory focus, I used Model 3 of the PROCESS macro (Hayes, 2013). The variables self- regulatory focus, DumTax (-1 = no tax, 1 = tax) and DumHec (-1 = no hec, 1 = hec), DV_Health_TargetCategory (i.e., the healthiness of the purchases in the categories with taxes and healthy eating calls), and Total_CAL_DV (i.e., the healthiness of the purchases in the overall shopping basket) were included. The variables were mean centered for interpretation purposes of the interaction effects (Robinson & Schumacker, 2009). For the results, please refer to Table VI.

The healthiness of the purchases in the target categories.

The overall model was significant (F (7, 261) = 5.08, p = .00, R

= .12). The main effect of taxing unhealthy food items was not significant (B = -.00, SE = .01, t (261) = -.13, p = .90), and the main effect of healthy eating calls in the form of a descriptive norm was significant (B

= -.06, SE = .01, t (261) = -4.51, p = .00). The interaction effect of taxing unhealthy food items

Predictors

B SE t p B SE t p

Taxing unhealthy food items -.01 .01 -.86 .39 -17.08 28.18 -.61 .55

Healthy eating calls in the form of a descriptive

norm -.05 .01 -3.64 .00** -58.17 28.51 -2.04 .04*

Age .00 .00 1.37 .17 6.69 3.01 2.18 .03*

Gender -.01 .03 -.45 .65 -28.06 56.40 -.50 .62

Education .01 .01 .54 .59 27.12 28.11 .97 .34

BMI -.00 .00 -1.64 .10 1.64 4.97 .33 .74

Hunger level -.01 .01 -.48 .63 -8.98 25.94 -.35 .73

Dietary restrictions (No = 0, Yes = 1) .01 .03 .35 .73 -55.97 68.38 -.82 .41

Health consciousness -.09 .02 -5.10 .00** -202.57 35.65 -5.68 .00**

Self-regulatory focus .06 .02 3.28 .00** 97.91 36.97 2.65 .01**

* p < .05; ** p < .01 df = 255

Healthiness of the purchases in the target categories Healthiness of the purchases in the overall shopping basket

Regression analyses for taxing unhealthy food items, healthy eating calls in the form of a descriptive norm, age, gender, education, BMI, hunger level, dietary restrictions, health consciousness and self-regulatory focus on the healthiness of the purchases in the target categories and overall shopping basket

Dependent variable

TABLE V

EFFECTIVELY PROMOTING HEALTHY FOOD CHOICES 20

with healthy eating calls in the form of a descriptive norm was not significant (B = -.00, SE = .01, t (261) = -.16, p = .87). This is in line with the previous results of the factorial ANOVA and regression analysis.

The main effect of self-regulatory focus was significant (B = .05, SE = .02, t (261) = 2.82, p = .01). This indicates that when values of self-regulatory focus are higher (i.e., when someone is more promotion-focused), the healthiness of the purchases in the target categories is lower. As shown in Figure IV, the interaction effect of self-regulatory focus with taxing unhealthy food items was not significant (B = .03, SE = .02, t (261) = 1.63, p = .10). Moreover, there was a significant interaction effect of self-regulatory focus with healthy eating calls in the form of a descriptive norm (B = .04, SE = .02, t (261) = 2.36, p = .02). This indicates that when values of self-regulatory focus are higher (i.e., when someone is more promotion- focused), the effect of healthy eating calls in the form of a descriptive norm on the healthiness of the purchases in the target categories is weakened (see Figure V). The interaction effect of self- regulatory focus with taxing unhealthy food items and healthy eating calls in the form of a descriptive norm was not significant (B = .02, SE = .02, t (261) = 1.02, p = .31).

The healthiness of the purchases in the overall shopping basket.

The overall model was significant (F (7, 261) = 2.84, p = .01, R

= .07). The main effect of taxing unhealthy food items was not significant (B = 9.27, SE = 28.57, t (261) = .32, p = .75), and the main effect of healthy eating calls in the form of a descriptive norm was significant (B=

-85.35, SE = 28.51, t (261) = -2.99, p = .00). The interaction effect of taxing unhealthy food items with healthy eating calls in the form of a descriptive norm was not significant (B = -10.52, SE = 28.57, t (261) = -.37, p = .71). This is in line with the previous results of the factorial ANOVA and regression analysis.

The main effect of self-regulatory focus was significant (B = 85.41, SE = 36.18, t (261)

= 2.36, p = .02). This indicates that when values of self-regulatory focus are higher (i.e., when someone is more promotion- focused), the healthiness of the purchases in the overall shopping basket is lower. As shown in Figure VI, the interaction effect of self-regulatory focus with taxing unhealthy food items was not significant (B = 46.98, SE = 36.44, t (261) = 1.29, p = .20).

There was a significant interaction effect of self-regulatory focus with healthy eating calls in

the form of a descriptive norm (B = 77.22, SE = 36.13, t (261) = 2.14, p = .03). This indicates

that when values of self-regulatory focus are higher (i.e., someone is more promotion-focused),

the effect of healthy eating calls in the form of a descriptive norm on the healthiness of the

purchases in the overall shopping basket is weakened (see Figure VII). The interaction effect

of self-regulatory focus with taxing unhealthy food items and healthy eating calls in the form of a descriptive norm was not significant (B = 26.64, SE = 36.39, t (261) = .73, p = .46).

Predictors

B SE t p B SE t p

Taxing unhealthy food items -.00 .01 -.13 .90 9.27 28.57 .32 .75 Healthy eating calls in the form of a descriptive

norm -.06 .01 -4.51 .00** -85.35 28.51 -2.99 .00**

Taxing unhealthy food items × healthy eating calls

in the form of a descriptive norm -.00 .01 -.16 .87 -10.52 28.57 -.37 .71

Self-regulatory focus .05 .02 2.82 .01** 85.41 36.18 2.36 .02*

Self-regulatory focus × taxing unhealthy food

items .03 .02 1.63 .10 46.98 36.44 1.29 .20

Self-regulatory focus × healthy eating calls in the

form of a descriptive norm .04 .02 2.36 .02* 77.22 36.13 2.14 .03*

Self-regulatory focus × healthy eating calls in the form of a descriptive norm × taxing unhealthy

food items .02 .02 1.02 .31 26.64 36.39 .73 .46

* p < .05; ** p < .01 df = 261

Healthiness of the purchases in the target

categories

Healthiness of the purchases in the overall shopping

basket Dependent variables

Assessing the interaction effect of self-regulatory focus with taxing unhealthy food items and healthy eating calls in the form of a descriptive norm

TABLE VI

EFFECTIVELY PROMOTING HEALTHY FOOD CHOICES 22

Follow-up Analysis - The Moderating Role of Health Consciousness

Based on previous results of the regression analysis that showed a significant effect of health consciousness on the healthiness of the purchases in the target categories and the overall shopping basket, a follow-up analysis was conducted to analyze possible interaction effects with health consciousness. To analyze the interaction effect, I used Model 3 of the PROCESS macro (Hayes, 2013). The variables health consciousness, DumTax (-1 = no tax, 1 = tax) and DumHec (-1 = no hec, 1 = hec), DV_Health_TargetCategory (i.e., the healthiness of the purchases in the categories with taxes and healthy eating calls), and Total_CAL_DV (i.e., the healthiness of the purchases in the overall shopping basket) were included. The variables were mean centered for interpretation purposes of the interaction effects (Robinson & Schumacker, 2009). For the results, please refer to Table VII.

The healthiness of the purchases in the target categories.

The overall model was significant (F (7, 261) = 5.87, p = .00, R

= .14). The main effect of taxing unhealthy food items was not significant (B = -.01, SE = .01, t (261) = -.85, p = .39), and the main effect of healthy eating calls in the form of a descriptive norm was significant (B

= -.05, SE = .01, t (261) = -4.11, p = .00). The interaction effect of taxing unhealthy food items with healthy eating calls in the form of a descriptive norm was not significant (B = -.00, SE = .01, t (261) = -.03, p = .98). This is in line with previous results.

The main effect of health consciousness was significant (B = -.06, SE = .02, t (261) = - 3.79, p = .00). This indicates that when values of health consciousness are higher, the healthiness of the purchases in the target categories is higher. The interaction effect of health consciousness with taxing unhealthy food items was not significant (B = .00, SE = .02, t (261)

= .19, p = .85). Moreover, there was no significant interaction effect of health consciousness with healthy eating calls in the form of a descriptive norm (B = .01, SE = .02, t (261) = .90, p = .37). The interaction effect of health consciousness with taxing unhealthy food items and healthy eating calls in the form of a descriptive norm was not significant (B = .01, SE = .02, t (261) = .72, p = .47).

The healthiness of the purchases in the overall shopping basket.

The overall model was significant (F (7, 261) = 5.24, p = .00, R

= .12). The main effect

of taxing unhealthy food items was not significant (B = -13.07, SE = 28.11, t (261) = -.46, p =

.64), and the main effect of healthy eating calls in the form of a descriptive norm was significant

(B= -71.81, SE = 28.04, t (261) = -2.56, p = .01). The interaction effect of taxing unhealthy food

EFFECTIVELY PROMOTING HEALTHY FOOD CHOICES 24

items with healthy eating calls in the form of a descriptive norm was not significant (B = -5.89, SE = 28.12, t (261) = -.21, p = .83). This is in line with previous results.

The main effect of health consciousness was significant (B = -155.14, SE = 33.40, t (261) = -4.65, p = .00). This indicates that when values of health consciousness are higher, the healthiness of the purchases in the overall shopping basket is higher. The interaction effect of health consciousness with taxing unhealthy food items was not significant (B = 9.01, SE = 33.78, t (261) = .27, p = .79). There was no significant interaction effect of health consciousness with healthy eating calls in the form of a descriptive norm (B = 23.87, SE = 33.32, t (261) = .72, p = .47). The interaction effect of health consciousness with taxing unhealthy food items and healthy eating calls in the form of a descriptive norm was not significant (B = 19.87, SE = 33.70, t (261) = .59, p = .56).

Predictors

B SE t p B SE t p

Taxing unhealthy food items -.01 .01 -.85 .39 -13.07 28.11 -.46 .64 Healthy eating calls in the form of a descriptive

norm -.05 .01 -4.11 .00** -71.81 28.04 -2.56 .01*

Taxing unhealthy food items × healthy eating calls

in the form of a descriptive norm -.00 .01 -.03 .98 -5.89 28.12 -.21 .83 Health consciousness -.06 .02 -3.79 .00** -155.14 33.40 -4.65 .00**

Health consciousness × taxing unhealthy food

items .00 .02 .19 .85 9.01 33.78 .27 .79

Health consciousness × healthy eating calls in the

form of a descriptive norm .01 .02 .90 .37 23.87 33.32 .72 .47

Health consciousness × healthy eating calls in the form of a descriptive norm × taxing unhealthy

food items .01 .02 .72 .47 19.87 33.70 .59 .56

* p < .05; ** p < .01 df = 261

TABLE VII

Assessing the interaction effect of health consciousnesss with taxing unhealthy food items and healthy eating calls in the form of a descriptive norm

Dependent variables Healthiness of the

purchases in the target categories

Healthiness of the purchases in the overall shopping

basket