Red Light for Meat consumption

Red Light for Meat consumption

Can a color-coded carbon footprint label influence

consumers towards meat-alternative buying behavior?

Thesis for obtaining the Master of Science in Marketing

By:

Sören Köppen

UNIVERSITYOF GRONINGEN


FACULTYOF ECONOMICSAND BUSINESS ADMINISTRATION CHAIROF MARKETING

Red Light for Meat consumption

Can a color-coded carbon footprint label influence

consumers towards meat-alternative buying behavior?

– MASTERTHESIS – Author: Sören Köppen MSc. Marketing Student Number: 3168913 1st Supervisor: Dr. Wander Jager,

Associate Professor and Managing Director of the Groningen Center for Social Complexity Studies

2nd Supervisor:

Executive Summary

The need to act upon climate change is urgent. An often overlooked behavior in this field is the consumption of animal based proteins, particularly in the form of meat. Today’s livestock contributes approximately 18% to total greenhouse gases worldwide. The increase in population and per capita income in developing countries is accompanied by a higher demand for these animal proteins. In the developed world overindulgence, especially of meat and dairy products, leads to widespread diseases such as diabetes and cancer. It is therefore important to change consumers’ behavior with respect to these products and raise awareness about their impact on the environment and health.

One way to provide such information is through the use front-of-pack eco-labels. However, thus far no clear and standardized assessment for the measurement of a product’s environmental lifecycle impact has been put in place. The European Commission conducted multiple pilot programs between the 2013-2016 to provide a uniquely and scientific approach to this problem, with results to arrive in 2017. Yet it remains unclear how to present the obtained information to consumers so that they may benefit from it. Even more so, it is uncertain whether a newly created carbon footprint label as proposed by the EU Commission’s researchers will affect consumers’ buying decisions. The present study tries to give an answer to this question.

could, for some choice sets, explain subjects’ choices of meat-alternative products. The specific choice sets to which these results apply are discussed together with an attempt to find arguments for their appearance. Two dominant schemes that emerge are the psychological concepts of cognitive dissonance and that of social desirability. Concerning the independent variables, altruistic values, egoistic values, social norms, knowledge of meat-alternative recipes, label skepticism, and price perception of meat-alternative diets did not yield any statistically significant result in influencing consumers’ choices towards meat-alternative products. This suggests that other factors ought to exist that could potentially explain this specific type of consumption behavior. The present paper makes some assumptions as to what those could be.

Looking at the second, and main goal of this research – the effect of a color-coded eco-label on the choice for meat-alternative products – the results are somewhat counterintuitive. A significant effect between the label vs. no-label condition was found, albeit not in the expected way. The findings reveal that participants in the label condition chose, on average, one more meat product than participants in the control condition. Again, the two proposed psychological concepts mentioned above were used in an attempt to explain this paradoxical finding. These concepts presume that products and consumption foster a person’s self-concept which, once threatened, is defended rigorously to defuse upward moral comparison. One of the strategies to defuse this threat is resentment (i.e. disliking and distancing) to the perceived threatening object. The environmental impacts as well as the correct moral behavior was clearly visible in cases where the eco-label was displayed. This may have led respondents to resent the meat-alternative due to a conflicting moral comparison between their own preferences and those of vegetarians or vegans.

Preface

This thesis has been written as part of the graduation requirement for the Master in Science of Marketing Management at the University of Groningen, The Netherlands. I was involved with the research and writing of this paper from February to June 2017.

The present work deals with an important yet often overlooked issue in the public debate for climate change – meat consumption. Its title “Red light for meat consumption: can a color-coded carbon footprint label influence consumers towards meat-alternative buying behavior?” tries to answer the posed research question with two goals in mind. First, whether such an eco-label can, in fact, influence consumers’ buying behavior and secondly it investigates new and existing drivers behind meat consumption.

My motivation for this research stems from my deep appreciation for the environment and the constant question in mind of why there is so little being done to protect it. After all, this is the only planet we inhabit so far and shouldn’t we be utterly careful not to upset the very foundation that gave rise to our species? For this reason I developed the above research question under guidance and with constructive feedback of my supervisor Dr. Wander Jager. Identifying the problem and obtaining the correct literature to investigate this issue was challenging, but at the same time it allowed me to use my skills acquired throughout my academic career and work on a project I was very passionate about.

I would thus like to thank my supervisor for the valuable guidance and feedback along this journey. I am also grateful to the many respondents without whose participation in my survey I would not have been able to finish this thesis.

To my family and friends who have proofread my work time and time again and who have supported me unconditionally throughout the writing of my work, I would like to say: thank you and I love all of you.

In this spirit, I hope you take as much pleasure in reading this article as I took in writing it. Sören Köppen

Table of Contents

List of abbreviations ...VII List of figures and tables ...VIII

1 Introduction ...1

2 Theoretical Framework ...4

2.1 Theoretical review ...4

2.2 Hypotheses ...7

2.3 Labels and product lifecycle assessment ...11

3 Research Methodology ...13

3.1 Experimental design ...13

3.2 Label construction and package design ...14

3.3 Questionnaire ...17 3.4 Data collection ...18 3.5 Sample ...18 3.6 Measurement ...19 4 Empirical Results ...21 4.1 Descriptive statistics ...21 4.2 Estimation results ...22 4.3 Discussion ...25

5 Conclusions & Recommendations ...30

5.1 Managerial and policy implications ...30

5.2 Limitations and Future Research ...32

5.3 Conclusion ...33

Appendices ...36

List of abbreviations

cf. confer (compare)

e.g. example given

EPA Environmental Protection Agency ETL Environmental traffic light

EU European Union

ff. following page(s)

FoP Front-of-pack

GHG Greenhouse gas

KMO Kaiser-Meyer-Olkin

LCA Lifecycle Assessment

n.s. not significant

OEF Organization Environmental Footprint

p. Page

PEF Product Environmental Footprint

PoP Point-of-purchase

URL Uniform Resource Locator

US United States

VIF Variance Inflation Factor

List of figures and tables

Figure Title Page

Figure 1 Conceptual Model 4

Figure 2 Meat choices of high vs. low behavioral intention among participants 22

Table Title Page

Table 1 Overview of the key criteria for constructing an ecological footprint-label 15 Table 2 Impact of meat and meat-alternative products 16

Table 3 Sample characteristics 19

Table 4 Measures and reliability 21

1 Introduction

Global meat consumption is at an all-time high. The past decades have seen rapid growth of 1 population and per capita income in developing countries which, accompanied by the greater demand for more animal proteins, are the main drivers behind this phenomenon and are predicted to continue to be in the future. At the same time overindulgence in developed countries, specifically in meat and dairy products, is on the rise. Yet throughout the EU a growing part of the population (80%) has an interest in reducing their meat consumption for environmental reasons (European Commission 2015; DG Environment 2013; FAO as cited by European Commission 2015; OECD/ FAO 2016). Indeed, livestock’s contribution of approximately 18% to total global greenhouse gas (GHG) emissions (Gerber et al. 2013), in combination with other environmental consequences such as eutrophication and acidification (Tukker et al. 2006) as well as its direct impact on human health (i.e. processed meat as cause of cancer) (IARC 2015) pose a real short-term and long-term threat to consumers. It is paramount to remember, however, that we make the choices and decide upon our consumption habits.

To this measure, previous research has attempted to explain pro-environmental behavior and its drivers for many years and across multiple scientific branches (see e.g. Bezawada and Pauwels 2013; Diamantopoulos et al. 2003; Fransson and Gärling 1999; Grunert and Juhl 1995; Steg, Dreijerink, and Abrahamse 2005; Verhoef and van Doorn 2015). For example, Grunert and Juhl (1995) examine values and environmental concern as factors of organic buying behavior. Fransson and Gärling (1999) identify knowledge, personal responsibility, and perceived threats to personal health among those factors. Further researchers have considered socio-demographics for profiling of green consumers (Diamantopoulos et al. 2003), and others focus on the acceptability of policies through the value-belief-norm theory by Stern (cf. Stern 2000; Steg, Dreijerink, and Abrahamse 2005). Finally, Bezawada and Pauwels (2013) shift their attention onto supply-side characteristics such as price, availability, and promotions. Whereas Verhoef and van Doorn (2015) conceptualize a framework for both supply-side and consumer-side characteristics. Many of the aforementioned deal with these issues in the context of organic product purchases, and while it is insightful to have this research, it does not necessarily address the serious problem mentioned here. Particularly so, as organic products do not always perform better in terms of their carbon footprint than regular

Meat throughout this paper is defined as all animal flesh. This also includes fish.

products (Seufert, Ramankutty, and Foley 2012; Tuomisto et al. 2012). This is accountable to more land use and lower yields as well as other negative aspects such as higher eutrophication and acidification potential per unit. The objective of this study is to extend the literature in the field of sustainable marketing by specifically looking at meat consumption and means to reduce it via the use of a label-system, introduced further below.

Previous research has found changes in purchasing behavior to create greater environmental benefits than the reuse or recycling of available products (Gardner and Stern 2002). As much of consumers’ daily behavior is automated through habits, so is the shopping for groceries. Considering meat consumption as a ‘bad habit’ implies that in order to break it, a substantial amount of effort is required. The literature therefore offers three possibilities to achieve this (Jager 2003). (1) Making the performance of the habit impossible by removing its stimuli. (2) Providing clear and direct information on the negative long-term outcomes of the habit, while simultaneously promoting positive short-term outcomes of an alternative behavior. (3) Offering short-term benefits of an alternative behavior which facilitate the acceptance and adoption for a new habit to emerge. With regards to the first point, it would be impossible to ban all types of meat products from current supermarkets immediately. This paper will therefore mainly focus on point two and to some extent point three.

introduction of an international, standardized measurement for products’ ecological footprints (cf. DG Health and Consumer Protection 2005).

To achieve this, the EU has commissioned multiple pilot programs between 2013-2016 (with results coming in 2017) to assess product life cycles in a uniquely robust and scientific approach (European Commission 2016). This assessment considers both the Product Environmental Footprint (PEF) as well as the Organizational Environmental Footprint (OEF). Thereby possibly initiating policy changes that could have an influence on large-scale behavioral tipping for the majority of EU consumers that already intend to reduce their meat consumption (Nyborg et al. 2016). Grunert and Wills (2007) however point towards the serious lack in literature as to how labelling information is used and how it will affect consumers’ dietary patterns.

The current work aims to contribute to this research from a Marketing perspective by addressing the issue of effectiveness and influence of a unique eco-footprint-label, as could be introduced by the EU Commission. Specifically, this research investigates consumer choices for eco-labels in the context of meat consumption. The goal of this paper is hence twofold. First, it aims to investigate new as well as already inspected drivers behind meat consumption and potential alternative products (e.g. Tofu, Vegan burger patties). Second, it examines if the use of such a unique eco-footprint-label in combination with simple traffic light color-coding could facilitate the reduction of meat consumption (negative environmental traffic light [ETL]) by offering an alternative product (positive ETL). The use of a color-coded scheme stems from research in the provision of nutrition information (BLL 2008; Department of Health / Food Standards Agency 2016; WWF 2016). The current study builds upon a survey conducted predominantly among 115 young, well-educated professionals ascertaining their values and preferences.

2 Theoretical Framework

2.1 Theoretical review

For the purpose of this research, the employed approach is an adaptation and reduced version of the conceptual model by Verhoef and van Doorn (2015) (see Figure 1). The simplification being that this research will mainly look at the consumer-side-characteristics and only partially include supply-side-characteristics through the price variable. As the authors point out, this is reasonable since different supply-side factors can be of less (or more) importance for different consumers. Specifically, the current model only assumes that price will have a negative impact on the choice for a meat alternative product. That is, the higher the perceived price for such a product, the more likely it is that consumers will opt for the meat product even when a negative ETL is displayed on the package. This reasoning is in accordance with standard micro-economic theory and thus no individual hypothesis is formulated.

Figure 1. Conceptual model.

As found by Verhoef and van Doorn (2015), biospheric values are the most important driver of organic purchases. Consumers’ willingness to pay (a price premium) as well as them being affected by poor availability of those products in stores is reduced with an increase in biospheric value orientation. In fact, the authors propose that an exponential effect towards growing purchase tendency of organic products exists in consumers with very high biospheric values. In their model, they further do not find an effect of altruistic values on organic purchases. Importantly though, this only holds true for as long as biospheric values are included. Once biospheric values are excluded from the model, altruistic values indeed serve as a driver for organic purchases. This is in line with previous research suggesting positive effects of altruism with regards to pro-environmental consumption (Nordlund and Garvill 2002; Thøgersen and Ölander 2002).

Concerning social norms, Verhoef (2005) finds no significant effect on choice or frequency of organic meat purchases. However, there is vast evidence in literature that proposes substantial impacts of social norms on consumer behavior. For example, Childers and Rao (1992) find significant influences of peers on an individual’s product and brand decisions. Moreover, Goldstein, Cialdini, and Griskevicius (2008) show how descriptive norms in a hotel setting (e.g. “the majority of guests reuse their towels”) can nudge consumers into displaying pro-environmental behavior without even being aware of it. Lastly, one group of researchers was able to demonstrate how injunctive norms could change consumers’ littering behavior for the better or worse depending on the environment they find themselves in (Kallgren, Reno, and Cialdini 2000: Study 1-2).

Research has shown that consumers’ claims about pro-environmental attitudes do not always follow suit with their actual behavior (Kollmuss and Agyeman 2002; Chatzidakis et. al 2004). This phenomenon has been coined “attitude-behavior gap” or “intention-behavior gap”. For example, one might be strongly concerned with the environment and engage in small eco-friendly behaviors such as recycling, but consequently fail to make bigger commitments such as reducing car usage or meat consumption. Literature proposes a basic cognitive progression in consumer behavior models as: (1) beliefs form attitudes, (2) attitudes dictate intentions, and (3) intentions influence behavior (Carrington, Neville, and Whitwell 2010). The authors thus reveal two potential gaps within this theoretical construct – the first one between attitudes and intention, and the second one between intention and actual behavior. In this paper, the second gap will be explored to a certain degree as the intention to reduce habitual meat consumption is measured and afterwards compared to actual choices made throughout the survey.

some researchers have enunciated intricacies for consumers to differentiate certified and non-certified labels (Leire and Thidell 2005). The result of which is potential confusion among consumers as well as a curtailed credibility for all labels (Leire and Thidell 2005). It is therefore investigated in this paper how label skepticism might affect pro-environmental choices in a meat consumption setting.

A prerequisite for cooking meals that do not contain meat or substitute it with an alternative is that one has adequate expertise in the preparation of these dishes. Previous literature assumes a relationship between knowledge and behavior, where increases in knowledge imply greater leverage of attitudes on behavior (Fabrigar et al. 2006). Previous research has made the distinction between applicable knowledge (e.g. “What is the harmful effect of X on Y?”) and knowledge for action (e.g. “How can one save water?”) in the field of environmental behavior (Schahn and Holzer 1990). Clearly the current study refers more to the latter, although applicable knowledge might play a bigger role in the consciousness of consumers for environmental problems. As Tanner and Kast (2003) point out, knowledge for action is probable to exert greater influence on behavior. Hence, the broad scope of cooking can be related to the structure of habits which entails the purchase and preparation of food. Indeed, previous research suggests that our daily cooking is guided by scripts that are sequentially ordered and unexpectedly meticulous, comprising the act of purchase up to the very cooking, eating, and cleaning of dishes. Therein, at least eight different scripts have been identified (Blake et al. 2008).

As a closing aspect of the conceptual framework at hand, sociodemographic variables are included in the model for control purposes.

2.2 Hypotheses

Other-Oriented Consumer Characteristics

Biospheric and altruistic values. Biospheric values refer to the value orientation that reflects

H1. Biospheric values have a positive effect on the choice for a “green” alternative product.

Altruistic consumers tend to place the interests of others before their own and the influence of altruistic values on pro-environmental behavior is less apparent (Verhoef and van Doorn 2015). As previous research points out, altruistic values should lead to higher sustainable demeanors (Steg, Dreijerink, and Abrahamse 2005), although empirical evidence for this cannot be found very often (Nordlund and Garvill 2002; Schultz 2001). Nonetheless, the prevailing contention is that altruism positively influences pro-environmental behavior. Thus:

H2. Altruistic values have a positive effect on the choice of “green” alternative products. Social norms. People’s behavior has been found to be influenced by social norms even in the

environmental aspects of daily life (Goldstein, Cialdini, and Griskevicius 2008; Kallgren, Reno, and Cialdini 2000; Schultz 1998). Research thereby distinguishes between injunctive norms which refer to what is typically being approved of and descriptive norms which refer to what is actually being done (Cialdini and Goldstein 2004). Overall, norms in reference groups tend to play an important role for consumer decision making (Childers and Rao 1992; White and Dahl 2006). Hence, depending on the reference group a consumer belongs to and the social norms that are prevalent within that group, (s)he might choose to eat food in favor of the group in order to comply and avoid resentment. It is therefore possible that social norms have a negative effect on “green” alternative choices if a meat-based diet is dominant in a consumer’s daily social surroundings. Conversely however, and this is assumed here, a compelling attitude towards meat alternatives in one’s reference group affects the “green” alternative choice positively. It is therefore proposed that:

H3. Social norms positively affect the choice for “green” alternative products when the

reference group supports meat-alternative consumption

Self-Oriented Consumer Characteristics

Egoistic values. Consumers with an egoistic value orientation usually try to maximize individual

empirically supported by all researchers (Stern, Dietz, and Kalof 1993). On the contrary one could also argue that an egoistic value orientation could potentially advance pro-environmental behavior if a consumer were to change their diet to a meatless one out of pure egoistic motivation for their own health benefit (Schifferstein and Ophius 1998). Nonetheless, from the first theoretical standpoint the following hypothesis is derived:

H4. Egoistic values negatively affect the choice for a “green” alternative product.

Intention to reduce meat consumption. As awareness towards the consequences of consumption

become more prominent in the minds of consumers in the developed world, so are recommendations to dietary changes (Reynolds et al. 2014), reduced portion sizes, and suggestions to reduce the individual frequency of meat consumption (e.g. through “meatless” days) (Dagevos and Voordouw 2013; de Boer, Schösler, and Aiking 2014). This new consciousness may lead consumers to change their behavior and subsequently reduce intake of daily meat consumption. At the beginning of this stands the intention of consumers to change their behavior and possibly replace current meat products with meat-alternative ones. This potentially leads to the formation of new habits long-term (cf. Jager 2003). The following hypothesis therefore arises:

H5. An intention to reduce meat consumption will positively affect the choice for the “green” alternative product.

Meat alternative cooking skills. Just as buying meat in the supermarket can be considered habitual

behavior (cf. Jager 2003), so can its preparation and cooking. It goes without saying that people tend to rely on a set of different dishes that they are able to cook and usually do not look for new creations each week as this requires energy, time, labor, and skill (cf. Leroy and Degreef 2015). As research points out, consumers tend to stick to what they know, which implies that they prefer foods they are used to for convenience reasons (Nyborg et al. 2016). It can therefore be assumed that the more knowledge a consumer has about meat alternative recipes and the skills to cook those, the more likely (s)he will be to accept “green” meat alternatives. Thus:

H6. Meat alternative cooking skills positively influence the choice for “green” alternative

Label skepticism. Previous research has vastly studied the effects of nutrition and health claims in

the context of package design (Ford et al. 1996; Keller et al. 1997), advertising (Andrews, Netemeyer, and Burton 1998), and even restaurant menus (Kozup, Creyer, and Burton 2003). This has not been the case, however, for labels concerned with the environment, although some of these are likely to be transferrable. Yet, as mentioned in the beginning, about half of EU consumers raise their doubts about the credibility of eco-friendly product claims (DG Environment 2013). Therefore, consumers who are highly skeptical towards labels are assumed to be less likely to make a pro-environmental choice and rather stick to their intrinsic belief about a product’s attributes. These consumers might thus choose a green ETL label or even go as far as trusting non-labelled products more than any sort of non-labelled food. This leads to the following hypothesis:

H7. Label skepticism negatively influences the choice for “green” alternative products. Interaction Effects

As pointed out by Verhoef and van Doorn (2015), interaction effects could occur between supply-side factors and some of the consumer characteristic variables. Accordingly, the current paper will examine the interaction effect between price perception and biospheric values. The reasoning follows from Bezawada and Pauwels (2013) who found that for consumers with high biospheric value orientations the higher price of organic products is less relevant. Thus, it is tested whether this moderation effect will apply for the present case of vegan products, as well.

Control Variables: Demographics

meat-heavy diets have a significant effect on perceived masculinity (Rozin et al. 2012). Concerning the age variable, there is no clear empirical evidence about an age–pro-environmental behavior relation (Dietz, Stern, and Guagnano 1998; Thompson 1998). Moreover, income plays an important role in consumers’ purchasing decisions and those with high income should be better equipped to engage in pro-environmental purchase behavior, due to the common premium attached to these products (Verhoef and van Doorn 2016). Additionally, since the complexity and scope of environmental issues and the research thereof are not always easily intelligible, consumers with higher education might be inclined to appreciate pro-environmental behavior more readily (Dietz, Stern, and Guagnano 1998; Ngobo 2011). Finally, sociodemographic variables might mediate and thus detriment or amplify the effect that the consumer characteristic variables have on the outcome of the model (Ailawadi, Neslin, and Gedenk 2001).

2.3 Labels and product lifecycle assessment

Labels. Labels in general can be categorized according to three different criteria (Mudgal et al.

information fully or apply other processing methods such as heuristics to form an attitude towards a label. Consumer characteristics as the ones described in the current model will likely be a determinant of the amount of elaboration an individual will exert.

With the vast tally of labels currently on the market it appears practically impossible for consumers to fully comprehend and have extensive knowledge on all of them. Yet this understanding is vital for the efficacy of any eco-label should consumers not resort to heuristics. Discerning the meaning of an eco-label is dependent on an individual’s existing knowledge and the coherence with personal goals associated to it (Grunert 2011). That is, the more a person’s knowledge about ecological issues coincides with the pursuit of their inner goals, the more likely it is that a cognizant meaning will be attached to the label. Because the information behind an eco-label is not always easy to fathom it sometimes requires an deliberate act on the side of the consumer to find information elsewhere (Thøgersen, Haugaard, and Olesen 2010).

Nevertheless, none of the above matters for the objective of using a label if consumers do not trust the label they are exposed to. This is paralleled by the perceived credibility mentioned earlier, to which the copious quantity of labels is counteracting. “Greenwashing” describes the phenomenon of consumers distrusting or suspecting the advertising message as being deceptive (Atkinson and Rosenthal 2014). The ramifications of which are elevated levels of skepticism and lower purchase likelihood for products with pro-environmental claims in general, according to the same authors. Trust is therefore a pivotal factor before fully engaging in the adoption of eco-labels, which happens when a product containing the label is actively, repeatedly, and consistently considered (Thøgersen, Haugaard, and Olesen 2010). Once this adoption has taken place, consumers are likely to fall back to a state of low involvement and heuristic decision making with regard to that label (i.e. choosing a product with that particular label because they are familiar with its implications).

Lifecycle Assessment (LCA). As the ongoing studies by the European Commission show, assessing a

authors reveal that packaging, business-to-business products, human and ecotoxic impacts, waste, as well as impacts on biodiversity, among other outcomes are often not considered in studies assessing a product’s environmental impacts.

Recent investigations in the field of sustainable production have attempted to evaluate a basket of average food and beverage items in the European market (Notarnicola et al. 2017). Their results suggest that the food categories with the highest environmental impact are meat and dairy products, concluding that the agricultural phase of a product’s lifecycle has the highest impact of all foods investigated. This falls strongly in line with prior research on the matter, all indicating the same elevated impact for these two product types (e.g. Tukker et al. 2006; Fisher et al. 2013). The reason for dairy’s large impact is the fact that it is a byproduct of meat production. New calves must be born and raised in order to keep up with demand of meat. In order to feed their heifer the mother cows thus produce milk. The two are then separated to extract the milk from the mother for further processing in dairy products, while the offspring receives feedstuff. To clarify things amidst this potential confusion, the present paper will only consider the impacts of meat consumption specifically.

However, combining the findings from previous literature of LCA into a single coefficient for appropriate usage in an eco-label is a goal yet to be attained. If one considers the chain of causation: that the creation of products entails an environmental impact, which then can be measured and combined into a single coefficient that is subsequently put on a label for consumers to make a meaningful choice – then this paper is trying to investigate the very end of that chain. Namely, whether the use of such an eco-label is effective or not. In that regard, it is not an objective of this research to constitute a correct environmental coefficient and it seems only reasonable that the labels created throughout this work are solely based on the most recent and scientifically accepted data available. They may hence not represent the true depth of environmental implications.

3 Research Methodology

3.1 Experimental design

To test the hypotheses and in consultation with fellow researchers , it was decided that an online 2 survey would be the most efficient means of gathering the relevant data for the purpose of this

Thanks to Wander Jager for guidance in this decision.

research. This is underlined by the fact that sufficient amounts of data can be acquired in a relatively short period of time to reach statistical power in later analyses. A further benefit of this setup is the convenience for participants to fill out the questionnaire without having to stick to a certain timeframe. At the beginning of the survey, respondents were randomly assigned to one of two experimental conditions by the computer software: control group vs. label condition. In the control group, participants were simply confronted with a choice between a meat vs. a meat-alternative product (in this case always vegan). This condition solely displayed the two product packages side by side and respondents could select the one they preferred most. Participants in the label condition, on the other hand, were confronted with the same side by side product choices, albeit seeing one important change. In this condition, each of the product packages was accompanied by an ecological footprint-label, showcasing the environmental impact that this product would have. In total, each participant of an experimental group was exposed to five different choice sets, each displaying two options (meat vs. vegan). These dichotomous decisions thus served as the dependent variable of this research. In addition, participants of the survey were asked to answer questions pertaining to their values, as well as other individual consumer characteristics and demographical backgrounds. These scale measurements, taken from previous literature and sometimes created by the author of this article, served as independent variables.

3.2 Label construction and package design

features of this label is the application of a color-coding scheme or traffic light system, classifying the environmental impact of a product into red (large impact), yellow (medium impact), and green (low impact), as suggested by Mudgal et al. (2012).

In preparation for the choice part of the survey, product packaging of meat as well as their vegan complements had to be obtained. This was achieved through several Google Image search queries. The criteria for a food packaging to be chosen were multiple subjective, yet logical prerequisites. First, clear visibility of the product was important. Second, both meat and vegan packaging had to be roughly the same shape. Third, largely unfamiliar brands within the European market were chosen as to avoid any confounding biases that a participant might have towards a product’s brand. Fourth, relatively equally appealing products had to be displayed on the packaging to avoid possible favorability due to appearance. Fifth, popular food items were chosen to increase familiarity with the products and prevent uninformed product choices. Sixth, different types of meat were selected as to thwart any preference bias. This resulted in a total of five products (one beef, one chicken, one fish, and two pork) being selected with their respective vegan counterparts, mainly consisting of soy. Finally, due to difficulties arising from the restrictions in finding adequate product packages that met these criteria, it should be mentioned that all vegan products were from the brand Quorn, while no meat product had the same brand.

To prepare the product packaging for the label condition, each product had to be assessed regarding its environmental impact for the purpose of being able to put an environmental label on it. Assessing a product’s environmental impact throughout its lifecycle is one of the major challenges researchers are facing today, as pointed out in the beginning. To tackle this challenge, it was referred to Fisher et al. (2013) who investigated more than 150 studies on the environmental impact of grocery products and summarized data from the European Commission research project into the Environmental Impact of Products (EIPRO). From their carbon footprint data evaluating 217 food,

Table 1

Overview of the key criteria for constructing an ecological footprint-label, based on findings by Mudgal et al. (2012) • One single aggregated metric with up to three individual indicators.

• The information should come from a trusted source, ideally, a third-party and not the manufacturer.

• Information made available in brochures and websites can support the limited information on a product-label. • Quality of the information is more important than the quantity, as too much information inhibits decision-making. • General terms for indication are preferred over technical ones (e.g. “climate change” vs. “CO2-equivalent”).

drink, and other grocery data based on GHG emissions, all meat and meat-alternative products were extracted and compiled in table 2. The table shows these products and their annual median GHG emissions as well as their contribution to GHG emissions. These are ranked in descending order from rank 1 having the largest impact, to rank 217 having the smallest impact. Defining the threshold and deciding which of the selected products from the survey belonged to which category was not a trivial task. Ultimately it was decided to classify all those product as having a large impact that individually contributed at least 1 or more percentage points to total annual GHG emissions. This included all products from rank 1 to rank 36. Products falling into the category of having a medium impact on the environment were all those, where at least two or more product categories were necessary to evoke a 1% contribution to total annual GHG emissions. This concerned products ranking from 37 to 80. Lastly, products with a low environmental impact were all those, where at least 5 product categories or more were necessary to constitute a 1% contribution to total GHG emissions. This was the case for ranks 81 to 217. It should be mentioned that these cut-off values have been chosen from a subjective standpoint by the author of this paper and do not

Table 2

Impact of meat and meat-alternative products, extract based on the research by Fisher et al. (2013) Product name Annual GHG emission _{(Mt Co}

2e) - Median

% GHG Contribution

(based on median) Rank (out of 230) Beef, fresh 5.12 7 % 2

Red meat, frozen 2.31 3 % 4 Poultry, fresh 1.66 2 % 11 Lamb, fresh 1.59 2 % 12 Chilled fish/seafood 1.37 2 % 15 Canned fish/seafood 1.21 1 % 20 Pork, fresh 1.08 2 % 24 Frozen fish/seafood 0.88 2 % 29 Canned meat products 0.58 1 % 37 Poultry, frozen 0.23 0.5 % 74 Meat substitutes, frozen 0.09 0.1 % 123 Meat substitutes, fresh 0.07 0.09 % 134 Total 16.19 22 %

reflect a scientifically agreed upon categorization. Yet, as it was not the goal of this research to propose specific coefficients, but rather to test the effectiveness of an ecological footprint-label on product packaging, the applied method was deemed satisfactory.

In a final step the selected products for the survey were prepared in a design program (Sketch Version 43.2) by editing the correct eco-label both onto the package and in a magnified version below it for better visibility. This was done according to the classification of its environmental impact.

3.3 Questionnaire

dichotomous. For a full overview of the items used, including the literature they were taken from, see Appendix C. It is also noteworthy that the questionnaire was available in both English and German to facilitate comprehension for potential respondents.

3.4 Data collection

This research employed one type of data, namely: individual-level survey data concerning consumer characteristics and sociodemographic information. Before collection of the official data began, pretesting of the survey was conducted with a total of six acquaintances of the author (3 German, 3 English). These respondents were asked to offer their constructive and objective feedback regarding (a) orthography, (b) intelligibility, (c) visibility of the products and labels, and (d) to estimate the time for the questionnaire. This resulted in useful improvements for the actual study. Once ready, a generic URL from Qualtrics was used to distribute the survey via social networks and private mailing lists of the author in May 2017. Recipients of the link were kindly asked to share the survey with their friends, family and acquaintances to create a certain network effect and thereby accelerate data collection.

3.5 Sample

educated ones are under-represented in this research. These demographics could possibly be explained by the fact that the survey was distributed via the Internet and likewise somewhat biased through the author’s network effects. This implies that the results of the present research will be meaningful especially for young and well-educated consumers, while presumably being less convincing with respects to the elderly or less educated consumers. In total, 55 respondents (48%) were randomly allocated to the control condition, while 60 respondents were placed in the experimental condition (eco-label on product packaging). Multiple t-tests on the mean of the sociodemographic variables (e.g. age, income) between the two groups were conducted, none of which succeeded in reaching significance. Thus, there was no significant difference in sociodemographic properties between the two conditions (p < 0.05). Lastly, an assessment for relations between gender and education was conducted in form of a chi-square test. The results were not significant with X2 _{(4, N=115) = 3.36, p = .50, indicating that no significant educational} differences existed between the genders.

3.6 Measurement

For the purpose of measurement, SPSS (Version 23.0) was used. After cleaning the dataset of incomplete surveys and respondents who indicated to abstain from eating meat, as described above, some variables where recoded to better serve the analysis.

Reliability and validity of measures. To assess the different multi-item constructs of the

questionnaire towards their validity, the inter-item correlations were calculated. Those items with low coefficient alphas due to low inter-item correlations were hereby removed. While the majority of alphas exceeded the critical threshold of .7 (Nunnally and Bernstein 1994), two items on the Social norm scale were deleted in two consecutive steps. This increased internal consistency of the scale from .516 to .742 and thus replicates findings within literature (Verhoef 2005). Table 4 reports

Table 3

Sample characteristics (n = 115)

Age Highest education Monthly net income

the measures and reliability for the attitudinal measures throughout the survey. In continuation, principal component analyses were carried out which resulted in eight factors for the eight different constructs. These factors all had eigenvalues greater than 1, a KMO index greater than .5, and communalities greater .4. They explained approximately 60% of the variance, and scree-plots indicated that each single factor did, indeed, measure the right construct. Thus, all items loaded on their respective constructs and all scales were one-dimensional.

Analysis of variance. To investigate significant differences between the control group and the

experimental group, the number of times a respondent opted for a meat product in the choice experiment were counted and combined within a single variable. A one-way analysis of variance (ANOVA) was then carried out between the two independent samples, together with a test for homogeneity of variance. The fact that the two sample sizes were unequal in their population did not hinder the analysis, as this is not a requirement (Malhotra 2010).

Econometric models. Multiple binary logistic regressions were conducted to examine the effect of

4 Empirical Results

4.1 Descriptive statistics

The mean averages for weekly meat and dairy consumption were 5.23 and 6.07, respectively. This implies that respondents consumed these types of products on the majority of days per week (5 days and 6 days). Regarding the consumption of meat-alternative products, Appendix E depicts participants’ answers in a graphical way. The overall plurality of respondents indicated to ‘not know the given product’ (16%) or to ‘never eat it’ (45%). ‘Rarely’ eating them were 9%, while 4% admitted to eating them ‘occasionally’. Approximately one-fifth of participants (22%) stated to eat the mentioned products ‘sometimes’, whereas a small minority confessed to ‘frequently’ (2%), ‘usually’ (1%), and ‘always’ (1%) eating them. The most unknown product, by far, was Tempeh for which 78 (68%) respondents claimed they did not know the product. The most known product, on the other hand, was Tofu of which only 2 respondents reported to not know the product. The most popular meat-alternative appeared to be Falafel, of which 77 (67%) subjects disclosed to eat it at least rarely or more. Appendix F shows the frequency table for the choices between meat and meat-alternative options made by respondents for every product.

With regards to the theoretical gap between intention and behavior (Carrington, Neville, and Whitwell 2010) as mentioned earlier, figure 2 demonstrates the empirical findings. Participants ranking high on the behavioral intention scale (Likert-scale points 5-7) (N = 54) were compared with those ranking low on the scale (points 1-4) (N = 61). The low-ranking group included the neutral point 4 to simplify examination of the two groups with respect to their potential behavioral gap. As figure 2 clearly depicts, there does not seem to exist any major difference between those

Table 4

Measures and reliability.

respondents who stated that they had an intention to reduce their meat consumption in the future compared with those that did not make this claim. This is with regard to the subsequent choices between meat and meat-alternative products made in the experiment. The only striking thing appears to be the fact that some subjects with low behavioral intention selected meat in five out of five times, whereas this never occurred among subjects with high behavioral intention.

4.2 Estimation results

Analysis of variance. An ANOVA was conducted to assess the main question of this paper as to

whether significant differences in choices of meat-alternative products exist between the two experimental conditions or not. The results indeed show a statistically significant difference between the groups, albeit not in a way that was expected. The mean of respondents opting for a meat instead of a vegan option in the control condition was 2.2, while the mean in the label condition was 3.0. This implies that individuals in the experimental condition chose, on average, almost one more meat product than those in the control condition – which is somewhat paradoxical and leaves room for discussion in the appropriate section below. To elaborate upon this result, multiple ANOVAS for the consumer characteristic variables were conducted, which all failed to show significant differences between the two experimental conditions. This re-affirmed the initial finding by showing that no other group differences potentially distorted the result. It can thus be concluded that an environmental impact-label on product packaging does not yield the desired result of altering people’s choices towards a meat-alternative option when faced with a decision between meat and vegan options in an experimental setting. Therefore, other factors may be at play.

Binary logistic regression. To assess the effect of individual consumer characteristics on the choices

between meat and meat-alternative products, multiple binary logistic regressions were performed. This implied regressing the consumer and control variables on the dichotomous choice variables for each choice set in both conditions. Thus, a total of ten binary logistic regressions were conducted. Looking at the classification tables, it was discernible that the predictive capacity of each of the models was positive. They all increased in their correct classifications of predicting meat-alternative

Figure 2. Meat choices of high vs. low behavioral intention among participants.

Re sponde nt s 0 3 6 9 12 15 18 Meat choices 0 1 2 3 4 5

choices as opposed to the null-model with no variables except the constant. Nonetheless, the results were unexpected, as the bulk of the consumer variables failed to indicate any significant impacts on the choice for meat-alternative products, based on the Wald-statistic and p-values (p < .05). The only exception in the label condition was behavioral intention for the chicken product (β = 1.41, p = .013), indicating that respondents with higher behavioral intention to reduce their meat consumption in the future chose the vegan chicken more often. In the control condition, behavioral intention was significant for the burger product (β = –1.720, p = .005), which suggests that every increase in scale points of behavioral intention by one unit leads to an average increase in meat-product selection by a factor of almost two. This is somewhat counterintuitive and could potentially be attributed to the phenomenon of cognitive dissonance (Kunst and Hohle 2016). Further effects in the control condition were found for the sausage choice set. As biospheric values (β = –1.08, p = .046) and behavioral intention (β = –1.01, p = .026) both negatively, but statistically significant, affected the dichotomous decision between meat and vegan option. This finding suggests that consumers with a higher biospheric value orientation and a higher behavioral intention were less likely to choose a vegan option, respectively. Again, this effect could conceivably be ascribed to cognitive dissonance. In summary it can therefore be concluded that H1 and H5 are partially supported, although very weak and only in certain cases, whereas H2-H4 and H6-H7 are not supported. The overall estimation results for these regressions in combination with their hypotheses are summarized in table 5. The findings hence stand in stark contrast to the results of Verhoef and van Doorn (2015) who find significant effects for biospheric, altruistic, and egoistic values, as well as education and gender. Notwithstanding the fact that these authors were examining purchases of organic products.

Interaction effects. As pointed out by Verhoef and van Doorn (2015) however, consumer

perception and biospheric values. This holds true for all ten binary logistic regressions, as well as the linear regression model investigated below.

Robustness checks. Two robustness checks were performed to validate the previous findings. First, a

linear regression analysis was executed with the dependent variable defined as the total number of times individuals of both groups chose a meat-alternative (vegan) product. The same independent variables as before were retained. The analysis again was unsuccessful in reporting significant effects of the consumer variables (all p-values > .05). Thus reinforcing the result that the given consumer characteristics are not suited predictors for whether an individual chooses a vegan product or not. Secondly, the variation inflation factor (VIF) scores of the linear regression model were examined. These ranged between 1.12 and 1.90, alluding that multicollinearity did not appear to be an issue (Hair et al. 2009). Because some variables were high in their correlation (i.e. altruistic and biospheric values, intention and knowledge), further models were computed in an iterative process, where one of these variables was left out (cf. Verhoef and van Doorn 2015). This yielded very similar results as the ones before. However, in the model without behavioral intention the effect of label skepticism became significant (β = .259, p = .047). This indicates that behavioral intention and label skepticism might coincide to a certain degree.

Sociodemographics and attitudes. As previous research revealed, there might be a mediating role of

psychographic variables for the effect of sociodemographics on purchase behavior (Verhoef and van

Table 5

Estimation results for the binary logistic regression models.

Variable (expected sign) Hypothesis Main effect Conclusion Biospheric values (+) H1 partially significant* partially supported

Altruistic values (+) H2 n.s. not supported

Social norms (+) H3 n.s. not supported

Egoistic values (–) H4 n.s. not supported

Behavioral intention (+) H5 partially significant* partially supported

Knowledge (+) H6 n.s. not supported

Label trustworthiness (–) H7 n.s. not supported

Price perception n.s. Price perception × biospheric n.s.

Doorn 2015). Therefore, the sociodemographic variables were excluded from the model to test the findings for any mediating effects that these variables may exert between the relation of the independent variables on the dependent variable. This procedure did not result in any considerable changes among the effects of the independent variables, neither in the binary logistic regressions nor in the linear regression. This implies that the sociodemographic variables did not exert any influence on the relation between the consumer characteristic variables and respondents’ choices for meat-alternative products in the current model.

4.3 Discussion

The need for change in our environmental behavior is irrefutable. This research focused on the behavioral patterns of habitual meat consumption. The first goal was to investigate new and old drivers behind meat and meat-alternative consumption. The second goal was to investigate whether the use of an ecological footprint-label could influence consumers’ choices towards meat-alternative products when faced with different choices between the two types of products. Of the 7 proposed hypotheses, only 2 were partially confirmed. The remaining majority did not find significant support in the applied model. This stands in stark contrast to the findings of Verhoef and van Doorn (2015), who investigated some of the variables used throughout this paper in the context of organic purchases.

Drivers behind meat consumption.

As for the first goal of this article, and partly in line with prior research (e.g. Steg, Dreijerink, and Abrahamse 2005), biospheric values were one of the variables that could explain choices towards meat-alternative products to a small, but significant extent. Although this was only the case for the choice of sausages, and only in a condition where an ecological footprint-label was absent. It is assumed that this effect will be a rather random incident among the various portrayed choice sets. Certainly, the results measure significance from a statistical point of view, yet given the context and the one-time occurrence it is not easily intelligible to find a logical explanation for this sensation. This is further undermined by the fact that various choice sets with different kinds of meat were at display during the experiment. Hence, presuming that biospheric values would only have an effect on one specific type of meat in a real-life consumer setting deems fairly unlikely.

for the chicken choice set in the label condition, and negative main effects on the burger and sausage choice sets in the control condition. In spite of the obvious inconsistencies, arguments attempting to explain the chicken case can be found. It is possible that a behavioral intention to reduce meat consumption, in combination with an eco-label reinforcing this desire, is first tried out with a meat alternative product that is believed to be similar to the original. Most consumers would agree that the taste of chicken is far less intense than that of beef or pork and additionally consists of a more tender structure. This may give consumers reason to believe that both the taste and structure of chicken are easier to emulate in form of a meatless counterpart. As a consequence, chicken might constitute as a gateway product for meat-alternative consumption.

found statistically sound effects of altruistic values. These findings could not be replicated in the present study, however. While it is true that both biospheric and altruistic values are highly correlated in a positive way, indicating a good theoretical base for further analysis, their failure to reach statistical significance post hoc implies that these variables are not suited predictors for meat-alternative choice behavior.

With regard to social norms, previous findings investigating their impact on the choice for organic meat could be confirmed in the present study (Verhoef 2005). Both papers find no significant effects of social norms on their respective choices of interest (here: vegan). This implies that social norms might not play such an important role in influencing individual meat consumption behavior. It therefore stands in contrast to other types of environmental behavior where social norms do play a role (Goldstein, Cialdini, and Griskevicius 2008; Kallgren, Reno, and Cialdini 2000; Schultz 1998). Another argument for the insignificant effect of social norms might be that an individual’s own preferences towards food, and especially meat consumption, are simply too strong to be influenced by their peers. It is noteworthy though that the scale applied here was taken from previous literature and might not be optimal in measuring its desired construct. This is backed up by the fact that the reliability measures in form of Cronbach’s alpha for the factor score of this scale was the lowest among the employed scales in this research. Furthermore, two items had to be deleted from this scale in order to reach a Cronbach’s alpha > .7, due to low inter-item correlations. The main effects of the collapsed factor might thus not represent a true image of the influence that social norms exert on respondents’ choices.

Interestingly, knowledge about preparation and cooking of meat-alternative dishes had no significant effect on the choice for vegan products. One possible explanation for this outcome might be the fact that in the present experiment only popular food items were displayed. This may have led respondents to make the correct assumption that no particular skills were required in preparing these dishes, as these were rather easy in their nature to cook (e.g. burger, chicken nuggets). Another possible explanation for the absence of significant effects might include the general lack of knowledge for cooking among today’s consumers (Leroy & Degreef 2015). The authors point towards the industry trend in creating “convenience” food items that include instant food, ready meals, and the expansion of urban fast food cultures. As drivers behind these occurrences they mention altering lifestyles, such as variable family eating times, the emergence of small and single-households, impulsive consumerism, and consumer deskilling with respect to cooking abilities, among others. This relates well to the study at hand since a disproportionately high number of respondents consists of young consumers who fall within these categories and hence are likely to make use of such “convenient” products. With regard to the future, these symptoms should be of concern, as consumers with little knowledge of cooking might discard information on the content and ingredients of their processed meals altogether. Thereby conceivably fostering the consumption of meat, sugar, and other harmful foodstuffs that could lead to cancer, diabetes, and other illnesses (IARC 2017).

Relating these findings to the ones at hand, a clearer picture might emerge. Although the eco-label shown in this study claimed its approval to be of an independent party, unfamiliarity with the label itself as well as a general disassociation regarding meat consumption and its environmental impact may have led participants to ignore the information they were presented. This finding might even be subject to a more systemic problem, as distrust in science and official institutions coupled with the spread of false knowledge in the information age fortify untenable prejudices and discredit true assertions.

Instead of the actual prices, price perceptions of consumers with regard to vegetarian and vegan diets compared to meat-based diets were measured. The results are, again, insignificant for this variable, suggesting that participants in the study did not make their choices between meat and meat-alternative products based on perceived price differences. This is promising to a certain extent as it implies that, although usually priced higher than their original counterparts, meat-alternative products have the same chance as meat-products to be bought by consumers with respect to the price variable. Granted, this deduction is made from a very theoretical standpoint. Yet environmental marketing trends as discussed by Menon and Menon (1997) suggest that businesses should seize the opportunity to launch even more aggressive marketing campaigns promoting the meat-equal characteristics as well as personal and global benefits of their meat-alternative products, to convince consumers into buying them. Indeed, surveys of EU citizens reveal that more than 77% of them have a higher willingness to pay for eco-friendly products, provided they are convinced about the products true environmental merits (DG Environment 2013).

Effects of an eco-footprint label.

goes on to explain that one of the strategies to defuse this threat is resentment (i.e. disliking and distancing) to the perceived threatening object. As the environmental implications of the choices were clearly visible when an eco-label was displayed, the correct moral behavior was also in plain sight. This may have led respondents to resent the meat-alternative due to a conflicting moral comparison between their own preferences and those of vegetarians or vegans. Indeed, the same previous research has found that omnivores tend to think they are seen as morally inferior by vegetarians and thus develop a negative valence of words associated with vegetarians (moral do-gooder derogation) – despite the fact that actual morality ratings of omnivores given by vegetarians were significantly less negative (Monin 2007). On another note, research by Cowburn and Stockley (2004) provides evidence that consumers tend to report high usage of nutrition labels, although their actual usage of them is often much lower. As reasons fort this the authors mention that consumers frequently do not understand the information that is given to them on the label. This finding could be applicable in the current article, though questions ascertaining this were not posed during the survey. It is possible that consumers in general proclaim usage of eco-labels to pursue social desirability (i.e. a better self-image of themselves among their peers) while de facto failing to understand the information they are provided. The concepts of social desirability and cognitive dissonance might thus be closely interlinked in the case at hand.

5 Conclusions & Recommendations

5.1 Managerial and policy implications

Implications for managers

Given the fact that there were almost no significant effects found for the variables used in the present study, this indicates that other factors influencing consumers’ meat choices seem to be at play. The author of this paper expects that the issue of taste might be an important one and thus recommends retailers to offer customers free taste samples to generate interest and demystify the skeptical attitudes that they might old.

change (e.g. OECD/FAO; WWF 2016) there is a heightened awareness among the general public that could be used to the manufacturer’s as well as the people’s advantage. This could ultimately lead to a win-win situation.

Managers are further pointed towards the assumption that meat-substitute products whose original counterparts are less intense in taste or not as culturally anchored might be more effortless to sell. The example of vegan chicken shows that consumers might be easier to persuade to purchase products of which they believe that they are easier to emulate artificially. This might be due to the less intense taste and a more tender structure. Some products may thus serve better as gateway products than others to convert consumers into repeat-purchase customers of meat-alternative groceries.

Implications for Policy makers

In light of the findings in this paper, policy makers such as the European Commission are cautioned with regards to the effectiveness of their eco-labels. The present results show significant differences between the investigated groups in the experiment. Yet these results show an average increase in meat product selection for those respondents who were exposed to an eco-label as opposed to those that were not. This could be due to psychological self-protection mechanisms as mentioned by Monin (2007). It could also be that the other designed labels by the EU Commission might show a more promising effect. The author therefore recommends to conduct extensive testing with focus groups across the EU to fully take advantage of the important results from the 5-year research program for developing environmental impact coefficients.

It might further be advisable to work more closely with retailers and manufacturers to elevate the efforts of bringing sustainable products to the market by utilizing the current trends in environmental marketing. This could potentially be achieved via subsidies and programs helping to defraud labels that do not live up to the high standards set by the European Commission. This could be especially pivotal as there are no clear guidelines for products that have to carry eco-labels and those who don’t, since this is a voluntary decision made by producers.

extent on these issues. This might be a challenge, as lobbyists from the meat industry will likely work against these measures, yet with regard to the current ecological situation it appears to be overdue.

5.2 Limitations and Future Research

This study has several limitations. First, it is apparent that the acquired data is a non-representative sample. Reasons for this are the relatively small sample size, as well as the skewed distribution of participants’ age. Thus, trying to draw general conclusions for larger populations through the outcomes of this study is an attempting task at best. It is further important to recognize that the data did not represent actual purchase data, but rather self-reported features of psychological variables as well as hypothetical product choices. Future research is therefore advised to conduct this experiment under more realistic circumstances, such as physical or online retailing with the presented eco-label attached to products.

Second, multiple limitations arise with regard to the questionnaire. It is possible that the obtained insignificant results for label effectiveness are specifically ascribable to the selection of the label design. This is due to the fact that the particular design used throughout this study was only one of three possible iterations recommended by previous literature (cf. Mudgal et al. 2012). Future research should address this issue by testing the effectiveness of the other potential labels. The fact that respondents were informed to participate in research for a master thesis might have further decreased label credibility, as subjects were unfamiliar with the displayed label, even though it claimed to be certified by an independent third party. Moreover, although the survey was offered in English and in German (including the eco-labels), the product packages remained in English at all times. This could have been detrimental, although members of the pre-survey that were illiterate in English did not remark this as something that altered their choices upon questioning. Prospective research could thus make a sophisticated attempt in including all these recommendations with respect to the survey in their examinations.