WHAT THE FOP: EXAMINING THE IMPACT OF FRONT-OF-PACKAGE SERVING SIZE LABELS ON PURCHASE INTENTION THROUGH ANTICIPATED

WHAT THE FOP: EXAMINING THE IMPACT OF FRONT-OF-PACKAGE SERVING SIZE LABELS ON PURCHASE INTENTION THROUGH ANTICIPATED

GUILT

by

SISSEL OPHOF

University of Groningen Faculty of Economics and Business

MSc Marketing Management June 17, 2019

Kleine Gelkingestraat 7C 9711 NG Groningen

(06) 22164439 sisselophof@gmail.com

s2760223

TABLE OF CONTENT

Introduction pp. 4 – 7

Theoretical Background pp. 7 – 13

Method pp. 14 – 16

Results pp. 17 – 24

Discussion pp. 24 – 31

References pp. 32 – 36

Appendices pp. 37 – 38

Abstract

This research is among the first to examine the effect of FOP serving size labels on purchase intention through anticipated guilt, and to examine the moderating influence of health consciousness and product category (i.e., vice vs. virtue). In order to do so, an online experiment with a mixture of between-group and within-group factors was conducted. The corresponding results demonstrate that FOP serving size labels do not affect purchase intention via anticipated guilt and neither health consciousness nor product category (i.e., vice vs. virtue) moderates the mediating effect of anticipated guilt on the relationship between FOP serving size labels and purchase intention. However, the results show that (1) anticipated guilt has a significant negative effect on purchase intention, (2) health consciousness has a significant positive effect on anticipated guilt, and (3) product category (i.e., vice vs. virtue) has a significant effect on anticipated guilt. These findings add to the current understanding of the drivers of consumers’

healthful decision-making and have important implications for public policy, managers and science.

Key words: FOP labeling; anticipated guilt; consumer decision-making; health consciousness; vices; virtues; serving size

Research theme: Consumer decision-making

Seminar: Master’s Thesis Marketing Start February 2019 First seminar supervisor: M.T. Van der Heide

Second seminar supervisor: L.M. Sloot

1. INTRODUCTION

Over the past few decades, and worldwide, overweight and obesity rates have been continuously increasing to great extent, forming the main risk factor in noncommunicable diseases, such as diabetes, that cause over more than half of all deaths worldwide (World Health Organization, 2003). The crux of this ever increasing global problem consists of an unhealthy diet. Worldwide, there is a shift in diet, typified by the consumption of more calorie-dense foods, that contain much fats and sugars, but little vitamins and minerals (World Health Organization, n.d.). In addition, there has been a recent trend within the marketplace, namely that of increasing the serving size of packaged foods. This trend has started since the 1970s and currently even exceeds the prescribed standards for dietary guidance, which considerably contributes to the prevalence of overweight and obesity (Young & Nestle, 2002).

As a result, there have been recent public policy interventions addressing these issues.

In the Netherlands, current policy obliges the depiction of nutrition labels mentioning the number of calories, fats, saturated fats, carbohydrates, sugar, protein and salt on almost all packaged foods. Besides, there are rules for the specific notation of these amounts, and amounts must be expressed per 100 grams or 100 milliliters. Remarkably, these values are also allowed to be expressed per serving size, when the serving size is clearly quantified and easily recognized (Nederlandse Voedsel- en Warenautoriteit, n.d). Likewise, in the United States, the Food and Drug Administration (FDA) designed a similar compulsory nutrition label for packaged foods, on which the nutritional amounts are also expressed per serving size (Food and Drug Administration, 2019). Public policies on nutrition labels differ throughout the world, and specifically, policies regarding nutrition labels expressed per serving size differ, mainly with regard to the flexibility of defining a serving size (Roberto & Khandpur, 2014).

Although research has shown that accurate and clear nutrition labels are indeed a way

to guide consumers in their choices, many consumers fail to properly interpret nutrition

information (Rothman et al., 2006). Yet, adequately interpreting nutrition labels is at the heart

of tackling the obesity epidemic. Research pinpointed some striking aspects. Firstly, consumers

consult nutrition labels only some of the time. Secondly, consumers are often unable to interpret

nutrition information in the context of their whole diet. Lastly, and most relevant in this

research, there are many consumers lacking the ability to estimate the nutritional amount of a

normal serving size, which is allowed to be depicted on the compulsory nutrition labels

(Tarabella & Voinea, 2013; Lando & Lo, 2012; Huizinga et al., 2009; Nederlandse Voedsel-

en Warenautoriteit, n.d). These findings exactly highlight the peculiarity of the current flexible

policies on nutrition labels expressed per serving size, namely, that of serving size

discrepancies, potentially influencing consumers’ decision-making. Marketers can adopt a relatively small serving size, to decrease reported nutritional values, whereby consumers may be influenced. To illustrate, consider the examples in figure 1. Dutch supermarket Albert Heijn defines a serving size of its fresh pizza ‘prosciutto’ as 215 grams, which contains 450 calories, while the whole pizza weighs 433 grams and contains 909 calories. In contrast, Albert Heijn defines a serving size of its frozen pizza ‘formaggi’ as 415 grams, containing 1080 calories.

However, in the latter instance, the serving size corresponds to the whole pizza. Hence, this research concerns serving size discrepancies (i.e., large serving size vs. small serving size), since rightful interpretation of nutrition labels and proper estimation of a normal serving size is a major component of the prevention and treatment of not only overweight and obesity, but also of the prevention and treatment of the cardiovascular diseases, which may be the consequence of overweight and obesity (Huizinga et al., 2009).

FIGURE 1

Example of serving size discrepancies in nutrition labels

Therefore, there is a global call for providing consumers with nutrition labels that are easy to understand: consumers are better able to make healthful choices when nutrition labels are clear and concise, and so, rapid to evaluate (Möser, Hoefkens, Van Camp & Verbeke, 2010;

Tarabella & Voinea, 2013). One of the initiatives proposed in order to address this problem, is the displaying of crucial nutrition information on the front of the package in different formats.

This type of labeling is generically called front-of-package (FOP) labeling, and gained

considerable attention worldwide. Many countries are experimenting with different FOP

labeling systems, such as Multiple Traffic Light labeling in the United Kingdom and the

depiction of the Choices logo in the Netherlands (Roberto & Khandpur, 2014; Tarabella &

Voinea, 2013). FOP labels are not only a way to simplify and encourage healthy decision- making, but it may also be a way to address serving size discrepancies, and educate individuals about adequate serving sizes (Roberto & Khandpur, 2014). Yet, there is little research concerning the effectiveness of FOP serving size labels in addressing serving size discrepancies, and thus, it constitutes a topic that requires attention.

Still, the question remaining unanswered is: how will large serving size versus small serving size FOP labels influence consumers’ purchase intention? The answer to this question is inevitably linked to the process of consumers’ self-regulation, in which emotions are considered to be an important driver (Tracy & Robins, 2004; Hofmann & Fisher, 2012).

Particularly, in the context of this research, there is built upon the premise that anticipated guilt is antecedent to purchase intention. This premise is supported by Wansink & Chandon (2006), who demonstrate that ‘low-fat’ labels increase consumption, since these labels decrease anticipated guilt associated with consumption. In addition, amongst others, Lindenmeier, Lwin, Andersch, Pau & Seemann (2017) and Rozin, Fischler, Imada, Sarubin & Wrzeniewski (1999), find that anticipated guilt mediates the impact of its antecedents (e.g., FOP serving size labels) on purchase intention. For instance, think about when you are trying to lose some weight, but in the supermarket, you struggle to choose between a tasty pizza and a healthy salad for dinner.

In such a situation, you may consult both FOP serving size labels, and when you see that a serving of the pizza has more calories than a serving of the salad, you may anticipate to feel guilty when intending to choose the pizza, and as a result, intend to choose the salad instead.

So, anticipated guilt is likely to be an important component in the effect of FOP serving size labels on consumers’ decision-making. However, it is plausible to assume that individual and contextual differences moderate the mediating influence of anticipated guilt. Firstly, since not all consumers rely on nutrition information similarly, the mediating effect of anticipated guilt is likely to be moderated by the extent to which consumers are health conscious. Health consciousness relates to the motivational factors that encourage people to make healthful decisions (Michaelidou & Hassan, 2008). Secondly, product category potentially moderates the mediating effect of anticipated guilt. Studies demonstrate that people associate energy-dense foods with tastiness and pleasure, and thus these foods are considered to be relative vices, compared to less energy-dense foods, which are associated more with long-term health goals and less tastiness, and thus are considered to be relative virtues (Van Doorn & Verhoef, 2011;

Raghunathan, Naylor & Hoyer, 2006; McCory, Fuss, Saltzman & Roberts, 2000). Prior

research suggests that product category moderates the effect of nutrition labels (e.g., FOP

serving size labels) on anticipated guilt. The effect of FOP serving size labels on anticipated guilt is likely to be more pronounced for vice foods than for virtue foods. FOP labels displaying a smaller serving size and thereby reduce reported nutritional values, increase consumers’

purchase intention, because it allows them to feel less guilty while enjoying their vice food (e.g., Wansink & Chandon, 2006; Kivetz & Keinan, 2006; Strahilevitz & Meyers, 1998).

There is an extensive body of research investigating the drivers of healthy food choices, but this research only considers some of the phenomena central in this present research. In addition, little research has been conducted concerning the effectiveness of FOP labels to address serving size discrepancies. As such, there is no comprehensive theoretical account particularly examining how FOP serving size labels affect consumers’ decision-making and what the influence of certain circumstances is. This present research contributes to literature by being among the first to integrate FOP serving size labels, anticipated guilt, purchase intention, health consciousness and product category (i.e., vice vs. virtue) into one study. The purpose of this research is to develop novel insights into how FOP serving sizes labels guide consumers’

decision-making through anticipated guilt. Such insights are of pivotal importance to design effective health interventions to tackle the obesity epidemic.

In sum, the research question of this thesis is: Do FOP serving size labels influence purchase intention? And, does anticipated guilt mediate the effect of FOP serving size labels on purchase intention? And, is this mediating effect of anticipated guilt moderated by health consciousness? And lastly, is this mediating effect of anticipated guilt moderated by product category (i.e., vice vs. virtue)? In order to provide an answer to this research question, this paper is structured as follows. It starts with a review of existing literature to explain how FOP serving size labels may influence purchase intention through anticipated guilt. Next, it continues by discussing the method, which explains how the formulated hypotheses are tested. Subsequently, the result section assesses the findings of the study. Finally, the discussion considers implications, limitations, recommendations for future research and ends with a conclusion.

2. THEORETICAL BACKGROUND Front-Of-Package Labeling

In the recent overweight and obesity epidemic, there are many interventions called into

life aimed at addressing this epidemic, aiming to assist consumers in making informed and

healthy choices (Brownell & Koplan, 2011). One of the interventions that received considerable

attention, which is the intervention central in this research, is FOP labeling. FOP labels, as the

name already tells, are labels that are displayed on the front of the package of products and that

display essential nutrition information. This kind of labeling, as an extensive review of relevant literature suggests, holds promise as a manner to encourage consumers’ healthy decision- making (Hawley, Roberto, Bragg, Liu, Schwartz & Brownell, 2012). Worldwide, policy makers are experimenting with several FOP labels, for instance Multiple Traffic Light Labeling in the United Kingdom, in which red, yellow and green symbols are indications of low, medium or high levels of fat, sugar and/or salt. Another example is the Choices symbol in the Netherlands, which is a single symbol that indicates certain standards for low amounts of calories, fats, salt and/or sugar (Roberto & Khandpur, 2014; Tarabella & Voinea, 2013).

Notably, Roberto and Khandpur (2014), suggest that FOP labels potentially are a way to educate people about adequate serving sizes. This touches upon an interesting issue. Namely, the issue that policies concerning serving sizes on nutrition labels differ a lot throughout the world, specifically with regard to the flexibility manufacturers have in defining a serving size, which may cause serving size discrepancies and influence consumers’ behavior. This issue in turn touches upon an aspect that is at the heart of tackling the obesity epidemic: rightful interpretation of nutrition labels, and in the context of this research, particularly nutrition information on FOP serving size labels (Huizinga et al., 2009).

With the aim of addressing the discrepancies in serving size, this research considers the effect of FOP serving size labels on consumers’ purchase intention. Existing literature has found that health claims and health framing influence consumers’ purchase intention. Mohr, Lichtenstein & Janiszewski (2012) call this phenomenon ‘health framing’, and demonstrate that people viewing products with smaller serving sizes are more likely to assess a product as healthier than a similar product with a larger serving size, which in turn increases consumers’

purchase intention. In addition, Wansink & Chandon (2006) suggest that relative nutritional claims, such as nutrition labels per serving size on the package, can cause deceitful ‘health halos’ in the case of presenting smaller instead of larger serving sizes, leading consumers to perceive the food as having less calories and as healthier, resulting in higher purchase intention.

Supportive of these findings, Kliemann et al. (2018), state that reporting smaller serving sizes and creating lower energy values, lead to serving size distortion and make consumers perceive products as healthier, which increases the likelihood of consumption.

Summarized, FOP serving size labels are likely to influence consumers’ purchase

intention. It is predicted that FOP labels reporting smaller (vs. larger) serving sizes increase

consumers’ purchase intention. Therefore, the first hypothesis, which is concerned with the

main effect of FOP serving size labels on consumers’ purchase intention, is stated as follows:

Hypothesis 1. FOP serving size labels reporting smaller (vs. larger) serving sizes increase consumers’ purchase intention.

Anticipated Guilt

As discussed, manipulation of FOP serving size labels results in different serving sizes and different calorie values, which are likely to affect purchase intention. In such decision dynamics, emotions, and specifically guilt, may play an important role, since guilt is considered as the emotion regulating many consumption processes (Antonetti & Baines, 2015).

Guilt is a self-conscious emotion and provides consumers with feedback on how well or how poorly they are regulating themselves. These kind of emotions tend to be elicited in several steps. First, there must be an attentional focus on the self and it must activate self- representations. Subsequently, there must be identity-goal relevance, and it must evoke identity-goal congruence. Lastly, there should be an internal locus of attribution, and there should not be globality and stability at stake (Hofmann & Fisher, 2012; Tracy & Robins, 2004).

For instance, imagine being on a strict diet because you want to fit your favorite pair of jeans again. However, when doing groceries, you come across your favorite chocolate bar, and consider buying it. In this kind of situation, your attention is presumably drawn towards yourself, while considering whether buying that chocolate bar matters for yourself and your goals. If you subsequently consider not acting consistent with your goals and blame yourself for the undesirable behavior, the experience of guilt is anticipated. Furthermore, according to Lewis (1995), guilt is the consequence of perceiving one’s (potential) behavior as a failure, whereby there is a focus on one’s particular characteristics that led to that (possible) failure. It results in the need to repair the intention and/or action. Attributional processes in consumption- induced problems, such as focusing on the characteristics that led to the (potential) failure, are likely to affect the formation of guilt. To exemplify, again using the chocolate bar example, guilt is anticipated to be experienced, when you consider acting not in line with yourself and your goals, causing a decrease in purchase intention (Lindenmeier et al., 2017).

Particularly, there is considerable research demonstrating that anticipated guilt mediates

the impact of its antecedents on purchase intention (e.g., Lindenmeier et al., 2017; Rozin et al.,

1999). In this present research, it is predicted that anticipated guilt mediates the effect of FOP

serving size labels on purchase intention. This reasoning is supported by Wansink & Chandon

(2006), who show that ‘low-fat’ labels increase consumption. These labels reduce anticipated

guilt of consuming the product, because such labels create ‘health halos’ that lead consumers

to perceive the product as healthier. Consistent with this finding, Mohr et al. (2012) demonstrate

that ‘health framing’, meaning reporting smaller serving sizes and calories, decrease anticipated guilt associated with consumption, and increase subsequent purchase intention. Besides, Choi, Li & Samper (forthcoming), suggest that small health-enhancing changes in the calorie content of foods, decrease anticipated guilt of consuming the product and increase purchase intention.

Hence, FOP serving size labels are likely to influence consumers’ purchase intention through anticipated guilt. Resulting from this, it is argued that manipulating FOP serving size labels in a way that they report smaller serving sizes and present lower calorie values, reduce anticipated guilt related to consumption and increase purchase intention. Taken together, the second hypothesis, which concerns the mediating effect of anticipated guilt on the relationship between FOP serving size labels and purchase intention, is stated as follows:

Hypothesis 2. Consumers’ anticipated guilt of consuming a product mediates the effect of FOP serving size labels reporting smaller (vs. larger) serving sizes on consumers’ purchase intention.

Health Consciousness

So far, it became clear that anticipated guilt is assumed to be the emotion driving the relationship of FOP serving size labels on purchase intention. Nevertheless, making healthy decisions is not equally important for every consumer. For that reason, the extent of anticipated guilt that consumers experience from exposure to FOP serving size labels, is likely to be dependent on the extent to which consumers are health conscious.

Health consciousness is defined as the extent to which people are preoccupied with their health. It is the motivational factor that stimulates people to make healthful decisions (Michaelidou & Hassan, 2008). Moreover, health conscious people are very aware and concerned about their well-being. They possess the motivation to maintain their health quality, or to even increase it and they possess the motivation to prevent the prevalence of cardiovascular diseases, by committing themselves to a healthy living and engaging in healthy behavior (Mai & Hoffman, 2012). Furthermore, health conscious individuals seem to be more critical in assessing nutrition labels. They are more likely to actively look for nutritional knowledge, or to even enhance this knowledge. Also, they are relatively more inclined to choose fruit, vegetables and low-fat alternatives (DiPietro, Rmar & Parsa, 2016).

It follows that generally, it is believed that consumers that are highly concerned about

engaging in healthy behavior, rely more nutrition labels than consumers that have little concern

about engaging in healthy behavior. Therefore, it is predicted that consumers high in health

consciousness are more likely to be affected by FOP serving size labels. There are potentially two opposing ways in which health consciousness moderates the effect of FOP serving size labels on anticipated guilt, resulting in two competing hypotheses.

The first is that the effect of FOP serving size labels on anticipated guilt is predicted to be less pronounced for consumers high in health consciousness. Burton, Garretson & Veliquette (1999) suggest that interpreting nutrition labels is a complex task, involving much cognitive effort and information analysis. Many consumers fail to properly interpret nutrition information, because they do not have the cognition to do so (Rothman et al., 2006). As a consequence, Burton et al. (1999) propose that most consumers are likely to rely on heuristics or mental shortcuts, like FOP serving size labels, in order minimize cognitive effort. For that reason, only consumers high in health consciousness possess the motivation or cognition to assess FOP serving size labels and adjust for possible serving size discrepancies. As such, health consciousness weakens the effect of FOP serving size labels on anticipated guilt.

The second possibility is that the effect of FOP serving size labels on anticipated guilt is more pronounced for consumers high in health consciousness. Papies & Veling (2013), demonstrate that individuals much concerned about their diet have a great focus on calorie content. Notably, however, Chernev (2011) suggests that such consumers (i.e., high in health consciousness) potentially use strategies leading them to perceive calorie content as lower, and thus are more susceptible to biases in a product’s calorie content. For example, LeBoeuf &

Shafir (2003) demonstrate that people high in need for cognition, which consumers high in health consciousness are, are susceptible to framing effects; they cannot adjust for such effects.

Also, Choi et al. (forthcoming) find that minute changes of calorie amounts in nutrition information, decrease anticipated guilt of consuming the product and increase purchase intention, but only when consumers have strong concerns about their health. Taken together, consumers high in health consciousness often rely on nutrition information, but are likely to be unable to adjust for biases in nutrition information. As such, health consciousness strengthens the effect of FOP serving size labels on anticipated guilt.

As a result, it is posited that consumers high in health consciousness are either more or less prone to the effect of FOP serving size labels. Hence, the third hypothesis, concerning the moderating effect of health consciousness, predicts that the extent to which people are health conscious influences the relationship between FOP serving sizes labels and anticipated guilt.

Given the two different ways in which health consciousness may moderate the mediating effect

of anticipated guilt, there are two competing hypotheses formulated:

Hypothesis 3a. The mediating effect of anticipated guilt on the relationship between FOP serving size labels and purchase intention is more (vs. less) pronounced for consumers high (vs. low) in health consciousness.

Hypothesis 3b. The mediating effect of anticipated guilt on the relationship between FOP serving size labels and purchase intention is less (vs. more) pronounced for consumers high (vs. low) in health consciousness.

Vice and Virtue Foods

Not only individual differences in health consciousness are likely to moderate the mediating effect of anticipated guilt. Namely, contextual differences are likely to moderate the mediating effect of anticipated guilt on the relationship between FOP serving size labels and anticipated guilt as well. Therefore, another potential moderator is product category.

Specifically, this research considers vices and virtues.

Vice foods, such as a pizza, are considered as unhealthy, indulgent foods. These products offer the consumer a direct pleasurable experience, but are associated with negative long-term consequences (Van Doorn & Verhoef, 2011). Wertenbroch (1998) states that vice products provide immediate gratification. Moreover, vices are suggested to address a taste goal instead of a health goal, as they are perceived as unhealthy, but tasty (Liu, Haws, Lamberton, Campbell & Fitzsimons, 2015; Raghunathan, Naylor & Hoyer, 2006). In contrast, virtue foods, such as a salad, are considered as healthy foods. These products do not offer consumers a direct pleasurable experience, but are less associated with long-term negative consequences (Van Doorn & Verhoef, 2011). Virtues provide consumers with delayed gratification (Wertenbroch, 1998). Besides, virtues are likely to appeal to health goals instead of taste goals, as they are perceived as less unhealthy, but also less tasty (Liu et al., 2015; Raghunathan et al., 2006).

Accordingly, studies imply that consumers relate energy-dense (i.e., containing much calories) foods, like donuts and chips, with immediate pleasure and tastiness, and that such products are considered vices. Conversely, less energy-dense (i.e., containing less calories) foods, like soups or salads, are in turn more associated with long-term health goals and less tastiness, and thus, such products are seen as virtues. As consumers’ decision-making often involves a dilemma between energy-dense vices, satisfying a short-term pleasure goal and less energy-dense virtues, satisfying a long-term health goal, these findings have implications for consumers’

purchase intention (Van Doorn & Verhoef, 2011; Raghunathan et al., 2006; McCory et al.,

2000).

Particularly, research suggests that healthful nutrition claims and framing on labels affect anticipated guilt and purchase intention more for vices than for virtues. For instance, Kivetz & Keinan (2006) demonstrate that consumers that choose vices instead of virtues experience more guilt. Additionally, Strahilevitz & Meyers (1998) find that contemplated consumption of vice foods increase anticipated guilt of consuming the product. Consistent with this finding, Choi et al. (forthcoming) demonstrate that minor adjustments in presenting calorie content on vice foods results in lower levels of anticipated guilt and higher levels of purchase intention. Based on these findings, it is predicted that FOP serving size labels adopting a smaller (vs. larger) serving size, increase purchase intention of vices more compared to virtues, since it allows consumers to anticipate less guilt while being able to enjoy their pleasurable vice food.

Therefore, in this present research, it is posited that the impact of FOP labels reporting smaller serving sizes on anticipated guilt is more pronounced for vices than for virtues. Hence, the fourth hypothesis, concerning the moderating effect of product category (i.e., vice vs.

virtue), is stated as follows:

Hypothesis 4. The mediating effect of anticipated guilt on the relationship between FOP serving size labels and purchase intention is more (vs. less) pronounced for vices (vs. virtues).

Figure 2 demonstrates the researched concepts, the according hypotheses and relations.

FIGURE 2

Conceptual model and hypotheses

3. METHOD

This paper presents the findings of an experiment, using an online study design, aiming to investigate the interplay between several phenomena. Firstly, to examine the main effect of FOP serving size labels on purchase intention (H

). Secondly, to investigate the mediating effect of anticipated guilt on the relationship between FOP serving size labels and purchase intention (H

). Third, to assess whether this mediating effect of anticipated guilt is moderated by health consciousness (H

H

). Fourth, to discover whether this mediating effect of anticipated guilt is moderated by product category (i.e., vice vs. virtue) (H

). Due to the online experiment design, item-nonresponse was prevented. Also, the online survey design allowed for unit- nonresponses to be excluded from the final sample, resulting in no missing data. As a consequence, the final sample consisted of 107 respondents, all representing valid results.

Participants were distributed across different age groups, ranging from 18 to 65 years old (M = 25.22; SD = 7.92). Moreover, the sample consisted of 63.6% females and 36.4% males. Lastly, most of the participants were of Dutch nationality.

Design

The study, which was an experiment conducted via an online survey, was distributed via social media channels and used a mixed design. On the one hand, it included a between- subjects factor, as participants were randomly assigned to either the small serving size FOP label condition or the large serving size FOP label condition. On the other hand, the study used a within-subjects factor, since all respondents were exposed to both virtues and vices. After each exposure of the respondents to either small or large serving size FOP labels of randomly ordered relative vice foods and relative virtue foods, they were asked to indicate their intentions to purchase the specific product and to indicate the extent to which they anticipate guilt of consuming the specific product. The last part of the experiment was constituted of questions concerning the survey experience, during which participants’ health consciousness and other socio-demographic measures (e.g., hunger, dietary restriction, nationality and age) were collected.

Procedure and Measurements

After the respondents were invited to take part in the online survey, they were informed

about the general purpose of the study and their anonymity was guaranteed. When the

respondents gave their informed consent to participation in the survey, they were asked to

imagine going to the supermarket, in order to buy a convenient meal, drink and snack. By

providing the online survey with some real-life context, the participants’ involvement in the

survey was potentially enhanced.

Consequently, respondents were assigned to one of the conditions, in which they were exposed to either small serving size FOP labels or large serving size FOP labels of both vices and virtues. The FOP serving size labels had a Facts Up Front design, which is a design widely used in the Netherlands. These labels contain calorie information in kilocalories and kilojoules of a serving size in grams or milliliters (Roberto & Khandpur, 2014). In order to determine small serving size and large serving size, provisions for determining a serving size were observed. As there are no mandatory guidelines for serving sizes to be adhered to in the Netherlands, the serving sizes were constituted based on real-life observations, taking into account several aspects. Firstly, the stimuli should consist of different categories of products that are commonly bought by consumers. Also, the stimuli should not already be proportionated, consumers should have to proportionate the product themselves. Besides, the package size of the product was taken into account. That is, both small and large serving sizes should be realistically related to the package size. Lastly, small and large serving sizes were established based on observations of serving sizes actually displayed on products in Dutch supermarkets.

Within the small serving size FOP label condition, respondents were presented six randomly ordered stimuli comprised of both vices and virtues. Vice stimuli were (1) a pizza, of which one serving of 175 grams contains 402 calories, (2) a butter cake, of which one serving of 25 grams contains 121 calories and (3) a bottle of Coca Cola, of which one serving of 150 milliliters contains 63 calories. Virtue stimuli were (1) a paella meal, of which one serving of 250 grams contains of 250 calories, (2) a gingerbread zero sugar, of which one serving of 23 grams contains 72 calories and (3) a package of ice tea, of which one serving of 150 milliliters contains 21 calories. Within the large serving size FOP label condition, respondents were presented six randomly ordered stimuli comprised of both vices and virtues too. Vice stimuli were (1) a pizza, of which one serving of 350 grams contains 805 calories, (2) a butter cake, of which one serving of 50 grams contains 242 calories and (3) a bottle of Coca Cola, of which one serving of 250 milliliters contains 105 calories. Virtue stimuli were (1) a paella meal, of which one serving of 500 grams contains of 500 calories, (2) a gingerbread zero sugar, of which one serving of 46 grams contains 144 calories and (3) a package of ice tea, of which one serving of 250 milliliters contains 35 calories. Taken together, the calories and grams of the small serving size FOP labels of the pizza, paella, butter cake and gingerbread represented 50% of the calories and grams of the large serving size FOP labels of these products. The milliliters and calories of the small serving size FOP labels of the ice tea and Coca Cola represented 60%

of the milliliters and calories of the large serving size FOP labels of these products, as these

reflect observed realistic milliliter amounts (i.e., 150ml and 250ml). In figure 3, an example of the stimulus materials is found. For graphical representation of all stimulus materials, see appendix 1.

FIGURE 3

Example of stimulus materials

After each exposure to a stimulus, while the exposed product and FOP serving size label remained on top of the screen, anticipated guilt and purchase intention associated with consuming the product were measured. In order to measure anticipated guilt, a scale was adapted from Mohr et al. (2012). Respondents were asked ‘how guilty would you feel after consuming the whole pizza/paella/gingerbread/butter cake/Coca Cola/ice tea?’. Participants could rate their anticipated guilt on a scale ranging from (1) ‘not guilty’ to (7) ‘very guilty’.

Also, purchase intentions were measured according to a scale adapted from Mohr et al. (2012).

Respondents were asked ‘I would consider purchasing this pizza/paella/gingerbread/butter cake/Coca Cola/ice tea the next time I needed a convenient snack/drink/meal’, and could rate their purchase intention on a 7-point scale ranging from (1) ‘strongly disagree’ to (7) ‘strongly agree’.

In the last part of the experiment, the participants’ health consciousness was measured.

This was done using a scale developed by Dutta-Bergman (2004). Respondents were asked to

rate five statements, for instance ‘Eating right, exercising and taking preventive measures will

keep me healthy for life’ and ‘I do everything I can do to stay healthy’. These five statements

were rated on a 5-point scale, ranging from (1) ‘strongly disagree’ to (5) ‘strongly agree’ ( a =

0.704; M = 3.59; SD = 0.56). Moreover, the last part consisted of questions concerning the

survey experience, during which socio-demographic measures (e.g., hunger, dietary restriction,

nationality and age) were collected.

4. RESULTS

The aim of this research was to assess the interplay between FOP serving size labels for virtues as well as for vices, anticipated guilt, purchase intention and health consciousness. To assess if FOP serving size labels affect consumers’ purchase intention through participants’

anticipated guilt, and to assess the moderating influence of health consciousness in this mediating effect of anticipated guilt, a moderated mediation analysis using model 7 (5,000 bootstrap samples, 95% confidence interval) of Hayes’ PROCESS macro was performed (Hayes, 2013). Additionally, a two-way mixed ANOVA was performed, aiming to determine whether the mediating influence of anticipated guilt is moderated by product category (i.e., vice vs. virtue products). A complete overview of the corresponding results, is found in figure 6.

Data Exploration

In order to gain an initial understanding of the data, the descriptive information concerning the main variables was explored. Table 1 shows the descriptive information of the seven continuous variables. The seven rows represent the main variables – anticipated guilt of vices and virtues, anticipated guilt of vices, anticipated guilt of virtues, purchase intention of vices and virtues, purchase intention of vices, purchase intention of virtues, and health consciousness, respectively – whereas the columns represent the location and dispersion of the main variables – mean and standard deviation, respectively – for both the small and large serving size FOP label condition. The location and dispersion of the scores seem reasonable, and the performed outlier detection procedure does not indicate the presence of any outliers.

TABLE 1

Location and dispersion of main variables

Location and dispersion

Condition: small serving size FOP label Condition: Large serving size FOP label

Mean SD Mean SD

1. Anticipated guilt

vices and virtues 4.10 1.33 4.04 1.31

2. Anticipated guilt

vices 4.66 1.50 4.71 1.55

3. Anticipated guilt

virtues 3.55 1.43 3.38 1.45

4. Purchase intention

vices and virtues 3.75 1.13 3.72 1.05

5. Purchase intention

vices 3.53 1.37 3.37 1.29

6. Purchase intention

virtues 3.98 1.33 4.06 1.30

7. Health

consciousness 3.51 0.62 3.67 0.49

In addition, with the aim of gaining a primary understanding of the associations between the main variables, the correlations between the main variables were explored, as represented in table 2. The seven rows represent the main variables – anticipated guilt of vices and virtues, anticipated guilt of vices, anticipated guilt of virtues, purchase intention of vices and virtues, purchase intention of vices, purchase intention of virtues, and health consciousness, respectively – whereas the columns represent the correlations between the variables.

TABLE 2

Correlation matrix of main variables

* indicates statistical significance at a = 0.05 (2-tailed)

** indicated statistical significance at a = 0.01 (2-tailed)

The Main Effect of FOP Serving Size Labels on Purchase Intention

Firstly, it was assessed whether FOP serving size labels reporting smaller (vs. larger) serving sizes increase consumers’ purchase intention (H

), which is the main effect (i.e., direct effect) of the model. The results, represented in table 3, suggest that there is no significant influence of FOP serving size labels on purchase intention (B = -0.05; t(107) = -0.24; p = 0.81).

Hence, the first hypothesis is not supported, implying that FOP serving size labels do not affect purchase intention.

The Mediating Effect of Anticipated Guilt

Subsequently, it was examined if consumers’ anticipated guilt of consuming a product mediates the relationship between FOP serving size labels and purchase intention (H

), which concerns the indirect effect of the model. With regard to the effect of the independent variable

Correlation

1. 2. 3. 4. 5. 6. 7.

1. Anticipated guilt vices and virtues

1 2. Anticipated

guilt vices 0.90** 1

3. Anticipated

guilt virtues 0.88** 0.59** 1

4. Purchase intention vices and virtues

-0.32** -0.24* -0.32** 1

5. Purchase

intention vices -0.40** -0.42** -0.28** 0.82** 1

6. Purchase

intention virtues -0.12 0.02 -0.25** 0.82** 0.35** 1

7. Health

consciousness 0.25** 0.27** 0.17 -0.22* -0.30** -0.06 1

on the mediator, as depicted in table 3, the results suggest that the effect of FOP serving size labels on anticipated guilt is not significant (B = 0.45; t(107) = 0.27; p = 0.79), suggesting that FOP serving size labels do not affect anticipated guilt. Concerning the effect of the mediator on the dependent variable, the results, as found in table 3, suggest that anticipated guilt has a significant effect on purchase intention (B = -0.26; t(107) = -3.42; p = 0.00). Lastly, there is no significant indirect effect of FOP serving size labels on purchase intention via anticipated guilt given different levels of health consciousness. As FOP serving size labels fail to affect purchase intention, there is no effect to mediated. Hence, the second hypothesis is not supported, meaning that anticipated guilt does not seem to account for the relationship between FOP serving size labels and purchase intention.

The Moderating Effect of Health Consciousness

Next, it was investigated whether the mediating effect of anticipated guilt is more (vs.

less) pronounced for consumers high (vs. low) in health consciousness (H

) and to assess whether the mediating effect of anticipated guilt is less (vs. more) pronounced for consumers high (vs. low) in health consciousness (H

). The results, found in figure 4 and table 3, suggest that there is no significant interaction between FOP serving size labels and health consciousness (B = -0.17; t(107) = -0.36; p = 0.72). Therefore, hypotheses H

and H

are not supported.

FIGURE 4

Non-significant interaction of FOP serving size labels and health consciousness on anticipated guilt

0,00 0,50 1,00 1,50 2,00 2,50 3,00 3,50 4,00 4,50 5,00

HC + 1 SD Mean HC HC - 1 SD

An tic ip at ed g ui lt

Health Consciousness (HC)

Small serving size FOP label Large serving size FOP label

Yet, the results do suggest that health consciousness has a main and significant effect on anticipated guilt (B = 0.67; t(107) = 2.39; p = 0.02). The results, found in table 3, imply that health consciousness has a positive relationship with anticipated guilt, meaning that the higher consumers’ levels of health consciousness, the higher consumers’ levels of anticipated guilt.

TABLE 3

Results of the moderated mediation analysis

B t p VIF R

Model 1: M ~ X + W 0.06

Constant 1.74 1.73 0.09

FOP serving size

label 0.45 0.27 0.79 1.02

Health

consciousness 0.67 2.39 0.02* 1.61

Interaction health consciousness and FOP serving

size label -0.17 -0.36 0.72 1.60

Model 2: Y ~ X + M 0.10

Constant 4.83 14.02 0.00**

FOP serving size

label -0.05 -0.24 0.81 1.00

Anticipated Guilt -0.26 -3.42 0.00** 1.00

* indicates statistical significance at a = 0.05 (2-tailed)

** indicated statistical significance at a = 0.01 (2-tailed)

The Moderating Effect of Product Category

Finally, to investigate whether the mediating effect of anticipated guilt on the relationship between FOP serving size labels and purchase intention is more (vs. less) pronounced for vice (vs. virtue) products (H

), a two-way mixed ANOVA was performed. As the spherity condition is not violated, since there is only one set of difference scores, results were based on the uncorrected F-ratio. Tests of the within-subjects effects suggest that there is no significant interaction effect of FOP serving size labels and product category (i.e., vice vs.

virtue) on anticipated guilt ( µ

= 0.00; F(1, 107) = 0.62; p = 0.43). This means that irrespective of the type of FOP label (i.e., small serving size vs. large serving size), levels of anticipated guilt of vices and virtues did not differ significantly, as demonstrated in figure 5 and table 4.

Therefore, the fourth hypothesis is not supported.

FIGURE 5

Non-significant interaction of FOP serving size labels and product category on anticipated guilt

However, tests of the within-subjects effects, as depicted in table 4, suggest that there is a significant main effect of product category (i.e., vice vs. virtue) on anticipated guilt ( µ

= 0.45; F(1, 107) = 87.10; p = 0.00). Additionally performed post-hoc analyses reveal that levels of anticipated guilt for virtues (M = 3.47; SD = 1.44) are significantly lower than levels of anticipated guilt for vices (M = 4.68; SD = 1.52). Thus, vices cause consumers to anticipate more guilt compared to virtues. Lastly, tests of the between-subjects effects demonstrate that there is no significant effect of FOP serving size label on anticipated guilt ( µ

= 0.00; F(1, 107) = 0.05; p = 0.82).

TABLE 4

Results of the two-way mixed ANOVA

µ

F p

Between-subjects effects of FOP serving size labels on anticipated guilt

Intercept 0.91 1018.27 0.00**

FOP serving size label 0.00 0.05 0.82

Error

Within-subjects effects of product category on anticipated guilt of vices and virtues

Product category (PC) 0.45 87.10 0.00**

Interaction PC and FOP

serving size labels 0.00 0.62 0.43

Error (PC)

* indicates statistical significance at a = 0.05 (2-tailed)

** indicated statistical significance at a = 0.01 (2-tailed) 0,00

0,50 1,00 1,50 2,00 2,50 3,00 3,50 4,00 4,50 5,00

Vice Virtue

An tic ip at ed g ui lt

Product Category

Small serving size FOP label Large serving size FOP label

FIGURE 6

Results of all performed analyses

* indicates statistical significance at a = 0.05

Follow-Up Analyses Including Control Variables

In order to examine whether the results are subject to external factors, additional analyses were performed, taking several control variables into account: hunger, diet (i.e., dieting to lose weight), dietary restriction (i.e., being vegetarian), gender and age. The prior analyses were performed once more, but now included control variables.

To start with, the aforementioned control variables were included in the moderated mediation analysis using model 7 (5,000 bootstrap samples, 95% confidence interval) of Hayes’

PROCESS macro (Hayes, 2013). Concerning the effect of FOP serving size labels and

anticipated guilt on purchase intention (i.e., model 2, see table 5), the results including the

control variables demonstrate that the effect of anticipated guilt becomes smaller and shifts

from significant to marginally significant (B = -0.16; t(107) = -1.73; p = 0.09). In addition, a

significant effect of dietary restriction is found, which is estimated to have a stronger effect on

purchase intention than anticipated guilt (B = -0.35; t(107) = -2.06; p = 0.04). As such,

anticipated guilt and dietary restriction may explain the same part of variance in purchase

intention. However, the variance in purchase intention is more adequately explained by dietary

restriction than by anticipated guilt. Concerning the effect of FOP serving size labels on

anticipated guilt, and the moderating influence of health consciousness (i.e., model 1, see table 5), the results including the control variables indicate that the formerly found main effect of health consciousness on anticipated guilt is smaller and shifts from significant to marginally significant (B = 0.43; t(107) = 1.76; p = 0.08). Furthermore, results demonstrate significant effects on anticipated guilt of diet (B = 0.58; t(107) = 2.06; p = 0.04), gender (B = 1.19; t(107)

= 5.36; p = 0.00) and hunger (B = -0.12; t(107) = -2.08; p = 0.04). As such, the variance in anticipated guilt seems explained to greater extent by diet and gender, than by health consciousness. Besides, hunger also explains some of the variance in anticipated guilt, but it does not explain it to greater extent than health consciousness.

TABLE 5

Results of the moderated mediation analysis including control variables

B t p R

Model 1: M ~ X + W + control variables 0.37

Constant 0.29 0.29 0.77

FOP serving size

label 0.31 0.21 0.83

Health consciousness 0.43 1.76 0.08

Interaction health consciousness and FOP serving size label

-0.09 -0.23 0.82

Dietary restriction 0.18 0.97 0.33

Diet 0.58 2.06 0.04*

Gender 1.19 5.36 0.00**

Age 0.02 1.47 0.14

Hunger -0.12 -2.08 0.04*

Model 2: Y ~ X + M + control variables 0.17

Constant 5.12 10.07 0.00**

FOP serving size

label -0.09 -0.46 0.65

Anticipated guilt -0.16 -1.73 0.09

Dietary restriction -0.35 -2.06 0.04*

Diet -0.21 -0.77 0.44

Gender -0.23 -0.96 0.34

Age -0.00 -0.51 0.61

Hunger -0.00 -0.15 0.88

* indicates statistical significance at a = 0.05 (2-tailed)

** indicated statistical significance at a = 0.01 (2-tailed)

Consequently, the control variables were included in the two-way mixed ANOVA. With regard to the between-subjects effects of FOP serving size labels on anticipated guilt, the results, as displayed in table 6, demonstrate a significant effect of hunger ( µ

= 0.04; F(1,107)

= 4.16; p = 0.04), diet ( µ

= 0.05; F(1, 107) = 5.43; p = 0.02) and gender ( µ

= 0.24; F(1, 107)

= 30.63; p = 0.00). Therefore, individuals’ hunger, diet and gender are likely to explain the

variance in individuals’ anticipated guilt scores. With regard to the within-subject effects of product category (i.e., vice vs. virtue), the results, as displayed in table 6, indicate that inclusion of the control variables shifts the effect of product category (i.e., vice vs. virtue) from significant to non-significant ( µ

= 0.00; F(1, 107) = 0.05; p = 0.83). However, a significant interaction effect of product category (i.e., vice vs. virtue) and diet is found ( µ

= 0.04; F(1, 107) = 4.49;

p = 0.04). Moreover, a marginally significant interaction effect of product category (i.e., vice vs. virtue) and gender is found ( µ

= 0.03; F(1, 107) = 3.08; p = 0.08). As a consequence, individual differences in diet and gender influence the effect of product category (i.e., vice vs.

virtue) on the difference in individuals’ scores of anticipated guilt for vices and virtues.

TABLE 6

Results of the two-way mixed ANOVA including control variables

µ

F p

Between-subjects effects of FOP serving size labels on anticipated guilt

Intercept 0.10 11.33 0.00**

Hunger 0.04 4.16 0.04*

Dietary restriction 0.01 1.21 0.27

Diet 0.05 5.43 0.02*

Gender 0.24 30.63 0.00**

Age 0.02 2.14 0.15

FOP serving size label 0.00 0.06 0.80

Error

Within-subjects effects of product category on anticipated guilt of vices and virtues

Product category (PC) 0.00 0.05 0.83

Interaction PC and hunger 0.00 0.04 0.85

Interaction PC and dietary

restriction 0.00 0.06 0.80

Interaction PC and diet 0.04 4.49 0.04*

Interaction PC and gender 0.03 3.08 0.08

Interaction PC and age 0.00 0.22 0.64

Interaction PC and FOP

serving size labels 0.01 0.78 0.38

Error (PC)

* indicates statistical significance at a = 0.05 (2-tailed)

** indicated statistical significance at a = 0.01 (2-tailed)

5. DISCUSSION

The purpose of the current study was to assess the mediating effect of anticipated guilt,

on the relationship between FOP serving size labels and purchase intention, and to examine the

moderating effect of health consciousness and product category (i.e., vices vs. virtues) on the

mediating effect of anticipated guilt. Specifically, by being among the first to investigate the

full interplay between FOP serving size labels for vice and virtue foods, anticipated guilt,

purchase intention and health consciousness, this research aimed to gain more understanding about the drivers of consumers’ decision-making.

Main analyses

To start with, the main effect of FOP serving size labels on consumers’ purchase intention was assessed. Particularly, it was predicted that FOP serving size labels reporting smaller (vs. larger) serving sizes increase consumers’ purchase intention (H

). Although research finds that health framing and health claims influence purchase intention, this hypothesis is not supported (Mohr et al., 2012; Wansink & Chandon, 2006; Kliemann et al., 2018). The results of the current research suggest that there is no significant of FOP serving size labels on purchase intention. According to Roberto & Khandpur (2014), FOP serving size labels hold promise to affect consumers’ behavior, but they also state that FOP serving size labels adopting a Facts Up Front design raise some concerns. As this is the type of FOP serving size label examined in this study, this could explain its failing effect. Fact Up Front FOP serving size labels can lead to confusion, because they depict nutritional information of serving sizes in for example grams, milligrams and liters, which possibly impedes rightful interpretation and in turn, does not affect consumers. Moreover, consumers are confronted with various types of FOP labels, all competing for the consumers’ attention. As argued by Bialkova & Van Trijp (2010), not all FOP labels receive equal attention of consumers. Their research demonstrates that familiarity with the type of FOP label greatly influences consumers’ attention and as a consequence, might influence consumers’ behavior. However, the results of the current study, which included a widely used FOP serving size label (i.e., Facts Up Front), indicates that FOP serving size labels to affect consumers’ decision-making. Soederberg-Miller et al. (2015) explain that even familiar FOP labels can confuse consumers when using them for decision- making, due to the lack of nutrition knowledge. In addition, the authors suggest that even though the amount of calories on Facts Up Front FOP labels receives most attention, it often fails to guide consumer behavior.

Next, the mediating effect of anticipated guilt on the relationship between FOP serving size labels and consumers’ purchase intention was examined. It was predicted that anticipated guilt mediates the relationship between FOP serving size labels and purchase intention (H

).

The results suggest that there is no significant effect of FOP serving size labels on anticipated

guilt. Again, this might be due to the fact that FOP serving size labels with a Facts Up Front

design cause confusion, as elaborated upon above. Yet, the results suggest that anticipated guilt

has a significant, negative effect on purchase intention. This finding is consistent with research

demonstrating that guilt regulates many consumption processes (e.g., Antonetti & Baines,

2015). For example, Chitturi, Raghunathan & Mahajan (2007) prove that guilt affects consumers’ decision-making. There is little literature examining a main effect of anticipated guilt on purchase intention, but Birkimer, Johnston & Berry (1993) demonstrate that consumers avoid risky health behaviors, because this is related to a prior judgment considering how guilty they would feel when committing the behavior. Furthermore, the findings suggest that there is no significant indirect effect of FOP serving size labels on purchase intention via anticipated guilt. Since FOP serving size labels fail to affect purchase intention, there is no effect to mediated, and therefore, the second hypothesis was not supported.

Subsequently, the moderating influence of health consciousness in the mediating effect of anticipated guilt was investigated. On the one hand, it was hypothesized that the mediating effect of anticipated guilt is more (vs. less) pronounced for consumers high (vs. low) in health consciousness (H

). On the other hand, it was hypothesized that the mediating effect of anticipated guilt is less (vs. more) pronounced for consumers high (vs. low) in health consciousness (H

). The results reveal that there is no significant interaction between FOP serving size labels and health consciousness. Again, this could be due to the failing effect of the stimuli, which were FOP serving size labels adopting a Facts Up Front design. Nevertheless, the results demonstrate that health consciousness has a direct, significant and positive effect on anticipated guilt. Consistent with this finding, Pelletier, Dion, Slovinec & Reid (1997) find that self-regulatory emotions (e.g., anticipated guilt) depend on the strength of motives, which possibly explains the positive effect of health consciousness on anticipated guilt. Besides, as health consciousness determines consumers’ motivation towards healthful behavior, research suggests that it affects anticipated guilt when assessing certain undesired or desired events (Michaelidou & Hassan, 2008; Onwezen, Bartels & Antonides, 2014).

Lastly, the moderating influence of product category (i.e., vice vs. virtue) in the

mediating effect of anticipated guilt was examined. It was posited that the mediating effect of

anticipated guilt is more (vs. less) pronounced for vice (vs. virtue) products (H

). Although the

results indicate that there is no significant interaction effect of FOP serving size labels and

product category (i.e., vice vs. virtue) on anticipated guilt, the results indicate that there is a

significant, main effect of product category (i.e., vice vs. virtue) on anticipated guilt. The latter

implies that vices cause consumers to anticipate more guilt compared to virtues. This effect is

in line with several studies suggesting that consumers anticipate to experience guilt when

considering to consume unhealthy foods or when they choose vices over virtues (e.g., Rozin et

al., 1999; Chitturi et al., 2007).

In sum, none of the four hypotheses are supported. FOP serving size labels do not affect purchase intention via anticipated guilt and neither health consciousness nor product category (i.e., vice vs. virtue) moderates the mediating effect of anticipated guilt. Nevertheless, the results demonstrate several alternative relationships. To begin with, the results suggest that anticipated guilt has a significant, negative effect on purchase intention. Moreover, the results suggest that health consciousness has a direct, significant and positive effect on anticipated guilt. Lastly, the results demonstrate that product category (i.e., vice vs. virtue) directly affects anticipated guilt, implying that anticipated guilt for vices is significantly higher than anticipated guilt for virtues.

Follow-Up Analyses

In addition to the main analyses, follow-up analyses were performed taking several control variables into account: hunger, diet (i.e., dieting to lose weight), dietary restriction (i.e., vegetarian), gender and age.

Firstly, the moderated mediation analysis using model 7 (5,000 bootstrap samples, 95%

confidence interval) of Hayes’ PROCESS macro (Hayes, 2013) demonstrates that the direct effect of anticipated guilt on purchase intention is smaller, when control variables are included.

The results indicate that dietary restriction (i.e., being vegetarian) has a greater effect than anticipated guilt on purchase intention. Existing literature provides several explanations for this effect. For example, Lindeman & Vaananen (2000) demonstrate that motives on ecological welfare, such as a vegetarian motivation, affect consumers’ food choices. Moreover, Orlich et al. (2014) suggest that vegetarian dietary patterns influence consumption behavior.

Also, the results suggests that the main effect of health consciousness on anticipated guilt becomes smaller. The results imply that diet (i.e., dieting to lose weight) and gender affect anticipated guilt to greater extent than health consciousness. The effect of gender might be due to the fact that women experience more guilt in response to food than men (e.g., Rolls, Fedoroff

& Guthrie, 1991; Narchi, Walrand, Boirie & Rousset, 2008). In addition, dieting in order to

lose weight could be considered as a specific health motivation. As such, the effect of diet, and

as a consequence avoiding risky health behaviors, is likely to affect anticipated guilt (e.g.,

Birkimer, Johnston & Berry, 1993; Pelletier, Dion, Slovinec & Reid, 1997). Furthermore, a

significant effect of hunger on anticipated guilt is found, although the influence of hunger is

not stronger than the effect of health consciousness. As hunger might cause food cravings,

leading to perceived overeating and unsuccessful dieting, it could influence anticipated guilt

(Velarde, Moore, Boakye, Parkhurst & Brewer, 2018).

Subsequently, the between-subjects effects of the two-way mixed ANOVA, including control variables, demonstrate that hunger, gender and diet significantly influence anticipated guilt. Therefore, individual differences with regard to hunger, gender and diet might be able to explain differences in consumers’ anticipated guilt. Explanations existing literature offers for these effects, are discussed in the former paragraph, as the same effects were found in the prior follow-up analysis as well. With regard to the within-subjects effects of the two-way mixed ANOVA, including control variables, results suggest that the influence of product category (i.e., vice vs. virtue) becomes non-significant. Yet, the results imply that there is a significant interaction of product category (i.e., vice vs. virtue) and diet. Being on a diet to lose weight, implies a certain health goal. As a consequence, this health goal might influence the relationship of product category (i.e., vice vs. virtue) on the difference in anticipated guilt for vices and virtues. To illustrate, energy-dense vices satisfy a pleasure goal, whereas less energy-dense virtues, satisfy a health goal. Being on a diet implies a health goal, which is associated with virtues, and contrasts with a pleasure goal, which is associated with vices. Hence, diet might explain why anticipated guilt scores of vices are higher compared to anticipated guilt scores of virtues (Van Doorn & Verhoef, 2011; Raghunathan et al., 2006; McCory et al., 2000). Also, a marginally significant interaction of product category (i.e., vice vs. virtue) and gender is found.

Gender might affect the influence of product category (i.e., vice vs. virtue) on the difference in individuals’ scores of anticipated guilt for vices and virtues, because females tend to have a greater interest in healthy and lower-calorie food than males (Rolls, Fedoroff & Guthrie, 1991;

Fagerli & Wandel, 1999).

Implications

This research examined the relationship between FOP serving size labels for vice and

virtue foods, anticipated guilt, purchase intention and health consciousness. As a result, this

research has certain implications. In the first place, this study adds to literature by finding that

FOP serving size labels with a Facts Up Front design fail to affect consumers. As such, FOP

labels with a Facts Up Front design seem unable to address serving size discrepancies and

educate consumers about normal serving sizes. FOP serving size labels contain complicated

numeric information, which consumers cannot adequately interpret, since they do not possess

the required nutrition knowledge. This finding highlights a problem that is not specific to this

study, it goes way beyond: rightful interpretation of nutrition labels is at the heart of tackling

the global obesity epidemic and its consequences. For that reason, public policy interventions

aiming to educate consumers about interpreting nutrition information hold great promise to

guide consumers towards healthy behaviors and address the obesity epidemic. For example,