Mixed Feelings, Mixed Baskets

(1)

Mixed Feelings, Mixed Baskets

How Shopping Emotions Drive the Relative Healthiness of

Sequential Food Choices

Master’s Thesis

Completion Date

14.08.2017

University of Groningen

Faculty of Economics and Business

Master of Science in Research

1

st

Supervisor

(2)

ABSTRACT

With the rise of the obesity epidemic, research has been abundantly addressing which factors influence the healthiness of isolated product purchases. However, our understanding of how the healthiness of the entire shopping basket during larger shopping trips comes about is still limited. Building on the literature on self-conscious emotions, we propose that the healthiness of sequential food choices is dynamically and non-linearly driven by shopper’s emotions of pride and guilt, which shoppers experience in response to their food choices. We conducted an online experiment with 401 US consumers, in which we assessed participants’ experienced emotions of pride and guilt during different stages of a simulated grocery shopping task. Although the results of our study do not provide evidence that shopping emotions account for the dependencies of shoppers’ sequential food choices, they indicate that these emotions play a significant role. We find that shoppers’ level of health motivation amplifies the experience of emotions of pride and guilt in response to the relative healthiness of a food choice. In turn, the shopping trip stage, during which these emotions are experienced, determines how shoppers react to emotions of pride. Specifically, we find directional support that pride generally positively affects the relative healthiness of the next choice. However, this effect is reversed later in the shopping trip, indicating a licensing response. These findings underscore the importance of understanding the dynamic interplay between the healthiness of multiple, consecutive food choices and the emotions shoppers experience in response to these choices during larger grocery shopping trips. Only when healthy shopping dynamics are fully understood, effective in-store health interventions can be designed to curb the obesity epidemic.

(3)

INTRODUCTION

According to the World Health Organization (WHO), obesity is ‘one of the biggest global health crises of the 21st_{century’ (Chan, 2016). Globally, obesity has more than}

doubled since 1980, with more than 39% of adults over 18 being overweight (i.e., with a Body Mass Index (BMI) > 25) and 13% being obese in 2014 (i.e. BMI > 30; WHO, 2016). The obesity epidemic is largely spurred by the (over)consumption of unhealthy foods that are energy-dense, and high in saturated fat, sugar, and salt (Swinburn et al., 2011). Obesity causes severe physical conditions such as diabetes, cardiovascular and musculoskeletal diseases, as well as some types of cancer, and is therefore considered as one of the main preventable causes of death (WHO, 2016). Apart from the negative consequences for individuals’ personal well-being, the obesity epidemic also creates substantial health care costs (Finkelstein et al., 2012) as well as economic productivity losses at the societal level (Cawley & Meyerhofer, 2012; Lim et al., 2012). Hence, it is not surprising that researchers, policy makers, and food manufacturers alike are looking for strategies to curb the obesity epidemic.

(5)

dynamically influence subsequent purchases, which implies that the shopping basket is more than the sum of single, isolated food choices (e.g., Sheehan & van Ittersum, 2016; van der Heide, van Ittersum, & van Doorn, 2016). Up to this point, however, there is no single comprehensive theoretical account in the literature that enables us to predict how the healthiness of an initial choice influences the healthiness of the next choice over the shopping trip and why this is the case.

(6)

2012; Williams & DeSteno, 2008). Our results indicate that healthy shopping dynamics are the result of an intricate interplay between shoppers’ level of health motivation, the relative healthiness of their purchases and the emotions experienced in response to these purchases. We provide evidence that shoppers’ level of health motivation amplifies their experienced levels of pride and guilt given the healthiness of a preceding choice. Further, we find directional support that emotions of pride generally have positive motivational consequences for future goal pursuit. However, this effect is reversed at a later stage of the shopping trip, suggesting that shoppers may construe the experience of pride as a justification to indulge as the shopping trip progresses. Our findings provide preliminary indications that self-conscious emotions can affect the relative healthiness of consecutive food choices in the grocery shopping context, although they fall short of providing statistical evidence that shopping emotions account for the relationship between the relative healthiness of two sequential choices.

THEORETICAL FRAMEWORK

Shoppers’ Self-Regulation Dilemmas – The Motor of Healthy Shopping Dynamics

(7)

high caloric foods (Rabia, Knäuper, & Miquelon, 2006). Hence, food choices frequently involve trade-offs between the pursuit of shoppers’ long-term health goals, and the attainment of instant reward at the cost of such long-term goals (Wansink & Chandon, 2006). We contend that these self-regulation dilemmas are the motor of potential healthy shopping dynamics and may cause dependencies between sequential food choices in the grocery store. Given the conflicts that shoppers face between their long-term health goals and their desire for indulgence, the question is how shoppers will pursue these two opposing goals when shopping.

The literature suggests that there are two ways, in which individuals can resolve competing goals in sequential choices: Consistency may occur when the goal that underlies an initial choice is reinforced by this choice, resulting in a subsequent choice that is congruent with the focal goal. Conversely, balancing occurs when the initial choice satisfies the goal underlying this decision, leading to a subsequent choice that is incongruent with this goal (Huber, Goldsmith, & Mogilner, 2008). In this research, we consider two sequential food choices and distinguish between two different types of consistency and balancing responses depending on whether the initial choice was healthy or unhealthy and whether it is followed by a healthy or unhealthy choice, in turn. Specifically, shoppers show ‘positive’ consistency when they engage in two consecutive healthy choices, whereas ‘negative’ consistency describes two sequential unhealthy choices (see Mullen & Monin, 2016). Similarly, shoppers can engage in ‘positive’ balancing when an initial unhealthy choice is followed by a healthier choice, whereas ‘negative’ balancing reflects a licensing response (e.g., Khan & Dhar, 2006), where an initial healthy choice leads to a subsequent unhealthy choice (see Mullen & Monin, 2016).

(8)

(Huber et al., 2008). For instance, research on the ‘shopping momentum effect’ (Dhar et al., 2007) draws on Gollwitzer’s (1990) theory of mindsets and suggests that spending on one product can motivate shoppers to continue purchasing a second, unrelated product due to a shift from a deliberative to an implementational mindset. Conversely, research on ego depletion shows that consumers engage in balancing responses by choosing vices over virtues once an initial choice depleted their finite pool of self-regulatory resources (Novemsky, Dhar, Schwarz, & Simonson, 2007). However, as suggested by the above examples, the mechanisms, which are studied in the literature on sequential choices, only explain either consistency, or balancing responses, but seldom both. Further, existing theories do not inform us when and why shoppers would consistently pursue or balance their health and indulgence goals throughout the shopping trip. Drawing on the literature on self-conscious emotions (e.g., Tangney & Tracy, 2011; Tracy & Robins, 2004; Tangney & Dearing, 2002), we contend that the emotions of pride and guilt, which shoppers may experience in response to a given purchase could be the underlying mechanism that predicts both consistency and balacing strategies in shoppers’ health goal pursuit in the grocery store.

(9)

different categories with each other, the comparison between vice and virtue categories may also be a relatively artificial choice task that does not necessarily reflect how shoppers make food choices in the grocery store. Research on shopping goals posits that shoppers tend to think of larger shopping trips (e.g., weekly trips with >10 purchse decisions; Kahn & Schmittlein, 1989) abstractly. Specifically, shoppers plan larger shopping trips in terms of the product categories from which they intend to shop, suggesting that the product category decision is often already made before shoppers enter the store (Sarantopoulos, Theotokis, Pramatari, & Doukidis, 2016). Hence, shoppers may be more likely to compare different product alternatives from one category against each other, rather than making trade-offs between two different categories. Put differently, standing in the snack aisle, shoppers may oftentimes not think of inherently healthy snacks, such as fruits or vegetables, but may consider the relative healthiness of different snack alternatives (e.g., deep-fried vs. oven-baked potatoe chips) to make their purchase decisions.

Shopping Emotions – The Mechanism Underlying Healthy Shopping Dynamics

(10)

actions, inactions, circumstances, or intentions’ (Baumeister, Stillwell, & Heatherton, 1994, p. 245). Pride and guilt belong to the class of self-conscious emotions that are unique to human nature because they are cognitively complex and require a high degree of self-reflection and self-evaluation (Tracy & Robins, 2004). Self-conscious emotions are experienced when a success or failure is attributed to the self, and perceived as (in)congruent to one’s sense of self, standards, or goals (Tracy & Robins, 2004). Since self-conscious emotions do not merely arise as a response to a self-relevant external event, but require the attribution of the event to an aspect of the self, they are also regarded as an individual’s reaction to its own characteristics or behavior (Tangney & Tracy, 2011): Hence, we may feel joy or contentment in response to a range of positive events. However, we may only experience pride when we credit the self for a postive outcome, or our own positive attributes and successes. Conversely, negative events can elicit a host of different negative emotions such as sadness, or frustration. Yet feelings of guilt only arise upon recognition of one’s own negative attributes or failures (Tangney & Tracy, 2011).

(11)

relatively healthy foods and the resistance of relatively unhealthy, and potentially tempting foods, requires shoppers to exercise self-control (Baumeister, Bratslavsky, Muraven, & Tice et al., 1998), we predict that shoppers also experience emotions of pride and guilt in response to relatively healthy choices (i.e., self-regulation successes) and relatively unhealthy choices (i.e., self-regulation failures) while shopping. According to Bisogni, Connors, Devine, and Sobal (2002, p. 129), ‘the body has become a marker for personal and social identity in Western societies, with a fit body being equated with self-control, self-denial, and will-power’. As the maintenance of a healthy diet is an integral part of achieving a fit and healthy body and regarded as the outcome of exercising self-control, shoppers may likely consider themselves as responsible for their food choices when shopping. That is, shoppers will likely credit or blame the self for the relative healthiness of their purchase decisions to the extent that these choices support or violate their health goals.

(12)

consider the purchase of relatively unhealthy and potentially health-compromising foods as a violation against their health goals. Since shoppers likely regard themselves as responsible for such self-regulation failures, they may experience feelings of guilt after they engaged in a relatively unhealthy food choice. In sum, we predict that emotions of pride and guilt are a function of the relative healthiness of shoppers’ preceding purchase decision: That is, shoppers experience greater emotions pride the relatively healthier their preceding choice was and greater emotions of guilt the relatively unhealthier their preceding choice was.

H1: The relative healthiness of a food choice positively (negatively) affects shoppers’

experience of pride (guilt).

(13)

related to this topic (Petty & Cacioppo, 1986). Individuals with a strong health motivation may therefore not only be more likely to engage in health-goal directed behavior, but may also show a higher inclination to engage in cognitions related to the relative healthiness of their choices. Consequently, we hypothesize that individuals with strong health motivations are more likely to critically reflect on the relative healthiness of their choices and evaluate their food choices against their health goals. Therefore, these shoppers experience stronger emotions of pride (guilt) after an initial relatively (un)healthy choice than shoppers with a weak health motivation. Accordingly, as shoppers with a weak health motivation may not consider the maintenance of a healthy diet as a goal that is relevant to their self-definitions, they may not pay sufficient attention to their purchase behavior, and therefore experience self-conscious emotions of pride and guilt to a lesser extent.

H2: Shoppers’ health motivation attenuates the relationship between the relative

healthiness of shoppers’ initial food choice and their experience self-conscious emotions of pride and guilt: Health consciousness positively (negatively) affects the relationship between the relative healthiness of a food choice and emotions of pride (guilt).

Shopping Emotions & Subsequent Health Goal Pursuit

(14)

regarding individuals’ goals, thereby helping them to navigate ‘through a landscape of short-sighted desires’ (Hofmann & Fisher, p. 682). Based on these findings, we argue that the emotions shoppers experience in response to the relative healthiness of a food choice can initiate or guide subsequent health goal-directed behavior (Tooby & Cosmides, 2008). Because emotions are not mere reactions to the healthiness of an initial food choice, but potentially also influence the healthiness of subsequent food choices, we consider emotions of pride and guilt as the psychological mechanism underlying healthy shopping dynamics.

While the literature is unanimous that emotions have implications for future goal-directed behavior, there is disagreement about when emotions of pride and guilt enhance or hamper the allocation of effort towards a focal goal. Will proud shoppers, who just made a relatively healthy choice, behave consistently, and make a healthy choice subsequently? Will guilty shoppers, who just made an unhealthy choice, try to make up for this initial violation of their health goals by choosing healthier products during the rest of the shopping trip? Or, will emotions of pride and guilt induce shoppers to disengage from their health goals leading to subsequent unhealthy purchases?

The Bright Side of Shopping Emotions – Positive Consistency & Balancing

(15)

Zhang, & Koo, 2009) and motivates individuals to act consistently with their long-term goals. Therefore, shoppers may react with positive consistency to the experience of emotions of pride.

Conversely, guilt in response to a transgression motivates reparative behavior to make up for previous self-regulatory failures (Tangney & Tracy, 2011). For instance, research on consumer guilt suggests that the experience of guilt can induce consumers to acknowledge their transgression and motivate them to make amends, enhance commitment to their long-term goals (Dahl et al., 2003), and reduce the consumption of unhealthy foods (Mohr et al., 2012; Mishra & Mishra, 2011). Such compensatory or reparative actions may serve to alleviate shoppers’ feelings of guilt after making an unhealthy food chioce, thereby stimulating healthier choices subsequently (Tangney & Dearing, 2002). For instance, Fishbach and Dhar (2005) demonstrate that dieters who believed they had made insufficient progress towards their weight loss goals were more likely to behave consistently with these goals (i.e., in choosing an apple instead of a chocolate bar as a parting gift) than dieters who perceived they had made sufficient progress towards their weight loss goals. Hence, the experience of guilt upon the purchase of a relatively unhealthy food may induce shoppers to react with positive balancing behavior.

The Dark Side of Shopping Emotions – Negative Consistency & Balancing

(16)

their long-term health goals by shopping for even unhealthier products subsequently. For instance, guilt experienced from actions contravening the achievement of a consumption goal, can induce consumers to respond with denial of their transgression without the stimulation of reparative actions (Dahl et al., 2003). Similarly, Nelissen and Zeelenberg (2009) found that guilt is likely to become maladaptive and lead to self-punishment, when individuals are under the impression that insufficient opportunities for reparation exist. This is in line with research on the ‘What-the-hell effect’, which shows that restrained eaters tend to mindlessly overeat, once they are induced to consume more than their allotted number of calories a day (Polivy & Herman, 1985). Since the consumption of calories cannot be reversed, restrained eaters perceive their dietary goal for that specific day as lost, which leads to complete disengagement from their dietary goals for that day. Hence, shoppers who experience guilt after a relatively unhealthy choice might also decide to disengage from the pursuit of their health goals to minimize any unpleasant feelings that could arise from goal failure and display negative consistency.

(17)

behavior. A summary of the four proposed behavioral responses to emotions of pride and guilt is provided in Table 1.

TABLE 1

Shoppers’ Responses to Emotions of Pride and Guilt

Relative Healthiness of the Next Choice Relatively Healthy Relatively Unhealthy Emotions

Pride Positive Consistency Negative Balancing

Guilt Positive Balancing Negative Consistency

The Role of the Shopping Trip Stage

We propose that these opposing predictions may be reconciled by taking the shopping trip stage, during which the shopper experiences emotions of pride and guilt, into consideration. Specifically, we predict that shoppers will respond differently to emotions of pride and guilt during different stages of the shopping trip (i.e., beginning vs. end).

As suggested by the self-regulation literature, consistently exercising self-control to abstain from tempting foods is an effortful process that taxes one’s self-regulatory resources over time (Baumeister et al., 1998). During larger grocery shopping trips, shoppers have to make multiple, consecutive purchase decisions, during which they will be confronted with the trade-off between relatively healthy versus relative unhealthy alternatives over and over again. Hence, the effortful nature of making relatively healthy food choices should be more pronounced later in the shopping trip (Patrick et al., 2009). Based on these findings, we argue that the positive motivational force of pride to motivate positive consistency in shoppers’ health goal pursuit weakens over the course of the shopping trip.

(18)

circumstances, emotions of pride may not suffice to motivate a subsequent relatively healthy choice anymore. On the contrary, we argue that emotions of pride may even be construed as a justification to indulge, and thereby serve as a license for shoppers to temporarily disengage from their health goals and give in to their need for instant gratification. Research suggests that individuals do not easily abandon their goals and are in need of justifications to choose hedonic over utilitarian alternatives (Okada, 2005). Similarly, Kivetz and Zheng (2006) argue that individuals frequently tend to rely on justification cues in the decision whether or not to indulge and find that self-control exercised to resist a temptation can be a sufficient justification to choose a hedonic over a utilitarian product, subsequently. Further, the licensing literature posits that licensing behavior can occur when shoppers’ prior restraint becomes salient as self-restraint is regarded as a justification to indulge (Mukhopadhyay & Johar, 2009).

Consequently, we predict that shoppers will behave consistently (i.e., make two sequential relatively healthy choices) upon the experience of pride early in the shopping trip since they still have sufficient self-regulatory resources at their expense to continue exercising self-control in pursuit of their health goals. However, at the end of the shopping trip, shoppers may not possess the necessary self-regulatory resources to restrain themselves any more. Since the persistent pursuit of their health goals may have become too effortful, shoppers may tend to construe emotions of pride as a justification to indulge. Hence, we suggest that shoppers display balancing responses (i.e., a relatively healthy choice followed by a relatively unhealthy choice) upon the experience of emotions of pride later in the trip.

H3: The relative healthiness of a food choice positively (negatively) affects the relative

healthiness of the next choice through pride early (late) in the shopping trip.

(19)

fail to find a path toward reparation of their previous transgressions (Tangney & Tracy, 2011; Tangney & Dearing, 2002), whereas guilt is most likely to lead to positive outcomes when individuals take responsibility for their transgressions and react to guilt by developing and carrying out ‘a reparative plan’ (Tangney & Tracy, 2011). We suggest that shoppers’ perceptions of whether opportunities for reparation still exist essentially depend on the shopping trip stage. That is, shoppers who made a relatively unhealthy choice at the beginning of the shopping trip, still possess the self-regulatory resources to resist possible temptations from subsequent encounters with relative unhealthy foods. That is, shoppers’ self-regulatory resources enable them to make up for previous relatively unhealthy choices by choosing relatively healthier products subsequently. At the end of the shopping trip, however, shoppers may have greater difficulties with identifying and following a path toward reparation. The preceding food choices may have already taxed shoppers’ self-regulatory resources (Baumeister et al., 1998), such that shoppers may feel fatigued and find it too difficult to make relatively healthy choices subsequently.

We propose that shoppers who experience feelings of guilt in response to a relatively unhealthy food choice late in the shopping trip, may perceive reparation for an initial unhealthy choice as too effortful or even impossible. This perception will lead shoppers to disengage from their health goal pursuit and induce them to continue shopping for relatively unhealthy products later in the trip (i.e., negative consistency). Hence, we predict that shoppers will behave consistently (i.e., make two sequential relatively unhealthy choices) when experiencing guilt at the end of the trip, whereas shoppers will display balancing responses (i.e., a relatively unhealthy choice followed by a relatively healthy choice) upon the experience of emotions of guilt early in the trip.

H4: The relative healthiness of a food choice positively (negatively) affects the relative

(20)

In sum, we contend that the emotions, which shoppers experience in response to the relative healthiness of an initial food choice, play a central role in allocating effort between shoppers’ long-term health goals and their desire for indulgence in a grocery shopping context and possibly account for the dependencies between two sequential food choices. Specifically, we suggest that the relative healthiness of a food choice positively affects shoppers’ experience of pride (i.e., greater relative healthiness leads to more pride), whereas it negatively affects shoppers’ experience of guilt (i.e., smaller relative healthiness leads to more guilt). Since shoppers with strong health motivations may be more inclined to direct attention towards the healthiness of their purchases and critically reflect on their choice behavior, such shoppers may be more prone to the experience of self-conscious emotions of pride and guilt during the shopping trip. We contend that whether shopping emotions sustain or hamper subsequent health-goal striving essentially depends on the the shopping trip stage, during which these emotions are experienced. Figure 1 visualizes how the relative healthiness of an initial food choice affects the healthiness of a subsequent choice through shoppers’ experienced shopping emotions.

(21)

METHOD

To test whether the healthiness of a given food choice is a function of the relative healthiness of the preceding choice, and the emotions related to this choice, we conducted an online experiment with a sample of 401 US consumers from Amazon’s MTurk consumer panel. The study involved a computer-simulated grocery shopping task where participants engaged in a sequence of 15 food choices from common product categories. An overview of these 15 product categories is provided in Appendix 1. The average age of participants was 39 years (age range: 20–76 years), and 55.4 % of participants were female. Since the study involved a grocery shopping task, participants were only eligible to the study if they were the primary grocery shopper within their household. The average household size was 2.6 people (size range: 1–9 people).

Experimental Design

The study was a 2 x 2 between-subjects design with the timing of two sequential choices (i.e., choices 5 and 6 vs. 10 and 11) and the measurement of emotions (i.e., yes vs. no) as experimental factors. Consequently, participants were assigned to one of four experimental conditions, either with or without emotion measures and with the two focal choices early (i.e., choices 5 and 6) or late (i.e., choices 10 and 11) in the shopping trip. The relative healthiness of the initial choice (i.e., choice 5 or 10, respectively), served as independent variable, whereas the relative healthiness of the next choice (i.e., choice 6 or 11, respectively) served as dependent variable. To rule out that any product category differences distort the relationship between the relative healthiness of these two sequential choices, all participants chose a product from the same product category (i.e., granola bars) as 6th_{or 11}th_{choice. For the}

(22)

presented to them randomly. A graphical illustration of the grocery shopping task for the four experimental conditions is provided in Figure 2.

FIGURE 2

Graphical Illustration of Grocery Shopping Task for all Conditions

We deliberately selected the product category granola bars as choice set for the dependent variable since we expected that participants would neither perceive this product category as very healthy, nor as very unhealthy, thereby allowing for less inferences on the healthiness of the product based on the perceived healthiness of the category. To verify this assumption, we assessed participants’ perceptions regarding the healthiness of all 15 product categories (‘Please indicate the extent to which you perceive the product categories from which you just shopped as healthy or unhealthy.’; 1 = ‘not healthy at all’, and 7 = ‘very healthy’; see Appendix 1). On average, participants perceived granola bars as moderately healthy (M = 4.37, SD = 1.42), suggesting that there was some ambiguity regarding the perceived healthiness of this choice set.

The two conditions, in which participants shopped without interception served as control conditions (i.e., with granola bars as 6th_{vs. 11}th_{choice, respectively) to verify whether}

(23)

behavior. In emotions research, the subjective experience of emotions is usually captured via self-reported measures. Such measurement instruments assume that individuals’ feeling states remain constant during the measurement, such that the process of comprehending and rating emotions does not change the experience at hand (Quingley, Lindqvist, & Barrett, 2014). However, subjective feelings, the conscious component of an emotional response, are the result of a complex cognitive process that requires focal attention and the interpretation of a wide range of information (Kassam & Mendes, 2013). Such cognitions can change how a participant feels (Quingley et al., 2014), can have a substantial effect on the body’s reaction to an emotional situation (Kassam & Mendes, 2013), and alter behavioral responses to emotions (Carver & Scheier, 1982). For instance, self-awareness can enhance cognitive processing, thereby increase the salience of an emotional experience, and change the relationship between the emotion and subsequent behavior (Carver & Scheier, 1982). Consequently, it may be possible that the assessment of participants’ experienced emotions of pride and guilt in the two emotion conditions might distort or diminish possible effects of participants’ experienced emotions on the relative healthiness of their next choice.

(24)

Moreover, participants were instructed to make choices that reflected their preferences during a regular grocery shopping trip. To motivate participants to choose products that closely reflected their usual choices, we used incentive-alignment (Ding 2007; Ding, Grewal, & Liechty, 2005). Participants were informed that one in fifty shoppers was eligible for a prize package worth $50. These prize packages consisted of all the products that participants selected during the grocery shopping task and some cash. If the total price of the items in the shopping basket was $35, participants would receive those groceries and $15 in cash ($50 – $35 = $15). We measured how realistically participants engaged in the decision-making process during the shopping task (‘The choices I made accurately reflect what I would do in my regular grocery store’; ‘The comparisons I made between products and prices closely reflected how I make decisions in my regular store’; 1 = ‘strongly disagree’, and 7 = ‘strongly agree’; = .76). On average, participants scored high on this measure (M = 6.12, SD = .99), suggesting that the incentive alignment was successful and that participants’ food choices were sufficiently realistic. There were no differences between the four conditions, F(3, 397) = .59 , p = .621.

Procedure

Participants were randomly assigned to one of the four conditions and asked to shop for products from the 15 pre-specified product categories. Further, participants were informed they would be asked some questions pertaining to their shopping experience after they finished the grocery shopping task. Participants shopped for one product category at a time, in random order, except for the granola bars which were presented as 6th_{or 11}th_{choice set (i.e.,}

(25)

of all the 15 product categories they would shop for during the grocery shopping task. This was to ensure that participants were aware of their concrete shopping goal.

The selected product categories entailed common categories (e.g., bread, yoghurt, peanut butter, cola) that differed in terms of their perceived healthiness (see Appendix 1). To assess participants’ within-category choice behavior, each product category consisted of four product options that differed in terms of their relative healthiness (i.e., calories/100g compared to the category average). For each option, participants were provided with a picture of the product, the product name, unit size, and price, as well as the nutritional information per 100g. An illustration of such a choice set is provided in Appendix 2. Participants in the two emotion conditions were interrupted after 5 or 10 food choices, respectively and responded to a measure of the self-conscious shopping emotions of guilt and pride. After the emotions were assessed, participants continued with the grocery shopping task. Participants in the two control conditions made the 15 subsequent food choices without interception. Upon completion of the grocery shopping task, participants responded to some socio-demographical measures.

Measures

Relative healthiness. We measured the relative healthiness of each food choice based

(26)

alternative based on its number of calories/100g. Further, using the number of calories/100g as a healthiness indicator allowed us to capture how participants’ selected product options differed from the remaining options within a given choice set in a more nuanced manner. That is, based on the calories/100g we could not only assess whether a given choice was relatively healthier, or unhealthier compared to the category baseline, but also how much healthier or unhealthier this choice was. For all participants, we recorded the calories/100g of each choice as well as the sequence, in which the different choice sets were presented.

To assess the relative healthiness of participants’ choices, we computed a health index (HI) for each choice based on the caloric information of the products. The HI was computed by subtracting the number of calories/100g of a given food choice from the average number of calories/100g of the four products in the respective choice set. If the average calories/100g for a choice set amounted to 100 calories/100g, and a given food choice within this category had 60 calories/100g, then the HI equals 40 (i.e., 100 calories/100g – 60 calories/100g = 40 calories/100g). A HI of 40 indicates that the choice made by a participant was 40 calories healthier compared to the average calories/100g within a certain category. Hence, a HI higher than zero indicates a relatively healthier choice compared to the category baseline, whereas a HI lower than zero indicates a relatively unhealthier choice within this category. The HIs of the choice which preceded the granola bar choice set (i.e., choices 5 or 10, respectively) served as independent variable (Mlast choice = -5.06, SD = 96.23; HI range: -308.25–234.75).

The health index of the granola bar choice served as dependent variable (Mgranola = 4.62, SD =

(27)

FIGURE 3

Average Health Indexes of Participants’ Choices

(28)

participants tended to feel relatively proud regarding their preceding choices (Mpride = 3.71,

SD = .86), but not very guilty (Mguilt = 1.32, SD = .63). Although pride and guilt are significantly negatively correlated with each other (r = -.24, p = .001; see Appendix 3 for the correlation table), the correlation coefficient is not very high, suggesting that pride and guilt may be regarded as two distinct, yet slightly related emotions. Participants’ experienced emotions of pride and guilt did not differ depending on the stage of the shopping trip (pride:

t(198) = -.47, p = .637; guilt: t(198) = -.15, p = .880).

Shopping trip stage. The shopping trip stage during which the granola bar choice set

was presented was recoded into a dummy variable, where a value of 1 indicates that the participant encountered the granola bar choice set late in the shopping trip (i.e., after 10 choices).

Health motivations. We selected health consciousness as operationalization of

(29)

RESULTS

Hypotheses Testing

The Relative Healthiness of the Last Choice & Shopping Emotions

We predicted that the relative healthiness of shoppers’ food choices prior to the granola bar triggers shopping emotions, which determine the relative healthiness of the chosen granola bar, in turn. Specifically, we hypothesized that the relative healthiness of the last choice prior to the granola bar (HI last choice) positively affects emotions of pride (i.e., higher relative healthiness leads to more pride), whereas it negatively affects emotions guilt (i.e., lower relative healthiness leads to more guilt, see H1). Further, we expected that this

effect is attenuated for health-conscious shoppers, such that health-conscious shoppers experience stronger emotions of pride after a relatively healthy choice and stronger emotions of guilt after a relatively unhealthy choice (see H2). To test H1 and H2, we conducted two

hierarchical ordinary least squares (OLS) regressions, where we entered HI last choice (i.e., Step 1) and health consciousness with the respective interaction term (i.e., Step 2) as predictors of either pride or guilt. Pride, guilt, and health consciousness were mean-centered prior to the analyses. Since the variance inflation scores in all of the following analyses were below 3, multicollinearity was not deemed an issue. The results of the regression analysis for outcome variable pride are displayed in Table 2 and for guilt in Table 3.

Pride. As displayed in Table 2, the relative healthiness of the last choice and health

consciousness explain 9.00 % of the variation in pride. The final model (i.e., Step 2) significantly predicts participants’ experience of pride (F(3, 196) = 6.488, p < .001). Contrary to H1, the relative healthiness of the last choice does not significantly affect shoppers’

(30)

pride than less health-conscious shoppers. Further, we find a marginally significant interaction effect between the relative healthiness of the last choice and health consciousness (B = .002, t(196) = 1.944, p = .053), providing preliminary support for H2. This finding is in

line with our assumption that shoppers experience significantly more pride after a relatively healthy choice the more health-conscious they are.

TABLE 2

Effect of the Relative Healthiness of the Last Choice Prior to the Granola Bar on Pride

Step 1 Pride N = 200 B SE(B) b t p Constant .000 .061 .002 .999 HI Last Choice .000 .001 .005 .072 .943 Step 2 Constant -.027 .060 -.454 .650 HI Last Choice .000 .001 -.055 -.785 .434 Health Consciousness .374 .093 .279 3.999 .000 HI Last Choice x HC .002 .001 .132 1.944 .053

Note. 95% Confidence Interval. HC = Health Consciousness.

Guilt. As the results in Table 3 suggest, the final model (i.e., Step 2) does not

significantly predict participants’ experience of guilt (F(3, 196) = 1.860, p = .138). Specifically, the relative healthiness of the last choice and health consciousness merely explain 2.8 % of the variation in guilt. Against our prediction that the relative healthiness of the last choice is negatively related to emotions of guilt, HI last choice does not significantly affect guilt (B = .001, t(196) = 1.231, p = .220). Hence, H1 is also not supported for guilt.

However, we find a marginally significant interaction between HI last choice and health consciousness (B = -.001, t(196) = -1.737, p = .084). This suggests that as the relative healthiness of the last choice decreases, health-conscious shoppers experience significantly more guilt than shoppers with lower levels of health consciousness. This finding provides further support for H2. Interestingly, health consciousness does not directly affect emotions of

(31)

suggest that while participants’ level of health consciousness appears to influence their general susceptibility to experience pride, it does not influence whether participants are generally more prone to experience guilt.

TABLE 3

Effect of the Relative Healthiness of the Last Choice Prior to the Granola Bar on Guilt

Step 1 Guilt N = 200 B SE(B) b t p Constant .001 .044 .034 .973 HI Last Choice .001 .000 .101 1.426 .155 Step 2 Constant .019 .045 .418 .676 HI Last Choice .001 .000 .089 1.231 .220 Health Consciousness .048 .070 .049 .682 .496 HI Last Choice x HC -.001 .001 -.122 -1.737 .084

Note. 95% Confidence Interval. HC = Health Consciousness.

Shopping Emotions & the Relative Healthiness of the Granola Bar Choice

We predicted that the emotions, which shoppers experience in response to the relative healthiness of their last food choice, subsequently affect the relative healthiness of the granola bar choice depending on the shopping trip stage. Specifically, we hypothesized that emotions of pride (H3) and guilt (H4) both positively affect the relative healthiness of the granola bar

(32)

TABLE 4

Effect of Emotions of Pride and Guilt on the Relative Healthiness of the Granola Bar

Step 1 HI Granola Bar N = 200 B SE(B) b t p Constant 5.300 2.189 2.421 .016 Pride 1.192 2.615 .033 .456 .649 Guilt -.304 3.594 -.006 -.085 .933 Step 2 Constant 5.872 3.120 1.882 .061 Pride 5.290 3.866 .148 1.368 .173 Guilt -.318 5.068 -.007 -.063 .950

Shopping Trip Stage -.919 4.390 -.015 -.209 .834

Pride x STS -7.535 5.260 -.155 -1.432 .154

Guilt x STS .212 7.204 .003 .029 .977

Note. 95% Confidence Interval. STS = Shopping Trip Stage (1 = late).

The final model (i.e., Step 2) only explains 1.3% of the variation in HI granola bar, suggesting that it does not significantly predict the relative healthiness of the granola bar choice (F(5, 194) = .498, p = .778). While there is some directional support that pride positively affects HI granola bar (B = 5.290, t(194) = 1.368, p = .173), there is no significant effect of guilt on HI granola bar (B = -.318, t(194) = -.063, p = .950). Hence, it appears that participants’ experienced emotions did not significantly affect the relative healthiness of the next choice. Further, the interaction effect between pride and the shopping trip stage is not significant (B = -7.535, t(194) = -1.432, p = .154). Nevertheless, the effect still points into the hypothesized direction, yielding directional support for H3, which predicted a negative

(33)

Mediation Analysis

As part of H3 and H4 we predicted that the relative healthiness of the last choice

affects the relative healthiness of the granola bar choice through emotions of pride and guilt. To test this hypothesized mediation, we conducted a mediation analysis using Model 29 (bias-corrected, 1,000 bootstrap samples, 95% confidence interval; see Appendix 4) of Hayes’ PROCESS macro (Hayes, 2012). The results of this analysis are displayed in Table 5 and Appendix 5.

TABLE 5

Mediation Through Pride and Guilt

Pride N = 200 B SE(B) t p Constant -.027 .060 -.454 .650 HI Last Choice -.001 .001 -.785 .434 Health Consciousness .374 .094 3.999 .000 HI Last Choice x HC .002 .001 1.944 .053 Guilt N = 200 B SE t p Constant .019 .045 .418 .676 HI Last Choice .001 .000 1.231 .220 Health Consciousness .048 .070 .682 .496 HI Last Choice x HC -.001 .001 -1.737 .084 HI Granola Bar N = 200 B SE t p Constant 6.600 3.031 2.178 .031 Pride 2.251 3.763 .598 .550 Guilt -3.222 4.892 -.659 .551 HI Last Choice .066 .027 2.424 .016

Shopping Trip Stage -1.564 4.192 -.373 .710

Pride x STS -5.999 5.050 -1.188 .236

Guilt x STS .255 6.903 .037 .971

Health Consciousness 9.673 3.481 2.778 .006

HI Last Choice x HC -.021 .034 -.598 .551

HI Last Choice x STS .012 .043 .278 .781

(34)

In line with the results of the preceding OLS regression analyses, the results in Table 5 indicate that the effect of the relative healthiness of the last choice on emotions of pride and guilt is amplified for health-conscious shoppers: That is, health-conscious shoppers experience (marginally) significantly stronger feelings of guilt as the relative healthiness of the last choice decreases (B = -.001, t(196) = -1.737, p = .084), and (marginally) significantly stronger feelings of pride as the relative healthiness of the last choice increases (B = .002,

t(196) = 1.944, p = .053). Although HI last choice neither significantly affects emotions of

pride (B = -.001, t(196) = -.785, p = .434), nor guilt (B = .001, t(196) = 1.231, p = .220), it significantly positively affects the relative healthiness of the granola bar choice (B = .066,

t(190) = 2.424, p = .016). This suggests that if participants’ initial choice was relatively

healthier, they also chose a relatively healthier granola bar, on average.

Further, the results of the mediation analysis reveal that the relative healthiness of the last choice, participants’ experienced shopping emotions, health consciousness, and the shopping trip stage significantly predict the relative healthiness of the granola bar choice (F(9, 190) = 3.035, p = .002) and jointly explain 12.57 % of the variation in the relative healthiness of this choice. Against our expectation, however, emotions of pride and guilt both do not significantly affect the relative healthiness of the granola bar choice (pride: B = 2.251, t(190) = .598, p = .550; guilt: B = -3.222, t(190) = -.659, p = .551). Appendix 5 displays a test of the conditional indirect and direct effects of HI last choice on HI granola bar for different levels of the mediators pride and guilt and the moderators health consciousness and shopping trip stage. Contrary to H3 and H4, the results reveal that the effect of the relative healthiness of the

(35)

the relative healthiness of the granola bar choice cannot be explained by participants’ experienced emotions.

In addition, the results in Table 5 reveal that also health consciousness is a highly significant predictor of the relative healthiness of the granola bar choice (B = 9.673, t(190) = 2.778, p = .006). That is, the more health-conscious participants were, the healthier was the granola bar they chose. Although the interaction term of HI last choice and health consciousness is non-significant (B = -.021, t(190) = -.598, p = .551), the test of the conditional direct effect in Appendix 5 indicates that the relative healthiness of the last choice significantly positively affects the relative healthiness of the granola bar choice for values of HC that are one standard deviation below the mean (BSTS=0 = .080, t(190) = 2.267, p = .025;

BSTS=1 = .091, t(190) = 2.242 p = .026), or equal to the mean (BSTS=0 = .066, t(190) = 2.424, p = .016; BSTS=1 = .078, t(190) = 2.350, p = .020). For participants with levels of health consciousness that are one standard deviation above the mean this effect is no longer significant at the 5% level (BSTS=0 = .053, t(190) = 1.499, p = .136; BSTS=1 = .065, t(190) = 1.655, p = .100). This seems to suggest that the relative healthiness of the last choice prior to the granola bar is a stronger predictor of the relative healthiness of the granola bar for participants with low to average levels of health consciousness.

Discussion

(36)

(37)

Although at least emotions of pride appear to affect the relative healthiness of the next choice, our mediation analyses do not provide evidence that emotions of pride and guilt also account for the relationship between the relative healthiness of two subsequent choices. Overall, it appears that health consciousness and the relative healthiness of the last choice are the strongest predictors of the relative healthiness of shoppers’ subsequent choice. Given that there is extant evidence for the mediating effect of emotions between past and future goal-striving in the literature (e.g., Hofmann & Fisher, 2012; Williams & DeSteno, 2008; Muraven et al., 2005), these results are somewhat surprising. We attribute the absence of a mediating effect of self-conscious emotions to the use of self-reported measurement instruments to capture participants’ subjective experience of pride and guilt. Although self-reported measures are the only way to ‘assess the content of subjective experiences of emotion’ (Quingley et al., 2015, p. 35), the use of self-reported measures to capture such emotional experiences is contested in the literature (Scherer, 2005): On the one hand, they are prone to reporting biases since individuals may not be willing or able to report their emotions (Paulhus & Reid, 1991). On the other hand, asking participants to reflect and report how they feel about past behavior might also change their subsequent choice behavior (Quingley et al., 2015; Kassam & Mendes, 2013). To account for the possibility that the measurement of emotions of pride and guilt itself may have distorted or diminished the dependencies between two subsequent choices, we compared the relationship between the relative healthiness of the last choice and the relative healthiness of the granola bar choice across the two emotion and control conditions. The results of this analysis are provided in the next paragraph.

Comparison Between Control Conditions & Emotion Conditions

(38)

Step 1), health consciousness (i.e., Step 2), and shopping trip stage (i.e., Step 3) including the respective interaction terms as predictor variables of HI granola bar. Thereby, one sample consisted of the two control conditions, and one sample consisted of the two emotions conditions. The results of these analyses are displayed in Table 6.

The final model (i.e., Step 3) explains 7.8 % of the variation in the relative healthiness of the granola bar in the two control conditions (F(5, 195) = 3.288, p = .007) and 11.5 % of variation in the two emotion conditions (F(5, 194) = 5.025, p < .001). In both samples, the relative healthiness of the last choice significantly positively affects the relative healthiness of the granola bar choice (control: B = .091, t(195) = 3.233, p = .001; emotion: B = .067, t(194) = 2.459, p = .015). Hence, the higher the relative healthiness of an initial choice was, the healthier the next choice became. However, this effect is slightly smaller in the two emotion conditions compared to the control conditions, indicating that the relationship between the relative healthiness of the last choice and the granola bar is slightly more pronounced in the two control conditions.

(39)

TABLE 6

Effect of HI Last Choice on HI Granola Bar – Control and Emotion Conditions Control Conditions

N = 201

Emotion Conditions

N = 200

Step 1 B SE(B) b t p B SE(B) b t p

Constant 4.534 1.971 2.301 .022 5.500 2.100 2.618 .010 HI Last Choice .077 .022 .245 3.558 .000 .084 .021 .276 4.040 .000 Step 2 Constant 4.077 1.998 2.041 .043 5.782 2.122 2.725 .007 HI Last Choice .074 .022 .235 3.361 .001 .071 .021 .233 3.387 .001 Health Consciousness 4.440 3.119 .102 1.424 .156 9.150 3.288 .192 2.783 .006 HI Last Choice x HC .040 .034 .085 1.202 .231 -.022 .033 -.044 -.657 .512 Step 3 Constant 3.623 2.802 1.293 .198 6.602 3.014 2.191 .030 HI Last Choice .091 .028 .289 3.233 .001 .067 .027 .219 2.459 .015 Health Consciousness 4.498 3.127 .103 1.438 .152 9.088 3.305 .191 2.749 .007 HI Last Choice x HC .030 .035 .064 .855 .394 -.023 .033 -.046 -.677 .499

Shopping Trip Stage .709 3.965 .012 .179 .858 -1.680 4.169 -.027 -.403 .687

HI Last Choice x STS -.045 .047 -.089 -.954 .341 .013 .042 .026 .300 .765

(40)

Discussion

The comparison between emotion and control conditions suggests that participants indeed appeared to shop differently when they were asked to reflect on their emotions regarding their initial choices. Across conditions, the relatively healthier participants’ initial choice was, the relatively healthier their subsequent choice became, in turn. However, this effect is less pronounced in the emotion conditions compared to the control conditions. We suspect that this difference can be attributed to the assessment of shoppers’ emotions of pride and guilt in the two emotion conditions. Specifically, we contend that asking participants to reflect on their preceding purchase decision may have weakened the relationship between the relative healthiness of the two sequential choices. In the literature, there is evidence suggesting that the reflection on and expression of emotions can alleviate the experience of an emotion and reduce the effect that this emotion has on subsequent behavior (Kassam & Mendes, 2013). This may also explain why we did not find evidence for a mediating effect of emotions of pride and guilt in previous analyses.

(41)

regular shopping trips. Hence, the mere existence of a health goal may not suffice to influence shopping behavior. Instead, this health goal also needs to be salient to the shopper when making purchase decisions in the grocery store.

GENERAL DISCUSSION

The results of this study reveal that the relative healthiness of a given choice depends on the relative healthiness of participants’ preceding choice, providing preliminary support for the existence of healthy shopping dynamics. Nevertheless, we did not find empirical evidence that shopping emotions account for this dependence between the relative healthiness of shoppers’ consecutive choices. This is surprising given that there is ample evidence for the mediating effect of self-conscious emotions between past and future goal-directed behavior in the literature (e.g., Hofmann & Fisher, 2012; Williams & DeSteno, 2008; Muraven et al., 2005).

We predominantly attribute the lack of a mediating effect to the use of self-reported measures for the assessment of conscious emotions of pride and guilt. Although self-reported measures are often still the only way to receive insights into individuals’ subjective experience of emotions, the literature suggests that inducing individuals to reflect on their emotions may change their emotional state and distort or diminish possible behavioral consequences of emotions (Quingley et al., 2014; Kassam & Mendes, 2013). For instance, merely writing about the experience of negative emotions for a few minutes every day is known to ameliorate the negative effects of these emotions and lead to significant physical and mental health improvements (Pennebaker, 1997).

(42)

conditions. Further, we found that this relationship was only significant for shoppers with below to average levels of health consciousness in the two emotion conditions. We suspect that this was the case since shoppers characterized by above average levels of health consciousness reported stronger emotions of pride and guilt. Hence, it may be relatively likely that the reporting of emotions of pride and guilt itself reduced the strength of the relationship between the relative healthiness of the two studied sequential choices for shoppers with above average levels of health consciousness. Further, our self-reported measures only tapped into participants’ conscious experience of emotions, such that any emotional states outside participants’ consciousness may not have been captured by our measures.

Alongside the measurement procedure, another possible explanation for a lack of a mediating effect of emotions of pride and guilt could be that we measured the post-decisional experience of these emotions. However, the literature suggests that emotions can arise both in response to, or in anticipation of decision outcomes (Bagozzi, Baumgartner, & Pieters, 1998). Anticipated emotions, also referred to as affective forecasts, ‘are predictions of the emotional consequences of decision outcomes’ (Patrick et al., 2009, p. 538) that can help individuals to order their preferences (Kahneman & Snell, 1992) and affect individuals’ self-control (Bagozzi et al., 1998). For instance, Patrick et al. (2009) find that the anticipation of pride from resisting to a temptation (i.e., a jar of cookies) can motivate individuals to continue exercising self-control over an extended period of time, whereas the anticipation of negative emotions such as shame (Patrick et al., 2009), or guilt undermines subsequent self-control (Chun, Patrick, & MacInnis, 2006). Hence, it may be possible that the dependencies between the relative healthiness of sequential choices could also be affected by shoppers’ anticipated, instead of post-decisional emotions.

(43)

still provides evidence that shoppers’ level of health consciousness affects the extent to which they experience emotions of pride and guilt in response to the relative healthiness of the preceding choice. That is, the more health-conscious shoppers were, the greater was their experience of pride and guilt given the relative healthiness of their previous choice. This aligns with the literature on self-conscious emotions, which predicts that emotions of pride and guilt may only be experienced to the extent that individuals attribute a self-regulation success or failure to the self and consider these successes or failures as relevant to their self-definitions (Tracy & Robins, 2004). Interestingly, the comparison between the two control and emotion conditions revealed that shoppers’ level of health consciousness only significantly affected the relative healthiness of the next choice when shoppers were induced to reflect on the healthiness of their preceding purchases in the emotion conditions. This suggests that health-conscious shoppers may only be able to make healthier choices when they are forced to reflect on the extent to which these align with their health goals. This finding underlines the cognitive complexity attested to self-conscious emotions in the literature (Tangney & Tracy, 2011; Tracy & Robins, 2004; Lazarus, 1991).

(44)

eventually reverse any positive motivational consequences of the experience of pride. However, our results provide no evidence for a corresponding effect for emotions of guilt. Given that participants generally reported relatively low levels of guilt, these findings may be attributable to a reporting bias rather than the mechanism itself. Table 7 provides an overview of the tested hypotheses and results.

TABLE 7

Overview of Hypotheses and Results

Dependent Variables

Predictor Variables Pride Guilt HI Granola Bar

HI Last Choice H1 increases decreases X

HI Last Choice x HC H2 increases increases ✔

Pride x STS H3 decreases (✔)

Guilt x STS H4 decreases X

Note. STS = Shopping Trip Stage (1 = late), HC = Health Consciousness. Parentheses indicate directional support for our hypothesis.

Implications for Marketing Theory

Our research provides preliminary evidence for the existence of dependencies between the relative healthiness of sequential choices and suggests that such healthy shopping dynamics are the result of a complex interplay between the relative healthiness of shoppers’ food choices, their health motivation, their experienced shopping emotions, and the shopping trip stage during which these emotions are experienced. We contribute to a growing body of research on the self-regulatory function of self-conscious emotions (e.g., Hofmann & Fisher, 2012; Williams & DeSteno, 2008) and dynamic in-store decision making (e.g., van der Heide et al., 2016; Sheehan & van Ittersum, 2016; van Ittersum et al., 2013; Dhar et al., 2007).

Self-conscious emotions. Although our study did not provide empirical evidence for

(45)

grocery store and may therefore not always be inherently adaptive (Tangney & Tracy, 2011). Specifically, our findings suggest that although pride may generally have positive motivational consequences, this effect can be reversed entirely during later stages of the shopping trip and induce shoppers to engage in licensing behavior. Thereby, we also inform the literature on the licensing phenomenon (e.g., Mukhopadhyay & Johar, 2009; Wilcox et al., 2010) and demonstrate that the motivational outcome of pride in sequential consumer decision-making depends on the number of choices that shoppers have already made. Overall, these findings imply that it is crucial to take the broader choice context, in which emotions are experienced, into account.

For future theory development, it will be critical to investigate whether there is a corresponding effect for guilt. Although also guilt is regarded as an inherently adaptive self-conscious emotion (e.g., Tangney & Tracy, 2011; Fishbach & Dhar, 2005; Dahl et al., 2003), research also suggests that guilt can become maladaptive when individuals feel that reparation for their initial transgressions is impossible (e.g., Nelissen & Zeelenberg, 2009; Tangney & Dearing, 2002). Pinpointing when and why in the shopping trip emotions of pride and guilt enhance or hamper subsequent health-goal striving, would enable the development of a comprehensive framework that explains both when shoppers behave consistently with their health goals, and when shoppers balance their health and indulgence goals in the grocery store.

Dynamic in-store decision making. Our research extends recent advances in the area of

(46)

not provide participants with real-time caloric feedback on their food choices, the assessment of participants’ shopping emotions may have induced them to provide shopping feedback to themselves by reflecting on their past food choices and their emotions related to these choices. We show that shoppers’ level of health consciousness may positively affect the relative healthiness of their food choices, but this is more strongly the case to the extent that they are forced to deliberate on their initial choices. This suggests that even health-conscious shoppers may be unable to relate their purchase decisions to their health goals while shopping, which may explain why so many shoppers fail to pursue their health goals during major grocery shopping trips. In moving forward, it will be critical for theory building to understand how health interventions affect shoppers’ emotions and cognitions, and how these emotions affect subsequent decision-making, in turn.

Implications for Marketing Practice

Marketers and policy makers are constantly looking for effective ways to influence in-store decision making to curb the obesity epidemic. In the last couple of years, we could observe increasing debates on different types of health interventions aimed at reducing the share of high-caloric and unhealthy foods in consumers’ shopping baskets. These are ranging from fat and sugar taxes for high-caloric foods, and drinks to traffic lights labelling for soft drinks, health labels, and smart shopping carts. Further, the rise of digital and mobile technologies has created endless opportunities for marketers and policy makers to interact with shoppers at the point-of-purchase in real-time.

(47)

interventions that support shoppers in exercising self-control to resist the purchase of relatively unhealthy foods later in the shopping trip. Nevertheless, further research is warranted to determine the exact point in the shopping trip at which shoppers should be intercepted. To determine when shoppers may be in need of some extra help to pursue their health goals while shopping, retailers could try to forecast shoppers’ choice behavior based on historic purchase data from hand-held scanners that record the sequence, in which purchases decisions were made in the grocery store. Further, our study suggests that shoppers’ health goals may not always be very salient while shopping. As implied by our study, a small intervention that induces shoppers to reflect on their preceding purchases and draws attention to their health goals may suffice to make shoppers’ health goals more salient and influential in their decision-making.

Mixed Feelings, Mixed Baskets