The more we care, the harder we snack

(1)

The more we care, the

harder we snack

How do in store display advertisements for healthy and

unhealthy snacks affect high and low health consciousness-

consumers’ baskets?

Author: Ghaith Adnan

Master Thesis – MSc Strategic Marketing

Completion Date: 19/06/2015

Address: 9721PS, Bilderdijklaan 79, Groningen

Tel(Mobile): +31 6 87935510

Email:

ghaith.adnan@gmail.com

Student Number: S2709317

Thesis Supervisor: Professor Koert Van Ittersum

Supervisor 2: Yannick Joye

(2)

1

1.5 - Contribution: ... 9 2. THEORETICAL FRAMEWORK ... 10 2.1 - Conceptual Model ... 10 2.2 - Theoretical Background ... 11 2.3 - Experimental Design ... 13 2.4 – Hypotheses ... 13 3 – METHODOLOGY ... 14 3.1 Sampling: ... 14 3.2 - Instrumentation ... 15 3.3 Analysis: ... 18 4. RESULTS ... 19 4.1 - Descriptive Statistics ... 19

4.2 Testing the Hypotheses: ... 21

4.3 - Testing For Moderation: ... 29

5. DISCUSSION ... 34

5.1 - H1: Impact of a healthy product advertisement in a retail setting on snack choice: ... 34

5.2 - H2: Impact of an unhealthy product advertisement in a retail setting on snack choice: ... 35

5.3 - Implications: ... 36

5.4 - Limitations and Future research ... 37

5.5 - Conclusion: ... 39

(3)

2

ABSTRACT

Extant literature on obesity highlights it as an issue that deserves attention as a problem that is forecasted to grow as a global pandemic in the future. This research sheds light on how the marketing field affects consumers’ obesity risk and tackles the issue by exploring influences on consumers’ decisions in a retail setting. The effects of two types of informative in-store advertisements – healthy snack advertisements and unhealthy snack advertisements – on consumers’ basket healthiness are studied. A survey based experiment is conducted with

(4)

3

PREFACE

‘This research topic was chosen due to its relevancy to the wellbeing of consumers, who can be affected by marketers’ activities beyond what maybe initially considered. We examine the effects of

in-store communications beyond the usual sales, turnover and customer satisfaction metrics, and delve into measures of basket healthiness which have a direct impact on consumers’ health. We extend our deepest gratitude to the University of Groningen and Professor Koert Van Ittersum for

(5)

4

1. INTRODUCTION

1.1 - Background:

The Pandemic

Obesity in the past decade has been recognized as a global problem to tackle and declare war on, due to the serious threats it poses; so much so, that by 2005 the ‘war’ on obesity superseded the war on tobacco (Gilman 2008). Obesity poses a series of grave health risks including increased proneness to diabetes, high blood pressure, high cholesterol, asthma, arthritis, and poor health status (Mokdad et al. 2003). Furthermore, a study on overweight female adolescents and young adults in the USA showed that they were more likely to complete fewer years of school, have lower household incomes and were less likely to be married. A similar study on men produced similar results with regards to their likelihood of marriage (Gortmaker 1993). The obesity pandemic is likely to continue; a study by Finkelstein et al. forecasts a 33% increase in obesity prevalence and a 130% increase in severe obesity prevalence by 2030 (Finklestein et al. 2012).

The Forces Behind It

(6)

5

a role in unhealthy food intake and subsequently obesity. Dhar and Baylis’ study on Quebec’s 1980 imposed fast-food advertising ban serves as testament to this, with the ban leading to an $88 million decrease in spending on fast food per year (Dhar and Baylis 2011). To summarize, an imbalance between food intake and physical activity is the main driver of obesity with marketing and consumer factors playing a key role in consumers’ food consumption decisions.

1.2 - The Problem

The Helpless Consumer & Resisting Temptations Our Minds Tell Us Not To Resist Misjudgements of healthiness -

Past research has displayed how even highly health conscious consumers who are willing to consume food in a healthy manner may find difficulty in making sound decisions. For example, consumers typically underrate the amount of calories contained in a meal they view as healthy (such as a meal from Subway) and overrate the caloric content in meals they perceive as unhealthy

(7)

6

Tricks of the mind

Finkelstein and Fishbach also show that forcing oneself to eat a product branded as ‘healthy’ induces feelings of hunger stronger than would have been had the meal been skipped (Finkelstein and Fishbach 2010). To add to the psychological difficulty of making goal consistent choices, Khan and Dhar introduced findings about an interesting concept which they called the ‘Licensing Effect’. Their studies showed that when making a series of choices, consumers who initially made a decision that fulfils a satisfaction of their self-concepts sequentially felt warranted in choosing an indulgent option when making their next selection (Khan and Dhar 2006). The studies were conducted in the context of luxury goods as the indulgent choice, however the same logic follows for healthy vs indulgent foods. If a consumer decides to purchase a healthy food option at first, this may reinforce his/her self-concept and make them feel like they contributed to their goal of living healthy. When coming to the next choice, a more indulgent decision is warranted, or more namely ‘licensed’. This

phenomenon highlights a major hardship that consumers face in exerting self-control and eating healthily, for if you reward yourself with a chocolate bar for every salad you eat, how much progress have you really made?

The Powerful Marketers, Retailers and Their Tools Tricky Products and Dodgy Shortcuts -

(8)

7

they will not extend enough cognitive effort to evaluate a food product’s health benefits objectively and instead rely on certain stimuli to form their assessments – a notion supported by Boyson et al.’s findings which were mentioned in the previous section. A simple inclusion of health symbolism such as depictions of sporty activities, athletic figures or vegetables on the packaging of a product can affect customers’ perceptions of the products healthiness even though packaging in no way plays a part in the product’s nutritional value. Irmak, Vallen and Robinson’s experiment on the effect of branding the same dish with a different name on taste ratings and purchase preferences supports this argument. They found that calling a pasta salad ‘pasta’ which holds relatively unhealthy

semantics, leads to dieters perceiving it as less healthy and tasty than do non dieters. This sheds light on dieters’ reliance on heuristic cues to make health evaluations of food products (Irmak, Vallen and Robinson 2011).

How Important Are Retailers When It Comes To Eating Healthily?

Retailers can play a major role and have a significant impact on the consumption habits of their customers. A study by Glanz, Bader and Iyer published in the American Journal of Preventive

Medicine found a rather straightforward link between healthier eating and availability, affordability, prominence, and promotion of healthful foods in retailers (Glanz, Bader and Iyer 2012). A more surprising finding came about by studying the correlation between Walmart’s proliferation and obesity rates and found that for every supercentre that opened per 100,000 residents, obesity rates rose by 2.3% (Courtemanche and Carden 2011). Rose et al. similarly found a positive relationship between shelf space allotted to energy dense products and BMI. Though this doesn’t constitute a causal relationship, it is certainly eye opening to the impact that retailers could potentially have. In the next section some retailing tools that have an impact on consumer decision making are

(9)

8

Role of Shelves

(10)

9

1.3 - Solution – An experiment to unravel the interplay between shelving positions,

retailer in store communications and healthiness of baskets

Given the major role retail settings play on influencing consumer well-being and decision making, this study delves into uncovering the interplay between the choice setting, retailer communications and consumer decisions in the context of healthiness of products purchased. An experiment involving the hypothetical purchase of snack foods after exposure to retailer communications supporting either healthy choices or sinful indulgences is carried out and its results analysed. The conceptual model is built on consumer goal activation and its effect on subsequent choice. A detailing of experiment design and results analysis follows in the subsequent sections.

1.4 - Novelty of the study – Why this relationship has not been examined before.

Research on the healthiness of consumer choice is scarce and studies of this in retail settings

accounting for retailer interference and assortment display have not yet been delved into. Hopefully, the unveiling of the results of this research will spur further interest in the topic and lead to further investigation on how retailers can influence consumers’ choices when it comes to judging,

purchasing and consuming healthy products.

1.5 - Contribution:

The study is expected to contribute with managerial and ethical implications as well as a basis for future research.

Managerial Implications:

(11)

10

of reverse licensing driven by guilt and incidental pride is investigated, shedding light on a possible brighter side of promoting unhealthy foods. Retailers could utilize these insights to understand consumer sequential choice and make more enlightened decisions when it comes to advertising the food products they have on offer.

Ethical Implications:

Displaying how in store communications can lead to less/more healthful consumption by consumers may be of great importance to retailers who emphasize consumer well-being. Knowing the potential ramifications of advertising a certain product within a store may lead retailers to sound and socially aware decisions. Also, consumers’ knowledge of influences they are susceptible to in a retail environment may help them make better goal consistent choices when shopping.

2. THEORETICAL FRAMEWORK

2.1 - Conceptual Model

Figure 2.1: Conceptual Model

Goal activation through priming with healthy product Vicarious licensing through consideration of goal consistent choice More Indulgent Choices Goal activation through priming with unhealthy product Guilt at succumbing to temptation, leading to better choices to curb guilt/ Incidental pride from goal consistent choices, encouraging

similar choice.

(12)

11

2.2 - Theoretical Background

On a grocery shopping trip, the typical consumer is faced with a multitude of decisions and exposed to unanticipated stimuli that trigger different consumer goals. Even shoppers who plan in advance are susceptible to temptation and impulse and will have to either resist or succumb to indulgent options. This section delves into the psychological forces that are at play in such decision situations.

After Considering Healthy Options:

Licensing Effect

Kahn and Dhar’s ‘Licensing effect’, which was previously discussed, factors in heavily in product choice sequences. The following example displays how this concept pans out in a food retail setting. A typical licensing effect would occur when considering either skimmed or full fat milk. When the healthier option is purchased to stay in line with an activated goal – in this case eating healthily – consumers become more likely to indulge cometh the next decision. So if a consumer who purchased skimmed milk had to decide between low fat or full fat cheese, it is likely that they will feel warranted to reward themselves with the more indulgent option due to their previous goal consistent choice (Khan and Dhar 2006).

Vicarious Licensing Effect

Spurred by Kahn and Dhar’s findings on licensing’s effects on subsequent choices, Wilcox et al. demonstrated that simply thinking about a self-concept/goal reinforcement choice leads to vicarious fulfilment of that goal. This means that consumers need only consider a healthy option for a

(13)

12

they acted in line with health goals and subsequently warrant less healthy choices upon deciding on the next product to purchase. It also means that increasing the variety of healthy options on offer may lead to even unhealthier consumption.

After considering unhealthy options:

Guilt effect

Vice foods are those that are associated with enticing instant pleasures but with possible long term adverse effects (Chen and Sengupta 2014). When consumers consider purchasing a vice food, a conflicting goal of eating healthily is likely to be activated. This follows from Fishbach, Friedman, and Kruglanski’s finding that an exposure to a temptation to stray from a goal may activate the goal itself (Fishbach, Friedman, and Kruglanski 2003). Should vices be succumbed to, it is likely that the

consumer will experience feelings of guilt (Chen and Sengupta 2014).

Incidental Pride Effect

(14)

13

2.3 - Experimental Design

To study the interplay between the above psychological effects and retail settings, an online survey based experiment was carried out. Respondents were randomly assigned to one of three conditions and asked to choose 100 snacks to purchase from a shelf with 15 brands of snacks varying in

healthiness. The control condition simply asked participants to make their purchase decisions. The second condition subjected the participants to an advertisement endorsing a healthy brand before being exposed to the shelf, with the third condition doing the same but subjecting respondents to an advertisement for an unhealthy brand. In all conditions where participants were primed with an advertisement, the advertised product was placed in the centre of the shelf to make use of central gaze cascade effect to supplement the priming.

2.4 – Hypotheses

Priming with a healthy brand:

When primed with a healthy brand communication right before encountering the snacks shelf, it is likely that goals related to healthy eating and food regulation will be activated. Also since the advertisement compels you to consider a healthy brand, it is likely that vicarious licensing will take place. Given, the onset of the vicarious licensing effect it is likely that the subsequent choices will be inconsistent with the goal licensed by consideration of the healthy product.

H1: Priming with a healthy brand will result in an unhealthier basket relative to the non-prime condition.

Priming with an unhealthy brand:

(15)

14

consistent behaviour will be experienced, aiding in making more goal consistent decisions and leading to purchasing healthier items.

H2: Priming with an unhealthy brand will result in a healthier basket relative to the non-prime condition.

3 – METHODOLOGY

To assess the hypotheses and differences between the four mentioned study conditions, a cross sectional online survey design was implemented and carried out on a sample of 68 participants. In this chapter, details of the sampling method, data collection, instrumentation, descriptive statistics pertaining to conditions and limitations of the methodology are divulged.

3.1 Sampling:

(16)

15

practicality constraints and the limited number of respondents at hand, the possibility of running a power analysis to determine an ideal sample size is foregone and the recommended minimum of 20 valid respondents per condition is accepted.

3.2 - Instrumentation

Measurement of Independent Variables:

The independent variable under consideration in this study is the type of communication a shopper is exposed to. The categorical variable named “Condition” has values ranging from -1, 0 and 1; 0 symbolizing the control condition with no exposure to communications with -1 & 1 symbolizing exposure to an unhealthy brand advertisement and a healthy brand advertisement respectively.

The choice sets in each condition:

In each condition the respondents were asked to choose 100 snacks from a set of 15 snack brands which were the same across conditions albeit their shelf positions. The chosen brands are detailed in table 1 in the appendix along with the caloric and nutritional information presented with them. A depiction of shelf arrangement for each condition is included in the appendix. The shelf display consisted of 3 rows of 5 products each in an attempt to emulate as closely as possible a shopping rack at a supermarket. The positions of all the brands except for the middle three in the centre row were randomized per respondent with three different randomizations that they may be exposed to. The purpose of this is to discard the shelf positioning of brands as a confounding variable and to utilize the central gaze halo effect in line with the advertisements that respondents were exposed to. Conditions 2 and 3 involved an advertisement for a healthy and unhealthy brand respectively. The brand advertised then subsequently was positioned in the very centre of the shelf and

(17)

16

each brand, caloric, sugar, saturated fat and recommended daily value content were displayed in boxes in order to guide participants’ decision making.

Condition 2 – exposure to an advertisement for an unhealthy brand – Inducing guilt and pride:

To investigate the effects of priming with an unhealthy brand, respondents in condition 2 were presented with an advertisement announcing the availability of Double Decker chocolate bars. Double Decker was chosen since it contains the highest amount of calories and sugar per serving as well as a high saturated fat content. The advertisement as well as the instructions given to the respondentsare depicted in the appendix. The advertisement created for this study is informative in nature and depicts the Double Decker as a guilty pleasure which is now available. After exposure to the ad, the respondents were asked to move on to the shelf display to make their selections.

Condition 3 - exposure to an advertisement for a healthy brand – Inducing vicarious licensing:

To investigate the effects of priming with a healthy brand, respondents in condition 3 were

presented with an advertisement announcing the availability of Kind Almond and Apricot bars. The advertisement contained the message ‘healthy, fruity, nutty…tastes good feels better’ to highlight the benefits of indulging in the focal snack. The focal brand was chosen based on these nutritional benefits and reasonable caloric contents. After exposure to the ad, the respondents were asked to move on to the shelf display to make their selections.

Measuring Dependent Variables:

The dependent variable investigated in this study is the healthiness of consumers’ baskets. Many measurement types are possible for this, promising interesting insights; all of which will be investigated to uncover potential relationships between the different communications that

participants are exposed to and their effects on subsequent choice. Table 3 in the appendix explains the different measures (Total calories, sugar, saturated fat, healthy product count and total

(18)

17

to account for consumers making decisions and healthiness judgements based on visual cues and previously held brand beliefs. A supplementary survey composed of a sample of 20 respondents asked participants to rate each brands healthiness from our selection on a one item likert scale. Mean scores of the supplementary study are reported in table 3 of the appendix. The items classified as healthy to compute the ‘healthy items picked’ measure were based on scores of 4 and higher on perceived healthiness.

Measuring Possible Moderators:

B.M.I: The survey asked respondents to report their weight and height as accurately as possible in order to allow for an investigation of body mass index as a possible covariant on the effects of different conditions on the dependent variable. Body mass index was then calculated for all respondents using the formula B.M.I = weight (kg)/ Height (m) ^2. B.M.I across conditions did not vary significantly with an F value of 0.17 and a significance level of 0.912. To attain more analysis flexibility, all B.M.I scores above 25 were set to 1 and all observations below that were set to 0 to symbolize high and low B.M.I.

Health Consciousness:

To study possible interaction effects of health consciousness on the relationship between the experiment conditions and the dependent variable, an extant scale for health consciousness was used in the study. Michaelidou & Hassan’s six item scale was included as part of the survey and respondents were asked to respond to the items after selecting brands for the basket (Michaelidou & Hassan 2008). Health Consciousness was observed only after choices of brands were made. This was done so as not to confound the dependent variable due to respondents catching on to the purpose of the study, which may have caused them to respond in a manner they perceive as desirable to the researchers. Differences in health consciousness across conditions were not significant with an F value of 2.228 and a significance level of 0.091. To verify the internal

(19)

18

was found. Factor analysis was also conducted with the items set to load on to one factor due to the scale’s previous application in research. The factor scores were subsequently saved and used for analysis. To enhance flexibility, another health consciousness scale was created setting all positive values of health consciousness (which range from -3 to +3) to 1 and all negative values to zero. Setting health consciousness values above our median score to 1 and below the score to zero produced the same results, thus signifying an even distribution between high and low health consciousness in our sample. This was done to facilitate subsequent running of analysis of variance tests.

Demographics:

Age and gender observations were also collected from respondents. Statistical differences in both were analysed across conditions without any significant finding. Age displayed an F value of 2.686 at a 0.052 significance level. The difference is close to significance due to condition 3 exhibiting lower age means than 1 and 2; 24 vs 30 and 32 respectively. This finding is not expected to negatively affect the analysis of this study due to all ages of participants (ranging from 15 to 70) being

appropriate. Gender differences across groups were also insignificant with an F value of 0.286 and a significance level of 0.835.

3.3 Analysis:

(20)

19

4. RESULTS

This chapter displays the analyses’ results. A general description of the collected survey data and respondents is presented. The results of the MANCOVA which was run to determine whether we can support H1 and H2 is presented. The MANCOVA is then analysed to uncover any interaction effects between health consciousness and the experimental conditions. Pairwise comparisons offer a closer look at simple main effects of health consciousness within conditions on the dependent variables, with significant and relevant non-significant findings reported. The chapter is then concluded with a comprehensive summary of findings.

4.1 - Descriptive Statistics

68 participants whose responses were deemed valid, make up our sample. As described in the methodology chapter, the survey was shared on facebook and respondents were randomly assigned to either the control, unhealthy advert prime and healthy advert prime conditions. Relevant

demographics and characteristics pertaining to the respondents are presented in table 4.1. The majority of respondents (60.3%) were between 25 to 50 years of age with a nearly even split between male and female respondents (48.5% / 51.5%). Most accepted respondents fell in the unhealthy prime condition (39.7 %) with the remaining rather evenly split between the control and healthy prime conditions (29.4% / 30.9 %). A perfect split is exhibited with 34 health conscious respondents and 34 non-health conscious respondents.

Table 4.1 – General Descriptive Statistics

N %

Age Lower than 25 22 32.4

25-50 41 60.3

Greater than 50 5 7.4

Gender Male 33 48.5

Female 35 51.5

Respondents Per Condition Unhealthy Prime 27 39.7

Control 20 29.4

(21)

20

N % Health Consciousness Health Conscious 34 50.0

Non Health Conscious 34 50.0

BMI <25 47 69.1

>25 21 30.9

Furthermore, the means of the dependent variable measures (healthiness of basket) across all sampling conditions are presented with their standard deviations in table 4.2.

Table 4.2 – Means of DVs Across Conditions

Dependent Variable Scale Mean Standard Deviation

Total Calories 9000 – 25000 _19485.82 _1649.99

Total Sugar 750 – 3000 _1461.37 _292.73

Total Saturated Fat 0 – 800 _296.22 _124.90

Total RDV 450 – 1250 _971.77 _82.76

Total Perceived Healthiness 100 -700 _372.47 _89.28

Total Healthy Items Picked 0 – 100 _40.25 _27.52

The scale column in table 4.2 displays the lowest and highest values respondents could have scored on each measure of the dependent variable by picking 100 items from the snack selection.

Subsequently, the means of each dependent variable measure within each experimental condition are displayed in table 4.3.

Table 4.3 – Means of DVs Within Conditions

(22)

21

DV N Mean Std. Deviation

Total Sat. Fat

Unhealthy Prime 27 327.08 139.74 Control 20 268.10 91.36 Healthy Prime 21 282.48 129.31 Total RDV Unhealthy Prime 27 977.52 83.22 Control 20 960.64 81.13 Healthy Prime 21 974.98 86.67 Total Perceived Healthiness Unhealthy Prime 27 360.38 88.72 Control 20 378.33 73.6 Healthy Prime 21 382.45 104.82

Total Healthy Items

Unhealthy Prime 27 35.93 27.53

Control 20 41.75 23.40

Healthy Prime 21 44.38 31.39

4.2 Testing the Hypotheses:

(23)

22

Table 4.4- MANCOVA – IV: Experimental Condition; COV: Gender, B.M.I; MOD: Health Conscious-ness; DV: Healthiness of basket measures

Overall Model Test Value F Hypothesis df Error df Significance IV - Condition Pillai’s trace 0.18 0.91 12 112 0.54 n.s. Wilks’ lambda 0.83 0.90 12 110 0.53 n.s. IV – Health Consciousness Pillai’s trace 0.23 2.70 6 55 0.02 * Wilks’ lambda 0.77 2.70 6 55 0.02 * Interaction: HC*Condition Pillai’s trace 0.19 0.95 12 112 0.50 n.s. Wilks’ lambda 0.82 0.94 12 110 0.51 n.s. COV – Gender Pillai’s trace 0.20 2.23 6 55 0.05 ** Wilks’ lambda 0.82 2.23 6 55 0.05 ** COV – B.M.I Pillai’s trace 0.10 1.00 6 55 0.43 n.s. Wilks’ lambda 0.90 1.00 6 55 0.43 n.s.

4.4 - Between Subjects Effects:

Source DV DF F Significance

Condition Total Calories 2 0.44 0.65 n.s.

Total Sugar 2 1.72 0.19 n.s.

Total Sat. Fat 2 2.32 0.11 n.s.

Total RDV 2 0.42 0.66 n.s.

Total Perceived Healthiness

(24)

23

Condition Healthy Items Picked

2 1.02 0.37 n.s.

HC Total Calories 1 2.31 0.13 n.s.

Total Sugar 1 7.58 0.01 *

Total Sat. Fat 1 11.26 0.00 *

Total RDV 1 2.25 0.14 n.s. Total Perceived Healthiness 1 14.55 0.00 * Healthy Items Picked 1 12.43 0.00 *

Condition * HC Total Calories 2 1.59 0.21 n.s.

Total RDV 2 1.57 0.22 n.s. Total Perceived Healthiness 2 1.33 0.27 n.s. Healthy Items Picked 2 2.05 0.14 n.s.

Gender Total Calories 1 4.54 0.04 *

Total Sat. Fat 1 4.95 0.03 *

Total RDV 1 4.58 0.04 * Total Perceived Healthiness 1 8.46 0.01 * Healthy Items Picked 1 11.36 0.00 *

B.M.I Total Calories 1 2.25 0.14 n.s.

(25)

24

B.M.I Total Perceived Healthiness

1 3.05 0.09 **

Healthy Items Picked

1 3.87 0.05 **

n.s.: not significant; *: p<0.05; **: p<0.1; HC: Health Consciousness

Pillai’s trace and Wilks’ lambda show that the analysis is overall insignificant for experimental conditions (p>0.10). Furthermore the experimental conditions show no significant effect on any measure of the dependent variable. As such we conclude that there is no statistically significant difference in means across conditions. Nevertheless, figures analysing means of a selection of the dependent variables versus experimental conditions are presented below (figure 4.5). As expected, gender was found as a significant covariant with Pillai’s trace and Wilks’ lambda with p values of 0.05. B.M.I Pillai’s trace and Wilks’ lambda show no significant effects for the overall MANCOVA, yet between subject tests show significant covariance of B.M.I with Total Sugar, Total Perceived

Healthiness and Healthy Items Picked with p values of 0.04, 0.09 and 0.05 respectively. Interaction effects and main effects of health consciousness are discussed in subsequent sections of this chapter.

(26)

25

Figure 4.5 – Experimental Condition VS Means of Dependent Variables

Testing Hypothesis 1:

It was hypothesised that the experimental group that was exposed to an advertisement for a healthy brand would have less healthy baskets on average than the control group that was not exposed to any prime. For the hypothesis to be supported, significance is needed for a main effect between the conditions variable and any one of the dependent variable measures with that dependent variable exhibiting a higher mean in the healthy prime condition than in the control. However, the MANCOVA results displayed in table 4.4 show no significant relationship between experimental conditions and any of the basket healthiness measures. Though means of all objective (thus excluding perceived healthiness and healthy items picked) nutrition measures of baskets are less healthy in the healthy prime condition than in the control, hypothesis 1 is not supported.

Testing Hypothesis 2:

(27)

26

support was initially found for hypothesis 2. In search for further evidence, a replica of the

MANCOVA test presented earlier was run on observations including only the unhealthy prime and control conditions. The results are presented in table 4.6.

Table 4.6- MANCOVA – IV: Experimental Condition; COV: Gender, B.M.I; MOD: Health Conscious-ness; DV: Healthiness of basket measures – Control & Unhealthy prime conditions only.

(28)

27

Table 4.6 - Between Subjects Effects:

Condition Total Calories 1 0.75 0.39 n.s.

Total Sugar 1 3.68 0.06 **

Total Sat. Fat 1 4.60 0.04 *

Total RDV 1 0.76 0.39 n.s. Total Perceived Healthiness 1 1.57 0.22 n.s. Healthy Items Picked 1 1.20 0.28 n.s. HC Total Calories 1 5.61 0.02 * Total Sugar 1 9.99 0.00 *

Total Sat. Fat 1 10.02 0.00 *

Total RDV 1 5.30 0.03 * Total Perceived Healthiness 1 12.40 0.00 * Healthy Items Picked 1 9.81 0.00 *

Condition * HC Total Calories 1 0.05 0.82 n.s.

Total RDV 1 0.02 0.88 n.s. Total Perceived Healthiness 1 2.02 0.16 n.s. Healthy Items Picked 1 3.72 0.06 **

Gender Total Calories 1 4.55 0.04 *

Total Sugar 1 2.90 0.10 **

Total RDV 1 4.19 0.05 *

Total Perceived Healthiness

(29)

28

Gender Healthy Items

Picked

1 3.77 0.06 **

B.M.I Total Calories 1 1.65 0.21 n.s.

Total RDV 1 1.56 0.22 n.s. Total Perceived Healthiness 1 4.09 0.05 * Healthy Items Picked 1 3.01 0.09 **

n.s.: not significant; *: p<0.05; **: p<0.1; HC: Health Consciousness

(30)

29

Table 4.7 – Planned Means Comparison – Unhealthy Prime VS Control Condition

Unhealthy Prime VS Control - DV Mean Difference: UH - Control Std. Error Sig

Sugar 143.84 75.02 0.06 **

Saturated fat 71.90 33.52 0.04 *

4.3 - Testing For Moderation:

To find out whether any interaction effects exist between health consciousness and the dependent variable measures, we take a look at the results of the MANCOVA summarized in table 4.4. Pillai’s trace and Wilks’ lambda for health consciousness showed significance (p=0.02 & p=0.02) leading to the conclusion that the model is significant for main effects of health consciousness on the

dependent variables. This can be seen in the between subjects effects test with health consciousness exhibiting a significant main effect on sugar (p=0.01), saturated fat (p=0.00), perceived healthiness (p=0.00) and healthy items picked (p=0.00). A visualization of this relationship with the mentioned dependent variables is presented in figure 4.8. It is obvious from the graphs that health conscious participants constructed significantly less healthful baskets on than non-health conscious

respondents. This surprising result will be discussed in the next chapter of the report.

(31)

30

Figure 4.8 – Health Consciousness VS Sugar; Saturated Fat; Perceived Healthiness; Healthy Items Picked

Interaction effects between experimental conditions and health consciousness were not shown to be significant with p>0.1 for dependent variable measures. As such, we discard the possibility of health consciousness moderating the experimental conditions’ effects on measures of basket healthiness. However, plots of conditions versus the dependent variable means across health consciousness levels interestingly were non-parallel (figure 4.9). This usually indicates interactions, and in this case led to suspicion that there may be simple main effects between health

(32)

31

Figure 4.9 – Conditions VS Sugar, Saturated Fat, Perceived Healthiness and Healthy Items Picked for different levels of Health Consciousness

(33)

32

baskets constructed by low health consciousness participants. Similar significant simple main effects of health consciousness on sugar content, saturated fat content and perceived healthiness scores of baskets in the control condition were found. However these were less pronounced and significant than health consciousness’ effect on dependent variables in the unhealthy prime condition hinting towards the advertisement for Double Decker augmenting health consciousness’ negative influence on snack choice. Health consciousness had no significant simple main effects on the dependent variables in the healthy prime conditions. This surprising finding is elaborated on in the subsequent chapter.

Table 4.10 - Pairwise comparisons – simple main effects of health consciousness within conditions: DV: Sugar

Condition (I) (J) Mean

differ-ence (I-J) Std. Error Significance Unhealthy Prime: High health consciousness Low health consciousness 296.34 99.61 0.00 *

Control: High health consciousness Low health consciousness 219.08 119.12 0.07 ** Healthy Prime: High health consciousness Low health consciousness 15.43 111.82 0.89 n.s.

Table 4.10 - Pairwise comparisons – simple main effects of health consciousness within conditions: DV: Saturated Fat

differ-ence (I-J) Std. Error Significance Unhealthy Prime: High health consciousness Low health consciousness 135.21 43.60 0.00 *

Control: High health consciousness Low health consciousness 104.31 52.14 0.05 * Healthy Prime: High health consciousness Low health consciousness 43.71 48.94 0.38 n.s.

(34)

33

Table 4.10 - Pairwise comparisons – simple main effects of health consciousness within conditions: DV: Perceived Healthiness

differ-ence (I-J) Std. Error Significance Unhealthy Prime: High health consciousness Low health consciousness -113.91 29.75 0.00 *

Control: High health consciousness Low health consciousness -59.20 35.58 0.10 ** Healthy Prime: High health consciousness Low health consciousness -46.60 33.40 0.17 n.s.

Table 4.10 - Pairwise comparisons – simple main effects of health consciousness within conditions: DV: Healthy Items Picked

differ-ence (I-J) Std. Error Significance Unhealthy Prime: High health consciousness Low health consciousness -35.57 8.88 0.00 *

Control: High health consciousness Low health consciousness -11.57 10.63 0.28 n.s. Healthy Prime: High health consciousness Low health consciousness -13.51 9.97 0.18 n.s.

n.s.: not significant; *: significant at the 0.05 level

4.4 - Summary of Findings:

 No support was found for hypothesis 1.

 No support was found for hypothesis 2 with some opposing evidence.  No support was found for health consciousness as a moderator.

 A significant main effect on saturated fat was found for health consciousness.

 A significant main effect on perceived healthiness was found for health consciousness.  A significant main effect on sugar was found for health consciousness.

 A significant main effect on healthy items picked was found for health consciousness.  Health conscious participants constructed baskets that were lower on perceived healthiness

(less healthy) and healthy items picked than low health consciousness participants.

(35)

34

 A significant simple main effect on sugar, saturated fat, healthy items picked and perceived healthiness was found for health consciousness in the unhealthy prime condition, but not in the healthy prime condition.

 Health conscious participants exposed to the unhealthy Double Decker chocolate advertisement constructed baskets lower in perceived health/healthy items picked and higher in saturated fat/sugar than low health consciousness participants.

5. DISCUSSION

5.1 - H1: Impact of a healthy product advertisement in a retail setting on snack choice:

(36)

35

unhealthy selection, implying that the effect would only be significant on a very limited number of choices as opposed to the 100 selections that they were asked to make.

5.2 - H2: Impact of an unhealthy product advertisement in a retail setting on snack

choice:

It was predicted that participants in this condition would on average select baskets that were healthier than those selected by the control group. The lack of significance in the MANCOVA (Table 4.4) does not provide support for the hypothesis. Furthermore, the results of the MANCOVA carried out with observations comparing only the unhealthy prime condition and control (summarized in table 4.5) offer opposing evidence; as such H2 is rejected. Incidental pride from temptation resistance and guilt effects from succumbing to temptations which were supposed to encourage healthier choices and deter from unhealthy selections respectively may not have played an

important enough part to lead to overall healthier baskets. Studying the interaction between health consciousness and the unhealthy product advertisement condition through pairwise comparisons yielded surprising results. Analysis of simple main effects in the healthy advertisement condition did not show any effect of the advertisement on health conscious participants. However, health

(37)

self-36

regulation availability leads to higher impulse buying probability which stems from lower temptation resistance (Vohs and Fabers 2007). It is safe to assume that health conscious consumers are more worried about resisting unhealthy foods than those with lower health consciousness. In hindsight, the advertisement for double decker bars (the least healthy choice in the experimental set) would have prompted more resistance initially from participants who are health conscious, thus leading to resource depletion which is not exhibited in non-health conscious subjects. The less healthy baskets constructed by the health conscious respondents is likely attributed to their drastic resource

depletion which carried on over their choices. Thus, though it was predicted that incidental pride and guilt would lead to better decision making, this was not the case since different effects might have come in to play with the health conscious respondents. An analysis of the effect of health consciousness on basket healthiness across all conditions also shows that health conscious respondents on average selected less healthy items and constructed baskets that were lower on perceived healthiness and higher on sugar and saturated fats as is visually depicted in figure 4.7. This surprising finding leads to the conclusion that health conscious consumers are more susceptible to over indulging in the snacks category.

5.3 - Implications:

Results of this study indicate that advertising or highlighting a healthy product next to a snack shelf did not have adverse effects on healthiness of choice. Retail managers should not worry about inducing harmful consumer choices through advertising of healthy products in their stores even though some theory hints at possible adverse effects of this. This finding may not hold for single decisions, however this research indicates that bulk decisions possibly hedge against vicarious licensing effects of advertisements for healthy products.

(38)

37

may be useful to policy makers to take into consideration when imposing regulations on media that targets such consumer segments. These results should also be of concern to retailers that target healthy lifestyle consumers, though it is unlikely in the first place for them to allow such features and advertisements of unhealthy products. The finding also raises questions regarding what elements of the adverts contribute to health conscious consumers’ subsequent poor decision making. For

instance, if an advertisement for a healthy brand positions a snack as a ‘reward’ would it have similar effects as one for an unhealthy brand that claims the same?

5.4 - Limitations and Future research

Limitations:

(39)

38

Future Research:

A replication of this study in a non-online setting may lead to more generalizable and useful insights. A controlled lab experiment where participants would get take their baskets back home to consume them for instance would possibly result in stronger guilt feelings from the choice of vice foods and more pride from resisting them. Participants could be asked to choose 10 items instead of 100 to increase the budgetary feasibility of the research. A field experiment in an actual supermarket would even allow for studying the effects of different communication on baskets constructed from multiple product categories as well. It would be interesting and significant to test the effect of actual

environments and to see whether the findings of these results hold as it would add more credibility to the conclusions made.

Another interesting topic that can be drawn from this research and studied in subsequent experiments is the effect of communication elements and the products being communicated on basket healthiness. For instance, testing which of the advertising elements or the product being advertised has the stronger effect on health conscious consumers’ less healthful choices, can have serious implications on branding and communication ethics. It is suggested to test whether healthy brands communicated as ‘treats’ would have similar effects to unhealthy products being positioned similarly. Given the explanation of health conscious consumers constructing less healthy baskets - after expending more energy resisting the unhealthy advertised products- than non-health conscious consumers, it would be interesting to see what happens when an unhealthy product is falsely

portrayed as harmless. If consumers are lead to lower resistance to the product and less self-regulatory energy is expended, their baskets overall could be healthier. This would pose ethical implications on whether it is really harmful to brand unhealthy products in a way that makes them appear healthier than they really are. Extant research on health branding mostly deals with

(40)

39

on healthiness of overall baskets. Thus the proposed research will possibly produce novel, unintuitive and important findings.

5.5 - Conclusion:

(41)

40

6. APPENDIX:

REFERENCES:

1. Anker, T. B., Sandøe, P., Kamin, T., & Kappel, K. (2011). Health branding ethics. Journal of business ethics, 104(1), 33-45.

2. Atalay, A. S., Bodur, H. O., & Rasolofoarison, D. (2012). Shining in the center: Central gaze cascade effect on product choice. Journal of Consumer Research, 39(4), 848-866.

3. Bauer, H. H., Heinrich, D., & Schäfer, D. B. (2013). The effects of organic labels on global, local, and private brands: More hype than substance?. Journal of Business Research, 66(8), 1035-1043.

4. Briesch, R. A., Chintagunta, P. K., & Fox, E. J. (2009). How does assortment affect grocery store choice?. Journal of Marketing Research, 46(2), 176-189.

5. Cacioppo, J. T., & Petty, R. E. (1983). Central and peripheral routes to persuasion: Application to advertising. Advertising and consumer psychology, 3-23.

6. Chandon, P., & Wansink, B. (2007). The biasing health halos of fast-food restaurant health claims: lower calorie estimates and higher side-dish consumption intentions. Journal of Consumer Research, 34(3), 301-314.

7. Chen, F., & Sengupta, J. (2014). Forced to Be Bad: The Positive Impact of Low-Autonomy Vice Consumption on Consumer Vitality. Journal of Consumer Research, 41(4), 1089-1107.

8. Courtemanche, C., & Carden, A. (2011). Supersizing supercenters? The impact of Walmart Supercenters on body mass index and obesity. Journal of Urban Economics, 69(2), 165-181. 9. Dhar, T., & Baylis, K. (2011). Fast-food consumption and the ban on advertising targeting

children: the Quebec experience. Journal of Marketing Research, 48(5), 799-813.

(42)

41

11. Finkelstein, S. R., & Fishbach, A. (2010). When healthy food makes you hungry. Journal of Consumer Research, 37(3), 357-367.

12. Fishbach, A., Friedman, R. S., & Kruglanski, A. W. (2003). Leading us not into temptation: Momentary allurements elicit overriding goal activation. Journal of personality and social psychology, 84(2), 296.

13. Fredrickson, B. L. (2001). The role of positive emotions in positive psychology: The broaden-and-build theory of positive emotions. American psychologist, 56(3), 218.

14. Gilman, S. L. (2008). Fat: A cultural history of obesity. Polity.

15. Glanz, K., Bader, M. D., & Iyer, S. (2012). Retail grocery store marketing strategies and obesity: an integrative review. American Journal of Preventive Medicine, 42(5), 503-512. 16. Gorn, G. J., & Goldberg, M. E. (1982). Behavioral evidence of the effects of televised food

messages on children. Journal of Consumer Research, 200-205.

17. Gortmaker, S. L., Must, A., Perrin, J. M., Sobol, A. M., & Dietz, W. H. (1993). Social and economic consequences of overweight in adolescence and young adulthood. New England journal of medicine, 329(14), 1008-1012.

18. Hofmann, W., Rauch, W., & Gawronski, B. (2007). And deplete us not into temptation: Automatic attitudes, dietary restraint, and self-regulatory resources as determinants of eating behavior. Journal of Experimental Social Psychology, 43(3), 497-504.

19. Irmak, C., Vallen, B., & Robinson, S. R. (2011). The impact of product name on dieters’ and nondieters’ food evaluations and consumption. Journal of Consumer Research, 38(2), 390-405.

20. Jayanti, R. K., & Burns, A. C. (1998). The antecedents of preventive health care behavior: An empirical study. Journal of the Academy of Marketing Science, 26(1), 6-15.

(43)

42

22. Khan, U., & Dhar, R. (2006). Licensing effect in consumer choice. Journal of Marketing Research, 43(2), 259-266.

23. Kozup, J. C., Creyer, E. H., & Burton, S. (2003). Making healthful food choices: the influence of health claims and nutrition information on consumers’ evaluations of packaged food products and restaurant menu items. Journal of Marketing, 67(2), 19-34.

24. Macdiarmid, J. I., Vail, A., Cade, J. E., & Blundell, J. E. (1998). The sugar-fat relationship revisited: differences in consumption between men and women of varying BMI. International journal of obesity, 22(11), 1053-1061.

25. Martinez, J. A. (2000). Body-weight regulation: causes of obesity. Proceedings of the Nutrition Society, 59(03), 337-345.

26. McFerran, B., Dahl, D. W., Fitzsimons, G. J., & Morales, A. C. (2010). I’ll have what she’s having: Effects of social influence and body type on the food choices of others. Journal of Consumer Research, 36(6), 915-929.

27. Michaelidou, N., & Hassan, L. M. (2008). The role of health consciousness, food safety concern and ethical identity on attitudes and intentions towards organic food. International Journal of Consumer Studies, 32(2), 163-170.

28. Mokdad, A. H., Ford, E. S., Bowman, B. A., Dietz, W. H., Vinicor, F., Bales, V. S., & Marks, J. S. (2003). Prevalence of obesity, diabetes, and obesity-related health risk factors, 2001. Jama, 289(1), 76-79.

29. Moorman, C., Diehl, K., Brinberg, D., & Kidwell, B. (2004). Subjective knowledge, search locations, and consumer choice. Journal of Consumer Research, 31(3), 673-680.

30. Mukhopadhyay, A., & Johar, G. V. (2007). Tempted or not? The effect of recent purchase history on responses to affective advertising. Journal of Consumer Research, 33(4), 445-453. 31. Muraven, M., Tice, D. M., & Baumeister, R. F. (1998). Self-control as a limited resource:

(44)

43

32. Seymour, J. D., Yaroch, A. L., Serdula, M., Blanck, H. M., & Khan, L. K. (2004). Impact of nutrition environmental interventions on point-of-purchase behavior in adults: a review. Preventive Medicine, 39, 108-136.

33. Vohs, K. D., & Faber, R. J. (2007). Spent resources: Self‐regulatory resource availability affects impulse buying. Journal of Consumer Research, 33(4), 537-547.

34. Wilcox, K., Kramer, T., & Sen, S. (2011). Indulgence or self-control: A dual process model of the effect of incidental pride on indulgent choice. Journal of Consumer Research, 38(1), 151-163.

35. Wilcox, K., Vallen, B., Block, L., & Fitzsimons, G. J. (2009). Vicarious goal fulfillment: When the mere presence of a healthy option leads to an ironically indulgent decision. Journal of Consumer Research, 36(3), 380-393.

(45)

44 Tables:

Choice Set And Nutritional Info – Table 1: ref. page 15

Brand Calories (Kcal) Sugar (g) Saturated Fat

(g)

RDV (% per 2000kcal/day)

Yorkie 249 26.9 8 12.0

Double Decker 250 30 5 12.5

Kinder Bueno 246 18 7 12.3

Chips Ahoy Cookies – 3 pieces 160 11 2.5 8.0

Fiber One 120 8 0 6.0

Crunchie 187 26 4.2 9.4

Nature Valley Honey ‘N Oats 190 12 0.5 9.5

Choco Liebniz 210 16.5 6.75 10.5

Digestives 210 7.5 0.75 10.5

Reese’s Buttercups 210 21 4.5 10.5

Bel Vita – 4 biscuits 200 9.6 1.9 9.5

Digestives Light 198 9 0.6 9.9

Oreo – 4 cookies 215 18.3 2.8 10.75

Rice Krispies Bar 90 8g 0.5 4.5

Kind – Almond & Apricot 180 13 3.5 9

DV Measures – Table 2: ref page 15

Total Calories Consists of the sum of caloric content of all items chosen per basket Sugar Content Consists of the sum of sugar content of all items chosen per basket Saturated Fat

Content

Consists of the sum of saturated fat content of all items chosen per basket

Total Healthy Products

Consists of a count of all healthy items chosen per basket

Total Perceived Healthiness

(46)

45

Table 3 – Perceived Healthiness Score of Brands – ref page 17

Brand Perceived Healthiness Score

Yorkie 2.3

Double Decker 2.3

Kinder Bueno 2.3

Chips Ahoy Cookies – 3 pieces 2.2

Fiber One 5.25

Crunchie 2.3

Nature Valley Honey ‘N Oats 5.35

Choco Liebniz 3.1

Digestives 4.9

Reese’s Buttercups 2.1

Bel Vita – 4 biscuits 5.55

Digestives Light 5.45

Oreo – 4 cookies 2.1

Rice Krispies Bar 4.45

(47)

46

Table 4 – Pairwise Comparison – Tests of Between Subjects Effects- 4 ANOVAs – DVs Sugar, Saturated Fat, Perceive Healthiness, Healthy Items Picked- IV Conditions, Health Consciousness (Mod) – COV Gender, B.M.I – ref. page 30

DV Source DF F Significance

Sugar BMI (COV) 1 4.48 0.04 *

Gender (COV) 1 7.76 0.01 * Condition (IV) 2 1.72 0.19 n.s. HC (IV) 1 7.58 0.01 * HC*Condition (MOD) 2 1.80 0.17 n.s.

Saturated Fat BMI (COV) 1 1.97 0.17 n.s.

(48)

47

Survey Images:

Instructions:

While doing your weekly grocery shopping in my-supermarket, you come across an advertisement poster placed right before a snack stand. You view the advertisement momentarily and continue to the snack stand to purchase 100 snacks that you plan to enjoy. Kindly scroll below the advertisement to view the snack stand.

While doing your weekly grocery shopping in my-supermarket, you come across a stand selling snacks. You decide to purchase 100 snacks that you plan to enjoy. Kindly observe the shelf and pick 100 snack items by indicating how many items you would like to order below. To do so, please fill out the number of snacks you would like to purchase in the small box next to the brand name. If you do not want to buy any snacks of a certain brand, please keep the number next to the brand name as 0.

(49)

48

Control condition shelf:

(50)

49

Unhealthy Prime Condition Shelf:

(51)

50

Healthy Prime Condition Shelf:

Perceived Health Measurement Survey Scale:

The more we care, the harder we snack