Consumers’ purchase decisions in a restaurant setting

(1)

Consumers’ purchase decisions in a restaurant setting

the effect of assortment organization, price discount, and prior attitude on purchases of non-branded organic and light products

(2)

Consumers’ purchase decisions in a restaurant setting

the effect of assortment organization, price discount, and prior attitude on purchases of non-branded organic and light products

Author: Faculty of Economics and business

Sanne Westdijk (1620142) Master thesis: Marketing Management

Prinses Marijkelaan 122 November 2010

7776 XH Slagharen Supervisors:

sanne1809@hotmail.com Dr. J.E.M. van Nierop

(3)

1

Management summary

This study builds on the prior work in the TransForum project. The main aim of this study is to investigate the use of assortment arrangement to stimulate consumers’ purchase of organic and light products in a real-life setting (the Restaurant of the Future). Furthermore, the potential moderating effects of price discounts will be examined. Moreover, the influence of the consumers’ prior attitude towards organic and light food products will be examined as well as the potential moderating effects of prior attitude.

Literature review revealed that arranging an assortment based on an organic/light attribute increases the salience of this attribute for consumers. This increased salience can stimulate the sales of organic/light products. The effects of assortment arrangement may also be stronger when the assortment arrangement is supported by price discount, because the discount further increases the salience of the organic/light attribute and the consumers may lessen the need for price comparison with the regular alternative. Moreover, the intention to purchase an organic/light food product is mostly determined by the consumers’ prior attitude towards these products. An assortment which is arranged on an organic/light attribute will enhance the relationship between prior attitude towards this organic/light attribute and sales of the product.

(4)

2

Although assortment organization does not have an influence on sales in this study, in other studies assortment organization did have an influence on sales (Van Herpen et al. 2005). This implies that self service restaurants should consider whether to stimulate sales by placing organic/light products apart. Self service restaurants should be aware of the fact that the distance between the organic/light products and the regular alternative should be large enough. Placing organic/light products apart increases the salience of the organic/light attribute and leads to fewer direct comparisons with the regular alternative. Increased salience stimulates sales, because consumers who are interested in and have a positive prior attitude towards organic and light food products are more likely to purchase these products when the assortment arrangement highlights them. Furthermore, self service restaurants should carefully consider if they want to offer price discounts because in this study price discount did not have an effect on sales.

(5)

3

Preface

This thesis is the final piece of work to finalize my master Business Administration with a specialization in Marketing Management. I had the opportunity to assist in a project named “TransForum” project. This project examines how retailers can stimulate the demand for sustainable products and services through the assortment decisions that they make. It was very interesting to get more insight in the way consumers make decisions and to find out how these decisions can be influenced.

I would like to take this opportunity to thank some people who made it possible for me to write this thesis and graduate from university. First of all I would like to thank Dr. L.M. Sloot for his feedback regarding the completion of my thesis. Furthermore, I want to show my gratitude to Dr. H.W.I. van Herpen for giving advice and feedback regarding my thesis. But most of all I would like to thank Dr. J.E.M. van Nierop for his feedback, insights, and advice regarding my thesis.

Last but not least, I would like to thank my parents and friends for their love and support during the years that I studied in Groningen.

Sanne Westdijk

(6)

4

1. Introduction

1.1 Background problem

The food consumption patterns of consumers are changing worldwide. Nowadays consumers are more concerned about their health, nutrition, environmental protection, and the quality of the food they eat. This resulted in a growth in organic food products as well as light food products . First we need to define what organic and light food products are . Green Earth Organics provides a clear definition of what an organic food products is: “organic foods are minimally processed to maintain the integrity of the food without artificial ingredients, preservatives or irradiation.” This definition focus mainly on the production process. Furthermore, food can be labelled as organic when the methods of production encourage environmental health and avoid destructive action which damage the environment. (Essoussi and Zahaf 2008). A light food product contains at least 30 percent less energy (kcal), fat or sugar than the regular alternative. What should be noticed is that in for example, light chips fat will be replaced by carbohydrates and proteins who also contain energy. If the human body does not need carbohydrates and proteins, the body will convert this in fat (voedingscentrum.nl).

(10)

8

worldwide consumer spending towards organic food grew from $46.1 billion in 2007 to $50.9 billion in 2008, which is an increase of 10.4% (Bio-Monitor 2009).

Nowadays, the low calorie food market is a multi-billion dollar industry because the production of low calorie food has expanded for the past decade (Krystallis et al. 2003). In the US, for example, the sales of low/reduced-calorie products has reached US$40 billion since the 1990s.

When looking at the sales channels in figure 1, it can be seen that all sales channels grew in 2008 in comparison with 2007, except for the category ‘rest of sales channels’. All sales channels grew in 2009 in comparison with 2008. The rest of sales channels consist of web shops, products which are sold directly at the farm, delivery services (e.g. Stichting Van Eigen Erf), and at bazaars especially for farmers. The growth in the supermarket channel is mainly due to the extension in the assortment.

Figure 1 Sales organic food per sales channel in the Netherlands (X million Euro)

(11)

9

which importance of location continues to decline and the Internet-based retailing is becoming more important (see Alba et al. 1997). Simonson (1999) however, suggest that these changes do not mean that assortment will become unimportant in the future. In fact, the arrangement of the assortment can satisfy customers’ wants as well as shaping those wants. In other words, arranging the assortment in the right manner can shape consumer preferences and affect the decision making process. Moreover, assortment organization can have a direct influence on the product comparisons that consumers make (Simonson 1999).

The marketing literature contains several papers writing about the effect of assortment organization on profit and sales levels. Drèze et al. (1994) have examined the effect of several display changes on sales levels. The outcome of this study revealed that a number of display changes decreased sales levels instead of increasing sales levels. Furthermore, Van Nierop et al. (2008) proposed a new method to measure the impact of shelf layout on sales and to optimize shelf arrangements. They found that the shelf layout has significant effects on baseline sales and marketing effectiveness. This means that when a retailer changes the assortment without a thorough understanding of the influence of assortment arrangement on the consumer decision-making process, there is a chance that their sales will drop.

Retailers have many options available which can be used to organize products. Two of the options will be discussed here. The first option is based on natural categorization (Rosch et al. 1976). This is the level at which consumers naturally organize items. For products this is the product type level (Sujan and Dekleva 1987). If retailers choose this type of assortment organization organic products will be placed next to regular alternatives (e.g. cauliflower, asparagus, carrots). The second option is an organization which is based on a less salient attribute, for example, organic nature or light nature. Retailers will place the organic products or light products on a separate shelf (e.g. all organic vegetables together or all light products together).

(12)

10

alternatives directly. This resulted in an increase in sales for organic products Another study revealed that when a relatively unknown organic brand is placed next to the same product with an A brand status, the participants favored the well-known A-brand over the relatively unknown organic brand (Razenberg, A. 2008).

However, in some categories, brand names do not apply. No research has been done in a real-life setting with a product category in which brand names do not apply, it is interesting to further examine the assortment organization within this field, which is the subject of this paper. Furthermore, most prior studies looked at organic food products. This paper will expand these results for light products because organic food products are good for the environment and light food products are good for yourself.

1.2 Problem statement

The situation described above leads to the following central problem statement:

“What are the effects of assortment arrangement and price differences on consumers’ purchase behavior of non-branded organic products and light products, when looking at a restaurant setting?”

1.3 Research questions

Addressing the following sub questions will support the answer to the central problem statement:

 What is the effect of different ways of assortment arrangement on the sales of the organic products and light products?

 What is the effect of a price discount for organic products and light products when taking into account the arranging of the assortment?

 What is the influence of prior attitudes towards organic food products /light food products on the sales of the organic products/light food products) when taking into account the arranging of the assortment?

(13)

11 1.4 Theoretical and practical relevance

1.4.1 Theoretical relevance

This study builds on the prior work in the TransForum project. The TransForum project examines how retailers can stimulate the demand for organic products and services through the assortment decisions they make. From an academic perspective, this research will contribute to a better knowledge about the use of product arrangement to stimulate consumers’ purchase of organic products in a real-life setting. The potential moderating effects of price discounts will also be examined. This research differs from prior research in two ways. First, in contrast to other studies where brand names often plays an important role, this study attempts to explain consumer purchasing behavior for an organic product category in which brand names do not apply. Second, prior results will be extended to the light product category. This will be examined in a real-life setting instead of a lab experiment, which has been usedin prior research.

1.4.2 Practical relevance

As mentioned before, understanding how assortment arrangement can be used to stimulate sales of organic and light products is very important. Effects of assortment arrangement on consumers’ evaluation of the assortment (Hoch et al. 1999) and on sales and profit levels (Drèze et al. 1994) can be substantial. This research will help self service restaurants with the placing of the assortment and the price which should be set.

1.5 structure of the thesis

(14)

12

2. Literature Framework

A consumer can be influenced to buy a product by several factors while making choices in the retail environment. One of these factors is environmental dimension (Bitner 1992). Environmental dimensions are ambient conditions (e.g. music tempo, temperature, odor), Space/function (e.g. layout, equipment), and signs, symbols and artifacts (e.g. style of décor, signage). Other factors which can influence the consumer decision making in a retail environment are the responses of the employees, other customers in the retail environment, assortment arrangement, and price. Another factor which determines whether or not to buy a product is the consumers’ prior attitude towards these products. This study will focus on three of these factors: assortment arrangement, price discount, and prior attitude towards organic and/or light products. It might be the case that these factors will not only influence the decision making of the consumer. It is also possible that these factors interact with each other. This means that the effect of one independent variable on the dependent variable depends on the value of the other independent variable.

Given that more and more consumers buy organic and light products, it is interesting to find out how these factors influence the purchase decision process of the consumer.

In this chapter relevant information and findings from existing literature are discussed. Combing all this information will lead to a conceptual model.

2.1 Assortment arrangement

The arrangement of the assortment can influence the consumers’ evaluation of the assortment, which in turn can affect consumers’ purchase intention and attitudes towards organic and light products.

2.1.1 Assortment evaluation

(15)

13

consumer to find their product which can cause frustration or dissatisfaction. This in turn can lead to a delay or forgo in purchase (Huffman and Kahn 1998). In addition, operational costs go up when the retailer enlarges the number of stock keeping units (SKU’s). However, retailers hesitate to reduce the number of SKU’s, since there is a change that the consumer will go to another store when products are not available which can have detrimental effect on sales (Borle et al. 2005).

A possible solution to this dilemma is given by Kahn and Wansink (2004). They show that the consumers’ perception of variety are not necessarily dependent on the actual number of options present. This means that the consumption can be influenced when the actual variety is unchanged. Kahn and Wansink (2004) manipulated the organization of the assortment through arranging M&M’s and jelly beans in a disorganized and organized manner. The outcome of the study revealed that increasing the actual variety of the assortment has less impact on consumption when the variety is increased in a disorganized manner than if the variety is increased in an organized manner which is appreciated by the consumer. In other words, increased sales of a particular product can be partly due to factors other than the actual variety. This is in line with Mogilner et al. (2008) who conclude that categorization of the assortment enables the consumer to obtain value from large assortments. By using categories retailers can highlight the existing variety in the assortment.

2.1.2 Retailers’ external organization of the assortment

(16)

14

The external organization of the assortment is an important aspect for retailers. It is also important for self service restaurants, which have numerous food counters where customers can put together a meal of their choosing.

2.1.3 Consumers’ internal organization of the assortment

The assortment which is organized externally by the retailer is not the only important factor for determining the perceived variety of the assortment by the consumer. It is also important to know the consumers’ internal organization structure. The internal organization structure is defined as the way consumers categorize items in their heads (Morales et al. 2005). Morales et al. (2005) examined how the internal organization structure of the consumer can influence their perceived assortment variety in a store or on a website. The outcome of this research shows that consumers’ perceptions of variety and satisfaction are dependent upon how the assortment is organized, both internally by the consumer and externally by the retailer. It is important to match the organization structure which is provided by the retailer with the consumer’s internal categorization structure for product categories which are familiar for the consumer. In contrast, congruency between the consumers’ structure in mind and the external store layout leads to lower perceptions of variety, however, increased satisfaction with the store’s assortment for unfamiliar product categories (Morales et al. 2005).

(17)

15

Consumers have limited capacity or ability to process information (Hoyer and Macinnis 2010). This means that consumers have to be selective when they shop for products. Simonson (1999) state that consumers do not use prior knowledge and do not evaluate all options which are available in the same category. Instead, they tend to focus on the set of options which are provided by the retailer. The retailer can therefore influence the information processing effort (Simonson 1999; Bettman and Kakkar 1977). Two main reasons for this will be discussed here. First, consumers would like to minimize the effort to process information. Using the structure that is provided by the retailer is the easiest way of processing information. Second, when a retailer presents his products in a certain manner, some aspects become more or less salient and more or less important for the final decision (Simonson 1999).

2.1.4 Assortment arrangement and self-regulation

Assortment arrangement can also be a factor that enhances self-regulation. Vohs and Faber (2007) defined self-regulation as the ability to overcome one’s first impulses in order to strive for a more abstract, higher order goal. It is thus important for a person to keep the desired end state in mind. Self-regulation seems to be extremely difficult in the area of eating and dieting behavior (Papies et al. 2008). The eating behavior of a dieter is strongly influenced by the attractiveness of the food in their environment. When the assortment is arranged a dieter has more self-regulatory resources available which makes it easier for them to choose light products (Papies et al. 2008).

Self-regulation is less important in case of choosing for organic products. A consumers will buy organic products when they are “environmental minded” because organic products are seen as good for the environment (Pieniak et al. 2010). Assortment arrangement will help these consumers in making their choice for organic food products.

2.1.5 Attribute salience

(18)

16

changed the assortment based on the production region of the wine. Consumers generally do not base their purchase decision on this attribute. The outcome of this study revealed that organizing wines into region categories increased the importance of this attribute and consequently influenced the purchase likelihoods by regions. Sales of wines from preferred regions increased whereas sales of wines from disliked regions decreased. In other words, when the assortment organization is based on a specific attribute, consumers will compare products on this specific attribute and the purchase likelihood of the product which is preferred on this specific attribute will increase. This specific attribute will become more important in the decision making process of consumers (Areni et al. 1999). Furthermore, the purchase likelihood is influenced more when the salience of the attribute is normally low rather than high and when brands have normally high rather than low purchase likelihoods (Areni et al. 1999). This means that when the assortment will be arranged on the basis of organic nature/light nature, organic products/light products will get more attention, and will also stimulate the comparison of alternatives based on organic nature/light nature.

One potential drawback of the increased salience of this attribute is the decreased salience of other attributes, like a decreased salience of price. There is a chance that consumers choose fewer options based on the prices of the products when the assortment arrangement is based on organic nature/light nature. However, it could also be that due to the clustering of organic/ light products the choice process of the consumers is simplified. Consumers can ignore half of the assortment (the organic products or the light products), and focus on the other half of the assortment when looking at other product attributes. In that case, consumers can make a better choice based on the price of the product. The influence of price will be discussed in more detail in paragraph 2.3.

2.1.6 Physical separation between products

(19)

17

products in the assortment when all organic/light products are placed together. When the organic/light products are placed next to the non-organic/non-light alternative, a consumer has to search for every organic/light products. Because of this, the variety of the organic/light assortment looks smaller.

In the beginning of the Transforum project, separating organic products from the regular alternative had little to no effect (van Herpen 2009) because all products were visible on one computer screen. Participants could easily compare alternatives across groups and did not have to focus on a subgroup. In a later computer-based experiment, the subgroups were placed on separate websites. This time, the manipulation of assortment organization had effect. This is in line with an experiment in a real-life store that was conducted earlier by Van Herpen et al. (2005). The outcome of this study revealed that consumers made fewer direct comparison between the organic products and the regular alternatives. Consumers choose more often for organic products when these were separated from the regular alternative. It is harder for consumers to compare these products on attributes like price and quality when organic products are placed separately in an assortment. Consumers will therefore make their choice based on the attribute organic versus non-organic products. It seems intuitive that consumers will choose more often for light products when these products are separated from the regular alternative based on the same reasoning as described above for organic products.

As a consequence, the following hypotheses are stated:

H1a: Arranging an assortment based on an organic attribute stimulates the sales of products with a positive value on this attribute (i.e., the organic products).

H1b: Arranging an assortment based on a light attribute stimulates the sales of products with a positive value on this attribute (i.e., light products).

2.2 Price discount

(20)

18

regular alternative (Verhoef 2005). It is likely that consumers want to compare prices between products from the same category before making the final purchase decision.

The relationship between assortment arrangement and sales may be moderated by the effect of price discounts. It could be that when the assortment arrangement is supported by price discounts the effect may be stronger. There are two reasons behind this. First, the salience of the organic/light attribute is further increased by these discounts. Second, when a discount is given on the organic/light product, consumers do not feel the urgent need to compare the prices with the regular alternative.

During the decision making process a consumer can make direct alternative comparisons and comparisons based on separate evaluations which is called indirect alternative comparisons. (Nowlis and Simonson 1997). With direct alternative comparisons, consumers compare products which are placed next to each other on attributes which are easily and precisely comparable for example the price. On the other hand, enriched attributes which are harder to compare such as brand name/equity, are more important when a consumer makes indirect alternative comparisons. A brand name on its owns says nothing about the product. A consumer has to know what the brand stand for. Each product will be evaluated separately in case of enriched attributes.

Price differences will become more visible when organic or light products are placed next to the regular alternative when a consumer makes direct alternative comparisons.

(21)

19

This is in line with another study done by Van Nierop et al. (2008). They showed that sorting the assortment on price has a negative effect on market share, because it becomes to visible that organic products in general have higher prices than regular products. Furthermore, when the organic product is more expensive than the market leader, a negative effect on market share is expected. Van Nierop et al. (2008) observed also that when an organic product is on discount, this will have a positive effect on the market share of the organic product.

Thus, the following hypotheses are stated:

H2a: A price discount for organic products will have a positive effect on the sales of organic products.

H2b: A price discount for light products will have a positive effect on the sales of light products.

H3a: Price discounts for organic products have a larger sales effect when the arranging of an assortment is based on an organic attribute.

H3b: Price discounts for light products have a larger sales effect when the arranging of an assortment is based on a light attribute.

2.3 Prior attitudes

People have attitudes regarding politics, clothes, food, and almost everything else. The term attitude is defined as an overall evaluation that expresses how much a person likes or dislikes an object, issue, person, or action (Hoyer and Macinnis 2010). Attitude strength matters, because strong attitudes are resistant to change, persistent over time and likely to predict behavior, while weak attitudes are less likely to affect behavior (Holland et al. 2002). Therefore, a company should try to fit its products into the formed attitudes of the consumer rather than trying to change the attitudes.

(22)

20

itself is not enough for predicting his/her behavior. It is also important to know how someone feels about buying a product. (Ajzen & Fishbein 1980).

2.3.1 The influence of prior attitudes on behavior

The relationship between attitudes and behavior has been researched extensively in social psychology, marketing and consumer research. Organic food products are being perceived by the consumer as more healthy, better for the environment, more nutritious and better tasting than non-organic food products (Pieniak et al. 2010). Consumers’ attitudes towards food and nutrition are important factors in influencing the choice to buy sustainable products and organic products in particular Pieniak et al. (2010) found a positive relationship between the attitude towards organic vegetables and the consumption of organic vegetables. Moreover, Chen (2007) found that the intention to purchase organic food products is mostly determined by the consumers’ prior attitude towards these products. In addition, according to the theory of planned behavior (Ajzen 1991), when a consumer’s attitude towards organic food is positive, the consumer is more likely to engage in that purchase behavior. Findings from all of these studies are in line with other studies about the relationship between attitude and behavior (de Magistris and Gracia 2008; Aertsens et al. 2009; Honkanen et al. 2006).

Light products are being perceived by the consumer as healthy but not pleasure giving. There is often a conflict between health and taste for light products (Bower et al. 2003). For example, consumers from the United Kingdom were most concerned about the reduction in taste quality when the product is light (Zandstra et al. 2001).

(23)

21

To summarize, based on those papers, it can be concluded that the decision-making process for organic and light food products is mostly determined by the consumers’ prior attitude towards organic and light food products. Thus the following hypotheses are proposed:

H4a: When a consumer has a positive attitude towards organic food products they will have a higher intention to purchase organic food products.

H4b: When a consumer has a positive attitude towards light food products they will have a higher intention to purchase light food products.

2.3.2 The influence of assortment organization on the relationship between prior attitude and purchasing behavior

(24)

22

making a choice. Furthermore, a lot of participants immediately noticed “Appelsientje” and did not even see the organic alternative. This could have influenced the consumers’ choice.

Based on that reasoning the following hypotheses are proposed:

H5a: Arranging an assortment based on an organic attribute will enhance the relationship between prior attitude towards this attribute (i.e. the organic products) and sales of the organic products.

H5b:Arranging an assortment based on a light attribute will enhance the relationship between prior attitude towards the light attribute (i.e. the light products) and sales of the light products.

2.4 Control variables

Several control variables are included in this study. These control variables are: neophobia, age, and gender.

A variable that has an impact on choice behavior is neophobia (Chen 2007). Food neophobia is defined as the extent to which consumers are reluctant to try novel foods (e.g. food products, dishes, cuisines). Consumers with higher food neophobia personality traits are more reluctant to try organic foods because they may not be familiar with organic foods (Chen 2007).

(25)

23

about issues concerning the world. The second group is Generation X. This group is born between 1965 and 1977, has experienced large-scale social, economic, and environmental changes during their lifetimes. The last group are the Baby Boomers. Members of this groups were born during the years 1946 and 1964. What is typical for this generation, is the greater concern for the environment. Barber et al. (2010) have done research relating this topic. They used wine as the product category. The outcome of this study revealed that there are differences between the three generations. The Millennial generation is the generation who has the most respondents with the strongest environmental attitude, followed equally by Baby Boomers and Generation X.

Another factor which has an influence on consumer attitudes is gender. Several studies show that women are on average more concerned about the environment than men because women are socialized towards caring for others and nurturance. These attitudes will carryover towards social issues as the environment (Blocker and Eckberg, 1989; Henderson, 1996) However, not every study support this (Palan 2001). Barber et al. (2010) concluded that females have stronger environmental attitudes than males.

2.5 Conceptual model

(26)

24 H1b + H1a + H3a + Assortment organization based on organic attribute* Price discount Sales of organic/light products Prior attitude H4a b + Assortment organization based on light attribute*

H3b +

H2a b +

* attribute is indicated by labels

H5a + H5b +

(27)

25

3. Research design

In this chapter how the actual research is performed and how hypotheses are tested and evaluated are described. Also, the questionnaire design and method of data collection will be discussed.

3.1 Research method

This experiment will be performed in a real life setting, namely in a restaurant named “The Restaurant of the Future”. This restaurant is a field laboratory where information can be collected about for example the influence of the variety in assortment on the choice behavior of the consumer. There are several advantages of a field experiment. First, the setting is more similar to a real shopping situation. Second, shoppers are in their natural environment. Finally, there is less evaluation apprehension. The disadvantages of a field experiment are less experiment control and there could be intervening variables (e.g. consumer characteristics). The advantages outweigh the disadvantages for this study. As already mentioned, prior lab experiments found that assortment organization is influential. Therefore, the organization in this study will be manipulated in a real life setting to find out if assortment organization is also influential in this type of setting.

(28)

26

Table 1 3X3 experimental design

Price discount

Arrangement No discounts Biologic discount Light discount

Mixed Week 1 Week 2 Week 3

Biologic separate Week 4 Week 5 Week 6

Light separate Week 7 Week 8 Week 9

3.1.1 Participants

The Restaurant of the Future is open for lunch to all employees of the Wageningen University. Consequently, all participants in this study are employees of the Wageningen University. They should always bring their personalized restaurant card with them, because they have to pay with the card. Otherwise the researchers are not able to track the behavior of the participants. Employees can bring a guest with them on the condition that they bring a signed consent form or complete one that is available in the restaurant.

3.1.2 Procedure

The experiment took place from March 8th till May 7th 2010. Data were collected from Monday until Friday, during the lunch. On three days no data were collected, because the restaurant was closed due to public holidays. These days were the fifth of April (Easter Monday), April the 30th (Queen’s day), and the fifth of May (Liberation day).

The product category that was chosen for this experiment is sandwiches. Sandwiches were selected because it is an organic product category in which brand names do not apply. As already noted, a prior experiment of the TransForum project concluded that strong brands may overwhelm effects of assortment organization. There are four types of sandwiches: regular sandwiches (A), regular-light sandwiches (B), organic sandwiches (C), and organic-light sandwiches (D). All sandwiches will be labeled according to their type. For every type 20 different sandwiches were made (1-20). See appendix section A.1 for the 80 different sandwiches. All sandwiches have similar prices except for the discount weeks.

Assortment organization is one of the manipulated variables. In this experiment the assortment is arranged in three different ways:

(29)

27

2) The assortment will be arranged on the attribute organic versus non-organic. Organic sandwiches will be placed separately on a table.

3) The assortment will be arranged on the attribute light versus normal. Light sandwiches will be placed separately on a table.

Every arrangement lasted three weeks. The sandwiches were spread over two tables which stood next to each other. There was an open space in-between the two tables, so that the participants were not able to see all the products at one glance. On every table were in total 12 sandwiches. See figure 3 for an example of a table.

C12 B4 A8 D2

Figure 3 Example of a table organization

The other manipulated variable is the price of the product. In the second and third week of each period, a discount was be given to respectively organic and light sandwiches.

Since participants could interpret discounts wrong, which can lead to wrong impression a small pre-test among 15 respondents (mainly Wageningen University and Research Centre employees) was conducted. Three photos of three different sandwiches (regular, organic, and light) with corresponding label were shown. After that, four different discount possibilities were discussed:

1) “Price: €3.30 20% off! Now €2.65” 2) “Action price: From €3.30 to €2.65” 3) “Now for an introductory price! €2.65”

4) “Only during the biological week: biological sandwiches for €2.65”

(30)

28

This is even more likely when the discount is noticed at the end of the day or week. Especially in case of the first option, respondents relate to the discounts offered in supermarkets. A product which is discounted in a supermarket is often near its expiration date. It can be concluded that the first two discount possibilities lead to an impression of lower quality.

The third option will only be credible for the first week, because of the fact that in the option is stated: “Now for an introductory price!”.

The last option does not give the impression of a low quality of the sandwich, because the discount will be given the whole week. A discount of 20% is experienced as reasonable or fair. Some respondents who participated in the pre-test suggested to include the normal price when the last option will be chosen.

The fourth option will be applied during the experiment. See appendix section A.2 for an example of a product label with “biological week” discounted.

3.1.3 Questionnaire design

(31)

29

on a seven point scale from “totally disagree” to “totally agree” developed by Michaelidou and Hassan (2008) and Dutta-Bergman (2005). Because these questions have already been tested, the validity is ok.

There is a tendency that respondents will automatically answer towards the end of the scale. A possible solution for this problem is phrasing questions positively and negatively. However, there is a chance that respondents do not see that some questions are positively and negatively framed which lead to a wrong circle. Therefore all questions in this questionnaire are formulated the same way. Some parts of the questionnaire contain questions about light as well as organic products. Half of the respondents have to answer the questions about light products first and the other half of the respondents have to answer the questions about organic products first to avoid that the respondents get confused and to prevent question order effects. Research on question order demonstrates that the context and order of questions can influence subsequent responses (DeMoranville and Bienstock 2003). Preceding questions perform an activation function for subsequent questions.

After the questionnaire was designed it was pilot tested. This pretesting was done to improve the questionnaire and to identify if the questionnaire was clear and understandable. After the pretesting several questions were deleted and others were formulated in a different way. For a complete overview of the final questionnaire used, see appendix section B.1.

3.1.4 Data collection

Data has been collected about which participants bought a sandwich and when he/she bought a sandwich. The data is collected with the help of the personalized restaurant passes of the employees. The employees have to pay with his/her card every time he/she buys something. This is stored in the database of the cash register.

In a prior survey data has been collected about customer characteristics. These characteristics are: age, gender, length, weight, BMI, job description, education, and neophobia score.

(32)

30

the email). The advantage of an online questionnaire is that it is convenient for the respondents because they can fill in the questionnaire at any time. However, the response rate for an online questionnaire is low. Because of that, five mp3 players will be raffled among all respondents to stimulate them to fill in the questionnaire. In total, 187 employees filled in the questionnaire. Of those 187 questionnaires 24 were incomplete. Thus the total sample size is 163.

3.2 Plan of analysis

To test the hypotheses the data program SPSS 17 was used. Several statistic tests will be conducted in order to test if the hypotheses will be rejected or not. Unless otherwise stated, the p-value should be smaller or equal to 0.05 (significance level), in order to report a difference.

Factor analysis will be used in order to create the prior attitude score. Cronbach’s alpha will be determined to assure the internal consistency reliability of the variables. The Cronbach’s alpha ranges from 0 to 1, with values of 0.60 to 0.70 deemed the lower limit of acceptability. (Malhotra 2007). To test if a factor analysis is appropriate the KMO measure of sampling adequacy has to be >0.5. Sampling adequacy predicts if data are likely to factor well, based on correlation and partial correlation. The Bartlett’s test of sphericity tests if the variables are uncorrelated. The principal components analysis is used as a method to extract the factors. Three criteria are used to select the number of factors: the eigenvalue >1, total explained variance >60%, and those factors that explain >5% each (Malhotra 2007).

Multicollinearity diagnostic statistics will also be used to see if there is any intercorrelations among all variables. Finally logistic regression will be used to examine whether or not assortment organization, price discount, and prior attitude have an effect on the choice made by the respondents for organic and light sandwiches. This test was chosen as method of analysis because the dependent variable is a nonmetric, dichotomous (binary) variable.(Hair J.F. 2010). When the respondent did buy a sandwich it is coded as “1”. When the respondent did not buy a sandwich it is coded as “0”.

(33)

31

4. Results

In this section issues of representativeness of the sample, validity, and reliability are addressed. After that, the relevant results are described and interpreted.

4.1 Generalizability, reliability and validity

Generalizability refers to the degree to which a study based on a sample applies to a universe of generalization (Malhotra 2007). This study is merely representative for the Dutch WUR employees because Dutch WUR employees filled in the survey and only data about their buying behavior in the restaurant is recorded. It is expected that employees from every department visited the Restaurant of the Future during the experiment, because the restaurant is situated on the campus. Therefore, it is representative for the WUR employee population.

Reliability is the extent to which the observed variable measures the “true” value and is “error free”. Thus a scale has to produce consistent results if repeated measurements are made on the characteristic (Hair et al 2010). Coefficient alpha (Cronbach’s alpha) will be used to measure the internal consistency reliability. As mentioned before in paragraph 3.2, multicollinearity diagnostic statistics will be performed to check if there are high intercorrelations among independent variables which indicate multicollinearity.

Finally, validity has to be considered. According to Hair et al. (2010) validity is the degree to which a measure accurately represents what it supposed to. Most questions in this study are obtained from other scales used in previous research which increases the validity of this research.

4.2. Demographic Data

(34)

32

warm meals) 790 times during the experiment. See table 2 for the purchase shares per category. The customer spending per transaction ranged from €0.18 till €12.48. Customers spent on average € 3.53 per transaction with a standard deviation of €1.85. See figure 4 for a histogram of the customer spending per transaction.

Table 2 purchase shares per category

Category Number of sandwiches sold Percentage of total sandwiches Percentage of total transactions Regular sandwiches 40 18,10 _3,96 Light sandwiches 62 28,10 _6,13 Organic sandwiches 58 26,20 _5,74

Organic + light sandwiches 61 27,60 _6,03

Total 221 100 _21.86

Figure 4 Customer spending per transaction

(35)

33

€2.37 and they spent a maximum of €12.48. The respondents who did buy a sandwich spent more per transaction than the non buyers. The reason for the higher average spending per transaction for this group is because the cheapest sandwich cost €1.55, which is higher than the cheapest product (€0.18) for non buyers.

Table 3 provides an overview of the mean age, BMI, and neophobia score for the respondents divided in all respondents, respondents who did buy a sandwich and respondents who did not buy a sandwich.

Table 3 Means of respondents

Age BMI Neophobia

Score

All respondents 43.71 24.07 25.06

Respondents did buy sandwich 41,29 24.81 23.18

Respondents did not buy sandwich 44.39 23.86 25.59

To test if there is a significant difference between the means in table 3, independent sample t-test are performed. The sandwich buyers are on average younger (M=41.29, SD=9.70) than the group who did not buy a sandwich (M=44.38, SD=9.32). The difference is significant (t(1009) = 4.335, p=0.000). The mean BMI for sandwich buyers is higher (M=24.81, SD=2.83) than the mean BMI for non buyers (M=23.86, SD=2.67). The difference is significant (t(1009) = -4.606, p=0.000). Finally neophobia was tested with an independent sample t-test. The mean neophobia score for sandwich buyers is lower (M=23.18, SD=6.68) than the mean neophobia score for non sandwich buyers (M=25.59, SD=6.13). The difference in means is significant (t(1009) = 5.064, p=0.000). The detailed output is presented in appendix C1.

Of all respondents 63.3 % is male and 36.7 % is female. From the respondents who did not buy a sandwich 63.4 % is male and 36.6 % is female. These percentages are 62.9 and 37.1 respectively for all the respondents who did buy a sandwich.

(36)

34

Science Group (PSG), Agrotechnology and Food Sciences Group (AFSG), Environmental Science Group (ESG), and WUR others (administration, FB, and staff). Figure 5, shows that the group “SSG” never bought a product in the Restaurant of the Future during the experiment. Furthermore, the groups “AFSG” has the most respondents who did buy something in the restaurant of the future.

Education is divided in four levels, namely voortgezet onderwijs, beroepsopleiding, universitair, and gepromoveerd. Figure 6 shows that education level ‘Universitair’ has the most respondents who did buy something in the restaurant of the Future.

A chi-square statistic is used to test the statistical significance of the observed association in a cross-tabulation. The relationship between buying a sandwich and profession level is significant (2= 18.751, p=0.000). The Cramer’s V is 0.136 which means that profession level has a small relationship to buying a sandwich. Moreover, education level is also significantly related to buying a sandwich or not 2=13.191, p=0.004). The Cramer’s V has a value of 0.114, which means that education level has a small effect on chance of buying a sandwich. See appendix C2 for the exact output.

Figure 5 Profession percentages ,0% 10,0% 20,0% 30,0% 40,0% 50,0% 60,0% Pe rc e n tage

Profession

Respondents who did not buy a sandwich

Respondent who did buy a sandwich

(37)

35

Figure 6 Education percentages

4.3 Prior attitude towards organic and light sandwiches

A factor analysis has been conducted twice to reduce the larger set of items into a smaller set of uncorrelated factors and to retrieve the underlying dimension of these items for prior attitude towards organic and light sandwiches.

4.3.1 Prior attitude towards organic sandwiches

Prior attitude towards organic sandwiches is measured in the questionnaire with seven items (See question 11 in appendix B.1). In order to test if the seven items measured in the questionnaire form the prior attitude towards organic sandwiches, a factor analysis has been used. Two factors were extracted. The item “rate your overall feeling about organic sandwiches in comparison with normal sandwiches, more expensive-less expensive”, is a factor on its own. Therefore, this item was excluded when the second factor analysis was conducted.

The KMO has an value of 0.841 and the Bartlett's Test of Sphericity is 0.000 which is significant (p<0.05). Based on the eigenvalue, one component was extracted, the second component is 0.780 which is <1. The total explained variance is >60% when two components are extracted. Five factors are extracted when looking at the % of variance per factor. One component was extracted, because the eigenvalue is greater than one when one component is extracted and the ultimate criterion for the "right" number of factors is the interpretability of the solution. See table 4 for the items. The output is presented in appendix C3.

,0% 10,0% 20,0% 30,0% 40,0% 50,0% Per ce n tage

Education

Respondents who did not buy a sandwich

Respondent who did buy a sandwich

(38)

36

Table 4 Items prior attitude towards organic sandwiches

In comparison with regular sandwiches, what are your overall feelings about organic sandwiches:

Loadings

Tastes worse - Tastes better 0.753

Unhealthier - Healthier 0.778

Worse quality - Better quality 0.869

More unattractive - More attractive 0.722

Worse for the environment - Better for the environment 0.666

I would not buy - I would buy 0.693

Cronbach’s alpha is used to measure the internal consistency reliability. The Cronbach’s alpha increased after deleting the item: “rate your overall feelings about organic sandwiches in comparison with normal sandwiches, more expensive-less expensive”. Table 5 shows that the Cronbach’s alpha is sufficient for prior attitude towards organic sandwiches.

Table 5 Cronbach’s alpha

Variable Cronbach’s alpha

Prior attitude organic sandwiches 0.824

4.3.2 Prior attitude towards light sandwiches

Prior attitude towards light sandwiches is measured with seven items in the questionnaire (see question 12 in appendix B.1). In order to test if the seven items measured in the questionnaire form the prior attitude towards light sandwiches, a factor analysis has been used. Two factors were extracted. After deleting the items: “In comparison with regular sandwiches, what are your overall feelings about light sandwiches: Worse for the environment-Better for the environment”, and “in comparison with regular sandwiches, what are your overall feelings about light sandwiches:-More expensive-Less expensive”, one factor was extracted.

(39)

37

component was extracted, based on the eigenvalue and the interpretability of the solution. See table 6 for the items. In appendix C.4 is the output presented.

Table 6 Items prior attitude towards light sandwiches

In comparison with regular sandwiches, what are your overall feelings about light sandwiches:

Loadings

Tastes worse - Tastes better 0.818

Unhealthier - Healthier 0.576

Worse quality - Better quality 0.802

More unattractive - More attractive 0.846

I would not buy - I would buy 0.713

Table 7 shows the Cronbach’s alpha for the variable prior attitude towards light sandwiches. The Cronbach’s alpha is sufficient for the variable.

Table 7 Cronbach’s alpha

Variable Cronbach’s alpha

Prior attitude light sandwiches 0.768

4.4 Results hypothesis

(40)

38

model. The method which is been used in this study to reduce multicollinearity is centering the variables, that is, setting their means to zero (Aiken and West, 1991). After centering the variables, the VIF values are below 2.5. Therefore, it can be concluded that there is no multicollinearity. The exact tolerance and VIF values can be found in appendix C5.

Four binary logistic regressions have been conducted with only data from sandwich buyers (N=221). In this case no variable was significant. Therefore data from sandwiches buyers as well as data from non sandwich buyers (N=1011) are taken into account in this study.

The Nagelkerke R Square for the five binary logistic regressions are presented in table 8. A Nagelkerke R Square of 0.221 means that the variables included in the regression predict in total 22.1 percent of the probability a customer will buy a sandwich.

Table 8 Nagelkerke R Square five categories (N=1011)

Category Nagelkerke R Square

All sandwiches 0.221

Regular sandwiches 0.111

Light sandwiches 0.133

Organic sandwiches 0.167

Organic + light sandwiches 0.103

A binary logistic regression has been conducted five times to test the hypotheses. The dependent variable for the five times were: sales sandwich yes/no, sales sandwich regular yes/no, sales sandwich light yes/no, sales sandwich organic yes/no, and sales sandwich organic + light yes/no. When a respondent bought a sandwich it has been coded as ‘1’ and when the respondent did not buy a sandwich it has been coded as ‘0’. See table 9 for the results. See also the output in appendix C6.

Based on the outcomes of the binary logistic regressions (N=1011), the hypotheses can be tested. Each hypothesis will be discussed separately.

4.4.1 Hypotheses 1a and 1b

(41)

39

hypothesis 1b states that arranging an assortment based on a light attribute stimulates the sales of products with a positive value on this attribute (i.e., light products). The results of the five binary logistic regressions show a p-value > 0.05. Therefore, it can be concluded that these results are not in line with hypotheses H1a and H1b. Therefore the hypotheses are rejected.

It was predicted that price discount for organic products would have a positive effect on the sales of organic products (h2a) and a price discount for light products will have a positive effect on the sales of light products (h2b). The results show a p-value > 0.05 for every type of sandwich. Therefore hypothesis 2a as well as hypothesis 2b are rejected.

Hypothesis 3a (3b) states that price discounts for organic (light) products have a larger sales effect when the arranging of an assortment is based on an organic (light) attribute. The p-value of all five binary logistic regressions are above 0.05. Therefore, hypotheses 3a and 3b are not supported.

It was predicted that when a consumer has a positive attitude towards organic food products the purchase intention of organic food products will be higher (H4a). The results of the binary logistic regressions show that prior attitude towards organic sandwiches has an significant effect on choice for the category organic sandwiches (B=0.538, p=0.011) as well as for the category organic + light sandwiches (B=0.527, p=0.004), regular sandwiches (B=0.436, p=0.053) and all sandwiches (B=0.564, p=0.000). It can be concluded that prior attitude towards organic sandwiches has a positive effect on choice for organic sandwiches. This means that if the consumers’ prior attitude towards organic sandwiches is more positive, the purchase intention of organic sandwiches will be higher. Therefore, hypothesis H4a is accepted.

(42)

40

effect on choice for all sandwiches (B=-0.207, p=0.018). Prior attitude towards light sandwiches has a negative effect on choice for a sandwich. This means that if a consumers’ prior attitude towards light sandwiches is more positive, the purchase intention of light sandwiches will be lower. The effect of prior attitude towards light sandwiches is not significant. Therefore, hypothesis H4b is rejected.

(43)

41

Table 9 results binary logistic regressions, unstandardized beta’s

Variable all

sandwiches

regular light organic organic + light

Intercept .424 -3.776 .754 -2.849 -2.268

Arrangement organic .107 -.038 -.031 .438 -.032

Arrangement Light .060 -.210 .022 -.110 .439

Price discount organic -.087 -.698 .096 -.465 .389

Price discount light .015 -.531 .014 .470 -.097

Prior attitude organic .564*** .436* .064 .538** .527***

Prior attitude light -.207** -.134 -.121 -.152 -.194

Age -.064*** -.040** -.056*** -.049*** -.042***

Gender (0=male, 1=female) .428** .810** .013 .871** -.144

BMI .195*** .158** .124** .191*** .125**

Profession -.619*** -.743*** -.355* -.601*** -.296

Education -.420*** -.138 -.634* -.367* -.078

Neophobia score -.080*** -.041 -.086*** -.060** -.045**

Price discount light * arrangement organic .614 -1.019 .513 .422 1.033 Price discount light * arrangement light .051 .718 -.139 -.695 .253 Price discount organic * arrangement light .128 1.447 -.656 .023 -.008 Price discount organic *arrangement organic .698 1.218 .048 1.313 .178 Prior attitude light * arrangement light .221 -.336 .717** -.414 0.494 Prior attitude organic * arrangement organic .110 -.256 .079 .276 -.032 ***p<0.01 **p<0.05 *p<0.10

4.5 Model fit

(44)

42

model is the model with the lowest AIC value. The AIC for each model (N=1011) is provided in table 10.

Tabel 10 AIC for each model

Model AIC

All sandwiches 942.729

Regular sandwiches 321.114

Light sandwiches 425.414

Organic sandwiches 399,261

Organic + light sandwiches 435,797

The model with regular sandwiches has the lowest AIC (321.114) followed by organic sandwiches (399.261). The model with the dependent variable “all sandwiches” has by far the highest AIC (942.729). It can thus be concluded that the model works the best for the dependent variable regular sandwiches followed by organic sandwiches. The high AIC value for all sandwiches is due to the fact that no difference is made between the four different categories of sandwiches which results in a lot of information loss when the model wants to predict if a respondents buys a sandwich yes or no.

4.6 Control variables

The control variables neophobia, age, gender, and BMI are also included in the binary logistic regression models.

As can be seen in table 9, neophobia is significant for all models (p<0.05), except for the model with the dependent variable: regular sandwich yes/no. The betas are negative, which means that the higher the neophobia score the lower the chance that a respondent will buy a light, organic, or organic + light sandwich. This is in line with findings in the literature (Chen 2007).

(45)

43

respondent will buy a sandwich. This conclusion correspond to findings in literature (see section 2.4).

Gender has a significant impact on the probability of buying all sandwiches, regular sandwiches, and organic sandwiches (p<0.05). The probability of buying an organic sandwich is higher when the respondent is female than male (B=0.871, p=0.011) This is in line with finding in the literature (see section 2.4).

(46)

44

5. Conclusion and recommendations

In this chapter the problem statement and corresponding research questions will be answered based on the findings of the research. In addition, recommendations for further research as well as limitation of this research will be given.

5.1 Conclusions

5.1.1 Conclusions hypotheses

This part of the conclusions will focus on the outcomes of the hypotheses. The first hypotheses (H1a, H1b) stated that arranging an assortment based on an organic/light attribute stimulates the sales of products with a positive value on this attribute (i.e., the organic/light products). This is based on prior research conducted by van Herpen et al. (2005). Their conclusion was that consumers choose more often for organic products when these products were separated from the regular alternative, because the consumer made fewer direct comparison between the organic product and the regular alternative. The hypotheses were not confirmed by the results of this research. A possible explanation for these results is that the distance between the products on the two tables was not large enough. It was still possible for the consumers to compare the organic sandwiches on one table with the regular alternatives on the other table.

Hypotheses 2a and 2b stated that a price discount for organic/light products will have a positive effect on the sales of organic/light products. The results do not correspond to findings in the literature (see section 2.2.1). A price discount for organic/light products does not have a significant effect on the sales of organic/light products. A possible explanation is that dummy variables are used when the data was analyzed. This means that a discount for organic sandwiches was codes as ‘1’ and no discount as ‘0’. It could be that although the organic sandwiches were on discount, an organic sandwich was still more expensive than another sandwich in the category regular sandwiches.

(47)

45

to make price comparisons with the regular alternative. Furthermore, van Razenberg (2008) found that consumers disliked the clustering of organic products when these had a premium price. This was not confirmed by the results of the current study. A possible explanation is that prices of organic/light products were equal to the regular alternative. Normally, prices of organic products are more expensive than the regular alternatives.

Hypothesis 4a stated that purchase intention of organic food products will be higher when the consumer has a positive prior attitude towards organic food products. This study is in line with prior studies about purchase intention of organic food products and prior attitude towards organic food products (Pieniak et al. 2010; Chen 2007). If a consumer has a positive prior attitude towards organic food products, the consumer will have a higher purchase intentions of organic food products. Hypothesis 4b stated that purchase intention of light food products will be higher when the consumer has a positive prior attitude towards light food products. This is based on articles from Solheim and Lawless (1995) an Aaron et al. (1994). This hypothesis was not confirmed by the results.

(48)

46

Figure 7 Conceptual model presenting the relationships found in the empirical research H1b + H1a + H3a + Assortment organization based on organic attribute* Price discount Sales of organic/light products Prior attitude Assortment organization based on light attribute*

H3b +

H2a b +

H5a + H5b +

H4b + H4a +

(49)

47 5.1.2 Overall conclusion

As already stated in the introduction, assortment arrangement can have a substantial effect on consumers’ evaluation of the assortment (Hoch, Bradlow, and Wansink 1999) and on sales and profit levels (Drèze, Hoch, and Purk 1994). This study was particularly interested in the effect of assortment arrangement on consumers’ purchase levels of organic and light products in a real-life setting. Furthermore we were interested in the effect of price discount. The effect of prior attitude towards organic and light products on sales of these products was also examined. Therefore, the following problem statement was formulated:

“What are the effects of assortment arrangement, prior attitude towards non-branded organic and light products and price differences on consumers’ purchase behavior of non-branded organic products and light products, when looking at a restaurant setting?”

The answer to the problem statement is as follows:

 Assortment arrangement has no effect on the consumers’ purchase behavior of non-branded organic products and light products.

 A price discount for non-branded organic and light products has no effect on consumers’ purchase behavior.

 The intention to purchase organic food products is higher when the consumer has a positive prior attitude towards these products. This is not the case for light food products.

5.2 Managerial implications

Consumers’ purchase decisions in a restaurant setting

Consumers’ purchase decisions in a restaurant setting

Consumers’ purchase decisions in a restaurant setting

Management summary

Preface

Table of contents

1. Introduction

2. Literature Framework

3. Research design

4. Results

Profession

Education

5. Conclusion and recommendations