The Power of Signs : How Recommendation Signs Affect Consumer Behavior at Supermarkets

Linköping University | Department of Management and Engineering Master Thesis in Business Administration, 30 credits | Business and Economics Programme Spring 2020 | ISRN-nummer: LIU-IEI-FIL-A--20/03358--SE

The Power of Signs

How Recommendation Signs Affect Consumer

Behavior at Supermarkets

Clara Mirbashiri

Linnea Möller

Supervisor: Hugo Guyader

Acknowledgement

We would like to send a huge thank you to all individuals who helped us in the process of writing the thesis. Thank you to all the opponents and the individuals in our seminar group for your constructive criticism and valuable insights. We would especially like to thank the manager Johan at the supermarket the experiment took place in. Without his help this experiment would not have been able to be performed. We would also like to express our gratitude to our supervisor Hugo Guyader for his support during the process but also his valuable suggestions which tremendously improved this thesis.

______________________ ______________________

Clara Mirbashiri Linnea Möller

Abstract

Background/Problematization: Many purchase decisions are made in-store and one way

of influencing the decision making process is through signs. Signs are said to draw attention to the product on which they are placed and result in increased sales. Sale signs are widely used to influence purchase decisions, but are there other types of signs that can impact the purchase decision? Recommendations have also shown to affect consumer’s purchase decisions. However, most of the research within the subject have been done on the power of social influences from friends and family, e.g. strong-tie relationships. How recommendations from experts (e.g. weak-tie relationships) can influence purchases is less researched, especially which type of recommendation sign have the most effect.

Purpose: The purpose is to examine and map how different types of signs on shelves

related to staff recommendations for specific products affect their sales.

Method: The scientific approach of the study is deductive and the corresponding research

method is a quantitative study with an experimental design and a descriptive focus. An experiment was executed in a supermarket on 28 products to test stimuli with different levels of information concerning staff recommendations with the use of signs.

Result/Conclusion: The result showed only one stimuli to have significantly increased

sales. The stimuli consisted of adding a name of a specific staff member to the

recommendation sign. The increase in sales might however have been caused by other factors than the applied sign as the outbreak of Covid-19 resulted in changed consumer behavior at the time of the experiment. The results showed that no significant differences in sales could be seen according to the number of alternatives each product had. This might however be a result of the tampered data caused by the changed consumer behavior.

Keywords: expert recommendations, influence, retail, signs, authority, sales, buying

Table of Contents

1. Introduction ... 1 1.1 Background ... 1 1.1.1 In-store Decision Process ... 1 1.1.2 Sale Signs ... 2 1.1.3 Source of Information ... 3 1.1.4 Recommendations ... 3 1.1.5 Expertise ... 4 1.1.6 Authority ... 4 1.2 Problematization ... 5 1.2.1 Purpose and Research Questions ... 7 2. Conceptual Framework ... 8 2.1 Buying Decision Process ... 8 2.1.1 Choice Complexity in the Buying Decision Process ... 10 2.2 Hypothesis Development ... 13 3. Methodology ... 18 3.1 Research Philosophy, Strategy and Approach ... 18 3.2 The Design of the Experiment ... 19 3.2.1 The Supermarket ... 19 3.2.2 Product Selection Process ... 20 3.2.3 Design ... 22 3.2.4 Placement ... 26 3.3 Limitations ... 27 3.4 Literature Review ... 27 3.5 Analytical Strategy ... 28 3.5.1 Reference Week of Sales ... 28 3.5.2 Calculations in SPSS Statistics ... 28 3.6 Quality Discussion ... 31 3.7 Ethical Considerations ... 33 4. Results ... 35 4.1 Descriptive Statistics ... 35 4.2 Assumption and Reliability Check ... 38 4.2.1 Outliers ... 38 4.2.2 Normal Distribution ... 38 4.2.3 Sphericity ... 39 4.2.4 Conclusion of Assumption and Reliability Check ... 39 4.3 Determining Differences in Sales ... 39 4.4 Determining Differences in Sales Between Pairs ... 40 4.4.1 Hypothesis H1 ... 40 4.4.2 Hypothesis H2 ... 41

4.4.3 Hypothesis H3 ... 41 4.4.4 Hypothesis H4 ... 41 4.4.5 Hypothesis H5 ... 42 4.4.6 Hypothesis H6 ... 43 4.4.7 Summary ... 43 4.5 Post Hoc Findings ... 44 5. Discussion ... 45 5.1 The Effectiveness of Sign 2 ... 45 5.2 Explanation to the Lack of Effectiveness of the Signs ... 46 5.3 Impact of Covid-19 ... 49 5.4 Post Hoc Findings ... 50 5.5 Limitations to the Study ... 51 5.6 The Difficulties in Performing an Experiment in a Real Supermarket ... 52 6. Conclusion ... 55 6.1 Knowledge Contribution ... 55 6.2 Suggestions for Future Research ... 56 7. References ... 58 8. Appendix ... 68

1. Introduction

This chapter provides a background to the topics of the study. The aim is to provide an overview of previous studies within the subject which will result in a problematization. Lastly, the purpose and research questions of the study are presented.

Consumers’ behavior can be influenced in multiple ways. Some of which affects them unintentionally, and some of which affects them intentionally (Shams, 2013). Store managers have the golden opportunity to exploit this in order to affect consumers’

behaviors in their favor. However, the question remains; how can customers be affected the most?

Well-known theories from Cialdini (2007) suggest six principles of ways of influencing which can all be leveraged in order to affect in-store decision making processes. One of which, authority, can be expressed through displaying staff recommendations. This is something that today is not used to a major extent in stores and if, it is often only expressed in relation to certain types of products. Maybe, the use of this type of influencing could result in positive gains for store managers if it is being applied on a wider product range. If so, how can these recommendations be phrased to be as effective as possible?

1.1 Background

This section aims to give a review of previous studies within the subjects of the in-store decision process, sale signs, source of information, recommendations, expertise and authority.

1.1.1 In-store Decision Process

Previous research has studied the level of unplanned buying behavior and the significance of in-store stimuli in the decision making process (Abratt & Goodey, 1990; Johnson & Williams, 1984; Kollat & Willett, 1967). Purchase decisions can be classified into four categories according to the intent behind the purchase; specifically planned purchase,

generally planned purchase, substitute purchase and unplanned purchase (Abratt &

Goodey, 1990). It has been stated that the level of in-store decisions is as high as 65% (Abratt & Goodey, 1990; Nordfält, 2011). Studies also claim that a major part relates to the category of unplanned purchases (Abratt & Goodey, 1990; Johnson & Williams, 1984; Kollat & Willett, 1967). Additionally, the studies state that differences in purchasing behavior can be seen between product categories (Abratt & Goodey, 1990; Johnson & Williams, 1984). In-store stimuli in the sense of promotional techniques, point-of-sale, shelf signs, end-of-aisles displays and special displays is found to affect purchasing behavior significantly (Abratt & Goodey, 1990). This is supported by other studies that conclude that in-store signage has a significant impact on customers’ purchase behavior and decision making (Chevailer, 1975; Clement, Aastrup & Forsberg, 2015) and that displayed products are often perceived by customers as “good deals” and result in more unplanned purchases and an increased average spending (Otterbing et al., 2014). In-store signs are said to draw attention to the product which have, although depending on placement and which type of signage, shown to result in increased sales (Clement, Aastrup & Forsberg, 2015).

1.1.2 Sale Signs

Previous studies have been made on the effects of sale signs (Anderson & Siemester, 1998; Anderson & Siemester, 2001; Chandrashekarana & Grewal, 2006). It has been established that sale signs are less effective at increasing sales when more products have them due to the credibility being reduced (Anderson & Siemester, 2001). Maximizing sales of total categories is furthermore reached through having sale signs on some, but not all, products in the category (Anderson & Siemester, 2001). The study also concludes that a sale sign on a product reduces the perceived likelihood that it will be available at a lower price in the future. This is however less effective when more products are labelled with sale signs (Anderson & Siemester, 2001). Conclusions are based on an equilibrium model presented in an earlier study arguing that customers who lack knowledge of market prices rely on point-of-purchase sale signs in their evaluation of the pricing (Anderson & Siemester, 2001).

1.1.3 Source of Information

During the decision making process, information processing is a key part (Newman, 1977; Newman & Staelin, 1972; Olshavsky & Granbois, 1979; Price & Feick, 1984). They have concluded that consumers often conduct limited external search and tend to rely on a single source of information in their information process. It is suggested that the level of search extensiveness is dependent on the trade-off between additional search and time spent and the expected benefit (Claxton, Fry & Portis, 1974). It is also suggested that the more

complex and expensive the purchase is, the more information gathering is conducted.Other studies have also contributed to the subject through dividing information into two different types; search information that is received through searching for information about the product’s quality and experience information which is received through trying or experiencing the product (Nelson, 1970). A study on informational signs consisting of shelf-labeling of organic foods concluded that the products’ sales volumes increased substantially when labelled with informational signs (Daunfeldt & Rudholm, 2014). Regarding shelf-labeling, it has also been concluded that this type of sign affects product categories differently (Curhan, 1974; Daunfeldt & Rudholm, 2014; Wilkinson, Paksoy & Mason, 1982). The studies have however only examined 3-4 products and therefore not been able to draw generalizable results over multiple product categories.

1.1.4 Recommendations

When faced with purchasing decisions, information in the sense of recommendations have shown to be able to influence the decisions (Berger, Sorensen & Rasmussen, 2010; Duhan et al., 1997; East et al., 2005; Hilger, Rafert & Villas-Boas, 2011; Kowatsch & Maass, 2010; Reinstein & Snyder, 2005; Srinivasan, Anderson & Ponnavolu, 2002). Most studies are primarily on the power of social influences from strong-tie relationships and word-of-mouth from people in the inner circle, as opposed to influences from weak-tie relationships that are less personal (Brown & Reingen, 1987; Chomvilailuk & Butcher, 2014; Duhan et al., 1997; East et al., 2005). It has been shown that strong-tie relationships are more influential than weak-tie relationships with the main argument being that it is due to it being more personalized (Brown & Reingen, 1987).

1.1.5 Expertise

Research has shown that weak-tie relationships are more likely to possess a greater

expertise than strong-tie relationships (Duhan et al., 1997) and that recommendations from people who possess knowledge within the subject of the purchase can facilitate the decision making process (Srinivasan, Anderson & Ponnavolu, 2002). Studies have also been made in the area of experience goods where an expert opinion have shown to affect purchasing decisions regarding wine, movies and books (Berger, Sorensen & Rasmussen, 2010; Hilger, Rafert & Villas-Boas, 2011; Reinstein & Snyder, 2005). It is for instance presented that expert opinion information increased sales of wine with 25% and that information

regarding quality resulted in an increasing demand. It has also been stated that expertise is more important regarding instrumental cues, relating to the technical- or performance-oriented aspects of the product, than affective cues, relating to the aesthetic aspects (Duhan et al., 1997).

1.1.6 Authority

Cialdini (2007) presents six forces that can influence and persuade people. These forces are reciprocity, commitment and consistency, social proof, liking, authority and scarcity. It is stated by the author that people create shortcuts as a way of quickly and simply knowing how to act and make decisions in different situations. People create these automatic shortcuts because it is efficient and sometimes necessary since they cannot always recognize and evaluate all aspects of a situation or choice, nor do they have the time, energy or capacity for doing so. These automatic shortcuts are thereby necessary for people in a world that is growing more complex. The shortcuts also make people vulnerable since knowledge of how they work can be exploited and trigger a desired behavior in people. The shortcuts are weapons of automatic influence and their power can be utilized to manipulate people without them feeling or being aware of the manipulation. Cialdini (2007) means that the six previously stated forces are different ways of triggering these automatic shortcuts humans have developed.

People can trigger a behavior of compliance easier by using authority as they follow the instructions of an authority figure voluntarily (Cialdini, 2007). If a person is instructed by an authority figure to perform a certain action most people will. This is demonstrated through an experiment performed at Yale University by Stanley Milgram where people were asked by a figure of authority to give increasingly strong electrical shocks to another person for every incorrect answer to a question. The experiment revealed that the

participants continued to give electrical shocks even if the person receiving the shocks begged them to stop or moaned out of pain because of the pressure from the figure of authority. Cialdini (2007) presents this example to demonstrate the power of authority figures and the strength of authoritarian pressure in controlling behavior.

1.2 Problematization

As stated in previous research on the topic of in-store consumer behavior and the decision making process at supermarkets, the level of in-store decisions is high (Abratt & Goodey, 1990). To take advantage of this phenomenon and understand how to influence these purchases through in-store stimuli is therefore a major advantage for retailers. Furthermore, it is essential to know which type of in-store stimuli generate the most effects.

It has been concluded that customers tend to rely on a single source of information in their decision making process (Newman, 1977; Newman & Staelin, 1972; Olshavsky &

Granbois, 1979; Price & Feick, 1984). This can in many cases be sale signs and it has been stated that customers who lack knowledge of market prices rely on sale signs in their

evaluation of the pricing (Anderson & Siemester, 2001). It has however not been concluded if this can be applied to other types of signs as well. As a significant amount of purchasing decisions are made in-store (Abratt & Goodey, 1990), there is the possibility that other types of signs might play a vital role in customers’ evaluations and decision making

processes as well. In fact, information in the sense of recommendations have also shown to be able to influence decisions (Berger, Sorensen & Rasmussen, 2010; Duhan et al., 1997; East et al., 2005; Hilger, Rafert & Villas-Boas, 2011; Kowatsch & Maass, 2010; Reinstein & Snyder, 2005; Srinivasan, Anderson & Ponnavolu, 2002). Another source of information

that can affect in-store decisions is, hence, recommendations. Perhaps, recommendation signs can be exploited and affect consumers’ decisions in the same way as sale signs often are used.

Previous studies have mainly focused on the influences of strong-tie relationships (Brown & Reingen, 1987; Chomvilailuk & Butcher, 2014; Duhan et al., 1997; East et al., 2005) and concluded that it is more influential than weak-tie relationships. Hence, a recommendation from a strong-tie relationship would affect customers’ decisions more than a weak-tie relationship. This is said to being caused by strong-tie relationships being more

personalized (Brown & Reingen, 1987), but there might be a way for weak-tie influences to become more personalized. This would likely increase its effect. Studies have also highlighted the impact of expertise regarding weak-tie relationships and its influence on decisions (Duhan et al., 1997; Srinivasan, Anderson & Ponnavolu, 2002). Even though it is evident that strong-tie relationships have greater influence than weak-tie relationships, there might be ways to utilize the effects of weak-tie relationships through exploiting expert influence and increased personalizing.

Studies on experience goods (Berger, Sorensen & Rasmussen, 2010; Hilger, Rafert & Villas-Boas, 2011; Reinstein & Snyder, 2005) have concluded that expert opinions have effect on purchasing decisions as a source of information. These types of products cannot be evaluated before they are tried, hence consumers are more recipient to in-store

recommendations and a target group for utilizing the power of influence from weak-tie relationships. Especially products with low complexity have shown to be more recipient to being influenced through limited search and single sources of information (Claxton, Fry & Portis, 1974; Newman, 1977; Newman & Staelin 1972; Olshavsky & Granbois, 1979; Price & Feick, 1984). More thought is being put into bigger investment purchases. Moreover, it is stated that products being purchased according to instrumental cues rather than affective cues are more likely to be influenced by expert inputs (Duhan et al., 1997). It can therefore be proposed that groceries are products suitable for being subjected to weak-tie influences in the sense of staff recommendations as they are non-complex experience goods.

It has been found that some products increase in sales when they have informational shelf-labeling (Daunfeldt & Rudholm, 2014). Other research has concluded that

recommendations are especially influential and helpful for consumers when they are faced with a larger assortment (Ying, Feinberg & Wedel, 2006). This brings up the question if any differences in the influential effect of recommendations can be detected in regards to how many alternatives the customer is faced with of a particular product. Choosing from a small assortment is not as cognitively challenging as choosing from a larger assortment and thereby signage has a simplifying effect on the decision when the assortment is larger (Goodman, et al., 2013). When obtaining a larger product sample, protruding differences could possibly be detected when analyzing the result from a product level. More

generalizable conclusions could also be drawn across subcategories of different staple products as some products could be more recipient to influences. Moreover, it would be of interest to examine more products than solely 3-4 products done in earlier studies (Curhan, 1974; Daunfeldt & Rudholm, 2014; Wilkinson, Paksoy & Mason, 1982).

1.2.1 Purpose and Research Questions

The purpose is to examine and map how different types of signs on shelves related to staff recommendations for specific products affect their sales.

Research questions:

· How are the sales of the products affected by the signs? · Which of the signs affect sales the most?

· Are the sales of the products affected differently according to how many alternatives they have?

2. Conceptual Framework

This chapter presents the conceptual framework of the study. Firstly, an overview of the buying decision process is presented with an in-depth discussion of choice complexity. Following, a hypotheses development is elaborated where the six hypotheses of the study are presented.

2.1 Buying Decision Process

The process of buying a product as a consumer is more than solely the purchasing moment. Events both before and after affect the consumer’s choice of brands, products and overall satisfaction of the products. This process is known as the buying decision process and entails five stages that consumers go through from when they realize a need for something to afterwards evaluating the purchase and product (Kotler et al., 2016).

The first step in the process is need recognition, where consumers realize they have a need for something. This feeling can be triggered by either internal or external stimuli. Internal stimuli refer to a person’s normal needs like hunger, whereas external stimuli refer to a need caused by a person admiring their neighbors’ new car for instance (Kotler et al. 2016). The second step of the process is information search. It is during this phase that consumers search for information. This phase can be different in length and level of involvement depending on the consumer and the product. During this phase the consumers become more receptive to information about products that can fulfill the need that was triggered (Kotler et al., 2016). Consumers create decision criteria of what kind of product they are looking for and what attributes it needs to have. How consumers frame these criteria affect their buying decision process since they search for information in different ways depending on the criteria. They impact which products are seen as options and which are to be regarded as alternatives in the next stage of the buying decision process (Shams, 2013). Information can come in many different forms, one of them being visual form. Consumers perform visual search, a voluntary search of the environment, to find relevant information. They also become involuntary conscious of their environment through the saliency effect for instance (Shams, 2013). For visual search or the saliency effect to influence the consumer,

they need to pay visual attention. Both the design and the saliency of a product’s package is of importance during the visual search stage. The design can help a product stand out on the shelf while the saliency is more dependent on the criteria the consumer has established. Certain information can have a saliency effect on one criterion and not on the others (Shams, 2013). Another aspect that has a visual impact on the consumer’s buying decision process is the placement on the shelves as products placed at eye-level have a perceptual advantage. Consumers use visual search to find relevant information regarding the criteria they have established (Shams, 2013).The information the consumer is exposed to during information search can come from many different sources, such as personal sources, commercial sources like advertisements or salespeople, or from public sources like social media and consumer-rating organizations (Kotler et al., 2016). The final source of

information is experiential which includes examining, handling and using the product (Kotler et al., 2016). Research conclude that consumers even choose the alternative they spontaneously find, and many consumers are content with the first option presented to them. Retailers can thereby help and remind consumers of their needs and present them with appropriate solutions during this phase. This can be done through displays,

mannequins or signs (Nordfält, 2011).

The third step in the process is the evaluation of alternatives. This is where the consumers evaluate the different options and make a final judgement on what their preferred option is (Kotler et al., 2016). Consumers evaluate how well the different options will satisfy their needs by looking at the different attributes of the alternatives and their benefits (Kotler et al. 2016). As a way of simplifying the final decision, consumers perform a set of

comparisons which results in eliminating some alternatives (Shams, 2013). They also use selective visual attention as a way of suppressing non-relevant information and accenting relevant information (Shams, 2013). When making the final decision, one of the major factors affecting the decision is the number of alternatives of the product. When the number of alternatives has reached six or more, choice strategies are adopted to quickly eliminate some alternatives (Payne, 1976). A small number of alternatives often results in a one-stage strategy, usually cost-depending, whilst a larger number of alternatives results in a

multistage strategy. When faced with many alternatives, other factors than solely compensatory are evaluated (Olshavsky & Granbois, 1979).

The fourth step is the purchase decision where the actual purchase takes place. Before this phase the consumer has formed preferences among the brands considered and usually concluded the preferred brand and product and forms a purchase intent. During this purchase it is more than just the choice of brand and product that is considered. Where to buy it, quantity, when to buy the product and what payment method should be used is also considered. There can be some intervening factors that might intervene the purchase intent from becoming the purchase decision such as the attitudes of others, unanticipated

situational factors and perceived risks. The attitudes of others consist of the extent that other peoples’ attitudes diminish the consumer’s preference for a brand and product. Here the evaluation and reviews from others can also have an impact if the purchase intent develops to a purchase decision. An unanticipated situational factor like being laid off from work can change the purchase intent (Kotler et al., 2016).

The fifth and final phase is the post-purchase behavior. This comes after the actual purchase and here the consumers might feel a satisfaction or dissatisfaction depending on how well their expectations were met but also what information they were exposed to about how satisfactory other alternative products and brands were. A satisfied consumer is more prone to buy the same brand and product again but also to speak highly of it to others. This phase also includes how the consumer uses and disposes of the product (Kotler et al., 2016).

2.1.1 Choice Complexity in the Buying Decision Process

Multiple factors can affect the buying decision process (Kotler et al., 2016). The buying decision process for consumers is a complex process which many actors like stores, brands and other organizations try to influence. One common way of attempting to influence the process is by providing information to the customers relating to products. Actors provide this information in an attempt to influence the customers, often as a way to try to make

them buy a certain product. However, providing this information about product attributes could have the opposite effect. When customers are faced with more information regarding product attributes, it affects which product attributes consumers take into account when making their decision. When information increases the number of attributes to consider, customers choose to ignore or use some in a very limited way. Customers choose to adopt a choice strategy as a way of simplifying the decision since the human brain have constraints relating to processing information. A choice strategy is a way of decreasing the cognitive strain (Olshavsky & Granbois, 1979). Research concludes that a certain variance in the decision making process can be seen according to the complexity of the attributes that the product possess, where an increasing level resulted in more complex strategies not only based on compensatory factors (Olshavsky & Granbois, 1979).

The choice is complicated further by the level of impulsiveness consumers have when shopping. Factors like store layout, reference group, availability of cash, mood of

consumer, POS terminal/ATM facility, price, availability of time, product promotion and store environment have significant effects on the impulsiveness of customers (Pradhan, 2018). The type of weather can also impact the purchasing behavior as sunny and warm weather typically increases the sales of items used for barbeques and ready-to-eat foods for example (Svensk Handel, 2019). It is thereby important to take these factors into account when trying to affect the decision making process of customers. As can be seen by examining the above mentioned factors, many of them relate to the store. To be able to understand how a customer’s choice was made and how to impact the impulse purchases a customer makes, these factors need to be evaluated.

The kind of retail format of the store has also been shown to have an impact on consumer behavior (Wang et al., 2010). This fact further complicates the understanding of

consumers’ decision making process since it implies that they do not make the same choice in all stores. It has been indicated that store formats such as supermarkets, discounter or organic retailers have an impact on purchase behavior even when attributes like product and price are the same (Baum & Weigelt, 2017). Customers are more likely to buy larger package sizes in mass merchandiser stores (retailers that offer products like food, furniture, clothing and accessories all in one store), but consumers are also more price sensitive in

these types of stores compared to normal supermarkets, that only offer food or drug stores, that offer a more limited assortment (Wang et al., 2010). The same product or promotional activity can thereby show a difference in efficiency depending on what type of store format it appears in.

Another factor affecting the impulsiveness and how much products are bought on impulse is what kind of product it is and if the consumers have tried the product before. The

influential effect of advertising is greater when a consumer has not tried the product before (Ackerberg, 2001). If a consumer has tried the product or brand before, advertisement providing information about attributes or characteristics can have an insignificant or even declining effect. How big of an influential effect signs or information have is also

dependent on what type of product it is. Products like grocery items, personal care or accessories, consumers buy more impulsively compared to products like electronics or kitchenware (Pradhan, 2018). This is because electronics and kitchenware are usually high involvement purchases while grocery items, personal care and accessories are usually low involvement purchases.

The understanding of the consumers’ buying decision process is even further complicated when the effects of semiotics and sensory marketing are taken into account. Depending on how a marketing activity like an advertisement or sign is designed, their effect on consumer behavior can be very different. It is important to understand that everything from shape, wording, color and even the hue of the color matter when trying to design an advertisement or sign that is effective. The relation between the amount of words and amount of pictorial content in an advertisement or sign should also be taken into consideration. Research suggest that advertisements with higher word content have a purpose of making customers read and examine the sign and the information, while advertisements with a higher picture content are the advertisements that are more likely to be viewed by customers. High word content versus high pictorial content thereby spark different reactions in consumers and have different strong suits (Aradhna, 2010). The research also shows that images in color versus monochrome images spark different reactions in the brain and different associations are made with the two. In practice, this can mean that using a monochrome color scheme for an advertisement can be effective when trying to create a nostalgic feeling. On the other

hand, it could be negative as well since some people might associate a monochrome color scheme with more high-end products. Consequently, using this color scheme on cheaper products could confuse the consumer but also make them change the associations they have with a monochrome color scheme (Aradhna, 2010). Research suggest that red colors tend to create a feeling of arousal in consumers while blue colors create a feeling of relaxation (Aradhna, 2010).However, there are also differences between background and text color. Many marketers use vivid colored imagery since these colors attract the visual attention of consumers (Theerathammakorn et al., 2019). Studies even suggest that the hue or saturation of the color can make a difference on consumers’ feeling and liking of an advertisement or image and in extension the product (Aradhna, 2010; Theerathammakorn et al., 2019). Brighter colors are also proven to lead to greater liking of the advertisement (Gorn et al., 1997).

It can be concluded that there are many factors affecting the choice complexity of a consumer’s buying decision process and understanding the underlying factors is hard. Understanding which factors plays a role in the consumer’s choice however is one step in the right direction to be able to affect their buying decision process.

2.2 Hypothesis Development

Research so far has concluded that signs have an impact on sales and on how consumers make their decisions in-store (Nordfält, 2011). Woodside and Waddle (1971) compared the effects of signs on products with ordinary price, cutting the price of the product, and cutting the price combined with having a sign. They concluded that all three options increase sales but that the sign in combination with a price-cut generates the most increase in sales. Their experiment was conducted on instant coffee and with having a 20% price cut increased sales by 34,1%, having a sign at ordinary price resulted in an increase in sales by 185,2% and by having a sign accompanied by a price cut increased sales by 380,7%.

There has also been done extensive research on how a sign receives and controls

placed on the sign, the consumers’ choice of brand and product differs. In this experiment when the consumer was introduced to unit-price information comparing products, the consumers bought more less-expensive products. Even though this experiment has a price focus, the fact that consumers alter their purchase behavior depending on what information they are faced with in-store is worth noting.

The idea that signs without a price cut still generates more sales can be explained by a few different aspects. Ackerberg (2001) explains that advertising has a few different effects, one being the so-called prestige effect. Nelson (1974) and Milgrom and Roberts (1986) along others explain that simply having advertisements on a product signals to the consumers that the product is of higher quality, especially when it is experience products, “products whose complete characteristics are not observable to consumers before purchase” (Ackerberg, 2011 s. 317). Food can be seen as an experience product since characteristics like taste or quality can´t be determined before experiencing the product. They determine that the advertising do not need to inform the customers in any way of the products characteristics, the advertising in itself signals that the product is of high quality or good taste (Nelson, 1974; Milgrom & Roberts, 1986).

Previous research has found that expert reviews do influence and persuade people. One experiment found that the sales of wine increased by an average of 25% when there was a label with an expert review next to the price (Hilger, Rafert & Villas-Boas, 2011). Another study was performed by adding an expert rating on movie releases. The study found that expert reviews did have a positive effect on sales, which increased with better ratings (Reinsteint & Snyder, 2005). In this study, a staff employee can be seen as an expert since they have a wider knowledge of the products available in their marketplace than the average consumer.

When keeping all of this in mind the conclusion is drawn that signs as a stimulus have a positive impact on sales, even if the stimuli does not contain a price cut or information about a products characteristic. It can also be determined that expert recommendations and reviews persuade and influence the consumers in their purchase decision which results in more sales of the recommended products. This provides the foundation for the first

hypothesis: Signs which states that the staff recommends the product will increase the sales of the product.

H1: Sign with “the staff recommends” will increase sales.

In a study performed by Brown and Reingen (1987) it is explained that recommendations from strong-tie relationships like close friends should have a greater influential impact on the receiver in their decision making process compared to weak-tie relationships. However, weak-tie relationships like acquaintances or a person one interacts with at times, like a store employee at the local shop, still has an impact when it comes to making referrals and recommendations. Based on this information, adding a name and job position to the sign, and thereby connecting the recommendation to a weak-tie relationship that many

consumers have to a store employee should increase the impact the recommendation has. Research has also been done on the matter of source credibility and the effect on consumer behavior. It has been concluded that source credibility affects the effectiveness of the communicated message (López & Sicilia, 2014). By increasing the source credibility, the effectiveness and persuasive effect of the message will increase. Therefore, adding the name of a member of staff to the signs should increase the source credibility since the customers know that the staff member works in the store and possess knowledge of the products. Adding the name to the sign increases the personalization of the sign and the influence should therefore more resemble that from a strong-tie relationship (Brown & Reingen, 1987). They thereby perceive the recommendation to be of higher credibility than the staff as a whole. The second hypothesis is hence that adding a name to the sign of a staff member should make the recommendation more personalized and thereby increase the sales of the product.

H2: Sign with “the staff recommends”, the name and job position of the staff member recommending the product will increase sales.

Research in the topic of advertising and how to optimize the attention signs and

advertisements get has been made with the goal to increase the attention they receive from consumers (Guyader, Ottosson & Witell, 2017). Research has concluded that text and pictorial both increases the consumers’ attention to the advertisement or sign (Aradhna, 2010). By adding a picture to a sign compared to only having text, the attention it gets from consumers will increase (Pieters & Wedel, 2004).It has also been shown that when people look at them longer, they are more likely to consider it in their final decision (Guyader, Ottosson & Witell, 2017). For this studyt it is thereby concluded that adding a picture to the stimuli will increase the attention it receives. This will in turn have a greater influential and persuasive effect on the consumers’ choice and will make them choose the advertised product.

H3: Sign with “the staff recommends”, the name, job position and picture of the staff member recommending the product will increase sales.

Even though a sign without any information about a product’s characteristics increases sales (Nelson , 1974; Milgrom & Roberts, 1986), research suggests that when the

advertisement include relevant and verifiable information about a product’s attributes, their confidence in assessing a product’s quality increases (Holbrook, 1978; Nelson, 1970; Nelson, 1974). Nordfält (2011) explains that people also put more trust in a message when it is more detailed, a sign with more information should thereby be more persuasive and have stronger effect on increasing sales of the product. Research have also indicated that when consumers receive information about the product or the way it was manufactured it increased their purchase intent (Abadio Finco et al., 2010). Due to this, the fourth

hypothesis is that stimuli with name, job position and picture of the staff member

recommending the product, including a personalized message of why they recommend it, should increase sales.

H4: Sign with “staff recommends”, the name, job position and picture of the staff member recommending the product, including a personalized message of why they recommend the product, will increase sales.

As previously stated Nordfält (2011) explains that people put more trust in a message when it is more detailed, a sign with more detailed information should thereby be more persuasive and have stronger effect on increasing sales of the product. The information can take different forms, but the amount of information can be seen as a moderator. A

moderator can be seen as changing the strength of the main effect, that the use of signs will lead to increased sales, without changing the direction of the effect. Consequently, as the amount of information on the signs increases, the sales should increase.

H5: As the amount of information increases on the signs, the sales will increase.

Ying, Feinberg and Wedel (2006) concluded that recommendations are especially helpful for consumers when they are faced with a larger assortment, which entails that there are more alternatives available that serve the same purpose and where differences in quality is not easily detectable. Choosing from a small assortment is not as cognitively challenging as when choosing from a larger assortment and thereby signs can have a simplifying effect on choice when the assortment is larger. This simplifying effect is especially prominent on people who have less developed preferences within the assortment (Goodman et al., 2013). It can thereby be concluded that signs will have a greater effect on sales on products that have many alternatives, compared to products with few alternatives, available in the supermarket.

H6: Sign placed on products with many alternatives will have a greater effect on sales.

3. Methodology

This chapter outlines the methodology of the study. The choice of methodology approach is presented and motivated as well as an in-depth description of the executed experiment. A review is presented of the supermarket in which the experiment was executed, the product selection process and the design and placement of the signs as well as the limitations. The approach to the literature review and the analytical strategy is discussed. Lastly, a quality discussion and ethical considerations are presented.

3.1 Research Philosophy, Strategy and Approach

The basis for the study is a realistic perspective as it aims to mirror a truthful picture of the reality and describe a specific in-store customer behavior. The basis of it is, hence,

behaviorism and more specifically stimuli and response.

The scientific approach suitable for the purpose of the study is deductive and the

corresponding research method is a quantitative study with an experimental design and a descriptive focus. Through conducting a quantitative study, structured and statistical data is received that, through its reliability and validity, later can be used to come to general conclusions (Rienecker & Stray Jørgensen, 2014). A quantitative study is further suitable for the deductive approach. The choice of an experimental design is suitable for an in-store study as it enables fast results which are easy to measure economically (Doyle & Gidengil, 1977). The choice of a real in-store experiment contrary to a laboratory experiment of hypothetical choice situations, is favorable for studying consumer behavior as it provides realistic purchasing conditions which contributes to greater possibilities to draw

conclusions from the results (Doyle & Gidengil, 1977). The choice of a real in-store experiment or a laboratory experiment also depends on the goal of the research. When the goal is to demonstrate a principle or the psychological underpinnings of it, a lab experiment could be suitable. A lab experiment also allows for more control as external factors can be eliminated while there are many, often unpredictable, factors that can affect a real in-store experiment. The measured effects of a real in-store experiment can therefore be less

which previously has not been researched or which is not well understood, a more realistic experiment could be more suitable. To deduct if a real experiment or lab-controlled

experiment is most suitable, two parameters should be analyzed: realism and behavior. Realism refers to how an experiment ranges on a scale from artificial to realistic. While behavior refers to what kind of behavior is measured ranging from hypothetical intention to actual behavior. Consumer research which is high in realism and use actual behavioral measures is more effective in giving insight into real consumer behavior (Morales, Amir & Lee, 2017). The goal of this study is to get a deeper understanding of the phenomenon of recommendation signs and its impact on a real behavior measure in the sense of actual sales. Hence, a high level of realism and behavioral measure seemed most appropriate and a real in-store experiment was selected.

3.2 The Design of the Experiment

This section aims to give a thorough description of the experiment and the supermarket in which it was executed in.

3.2.1 The Supermarket

The experiment took place in one of the largest supermarkets in Sweden. It is located in a shopping district outside the city center of a relatively large city. The supermarket is a stand-alone store with a large parking lot situated in front of it, making it convenient for customers to visit despite its location. As the supermarket is in a shopping district outside the city center not many customers live within walking distance. Customers can take public transport or go by bike to the supermarket, but the most common way for customers to get to the supermarket is by car. The supermarket’s sales area is 8000 square meters and approximately 660 000 items are sold every week. Except for food items, the supermarket also sells outdoor supplies, animal food, home décor, kitchen ware, clothes and stationery. The goal of the supermarket is to offer customers everything they need under one roof, as a way for customers to only have to make one stop. The supermarket has 80 000-90 000 visitors weekly where roughly 37 500 of them make a purchase. The customer clientele is mixed but the supermarket is especially popular amongst pensioners and families with

small children. A café is situated in the building to further incline customers to shop at the supermarket as it allows for the shopping trip to be a social event as well. The

supermarket’s food department offers everything from budget friendly options to more luxury goods to offer products for all types of customers. The supermarket offers a big variety and range within almost all product categories to meet varying customer demands.

3.2.2 Product Selection Process

The first step when preparing for the experiment entailed choosing the products that were going to be included in the experiment. All chosen products were staple commodities, as these are products most customers buy and consume and would therefore enable the experiment to be applicable to all types of customers. These products could also be seen as more homogenous and similar across brands which could make it easier to influence consumers since differences are less prominent compared to other types of products. The products were selected from the supermarket’s top selling list of staple products. The choice of only including staple products was made, as these are sold in larger quantities which entails that natural unexplainable fluctuations, where an increase or decrease of a few number of sold products, have relatively less of an impact on the total sales. Analyzing the result of the experiment would hence be more credible.

For a product to be included in the experiment a few requirements had to be met. The first requirement was that the sales the previous four weeks, before the selection took place, were relatively stable. The second requirement was that the sales from the last few months were relatively stable, when disregarding from sales during the holidays and various

campaigns. This was disregarded when evaluating stableness of the sales as they often have a big impact on them and therefore cannot be regarded as a normal sales quantity. Products were rejected if they were going to be included in any campaigns or reductions, or if a direct competitor was included in a campaign or reduction during the weeks of the experiment. A large number of products were chosen to be included in the experiment if campaigns or reductions on the products would become known later and they therefore had to be excluded. Since the aim of the experiment is to study how signs affect sales during

normal conditions, it was believed that excluding these abnormalities were the most accurate way of choosing products.

It was also taken into account to not have too many products from the same product category such as baking ingredients. This was considered as products from the same product category are usually placed close together on the shelves and having too many signs close together on the same shelf and aisle could be considered negative. If consumers noticed a lot of products being recommended at the same time, they could start to question the credibility of the signs. A mix of products from different subcategories of staple

products were also taken into account as the aim of the experiment was to be generalizable to all kind of basic food products. Having too many products within one product category could alter the results if one product type was affected differently by the signs compared to other product types. When considering these requirements 28 products were chosen. The chosen products can be seen in the Table 1 as well as the number of alternatives to each product.

Table 1: Chosen products and number of alternatives.

3.2.3 Design

Four different signs were placed on the shelves of the 28 products. The four signs had the same size, color scheme, background and font. The only things that differed between the signs were modifications in the wording on the signs and on some signs a picture or personalized messages. Every sign was a modified variation of the previous one including more information in the sense of a name of a certain staff member, a picture of the staff member or a personalized message of why the staff member recommended the product. All signs were in Swedish but the wording on the signs will be translated to English.

The first sign said “The staff recommends” (see Figure 1). The sign tested hypothesis H1 stating that a sign with “the staff recommends” would increase sales of the product.

The second sign said “N.N. from the staff recommends” (see Figure 2). The sign tested hypothesis H2 stating that a sign with “the staff recommends”, including the name and job position of the staff member recommending the product, would increase the sales the product.

The third sign said “N.N. from the staff recommends” accompanied with a picture of the staff member (see Figure 3). This sign tested hypothesis H3 stating that a sign with “the staff recommends”, the name, job position and picture of the staff member recommending the product would increase the sales of the product.

The fourth sign said “N.N. from the staff recommends” with the picture of the staff member along with a short, personalized message of why the staff member recommends the product (see Figure 4). This sign will test hypothesis H4 stating that a sign with “staff

recommends”, the name, job position and picture of the staff member recommending the product, including a personalized message of why they recommend the product, would increase the sales of the product. The personalized messages will be individually adapted to each product since the products are different from each other and favorable attributes differ. Having the same personalized message for all of them would hence not be suitable. Having the same personalized message on all of them could also have negative effects on the consumers’ perception of the credibility of the signs since they could perceive that the signs were placedwithout any deeper consideration making him less credible. All the

personalized messages were phrased in a way to describe that the specific product was of high quality and had the desired attributes for that specific product. The staff member was a part of the process of developing the personalized messages on the signs and was able to eliminate products or alter what the message said if he did not approve.

The chosen member from the staff is a male who has worked in the supermarket for many years and appears on the supermarket’s social media platforms as well as other forms of advertisements. This makes him a well-known and recognized individual by many

customers and therefore the choice was made to have him on the signs. Having other staff members who have not been employed in the supermarket for long or are less well-known to customers could be unfavorable since their effect and credibility would possibly be lower.

3.2.4 Placement

The signs were placed on the shelf next to the shelf-label of the products in the supermarket and placed in a sign holder making the sign stand out at a ninety-degree angle from the shelf (see Figure 1-4). The choice of sign holder was made to catch the attention of the customers since the signs otherwise would be harder to notice as the shelves already contain many signs. To ensure that the signs were noticed by customers, control questions were asked during the first day of the experiment where consumers were asked if they had seen them. The staff was informed of what products the signs were supposed to be placed on if they fell down and had additional signs available if the first ones broke.

All signs hung for one week each from Sign 1 to Sign 4 during 2nd _{to 29}th _{of March 2020.} The decision to have the same sign on all products the same week was made to limit the interference the different signs could have on each other. If for instance Sign 1 and Sign 4 was up at the same time the customers could think that the specific staff member

recommended the products even if Sign 1 only said “The staff recommends”. They could also question its credibility due to it not including a personalized message or picture. Consequently, to be able to measure the effect of one sign without any interference there was only one type of sign up in the supermarket each week.

3.3 Limitations

A limitation of the study is that a larger sample is not included. Even though the experiment includes more products than what has been in previous studies, the study could benefit from an even larger sample. The study could also benefit from including data of the same

experiment from multiple stores to make the results more reliable and generalizable. Furthermore, it would have been beneficial to collect data under a longer period to avoid sudden fluctuations from having significant impact on the results and hence receive more reliable results. This was however not possible due to time constraints.

3.4 Literature Review

Suitable theories and previous researchfor the study has been reviewed in the areas of the buying decision process, in-store consumer behavior, persuasion in relation to authority as well as signs. The aim has been to map out the fundamental theories and well-known authors within these areas to get a basis for the scope of the study and further deepen the knowledge through more area-specific studies suitable for the experiment. The aim of the literature review has been to use credible and well-known scientific journals with expertise within the field of marketing and consumer behavior. A further aim has been to opt for articles with a larger number of citations and articles that are mentioned as a basis in

multiple other studies. As the purpose of the study does not require any specific models, the theory basis is, except for the fundamental theories about the buying decision process and persuasion, mainly conducted from previous studies of observations on customer behavior.

Potential criticism to the literature review is the age of the studies as many of them could be considered old and outdated. However, many of the studies conducted within the relevant area were conducted in the 1970s and 1980s and even though there has been a strive to finding more contemporary studies, nowadays as the society has evolved, many studies are instead being made in areas related to digital solutions. Hence, many new relevant studies have not been able to be found.

3.5 Analytical Strategy

This section aims to describe the analytical process.

3.5.1 Reference Week of Sales

A reference week of sales was calculated to measure if and what effects the signs had. This was done through calculating an average number of sold items of each product during the equivalent month one year prior to the experiment period. The decision to use the same month a year prior was done because sales vary considerably depending on month and season which means using the same month a year prior had the most similar conditions. This comparison was thereby considered the most accurate for determining the effects of the signs. Promotions on the products or highly abnormal sales due to reasons like the substitute or competitor brand being on sale were excluded when calculating the sales of the reference week. If a product had a promotion during the month, the month prior was used since being relatively similar in sales and circumstances. To be able to compare if the signs had an effect it was appropriate to compare the sales to those of an average normal week of sales since these weeks can be regarded as not being manipulated and therefore depicting the normal demand of the product.

3.5.2 Calculations in SPSS Statistics

SPSS Statistics, a software package used for statistical analysis, was used to analyze the effects of the signs and conclude whether the results supported the hypotheses or not. In the tests performed, the products were set to independent variables and the sales numbers during the different signs were set to dependent variables.

A few conditions had to be examined prior to choosing which type of test was best suited for the data set. Firstly, the data set from the experiment was of a paired kind, which is defined as the same test subjects being measured on multiple occasions (Scheff, 2016). Consequently, statistical tests that are designed for unpaired data were inappropriate and not used in the analysis.Secondly, the normal distribution and a test of sphericity were

examined in order to determine if the data was parametric as it was desired to use parametric tests.

Parametric tests are considered to give more accurate and trustworthy results and have greater statistical power compared to non-parametric tests (IBM, 2018). It is thereby preferable to use parametric tests if possible, but they require data to be normally distributed. If the data is not normally distributed, but rather positively or negatively skewed, an attempt to improve it can be made by transforming the data (IBM, 2018). Determining if the data was normally distributed was done by examining the data visually when graphed in a histogram and by examining the Shapiro-Wilk test of normality. A histogram is beneficial to examine since it can give a clear indication if the data is positively or negatively skewed while the Shapiro-Wilk test is beneficial to draw clear conclusions if the data is normally distributed or not. The Shapiro-Wilk test is best suited to examine the distribution when the sample size is less than 50 (Thadewald & Büning, 2004). Since the sample size in this experiment is 18 after exclusion, the Shapiro-Wilk test was chosen. When studying the data in a histogram the conclusion can be made that if it looked approximately normally distributed and did not have too many deviant values it was

assumed to be normally distributed (Scheff, 2016). In the Shapiro-Wilk test, the data can be considered normally distributed if the p-value is above 0,05 (IBM, n.d.a). The data set was concluded to be positively skewed and therefore transformed using log10. Transforming data using log10 entails mathematically modifying the data to make the data set appear to be more normally distributed (Pallant, 2013).

The level of sphericity needs to be examined prior to analyzing the data using an ANOVA test (IBM, n.d.b). To examine the sphericity of variance, a Mauchly’s test was applied where the assumption of sphericity is considered violated if the level of significance is below 5% (IBM, n.d.b). When the assumption of sphericity is violated, the use of an

ANOVA test needs to be modified. An ANOVA test can still be used but the values need to be corrected through one of the methods; Greenhouse-Geisser or Huynh-Feldt or

Lower-bound (IBM, n.d.b). The Huynh-Feldt is the most suitable method for correcting the values

for smaller sample sizes (IBM, n.d.b). Since the sample size of the experiment was small, the Huynh-Feldt method was used when the data violated the assumption of sphericity.

A repeated measures ANOVA test was executed. The test determined whether the means of groups significantly differed from one another and whether the difference in means was a result of them originating from different populations or caused by some random variation (Scheff, 2016). A repeated measures ANOVA test was the most suitable type of ANOVA since it was used for paired data (Scheff, 2016), which the data set in the study was. An ANOVA test was used as it was the first step to conclude if there were significant differences in sales between the signs.

Paired samples T-tests were also performed. This specific T-test was chosen as it was

suitable for paired data (Scheff, 2016). This type of test determines possible differences between the mean of two groups and if one group’s mean is larger than another (Scheff, 2016). The tests were performed to be able to compare the sales during one sign to another as the test showed which mean of the two was larger and thereby, which sign resulted in the most sales.

To examine any differences depending on the number of alternatives the products have, a

two-way repeated measures ANOVA test was executed with the number of alternatives to

each product as the between-subjects factor. As described earlier, a Mauchly’s test of sphericity was examined prior to analyzing the results of the ANOVA test. It reveals the joint effect of two independent variables on one dependent variable through examining the p-value of the interaction effect (Scheff, 2016). The additional independent variable in the test was the number of alternatives each product had. The interaction effect exists when the effect of one independent variable on one dependent variable depends on the level of the second independent variable in the test (Scheff, 2016). This type of test was hence performed since it determined whether there were differences in sales, and thereby differences in the effects of the signs, depending on how many alternatives the products had.

All products were divided into one out of three categories depending on the number of alternatives they had available in the supermarket. “Few alternatives” was defined as 0-5 alternatives available, “some alternatives” as 6-15 alternatives available and “many

only include those that could be considered equivalent to the product in the experiment. The decision was thereby made to not view “light” products as alternatives.

A significance level of 5% was chosen for all the tests as it is the common standard of significance level. A significance level of 10% or 15% can be chosen when an experiment is performed on a small number of subjects since one data point can alter the mean more compared to when having a large number of test subjects (Scheff, 2016). This fact was taken into account since there were only 18 test subjects after exclusion. However, to keep a high level of credibility of the results, a 5% significance level was the benchmark for determining if there were any statistically proven differences between the sales of the signs.

3.6 Quality Discussion

The three most important criteria to ensure the quality of the method are reliability,

replicability and validity (Bryman & Bell, 2017). The first mentioned criterion, reliability, is especially important for quantitative studies and concerns whether the study would generate the same results if it were to be executed again or if other factors might have significant impact on the results of the study (Bryman & Bell, 2017). The aim is to reduce the impact from other factors as much as possible through thorough evaluations of

differences in sale patterns, other possible influences, external factors and possible points of failure. As the experiment was conducted in a real in-store environment opposed to a

laboratory setting, external factors were unfortunately more prominent even if they were reduced as much as possible. The objective of the study was however that true consumer behavior would be depicted.

The second criterion refer to the ability to replicate the study, which is especially important for quantitative studies when striving to generalize (Bryman & Bell, 2017). The aim was to give as thorough of an explanation of the experiment as possible, in order to make it

possible to replicate the study. The products included in the experiment, the design of the signs and the personalized messages on the signs are all disclosed. The brand and variant of the products could however not be disclosed due to confidentiality reasons.

Lastly, there is the criterion of validity which determines if the findings are generalizable where the most relevant types of validity for this study is internal and external validity. Internal validation concerns causality (Bryman & Bell, 2017). The abundance and geographical dispersion of stores are indicators of good control and thereby high internal validation (Doyle & Gidengil, 1977). There are two aspects of internal validation that are explicitly applicable to in-store experiments: history and instrumentation (Doyle &

Gidengil, 1977). The first refers to the interference of external events and the latter to error in the actual experiment like differences between the products’ shelf positions (Doyle & Gidengil, 1977). Regarding the aspect of history, the outbreak of Covid-19 and its

consequence of changed consumer behavior was evident at the time of the experiment. As to the aspect of instrumentation, the difference in shelf positions is a known source of error but was disregarded as the aim of the experiment was to be executed in a real in-store setting. External validation refers to if the results can be applied outside of the context of the study (Bryman & Bell, 2017). It is strengthened by the real environment of a store and the randomization regarding participants (Doyle & Gidengil, 1977). A risk however remains that the store is unrepresentative for the entire population (Doyle & Gidengil, 1977). Another factor that has had an impact on the external validation is the outbreak of Covid-19 as it changed the behavior of consumers and the experiment would likely show different results if it was to be executed again under normal circumstances. The criterion of validation is supported by a data set that is normally distributed and show sphericity of variance.

There are however still some risks concerning validation, for example how the effects on sales can be ensured to have been caused by the signs. There is always the risk of

influences from external and unpredictable factors even though the study was developed in a way that possible risks were thoroughly evaluated and taken into consideration when possible. To ensure that the customers had noticed the signs, control questions were asked during the experiment. However, since not any qualitative interviews were conducted, the customers’ behavior and thoughts during the buying decision process cannot be explained on a deeper level through the study. Moreover, it cannot be specified if the signs lead to a change of brand for an already generally planned or substitute purchase, or an unplanned

purchase. The intent of the study was however not to specify in which way the customers were influenced. It is also not possible to know if some of the customers are repeating customers that have seen the signs from the week or weeks before which could, as a consequence, lead to accumulating effects.

3.7 Ethical Considerations

The experiment resulted in some ethical dilemmas that needed to be considered. One of them being that the goal of the experiment was to persuade people into buying items. As it could not be concluded if the people influenced by the signs actually needed the items or if they had planned to buy the exact item or a similar item, it can be argued that in this time of climate change and over-consumption trying to influence people to buy more is unethical. However, since the products in the experiment were staple food products and the intent was not to encourage excessive spending, it was believed that these products were coming to use and would have been purchased anyway. Perhaps the purchase of the product was not planned to occur during the exact shopping trip, but most people consume staple food products regularly. Therefore, the impact on the climate and over-consumption should not have been increased by the experiment. The ethical concerns of influencing to buy the products in the experiment also decreases since the products were not alcohol, tobacco or unhealthy food items.

A conscious decision was also made to include products from all price points and not only promoting more expensive products. This decision was made partly because the aim was to be able to draw general conclusions about products from all price points, but also as a way to reduce the risk of influencing people to buy more expensive products than they normally would. The intent was not to affect peoples’ finances or food spending in any major way as that can be considered unethical and not align with the criterion stated by Swedish Research Council (2017) of protecting the participants. The choice was also made to not analyze the behavior of customers on an individual level through analyzing receipts, as a way of

protecting the customers’ integrity. Therefore, only aggregated results were presented. This choice was made to minimize the ethical concerns that arise when studying individual