When do customers choose an everyday low pricing store
above a promotional pricing store?
Rianne Kok
2
When do customers choose an everyday low pricing store
above a promotional pricing store?
Rianne Kok
University of Groningen
Faculty of Economics and Business
Msc Marketing Intelligence Master Thesis 22 – 06 – 2015 Nassauplein 15 A 9717 CM Groningen +31 616524409 r.kok.3@student.rug.nl Studentnumber: 2560372
Supervisor: Erjen van Nierop
3
Management summary
5
Preface
Half December 2015 I handed in my registration form for thesis topics. All topics were interesting, but I decided to sign up for a topic of which I do not know a lot about, so that I can learn the most out of my master thesis. Some topics overlapped with my bachelor thesis from the University of Applied Science and with my bachelor thesis from the University of Groningen. Therefore, the topic ‘The Price is Right! … Or Is It?’ was the topic that kept my attention right away. In most cases, price is the most important driver in buying a product or not. Therefore, it has a direct impact on the number of products sold and consequently on the profit of an organization. Every day I see product prices in groceries, retail stores and even online, however my knowledge about prices and price strategies is definitely not accurate. Accordingly, I decided to choose this topic as my number one choice. Now more than a half a year later, my thesis is finished.
When writing my thesis I did not have many troubles encountered until the results section. I had some troubles with working with Latent Gold, since most of the time all lectures are devoted to SPSS. Some methods and techniques were new for me so it took me some time to figure out how everything worked. In addition, I made use of a new software package, STATA. Now, I can say that I am satisfied with what I achieved with my thesis.
With the results of this thesis marketing managers can make better use of their pricing strategies in order to be preferred by their customer and even potential customers. Based on a segmentation, a supermarket can choose their most profitable customers and apply a pricing strategy on the wishes of these customers.
6 Table of contents Management summary ... 3 Preface ... 5 1. INTRODUCTION ... 7 2. THEORETICAL FRAMEWORK ... 10 2.1 Development of hypotheses ... 10
2.1.1 The effect of pricing strategy on customers’ store choice ... 11
2.1.2 Location of the store ... 11
2.1.3 Basket price ... 12 2.1.4 Price perception ... 13 2.1.5 Price knowledge ... 14 2.1.6 Type of shopper ... 15 2.1.7 Price sensitivity ... 15 2.1.8 Promotion sensitivity ... 16 2.1.9 Basket size ... 17 3. METHODOLOGY ... 18 3.1 Data collection ... 18 3.2 Measurements ... 18 3.2.1 Scale development ... 18
3.2.2 Reliability and validity ... 20
3.3 Method ... 22 3.3.1 Main model ... 22 3.3.2 Full model ... 24 4. RESULTS ... 25 4.1 Data exploration ... 25 4.2 Main model... 26 4.3 Full model... 27 4.4 Model validation ... 32 5. DISCUSSION ... 33
5.1 Summary of hypotheses testing ... 33
5.2 Managerial implications ... 37
5.3 Limitations and further research ... 39
6. REFERENCES ... 42
7. APPENDIX A ... 48
7
1. INTRODUCTION
The operating landscape of groceries has changed a lot nowadays. Groceries must compete with each other on many aspects such as price, service and assortment. Ellickson and Misra (2008) stated that the pricing strategy of retailers is one of the most important aspect to compete on. They provide two main types of pricing strategies: 1) offering low and constant prices for a longer period across a wide range of products, which is called an Every Day Low Pricing strategy, or 2) offering deep and temporary discounts on a smaller set of products, also
known as the Promotional Pricing strategy1. A retailer can generate more profit with an Every
Day Low Pricing (EDLP) strategy instead of maintaining a Promotional Pricing (PROMO) strategy (Lal & Rao, 1997). Some explanations for this effect are that an EDLP store has lower costs and serves customers who have less time better. Another explanation is that if customers have a large shopping basket, an EDLP store offers a greater price advantage than a PROMO store. One advantage of a PROMO store is that if products are offered at a promotional price, the price discount of these products is lower than the normal prices of these products in an EDLP store (Lal & Rao, 1997). This means that if customers only buy promotional products, the overall basket price will be lower at a PROMO store than at an EDLP store. One question that arises after knowing these pricing strategies is what do customers prefer? That is exactly what this study aims to discover.
Prior research (Alba, Broniarczyk, Shimp & Urbany, 1994) showed that customers can accurately distinguish price levels across different stores. Therefore, customers are able to compare product prices across groceries and this makes pricing strategies more important in customers’ store choice decisions. Store choice is not only based on the pricing strategy a grocery maintains (Ailawadi & Keller, 2004), but also on other characteristics of a grocery. For example, important criteria for store choice decisions are the distance to the store (Ailawadi & Keller, 2004), the assortment, service and the in store atmosphere. This study will focus on how pricing strategies will influence customers’ store choice and how some related variables may influence this relationship.
The contribution of this study to the literature is fourfold. First, while most studies have demonstrated the differences between an EDLP strategy and a PROMO strategy (Lal & Rao, 1997; Pechtl, 2004; Darke & Chung, 2005), little is known about the relationship between a grocery’s pricing strategy and customers’ store choice (Bell & Lattin, 1998). For
8 this reason, this study will focus on customers’ preference for a specific pricing format of a store. Groceries must adopt an effective pricing strategy to attract customers, increase shopping frequency and quantity (Blattberg, Eppen & Lieberman, 1981; Krishna, 1992; Lal & Rao, 1997). However, groceries do not know if customers prefer a specific type of pricing strategy. Therefore, it is important to know where customers base their store choice decisions on and which pricing strategy attracts what type of customers.
The second contribution to the literature is to include customers’ price perception and price knowledge in the conceptual framework. There is a lack of understanding how price perception and price knowledge can affect customers’ store choice. Monroe (2002) proposed that ‘‘prices influence choices because they constitute objective indicators of the monetary costs of purchasing’’. Bell and Lattin (1998) also indicate that prices are determinants of customers’ shopping behavior at a particular store and consequently of store selection. Customers’ price perception is determined as the customers’ level of awareness of the price of a product (Dickson & Sawyer, 1990). This will indicate if customers have a positive or negative attitude towards the price level of a store. Price knowledge indicates if customers can distinguish the sales price and the regular price of products (Krishna, Currim & Shoemaker, 1991; Lichtenstein & Bearden, 1989). This way the customer can judge whether a deal is of good value in relation to past offers. These definitions show that if customers have price perception and price knowledge, they are more likely to compare products. This might affect the relationship between a grocery’s pricing strategy and customers’ store choice.
9 different on prices and promotions. In addition, the existing literature misses information about the price sensitivity and promotion sensitivity of customers in relation to preference of a pricing strategy.
The last contribution to the literature is to extend the conceptual model by including the following variables: location of a store, basket price and basket size. The location of a store is one important aspect to base store choice on (Ailawadi & Keller, 2004) as well as the overall basket price (Nielsen, 2011). In general, customers prefer shopping at a grocery that offers the lowest basket price (Lynn, Flynn & Helion, 2013). However, which store do customers prefer if stores are located at different distances from a customer’s home location? This might results in a preference for the store that is the closest to the customer’s home location (Briesch, Chintagunta & Fox, 2009), regardless of the overall basket price. Ailawadi and Keller (2004) argue that customers’ store choice may not only depend on the location anymore, because there are more criteria to based store choice on. One such criterion is the nature of the trip. This variable can indicate if customers are willing to travel more to a store that is located farer away. For example, for small basket shoppers it is very unlikely that they will shop at distant groceries. In contrast to this, large basket shoppers prefer a convenient store that sells everything they need. Therefore, it might be that size of the shopping basket indicates if customers prefer an EDLP or a PROMO store (Bell & Lattin, 1998). This study will examine if customers are willing to travel more to a store of their preferences and if customers base their preference for a store on basket price and basket size.
Based on these contributions to the literature, the research question of this study will be: What is the effect of a grocery’s pricing strategy, location and basket price on customer’s
10 The remainder of this research is organized as follows: the next chapter presents the study’s conceptual framework, which contains existing literature to support the stated hypotheses with an underlying theory. The following chapter outlines the study’s approach, including data collection and method. The next chapter analyzes the data generated and presents the study’s results. The managerial implications and future research directions are presented in the discussion section of this study.
2. THEORETICAL FRAMEWORK 2.1 Development of hypotheses
Based on an extensive literature study, a conceptual model for this study is derived, see figure 1. In this model, the dependent variable is a customers’ store choice and the independent variables are the two pricing strategies (EDLP and PROMO), the store’s location and basket price. The moderators in this conceptual model are price perception, price knowledge, type of shopper, price sensitivity, promotion sensitivity and basket size.
11
2.1.1 The effect of pricing strategy on customers’ store choice
This study assumes that customers’ store choice is dependent on the grocery’s pricing strategy, because Ailawadi and Keller (2004) showed that a grocery’s pricing strategy, either EDLP or PROMO, affects customers’ shopping behavior and store choice. The existing literature suggests that a grocery’s pricing strategy influence customers’ store choice, however the direction and extent of this influence is not clear. Prior research (Omar & Abisoye, 2008) suggests that if customers are certain about their price awareness, customers are more likely to choose for an EDLP pricing strategy. Price awareness is the change in customer’s purchase behavior due to the customer’s focus on low prices. Therefore, if customers are more aware of the prices, they are also more likely to shop at low priced stores. However, if customers are status sensitive they are more likely to choose for a PROMO pricing strategy. Status sensitivity refers to the positive perception of price, where high prices are cues to reflect status. Therefore, status sensitivity will influence purchase behavior, because customers who are sensitive to status will consequently shop at stores that reflect their status. Another reason to base store choice on is showed by Bell and Lattins’ study. Bell and Lattin (1998) showed that the choice for a store with a specific price strategy might depend on the basket size, either a small or a large basket size. Based on prior research, this study assumes that there are different reasons to choose for a store with a specific price format. For this reason, the hypotheses in this study will distinguish the effects of the moderators on each of the two pricing strategies in relation to store choice. The main hypothesis is:
Hypothesis 1: A grocery’s pricing strategy, either EDLP or PROMO, has a positive effect on customers’ store choice.
2.1.2 Location of the store
12 the shopping trip. Small basket shoppers for example, will not travel a large distance to a store, because they have a small shopping list and therefore it will be inconvenient to travel more. However, large basket shoppers prefer a convenient store that sells everything they need. Even though there are arguments in favor for doing groceries at a further located store, this study assumes that in the end customers will prefer a store that is located the closest to their home location. Customers might prefer these stores more because is it less associated with risk and local stores have greater similarity of local customers, with more shared values and norms (Edwards, Kyun Lee & la Ferle, 2009). Based on these arguments the second and third hypotheses of this study are:
Hypothesis 2: Customers prefer the store that is the closest to their home location (minimum distance).
Hypothesis 3a: Customers are less willing to travel more to a store that is further located from a customer’s home location if the shopping goal is a small basket.
Hypothesis 3b: Customers are willing to travel more to a store that is further located from a customer’s home location if the shopping goal is a large basket.
2.1.3 Basket price
One of the most important factors to base store choice on is value for money (Nielsen, 2011). Zeithaml (1988) defines perceived value as ‘‘the consumer’s overall assessment of the utility of a product based on perceptions of what is received and what is given’’. Based on this definition, perceived value can be seen as a trade-off between spending money and getting benefits in return. This study suggests that customers strive to have the highest utility for the lowest value. Lynn et al. (2013) showed that in general customers prefer the alternative with the lowest price. So, retailers must engage in the cherry picking behavior of customers that aims at finding products offered at the lowest price. Retailers must provide customers with benefits such as offering promotions, price discounts or by creating an image of being the cheapest (Nielsen, 2011). Ailawadi and Keller (2004) also showed that the strongest retailers today are Walmart, Target and Aldi. These supermarkets have positioned themselves as a low priced grocery. This makes clear that customers do prefer a store with the lowest basket price. Based on this theory the following hypothesis is formulated:
13
2.1.4 Price perception
Customers’ perceptions are lasting (Gijsbrechts, 1993) and are the key drivers of store choice and spending (van Heerde, Gijsbrecht & Pauwels, 2008). As mentioned before, customers’ price perception is determined as the customers’ level of awareness of the price of a product (Dickson & Sawyer, 1990). In other words, price perception is the customers’ perception of the overall price level of the store. Based on prior research (Ailawadi & Keller, 2004; Feichtinger, Luhmer & Sorger, 1988; Büyükkurt, 1986) it can be argued that customers’ price perceptions do not come ‘out of the blue’. Instead, price perceptions are formed in a learning process of customers’ beliefs about the overall expensiveness of the store, which is based on the actual prices of products and the basket price. As Alba et al. (1994) say: ‘‘customers’ perceptions of store prices change with prior beliefs and information about how frequently a store has a price advantage on a set of products and the magnitude of that price advantage’’. They also find that the frequency of a price advantage mostly determines, beyond customers’ prior beliefs and the degree of a price advantage, customers’ store price perceptions. The study of Dickson and Sawyer (1990) showed that the exact price of a product was more easily perceived by customers who said that they have checked the price while making their choice, than customers who did not (70,7% against 17,3%). These customers were also more likely to indicate if a product was at a special price in comparison to other customers (60,4% against 25,7%). The average percentage error of the price estimates of products was 30 cent above or below the selling price. On average, price conscious customers have more accurate and favorable price perceptions. To link the theory to the context of this study, an EDLP store offers customers every day with low and stable prices, so the price advantage gets bigger at an EDLP store than at a PROMO store. Van Heerde et al. (2008) showed that dropping the prices improves store price perception. Therefore, the following hypotheses are being tested:
Hypothesis 5a: Customers prefer a store with an EDLP pricing strategy if customers have favorable price perception of a specific store.
14
2.1.5 Price knowledge
Price knowledge indicates if customers can distinguish the sales price and the regular price of a product, so that customers can judge whether a deal is of good value in relation to past offers (Krishna et al., 1991; Lichtenstein & Bearden, 1989). Prior research from Dickson and Sawyer (1990) showed that customers may have relatively poor price knowledge, but Monroe and Lee (1999) showed that customers do develop general price perceptions of products in a store. This way they can relatively evaluate the expensiveness of stores. Many studies have indicated that customers’ price knowledge is an important driver for how customers perceive and value prices. In addition, price knowledge also influences customers’ purchase decisions (Binkley & Bejnarowicz, 2003; Dolan, 1995; Mesak & Clelland, 1979; Monroe, 1973; Shapiro, 1968; Turley & Cabaniss, 1995; Vanhuele & Dréze, 2002). This study argues that if customers pay attention to prices, make the effort to compare prices and use this information to make purchase decisions, it must be that prices are important for these customers. Prior research showed that customers do have different levels of price knowledge (Estelami & de Maeyer, 2004; Magi & Julander, 2005). Price conscious customers, the customers that possess more accurate price knowledge, are not willing to pay higher prices for specific products (Link, 1997). This means that the range of acceptable prices for price conscious customers is relatively smaller than for price unconscious customers. In other words, customers who perceive prices more precisely are also the ones who find prices more important. In another study Ofir, Raghubir, Brosh, Monroe and Heiman (2008) argue that knowledgeable customers are more likely to recall product prices as being low if many products are offered at a low price. These customers will consequently perceive the overall price level of the store as low. On the other hand, less knowledgeable customers find it more difficult to recall lower priced products if many products have a low price. For this reason, these customers will perceive the price level of the store as high. Therefore, the more knowledgeable customers are the more certain they are about low priced products. Therefore, if customers have more accurate price knowledge they will prefer an EDLP store and if customers have less accurate price knowledge they will prefer a PROMO store. The proposed hypotheses are:
15 Hypothesis 6b: Customers prefer a store with a PROMO pricing strategy if they have less accurate price knowledge.
2.1.6 Type of shopper
This study will take into account differences between customers with regard to their shopping motives. Two types of customers are defined: hedonic and utilitarian shoppers. Babin et al. (1994) showed that ‘‘hedonic shopping motives are similar to the task orientation of utilitarian shoppers, only the ‘task’ is concerned with hedonic fulfillment, such as experience, fun, amusement, fantasy and sensory stimulation’’. One motive for hedonic shoppers is value shopping, which refers to shopping for sales, looking for discounts and hunting for bargains (Arnold & Reynolds, 2003). Hedonic value shoppers increase their sensory involvement and excitement through bargain perceptions. This is explained by the fact that customers are competitive in nature and seek for success and admiration to increase their self-esteem (Babin et al., 1994). Hedonic and utilitarian shoppers prefer different store formats (Westbrook & Black, 1985; Rintamäki, Kuusela & Mitronen, 2007). For utilitarian shoppers, store location and parking facilities might be more important than the atmosphere in the store, which hedonic shoppers will value. Hedonic shoppers look at the benefits they get from shopping, while utilitarian shoppers are looking for an efficient shopping process (Babin et al., 1994). In line with Babin et al. (1994), Teller et al. (2008) showed that hedonic shoppers are more tempted by prices. They enjoy the search of reduced prices and the fact of an accomplishment of saving money (Pechtl, 2004). PROMO stores offer hedonic shoppers with this benefit due to their temporary price discounts. On the other hand, utilitarian shoppers value store cleanliness, a moderated crowd of shoppers in the store and strive to minimize time and effort (Pechtl, 2004). Utilitarian shoppers do not want to compare prices across different stores, so they prefer one store to do their groceries. An EDLP store maintains constant low prices and as a result, there is no need for the utilitarian shoppers to look further. Based on these arguments the following hypotheses are developed:
Hypothesis 7a: Customers prefer a store with an EDLP pricing strategy if they are utilitarian shoppers.
Hypothesis 7b: Customers prefer a store with a PROMO pricing strategy if they are hedonic shoppers.
2.1.7 Price sensitivity
16 customers is lower and consequently the spending of customers is lower. This leads to a higher level of price awareness and a search for lower prices (Estelami, Lehmann & Holden, 2001). This means that customers’ behavior has changed since the economic crisis, in a way that customers are searching for information about the most suitable prices to pay for certain products. Customers are looking for price deals (Quelch, 2008) and switch more often to lower priced products (Lamey, Deleersnyder, Dekimpe & Steenkamp, 2007). The economic crisis also resulted in the fact that traditional groceries are cutting back on promotional discounts and moving towards the Every Day Low Pricing strategy (Bailey, 2008). For example, the American retail chain Walmart and the Dutch retail chain Jumbo both maintain an EDLP pricing strategy. With the entrance of an EDLP pricing strategy customers’ price perception and price knowledge has changed. Adamy (2005) suggests that ‘‘customers are conditioned to expect inexpensive goods every day and will respond to retailers that offer them these inexpensive goods’’. If customers are more price sensitive, they will also be more value-conscious (Garretson & Burton, 2003; Tietje, 2002) and will respond more favorably to the direct price advantage that an EDLP store offers (Bailey, 2008). Therefore the following two hypotheses are developed:
Hypothesis 8a: Customers prefer a store with an EDLP pricing strategy if customers are more price sensitive.
Hypothesis 8b: Customers prefer a store with a PROMO pricing strategy if customers are less price sensitive.
2.1.8 Promotion sensitivity
17 customers do. High sale-prone customers are more price consciousness and price sensitive than low sale-prone customers are. This way, high sale-prone customers spend more time on shopping in order to find the right deal and save money. Consequently, the following hypotheses are derived:
Hypothesis 9a: Customers prefer a store with an EDLP pricing strategy if customers are less promotion sensitive.
Hypothesis 9b: Customers prefer a store with a PROMO pricing strategy if customers are more promotion sensitive.
2.1.9 Basket size
18 Hypothesis 10a: Customers prefer a store with an EDLP pricing strategy if their shopping goal is a large shopping basket.
Hypothesis 10b: Customers prefer a store with a PROMO pricing strategy if their shopping goal is a small shopping basket.
3. METHODOLOGY 3.1 Data collection
To test the proposed hypotheses an online questionnaire was conducted. Qualtrics is the program used to make the survey. The respondents were randomly assigned to fill in the questionnaire. Randomly assignment means that every respondent has an equal chance of filling in the questionnaire (Aronson, Wilson & Brewer, 1998). Randomization is an equalizer, because differences between individual respondents are unlikely to influence the results. The questionnaire was distributed during three weeks, namely week 17, 18 and 19. The respondents were invited to fill in the questionnaire via email. Besides, to make the sample more representative the survey was also send to bigger households and families, via the acquaintances and colleagues of the researcher’s parents. The questionnaire was administered in Dutch to overcome possible language-based bias. From the 183 respondents who filled in the questionnaire 178 were used for analyzing purposes.
3.2 Measurements
3.2.1 Scale development
Whenever possible this study uses existing scale constructs from prior studies, in order to assess its reliability and validity. Some of these measurements are adapted to the study’s context. An eight-part questionnaire is used to collect data, see appendix A table A1 and A2. The respondents are informed how to answer the questions at each section.
Part one of the questionnaire covers general information about the demographic characteristics of the respondents, such as their age, education, residence and income.
19 three different groceries in the Netherlands. In the Netherlands, Jumbo maintains an EDLP pricing strategy and Albert Heijn is a PROMO store. The third grocery in this study is Deen that also maintains a PROMO pricing strategy, but has lower overall prices than Albert Heijn. These three groceries are the attribute levels of the attribute supermarket and represent the pricing strategy variable. The study will focus on the store Jumbo and Albert Heijn, since these supermarkets are well known, really promote their pricing strategy and these groceries are located throughout The Netherlands so that the findings are generalizable. The next attribute indicates the basket price that differs between small and large shopping baskets. GfK (2015) shows that the average value of a shopping basket in week 16 was €21,05. The small basket prices are therefore either €18,-, €21,- or €24,-. For the large basket sizes the prices are €60,-, €65,- or €70,-. The attribute location is measured as the distance between a customer’s home and the store. This could be either 1 kilometer, 3 kilometers or 5 kilometers distance. Different combinations of these attribute levels are shown to the respondents in order to discover whether customers prefer an EDLP pricing strategy or a PROMO pricing strategy. The conjoint analysis also indicates if customers are willing to pay more for a shopping basket and if they are willing to travel more to a specific store. In total, 12 different choice sets with each 3 alternatives are showed to the respondents. The selection of alternative options in the conjoint analysis is based on a program called ‘my preference lab’. My preference lab takes into account the efficient design criteria; balancy and orthogonality. Before the choices are showed to the respondents, they are informed about the purpose of the shopping trip, either a small basket (fill-in-trip) or a large basket (major trip). Second, this introduction will also explain that each shopping basket maintains the same a-branded products and that the differences in price are due to the groceries pricing strategy. In addition, the respondents are asked to assume that the three supermarkets are the same on every other aspect, for example assortment, service or parking facilities. This introduction must take away any doubts of the respondents regarding the different size of the shopping basket or prices.
The third part of the questionnaire covers the measurement of the variable price perception. Lourenço (2010) measured price perception by a scale ranging from 1 = most favorable to 9 = least favorable. However to make the scales used in the questionnaire consistent, Lourenço’s scale is adapted to a 7-point Likert scale, which ranges from 1 = most favorable to 7 = least favorable. The respondents must indicate their favorability towards the overall prices of Jumbo, Albert Heijn and Deen.
20 (1996) and Putrevu and Ratchford (1997) is used. The questionnaire contains the following items: ‘I read price tags of the grocery products I buy’, ‘I check the prices of the grocery products that I purchase’ and lastly ‘I know which stores have the best prices’. These items are measured by a 7-point Likert scale, ranging from 1 = totally agree to 7 = totally disagree.
The fifth part of the questionnaire indicates what type of shoppers the respondents are. A scale of Babin et al. (1994) is used to measure this variable. For the purpose of this study the scale is adapted in the sense that the items were transformed in the present tense and some items were left out in the questionnaire due to the length of the questionnaire. The items are measured on a 7-point Likert scale ranging from 1 = totally agree to 7 = totally disagree. The items used are: ‘I buy what I really need’, ‘I am disappointed when I need to go to another store(s) to complete my shopping’, ‘A shopping trip is truly a joy’ and ‘During a shopping trip, I feel excitement when hunting for bargains’.
The next session in the questionnaire will covers items developed by Muncy (1996), about the price sensitivity of the respondents. The scale is adapted to the context of this study. The items are: ‘I am willing to pay more to buy my regular brands than buy another brand’, ‘I generally buy the least expensive brand I can find’ and ‘If a brand other than the one I usually purchase is on sale, I will probably buy it’. These items are measured on a 7-point Likert scale ranging from 1 = totally agree to 7 = totally disagree.
Measuring the promotion sensitivity of the respondents is the next section of the questionnaire. For this variable, a scale of Lichtenstein et al. (1993) is used. The following items are shown to the respondents: ‘If a product is on sale, that can be a reason for me to buy it’, ‘When I buy a brand that’s on sale, I feel that I am getting a good deal’ and ‘I have favorite brands, but most of the time I buy the brand that is on sale’. These items are also measured on a 7-point Likert scale ranging from 1 = totally agree to 7 = totally disagree.
The last section of the questionnaire will ask questions about the respondents’ shopping behavior to get some more insights about the shopping behavior of the respondents. Such as: ‘Are you the one at home that usually does the groceries?’, ‘At which supermarket have you done groceries once or more in the past year?’, ‘At which supermarket do you shop the most (in terms of expenditures)?’ and lastly ‘Which supermarket has the second place when it comes to shopping (in terms of expenditures)?’.
3.2.2 Reliability and validity
21
Scale Item(s) Cronbach’s alpha Factor loadings
1 0,787 2 0,765 3 0,887 Utilitarian shopper 2 - 0,971 1 0,909 2 0,681 2 0,839 3 0,808 1 0,907 2 0,864 Price sensitivity 0,627 Promotion sensitivity 0,805 TABLE 1
Testing the construct’s reliability and validity
Price knowlegde 0,821
Hedonic shopper 0,632
alpha measures the internal consistency of each variable and consequently the reliability of the constructs. The results of the Cronbach’s alpha test are shown in table 1. Table 1 shows that each variable has a moderate till high degree of reliability following Hair, Anderson, Tatham and Black (1988) who state that the Cronbach’s alpha must exceed 0,6. The variable price perception is not included in the reliability test, since only one item measures the price perception of the three different stores. The variable price sensitivity had a negative Cronbach’s alpha with all three items included. Therefore the first item ‘I am willing to pay more to buy my regular brands than buy another brand’ of the variable price sensitivity is not used in the analysis in order to have a positive and moderate degree of reliability. From the correlation matrix (r = ,267 and r = ,274) and the Cronbach’s alpha (,613) it can also be concluded that item three of the variable promotion sensitivity ‘I have favorite brands, but most of the time I buy the brand that is on sale’ does not show high internal consistency with the other two items of that variable. As a result, this item is not used for the analysis. The items of the variable utilitarian shopper showed a negative and very low Cronbach’s alpha (-,178) and correlation (r = -,083). It can be concluded that these two items do not share any internal consistency. The solution for this problem is to keep working with one item for the variable utilitarian shopper for further analysis. The item ‘I buy what I really need’ of the variable utilitarian shopper is not used in the analysis.
22
3.3 Method
3.3.1 Main model
The test whether customers prefer an EDLP store above a PROMO store a conjoint analysis is conducted. By calculating the utilities of each attribute, it can be directly seen if Jumbo is preferred above Albert Heijn, which indicates that customer prefer an EDLP store above a PROMO store. The program Latent Gold is used to perform a conjoint analysis. In order to make use of Latent Gold the SPSS dataset that Qualtrics generated from the questionnaire was transformed. The selection of the alternatives is dummy coded, where 1 indicates that a specific alternative is chosen and 0 indicates that a specific alternative is not chosen. This way the dependent variable is a binary choice. This study makes a distinction between two datasets and estimate one model for the small basket size dataset and one model for large basket size dataset. The output of Latent Gold provides parameter estimates where the utility of each alternative from a choice set can be calculated. These utilities can be translated into probabilities. Ultimately, these probabilities are linked to customers’ store choice. In general, customers will prefer the alternative that has the highest utility and probability.
23 significant. For example, model 2 from the small basket size dataset does not outperform model 1 from the small basket size dataset because the Chi-Square statistic is lower than the critical value of 3,84 at p = ,05. Looking at the large basket size dataset, model 3 does not outperform model 2 at p = ,05, however at p < ,10 model 3 does outperform model 2. For the purpose of this study, this study is going to work with model 3 for both the small basket size and the large basket size dataset. Model 3 is preferred since this model has the least parameters while the fit is the same compared to the other models. Model 3 is therefore the most parsimonious model of the three models showed, which is important since simplicity is an important criterion in building the structure of a model (Little, 1970). Model 3 also significantly outperforms the null model and the other models. By calculating the utilities of each alternative and consequently the probabilities of preferring an alternative, the actual choices made by the respondents can be compared to the predicted choice probabilities. The mean absolute error (MAE) is calculated to demonstrate the error between the actual choices and predicted choice probabilities. Model 3 for the small basket size dataset has a MAE of 15,78% and model 3 for the large basket size dataset has a MAE of 14,48%. So, on average the prediction deviate about 14,48% for the small basket size dataset and 15,78% for the large basket size dataset from the observed values. This indicates a moderate degree of predictive
validity.
Model 1 Model 2 Model 3
All attributes nominal Supermarket and basket price nominal. Location numeric
Supermarket nominal. Basket price and location numeric
Log-likelihood -805,5511 -805,5537 -808,6128 Number of parameters 6 5 4 R2 0,31 0,31 0,31 R2adjusted 0,31 0,31 0,31 Chi-square 645,53* 0,01 6,12* Hit rate 65,14% 65,14% 65,14% Log-likelihood -884,1649 -890,9429 -892,4319 Number of parameters 6 5 4 R2 0,25 0,24 0,24 R2adjusted 0,24 0,24 0,24 Chi-square 144,58* 13,56* 2,98** Hit rate 58,05% 57,12% 57,58% TABLE 2
Large basket size
* p < ,05 ** p < ,10 The degrees of freedom is 1
Small basket size
24
3.3.2 Full model
Based on the Cronbach’s alpha reliability test and the factor analysis this study has less variables to work with, which benefits the analyses. After estimating the main model and testing hypotheses 1, 2 and 4, the full models, with interaction effects, are going to be estimated in order to test the remaining hypotheses.
To test hypotheses 3 a t-test is conducted that shows that the parameter estimate of the variable location is significantly different between the small and the large basket size datasets. When testing differences in parameters across two unrelated and different datasets, the parameter estimates and the variance of these parameters are needed in order to perform a t-test2.
Hypotheses 5 are tested with the software program STATA, because SPSS and Latent Gold are not able to test hypotheses 5. The hypotheses are tested via a conditional logit regression at store level, because the value of the variable price perception varies across the three different stores. The alternative selection is the dependent variable of this test, the three items of the variable price perception are the explanatory variables and the variable supermarket was the group variable in this regression. A group variable is required in a conditional logit model to specify a variable that identifies the groups. This way the data is grouped so that the parameters estimates of the explanatory variables are group specific.
To test hypotheses 6 till 9 new variables are computed in order to test the moderator effects. Concerning the interaction effect, the variable supermarket, which represents the pricing strategy variable, is recoded with effect coding. The reference level is Deen. New interaction variables are made in SPSS based on multiplying the effect coded supermarket variables with each moderator variable. In Latent Gold these interaction terms are added to the main model to test the remaining hypotheses. To test for multicollinearity among the independent variables, which Latent Gold cannot reveal, a linear regression in SPSS is performed. The results of this test show high multicollinearity between the independent variables (VIF > 5). To reduce multicollinearity the full model is estimated by including only the main effects (supermarket, basket price and location) and one interaction effect only
2 The calcutation of the t-test is as follows: t =
. Where b1 is the parameter estimate of the
25 instead of adding multiple interaction effects into one model. This way multiple models are going to be estimated.
To test the last two hypotheses, the t-test is used again in order to see whether the parameters estimates of the supermarkets in the small and large basket size dataset are significantly different from each other. The t-test is performed the same way as mentioned before when testing hypotheses 3.
4. RESULTS
This chapter will test the hypotheses. This section starts with the exploration of the data. When the data is explored, the main model is estimated and hypotheses 1, 2 and 3 are tested. Next, the interaction effects are added to the main model and the remaining hypotheses are tested.
4.1 Data exploration
26
4.2 Main model
Table 3 indicates the estimated parameters of the main model, where the variable supermarket is nominal coded and the variables basket price and location are numeric coded. From the choices made in the small basket size dataset it can be concluded that the majority of the sample preferred in 4 of the 6 choice sets, the alternative with the minimum distance to the store. From the choices made in the large basket size dataset it can be concluded that in 4 of the 6 choice sets, the alternative with the minimum distance to the store and the cheapest shopping basket price is chosen by the majority of the sample. This indicates that when the nature of the shopping trip is a large basket size, more variables can have an impact on customers store choice. Table 3 shows that location is indeed the most important attribute, 52,36%, to base store choice on when it comes to small basket size trips. However, when the purpose of the shopping trip is a large basket size, the price of the overall basket is the most important attribute, 50,41%, to base store choice on.
Hypothesis 1 is first tested. This hypothesis suggests that a grocery’s pricing strategy has a positive effect on customers’ store choice. When looking at the small basket size dataset it can be concluded that supermarket does influence customers store choice (p = ,000). The same holds for the large basket size dataset (p = ,000). The parameters in both datasets show that Jumbo (-0,0842 and -0,1256) and Albert Heijn (-0,2061 and -0,1236) have a negative sign and utility, while Deen (0,2903 and 0,2602) has a positive sign and the highest utility of the three supermarkets. Deen maintains a PROMO strategy and therefore it is suggested that a PROMO strategy has a positive effect on customers’ store choice, which is in line with
Attribute Attribute levels if nominal coded Wald statistic ß’ estimates Importance of attribute
Jumbo -0,0842 Albert Heijn -0,2061 Deen 0,2903 Basket price 184,7699* -0,2597 36,13% Location 349,7762* -0,5645 52,36% Jumbo -0,1256 Albert Heijn -0,1346 Deen 0,2602 Basket price 239,5138* -0,2218 50,41% Location 189,9025* -0,4467 40,61% TABLE 3
Outcome of main model conjoint analysis
Small basket size
Supermarket 29,2680* 11,51%
Large basket size
Supermarket 18,8113* 8,97%
27 hypothesis 1. However, Albert Heijn also maintains a PROMO strategy but the parameter estimate of Albert Heijn shows a negative effect on customers’ store choice. Jumbo maintains an EDLP strategy, however this supermarket also shows has a negative effect on customers’ store choice. The utilities of Albert Heijn and Jumbo contradicts with hypothesis 1, therefore it is concluded that hypothesis 1 is only partially supported. A reason why the supermarket Deen only has a positive utility might be due to a brand effect. This means that the brand of Deen maybe generated the positive utility effect instead of the pricing strategy of Deen.
The second hypothesis tests if store location influences customers’ store choice. Table 3 already reported that location is the most important attribute in the small basket size dataset and second most important attribute in the large basket size dataset. When looking at the Wald statistic and the p-value, both the small basket size and the large basket size dataset show a significant effect (p = ,000 and p = ,000). In both the small basket size dataset and the large basket size dataset the parameter estimate is negative (-0,5645 and -0,4467). It can be concluded that better locations leads to a higher utility. A better location in this study means that the location is closer to a customer’s home location, because when the location is further away from a customer’s home location the utility of that supermarket decreases. An increase in location with one unit will decrease the utility of a store with -0,5645 extra for the small basket size dataset and with -0,4467 extra for the large basket size dataset. To summarize, hypothesis 2 is supported by both datasets.
Hypothesis 4 suggests that customers will prefer the alternative with the lowest basket price, if the baskets are equally filled with the same a-branded products. The test results show that in both cases, the small and the large basket size dataset, basket price influences customers’ store choice significantly (p = ,000). The parameter estimate for this attribute is negative in both datasets, indicating that an increase in basket price would decrease the utility. An increase in basket price with one unit will decrease the utility of a store with -0,2597 extra for the small basket size dataset and with -0,2218 extra for the large basket size dataset. Accordingly, hypothesis 4 is supported by both datasets.
4.3 Full model
28 in more than one full model only. This way the VIF scores were reduced to acceptable VIF scores. Table 5 shows that the variable basket price and location are significant in every model, but the variable supermarket is not significant in every model. In this subparagraph, only the significant interaction effects are interpreted.
Hypotheses 3 state that customers are more willing to travel larger distances to a store when their shopping goal is a large basket, compared to the small basket size shopper who is less willing to travel larger distances to a store. Based on the main model parameter estimates of the variable location and on a t-test these hypotheses are going to be tested. If the outcome of the test shows a t-value larger than 1,96 the difference between the parameter estimates between the two datasets is significant at p < ,05. The t-test shows a t-value of 2,64. This indicates that there are significant differences between the parameter estimates of the variable location between the small and large basket size dataset. Table 3 shows that the utility for location for the small basket size dataset is more negative (-0,5645) than the utility for location for the large basket size dataset (-0,4467). Based on the t-test and the utilities of the variable location hypothesis 3a and 3b are supported.
Hypotheses 5 state that customers with favorable price perceptions prefer an EDLP store and customers with unfavorable price perceptions prefer a PROMO store. In order to test this hypothesis a conditional logit regression in STATA is performed. The outcome of the test shows that for the supermarket Jumbo the price perception of Jumbo is a negative coefficient for both small and large basket size dataset (-0,1318 and -0,2706). For supermarket Albert Heijn, the price perception of Albert Heijn is a negative coefficient for both small and large basket size dataset, however these coefficients are not significant and cannot be interpreted. The price perception of Deen also shows a negative coefficient in both datasets (0,1895 and -0,2694). These results show that when the price perception of a specific supermarket becomes more unfavorable, the probability of choosing that specific supermarket decreases. In other words, if customers possess more unfavorable price perception of specific store it is less likely that they will choose that supermarket. The stated hypothesis 5a and 5b are supported by both the small and large basket size datasets.
29 supermarket Albert Heijn increases with 0,1326 if the price knowledge of customers increases with one unit. The utility for the supermarket Albert Heijn in that situation becomes -0,4954 (-0,6280 + 0,1326). For the large basket size dataset, the interaction effect is not significant. To conclude, hypotheses 6a and 6b are suggesting the opposite of what the results show, accordingly hypothesis 6a and 6b are not supported.
Hypotheses 7 concern the interaction effect of type of shopper on a grocery’s pricing strategy and customers’ store choice. The variable type of shopper consists of two separate aspects, utilitarian shopper and hedonic shopper. From the analysis, see table 5, it can be argued that for the small basket size dataset type of shopper has no significant effect on the relationship between pricing strategy and customers’ store choice, since both utilitarian and hedonic variables are not significant. However, from the large basket size dataset the variable utilitarian shopper is significant for supermarket dummy 2 (p = ,001). Albert Heijn is the second supermarket dummy, so this means that utilitarian shopper type significantly moderates the effect of the pricing strategy of Albert Heijn on customers’ store choice. When a customer is a utilitarian shopper, the utility of the pricing strategy of Albert Heijn increases with 0,1159 for every unit increase in utilitarian shopper type. Therefore the utility of Albert
Attribute Attribute levels if nominal coded Wald statistic ß’ estimates
Jumbo 0,0337
Albert Heijn -0,6280
Deen 0,5943
Basket price 186,1410* -0,2626
Location 350,8817* -0,5698
Supermarketdummy1 x Price knowledge 0,9200 -0,0357
Supermarketdummy2 x Price knowledge 11,2900* 0,1326
Jumbo 0,0329
Albert Heijn -0,2516
Deen 0,2188
Basket price 239,5283* -0,2219
Location 189,7882* -0,4469
Supermarketdummy1 x Price knowledge 1,8451 -0,0509
Supermarketdummy2 x Price knowledge 1,2327 0,0373
*p < ,05
** p < ,10 Supermarketdummy 1 = Jumbo
Supermarketdummy 2 = Albert Heijn
Supermarket 4,9332**
TABLE 4 Testing hypotheses 5
Small basket size
Supermarket 25,7750*
30 Heijn become less negative -0,2755 (-0,3914 + 0,1159). However, these results are not in line with the stated hypotheses and consequently hypothesis 7a and 7b are rejected.
Price sensitivity became more important during the economic crisis. Therefore, it is needed to test whether price sensitivity has an effect on customers’ preferences for a specific store. The results for the small basket size dataset show no significant results, see table 5. Nonetheless, in the large basket size dataset price sensitivity has a significant moderator effect on supermarket dummy 2, Albert Heijn (p = ,040). When the price sensitivity of customers increases with one unit, the utility of Albert Heijn decreases with -0,0778. These results show that when customers get more price sensitive, they rather do not prefer a PROMO store. This result is in line with hypothesis 8b. Hypothesis 8b is only significant for the large basket size dataset, therefore hypothesis 8b is only partially supported. Hypothesis 8a remains unsupported.
Hypotheses 9 test if customers have different preferences for stores if they are promotion sensitive or not. According to the results of the analysis, see table 5, in both small and large basket size datasets there is no significant interaction effect. Hypothesis 9a and 9b are unsupported.
Hypo 1: Hypo 1: Hypo 1: Hypo 4: Hypo 2: Hypo 6: Hypo 6: Hypo 7: Hypo 7: Hypo 7: Hypo 7: Hypo 8: Hypo 8: Hypo 9: Hypo 9:
Attribute Jumbo Albert
Heijn Deen Basket price Location
Supdum1 x PK Supdum2 x PK Supdum1 x UT Supdum2 x UT Supdum1 x HE Supdum2 x HE Supdum1 x PS Supdum2 x PS Supdum1 x PR Supdum2 x PR Log-likelihood R 2 R2
adjusted Hit rate MAE
Main model -0,0842* -0,2061* 0,2903* -0,2597* -0,5645* -808,6128 0,31 0,31 65,14% 15,78% Model 1 0,0337* -0,6280* 0,5943* -0,2626* -0,5698* -0,0357 0,1326* -802,6602 0,32 0,31 64,68% 24,53% Model 2 -0,0193** -0,1949** 0,2142** -0,2598* -0,5649* -0,0301 -0,0050 -808,1367 0,31 0,31 65,14% 16,01% Model 3 -0,2222* -0,2065* 0,4288* -0,2599* -0,5650* 0,0383 0,0002 -807,9108 0,31 0,31 65,14% 17,28% Model 4 -0,0712 -0,2343 0,3055 -0,2597* -0,5646* -0,0032 0,0069 -808,6128 0,31 0,31 65,14% 15,75% Model 5 -0,0488 -0,2196 0,2684 -0,2597* -0,5646* -0,0151 -0,0057 -808,5796 0,31 0,31 65,14% 15,80% Main model -0,1256* -0,1346* 0,2602* -0,2218* -0,4467* -892,4319 0,24 0,24 57,58% 14,48% Model 1 0,0329** -0,2516** 0,2188** -0,2219* -0,4469* -0,0509 0,0373 -891,3293 0,24 0,24 58,52% 17,19% Model 2 -0,0316* -0,3914* 0,4230* -0,2224* -0,4484* -0,0422 0,1159* -886,6721 0,24 0,24 57,40% 20,19% Model 3 -0,0031 -0,2319 0,2350 -0,2218* -0,4468* -0,0342 0,0271 -891,9160 0,24 0,23 57,21% 16,28% Model 4 0,0775 0,1830 -0,2605 -0,2237* -0,4508* -0,0494 -0,0778* -887,7109 0,24 0,24 60,01% 27,02% Model 5 -0,1146** -0,2133** 0,3279** -0,2218* -0,4468* -0,0046 0,0333 -892,2286 0,24 0,23 57,49% 16,05% TABLE 5 Outcome of the main model and full models
Model fit
* p < ,05
** p < ,10 Supdum1 = Supermarket dummy 1 Jumbo
4.4 Model validation
The models build in this study are used for predictive purposes, namely to explain customers’ store choice. In order to conclude that the models are accurate and consequently useful, the predictive validity of the main models is tested. To validate the models a validation sample of 5%, 18 respondents, is randomly selected from both datasets. A random number generator is used to select the respondents that will belong to the validation sample. It is expected that the estimation sample has the same characteristics as the validation sample (Leeflang, Wieringa, Bijmolt & Pauwels, 2015). The validation tests are based on the main models of the small and large basket size datasets. Two different validation tests are conducted. The first validation method shows the MAE between the actual choices of the respondents in the validation sample and the models’ predicted choices based on the parameters estimates of the models. The second validation method provides the MAE between the actual choices of the respondents in the validation sample and the models’ predicted choices based on the parameter estimates of the models. What is different from the first validation method is that the second validation method accounts for no choices made by the respondents in the validation sample. One choice set of each respondent in de validation sample is changed in a way that there was no selection made by the respondent for that choice set, indicated by 0. Latent Gold will take into account these empty choice sets and estimate new parameters for each variable. These parameter estimates are used to calculate the utility of each alternative in every choice set and the probability of choosing an alternative in every choice set, in order to compare these predicted probabilities with the actual choices of the respondents in the validation sample. The following formulas are used for these calculations. Formula 1 shows the utility for customer n for alternative i. Formula 2 shows the probability that alternative i is chosen from choice set J:
33 In order to see the true predictive power of the models a second validation test is done. For the respondents in the validation sample, one choice selection is made empty. When estimating the main model again with this new dataset, new parameters are estimated by Latent Gold. Based on these parameter estimates the predicted probabilities of choosing an alternative are calculated. These predicted probabilities are compared with the actual choice frequencies of the validation sample. The results of the validation show a MAE of 24,22% for the small basket size dataset and 34,95% for the large basket size dataset. This validation method also shows that the predictive validity of the main models is not very strong.
To summarize, the statistical accuracy and quality of the main models is low. The difference between the predictive choices and the actual choices is large. Due to limited time, the models cannot be adjusted in order to have a higher degree of predictive validity.
5. DISCUSSION
5.1 Summary of hypotheses testing
This study investigates what the effect of a grocery’s pricing strategy, basket price and location on customers’ store choice. Some moderator variables are added to conceptual model to test whether the preference for a specific pricing strategy might depend on other variables. Table 6 reports a summary of the results of the tested hypotheses. In this chapter the most important findings, contributions, implications and limitations are discussed.
Variable Hypothesis Conclusion
Pricing strategy H1 Partially supported
Location H2 Supported
Location x Basket size H3a H3b
Supported Supported
Basket price H4a Supported
Pricing strategy x Price perception H5a H5b
Unsupported Unsupported
Pricing strategy x Price knowledge H6a H6b
Unsupported Unsupported
Pricing strategy x Type of shopper H7a H7b
Unsupported Unsupported
Pricing strategy x Price sensitivity H8a H8b
Unsupported Partially supported
Pricing strategy x Promotion sensitivity H9a H9b
Unsupported Unsupported
Pricing strategy x Basket size H10a H10b
Unsupported Unsupported
TABLE 6
34 The main question of this study is: What is the effect of a grocery’s pricing strategy,
location and basket price on customers’ store choice? By studying a grocery’s pricing
strategy the results show that a grocery’s pricing strategy does have an impact on customers’ store choice. The exposure of a PROMO store has a positive effect on customers’ store choice. Managers need to understand the role of the pricing strategy in customers’ store choice. However, the exposure to an EDLP store is negative and not significant, which might indicate that customers would not prefer an EDLP store. Bailey (2008) showed that store loyalty, price sensitivity and income could influence a customers’ response to an EDLP pricing strategy. Bailey showed that when customers are loyal to a specific store, customers are more likely to respond negatively to information of competitors’ stores. Since 51% of the respondents of the sample perceive Deen as their primary supermarket and only 8% of the respondents of the sample perceive Jumbo as their primary supermarket, loyalty could be of influence in this study. In addition, customers with a higher income do not specifically need the price advantage that an EDLP store generates and are less price sensitive. Therefore, this target group might be less likely to do groceries at an EDLP store (Bailey, 2008). The contribution of the finding that a PROMO store positively affects customers’ store choice helps the understanding of customers’ preferences for a specific store. Supermarket managers can generate more traffic to the store if they know which customers prefer a PROMO store. The results of this study show that customer who are less price sensitive are more likely to do their groceries at a PROMO store.
In order to answer the main question completely the effects of location and basket price are discussed next. This study explicitly mentioned location and basket price as the main drivers for customers’ store choice and the expected utility is a negative parameter estimate for both variables. From prior research, it is known that the location of the store is an important aspect to base store choice on (Ailawadi & Keller, 2004) and the same holds for basket price (Nielsen, 2011). This study shows that customers prefer the grocery that is located the closest to the customer’s home location. In addition to this, when a grocery offers the lowest basket price customers will prefer the grocery even more. Therefore, managers must clearly pay attention to the location of the store and the pricing strategy of the store in order to offer customers with the lowest basket price.
The sub questions of this study are discussed next, starting with the first sub question:
How does price perception influence the relationship between a grocery’s pricing strategy and customers’ store choice? Price perception significantly influences the preference for a
35 perception of a store becomes more unfavorable and negative, the probability of preferring that store above other stores decreases. Supermarket managers must keep in mind that price perception is an important aspect to base store choice on and that customers perceptions are lasting (Gijsbrechts, 1993). Managers must try to create favorable price perception, such that customers have a positive price awareness of a specific store.
The second sub question is: How does price knowledge influence the relationship
between a grocery’s pricing strategy and customers’ store choice? The results show that price
knowledge does not significantly contribute to this study. Prior research already suggested that customers have poor price knowledge (Dickson & Sawyer, 1990). Munnukka (2008) explains this by stating the following: ‘‘a high level of variability in prices may lead to low quality knowledge’’. The variability in prices comes from the constant change in product differentiation and the actual product quality variations of groceries (Estelami, 1998). Vanhuele and Dreze (2002) also showed that price knowledge dependents on several factors such as the number of brands in a product category, the frequency of promotions, the difference in price between the highest priced product and the lowest priced product and it might even depend on the pricing strategy of the grocery. If the influences of these factors are low customers can have accurate price knowledge. However, nowadays supermarket managers look very often at the prices and promotional activities of their competitors and often respond to them (Levy, Dutta, Bergen & Venable, 1998; Steenkamp, Nijs, Hanssens and Dekimpe, 2005). For example, look at the supermarket Jumbo who advertises a lot. In their ads, Jumbo states that if customers find a product priced lower somewhere else than at Jumbo, Jumbo will reduce the price of that product immediately. Therefore, there is more variability in the prices of groceries so that the price knowledge of customers become less accurate or even inaccurate. These arguments show explanations why hypotheses 6 are not significant.
The third sub question is: How does type of shopper influence the relationship between
a grocery’s pricing strategy and customers’ store choice? Hypotheses 7 are not significant
36 Sixth, the results of the study show that less price sensitive customers are more likely to choose a PROMO store above an EDLP store. Therefore, the fourth sub question, how does
price sensitivity influence the relationship between a grocery’s pricing strategy and customers’ store choice, is answered this way. This finding is important and contributes to the
literature because if customers are less price sensitive, they are more willing to pay for a specific product, which can generate more profit for a grocery. It is important that groceries know how to appeal this target group. Prior research (Munnukka, 2008; Degeratu, Rangaswamy & Wu, 2000) shows that price sensitivity is influenced by the customers’ demographic characteristics such as gender, age and income. For example, employed adults have less time, are more likely to shop at one store and are less price sensitive due to a higher overall income. On the other hand, hypothesis 8a was rejected. More price sensitive customers do not prefer an EDLP store. An explanation why hypothesis 8a is rejected might be that customers who are price sensitive are more likely to be influenced by price promotions, coupons and cheaper private brands. Because an EDLP store offers low regular prices, price promotions and coupons are not frequently used in these groceries. Price sensitive customers are therefore more likely to shop at a so called ‘no frills store’ such as Aldi or Lidl. These stores offer their customers lower overall prices than a PROMO or an EDLP store, because these stores do not sell A-branded products.
Krishna et al. (1991) suggested that many customers have accuracy about promotional activities of groceries. These customers know the timing of the deals and the sales prices. However, this accuracy is higher for customers who have a larger household, for those who purchase specific package sizes more frequently and for those who read weekly advertisements for products on sale. This study did not distinguish customers who are promotional sensitive and customers who are less promotional sensitive and their preference for store choice. Related to this, the fifth sub question needs to be answered: How does
promotion sensitivity influence the relationship between a grocery’s pricing strategy and customers’ store choice? Promotions try to induce trail of new products, selling more
37 money and effort associated to the change in service providers’’. The hypotheses 9 are not significant maybe due to loyalty to specific products and high perceived switching costs.
The sixth sub question is: How does basket size influence the relationship between a
grocery’s pricing strategy and customers’ store choice? The last hypotheses, 10a and 10b, are
also rejected. The nature of the shopping trip, either a small or a large basket size, does not influence the choice of a specific pricing strategy of a grocery. This might be because the distinction between small and large basket size shoppers is not clear, since some customers show unusual shopping behavior (Desai & Talukdar, 2003). For example, there are customers that buy many products on a given shopping trip, but do their groceries more frequently per week instead of once. Another example, some customers purchase a few products on a given shopping trip and make very few shopping trips to the grocery. This unusual shopping behavior makes the distinction between the fill-in trips (small basket size) and the major trips (large basket size) unclear. In addition, if customers know the price advantage that an EDLP store can generates for large basket sizes and the price advantage that a PROMO store can offer for small basket sizes, they probably shop at the specific store that generates the greatest price advantage. However, if they customers possess inaccurate price knowledge they will also be less likely to know that some stores offers price advantages for a given shopping trip.
Finally, the last sub question is: How does basket size influences the relationship
between the location of the store and customers’ store choice? This study also indicates that
hypotheses 3 are significant meaning that the nature of shopping, either a small or a large basket size, indicates if customers are more willing to travel to the store. Customers are less willing to travel more to a store if their shopping goal is a small basket. If customers do their major trips (large basket) they are more likely to travel more to a store that is further located from a customer’s home location. When combining location with basket size, location becomes less important for the major shopping trips. Managers must know that customers are willing to travel more when doing their major groceries. They need to attract customers from further away, by providing these customers with a good service, pleasant deals and broad assortment so that customers can find all products that they need in one shopping trip.
5.2 Managerial implications
