The difference in willingness to pay a
price premium for national brands
between store formats
Date:
15-01-2018
2
The difference in willingness to pay a price
premium for national brands between store
formats
Master thesis, MSc Marketing,
Marketing Intelligence and Marketing Management University of Groningen, Faculty of Economics and Business
3
Management summary
Nowadays, it is not uncommon anymore that discounters have national brands (NB) in their assortment. Brand manufacturers would like to add their NB for several advantages such as increase in turnover and profit. Nevertheless, this may also have negative consequences. This study focuses on the willingness to pay a price premium for national brands over private labels at discounters compared to service supermarkets. It can have a negative effect on the assortment image and value image of the discounter (Lourenco & Gijsbrechts, 2013). Furthermore, several studies found that the WTPPP (willingness to pay a price premium) is lower for NBs at discounters compared to service supermarkets (Deleersnyder & Koll, 2012). Thus, manufacturers of NBs may set lower prices and may therefore acquire lower margins as a negative consequence.
Prior studies did not focus on the effect of store perception on willingness to pay a price premium for national brands and how this differ. Furthermore, past research lacks in the effect that different product category characteristics have on the difference in WTPPP between store formats. Also the store perception, being hedonic or utilitarian, is a research gap that might provide managers with interesting findings that help them to better set prices and make better managerial decisions in general. Therefore this study examines the difference in WTPPP between store formats, the influence of the perception of the store format and product category, the price image of the stores and the congruity between the hedonic/utilitarian perceptions.
An online survey has been distributed in order to collect data. The stores that were rated on hedonism, utilitarity and price image were Aldi (discounter) and Albert Heijn (regular supermarket). Several questions were asked about the product categories: coffee (perceived as a hedonic and utilitarian product category), milk (utilitarian), detergent (utilitarian), beer (hedonic) and chips (hedonic). Participants had to rate these product categories on hedonism, on being utilitarian and had to mention their WTPPP for a national brand in those product categories in the Aldi and the Albert Heijn. A model per product category and per chain has been created in order to predict the WTPPP.
This study provides managers with insights in the differences in WTPPP in different store formats and the difference among several categories due to perceptions. Managers can use these insights in setting their prices for NBs in different store formats and categories. Setting prices more strategically might improve the performances.
4
brands. On the other hand, for several product categories, the extent to which the consumer perceives the Aldi as utilitarian has a negative effect on the WTPPP. Therefore, it is of importance to take into account the different perceptions consumers have of discounters. Consumers that see the Aldi as hedonic are less interesting to target for national brands. Overall, consumers perceive the Aldi more as utilitarian and therefore it is of importance to keep in mind, as a marketeer, that consumers do not want to pay a higher price premium for a national brand.
5
Preface
This is the last step in finishing my masters in Marketing Management and Intelligence. First of all, I studied HBO communications. Later on I decided to do the marketing pre-master and master marketing (both tracks). This is one of the best decisions I have made in my life. The combination of practical insights (HBO) and theory (Masters) made it possible to become ready for the working life. Especially the data courses during my masters made my study really interesting and made clear what kind of work I would like to do in the following years.
6
Index
Management summary ... 3
Preface ... 5
1 Introduction ... 8
1.1 National brands at discounters ... 8
1.2 Relevance ... 10
2 Theoretical Framework ... 12
2.1 Willingness to pay a price premium ... 12
2.1.1 Definition and measurement of willingness to pay a price premium ... 12
2.1.2 Influences on willingness to pay a price premium ... 12
2.2 Price image of different store formats ... 13
2.3 Utilitarian / Hedonic Perception of Store Format ... 16
2.3.1 Hedonic Perception of Store Format ... 16
2.3.2 Utilitarian Perception of Store Format ... 17
2.4 Utilitarian / Hedonic Perception of Product Category ... 17
2.4.1 Hedonic Perception of Product Category ... 17
2.4.2 Utilitarian Perception of Product Category ... 18
2.5 Store-to-Product Category (SPC) Congruity ... 19
2.6 Overview hypotheses ... 21
2.7 Conceptual model ... 21
3 Research Design ... 22
3.1 Data collection ... 22
3.2 Measures ... 22
3.2.1 Store Price Image... 22
3.2.2 Hedonic and Utilitarian Perception of Store Format and Product Category ... 23
3.2.3 Store-to-Product Category (SPC) – Congruity ... 24
3.2.4 Willingness to Pay a Price Premium ... 24
7
3.3 Analysis Procedure ... 26
4 Results ... 27
4.1 Descriptive statistics ... 27
4.1.1 Store price image ... 28
4.1.2 Utilitarian / Hedonic Perception of Store Format ... 28
4.1.3 Utilitarian / Hedonic Perception of Product Categories... 29
4.1.4 Willingness to Pay a Price Premium ... 29
4.2 Pooling of the Product Categories ... 30
4.2.1 Pooling of the product categories related to the Aldi ... 30
4.2.2 Pooling of the product categories related to the Albert Heijn ... 32
4.3 Violations of assumptions ... 32
4.4 Testing the hypotheses ... 33
4.5 Latent class linear regression analysis ... 35
Product category coffee within the Aldi ... 36
5 Discussion ... 40
6 Recommendations and implications ... 43
7 Limitations and future research suggestions ... 44
8
1 Introduction
For several years, private labels were the only types of brands that were available at hard discounters. Nowadays, it is common that hard discounters have national brands (NBs) listed in their assortments. Aldi for example, sells NBs such as Coca Cola, Heineken and Nutella (Aldi, 2017). At the moment of writing, Aldi has listed around 100 products of NBs in their assortment in the Netherlands (Aldi, 2017). Listing a national brand at a discounter as well, has several consequences. In section 1.1 the consequences and research gaps are described. The research question will be stated at the end of the section. In section 1.2 the relevance of this study is discussed.
1.1 National brands
at discounters
For the discounters the act of listing NBs to their assortment is a way of differentiating themselves (Deleersnyder, Dekimpe, Steenkamp & Koll, 2007). According to Deleersnyder & Koll (2012), listing NBs makes the assortment more attractive. Listing NBs can result in a win-win situation; for the NB and the discounter (Deleersnyder et al., 2007). Deleersnyder & Koll (2012) found that both parties can increase their total performances, while offsetting the losses of potential cannibalization. NBs might gain around 80% of additional sales from new brand buyers in discount stores. Next to this, discounters might acquire around 29% of additional sales from new category buyers (Deleersnyder & Koll, 2012). On the other hand, listing NBs might also have negative consequences.
Firstly, listing NBs can have a negative effect on the assortment image and value image of the discounter (Lourenco & Gijsbrechts, 2013). The authors found that adding non-category leaders and/or choosing wrong NBs additions could harm the image. This is especially the case when the quality gap is small, the NB is less-innovative and less-advertised. The number of SKUs of the NB also has an influence on the image of the discounter. Serving only one SKU of a NB might emphasize the limited assortment of a discounter compared to a service supermarket. Therefore, a lower number of SKUs might deteriorate the assortment image of the discounter.
Secondly, several studies found that the WTPPP (willingness to pay a price premium) is lower for NBs at discounters compared to service supermarkets (Deleersnyder & Koll, 2012). Thus, manufacturers of NBs may set lower prices and may therefore acquire lower margins as a negative consequence. Prior research has empirically proven that the decrease in WTP can be related to the retailer’s price image (Rhee & Bell, 2002). The consumers expect that the price of NBs is lower at discounters than service supermarkets due to the price image of discounters (Bell & Lattin, 1998).
9
As mentioned in the previous paragraph, a perceived quality gap between products might affect the WTPPP. Steenkamp (2010) found that a larger perceived quality gap between NBs and private labels results in a higher WTP for NBs over private labels. This perceived quality gap is affected positively by factors such as distinctive packaging, advertising and innovation. Furthermore, manufacturing factors also play a role of importance. The belief that the product is difficult to make and the belief that private labels are not manufactured by the NBs result in a larger perceived quality gap (Steenkamp, 2010). On the other hand, several studies found that the perceived quality and the objective quality differential is eroding (Nielsen, 2014; Apelbaum, Gerstner & Naik, 2003). The quality gap gets smaller and is therefore losing its influence in affecting the WTPPP (Sethuraman, 2003).
Past research also focused on other determinants of price premium, such as: uniqueness, brand associations, awareness and loyalty (Anselmsson, 2007). These dimensions are part of the customer based brand equity. Customer based brand equity is the differential effect that brand knowledge (awareness and image) has on the way customers act to the marketing of that brand (Keller, 2013). The dimension awareness is for example: the extent to which consumers are able to identify a brand under varying circumstances (Keller, 1993). Loyalty can be defined as the extent to which consumers are attached to a brand (Aaker, 2009). An increase in these dimenions may affect the WTPPP positively. Which variables have not/to a fewer extent been studied?
Prior research that focused on WTPPP for NBs did not take into account the difference in WTPPP between store formats. As mentioned before, price image may play a role in the amount that a consumer wants to pay for a national brand (Rhee & Bell, 2002). In other words, a higher or lower price image might affect the price strategy that a NB should apply for different stores. Besides this, several studies found that the characteristics of a product category play a role in affecting the WTPPP for a national brand (Sethurman & Cole, 1999; Lee & Hyman, 2008). Nevertheless, these authors did not examine the differences between store formats. It has not been studied if the findings hold for discounters and service supermarkets and to which extent. What did those prior studies actually found?
Sethuraman (2003) stated that products can be perceived as more utilitarian than hedonic and vice versa (characteristics of product categories). Utilitarian products are described as “fundamental necessities”(Lee & Hyman, 2008). An example is in this case: toiletpaper. Hedonic categories possess a higher consumption pleasure experience such as cookies. The difference in WTPPP can be related to which extent a product category is perceived as hedonic and/or utilitarian. Product categories that are perceived as more hedonic, are able to obtain a higher price premium (Sethurman & Cole, 1999). Even though some studies focused on these characteristics, Sethuraman & Gielens (2014) argue that these characteristics require more research.
10
This store perception may have an influence on the WTPPP as well. Furthermore, customers perceive some product categories differently due to the store format (Inman, Shankar & Ferraro, 2004). So, if a product category is perceived as hedonic and the store (service supermarket) as well, does this lead to a substantial difference in the WTPPP than when a NB is listed in a utilitarian store (discounter)? The former question will be addressed in this study as well. Therefore this study examines the difference in WTPPP between store formats, the influence of the perception of the store format and product category, the price image of the stores and the congruity between the hedonic/utilitarian perceptions.
Based on the arguments stated above, the following research questions are formulated: Main research questions:
What is the relationship between store perception and willingness to pay a price premium for national brands at hard discounters and regular supermarkets and which factors influence this?
Sub research questions:
1. To which extent does the level of perceiving a store-format as hedonic/utilitarian affect the willingness to pay a price premium for a national brand in that store?
2. To which extent does price image of a store format affect the willingness to pay a price premium for a national brand?
3. To which extent does the level of a hedonic and/or utilitarian category play a role in influencing the willingness to pay a price premium for a national brand?
4. To which extent does store-to-product category congruity, with regard to hedonism and utilitarian, affect the relationship between perceiving a store format as more utilitarian or hedonic and willingness to pay a price premium?
1.2 Relevance
As mentioned above, this study focuses on the difference in WTP a price premium for NBs in different store formats. Steenkamp & Koll (2010) studied the factors that make consumers willing to pay a price premium for NBs over private labels. The mentioned study does not take into account the different store formats. Besides this, Lourenco & Gijsbrechts (2013) focused only on the effects of listing NBs in the discounter format. This study complements the mentioned studies.
11
12
2 Theoretical Framework
In this chapter the literature will be discussed. It starts with explaining what willingness to pay a price premium entails. Next, variables will be discussed that influence the WTPPP. The sections after 2.1, will elaborate on the following variables: store price image, hedonic/utilitarian perception of the store format and product category. After these sections, the (in)congruity between perceiving a store format and/or product category as hedonic or utilitarian will be discussed. Last of all, an overview of the hypotheses is given and the conceptual model is presented to give a visual representation of the literature.
2.1 Willingness to pay a price premium
2.1.1 Definition and measurement of willingness to pay a price premium
Sethuraman & Cole (1999) define price premium as the maximum price consumers will pay for a NB over a store brand, expressed as the proportionate price differential between a NB and a private label. Likewise, Aaker (1996) stated that price premium is obtained when the sum that consumers are willing to pay for a specific brand is higher than of another brand that is part of the same product category. Moreover, WTPPP is a good indicator for explaining brand choices (Agarwal & Rao, 1996). Price premium is also an appropriate measurement metric for brand equity (Aaker, 1996). According to Ailawadi, Lehmann & Neslin (2003) price premium is a relatively stable indicator over time. Nevertheless, this indicator captures also fluctuations in brand equity over time. It should be noted that the willingness to pay is not the same as real prices. Moreover, Sattler and Nitschke (2003) examined the difference between WTP when the respondents of their study did not have to buy a product and the WTP of respondents that actually bought a product in the end. These authors found that the willingness to pay for a product is significantly lower when customers did not have to buy a product in the end, than the customers that had to buy the product.
As mentioned before, Aaker (1996) stated that willingness to pay a price premium is a metric for measuring brand equity. Other authors stated that WTPPP is an outcome of brand equity (Lassar et al., 1995). Brand equity is a result from the greater confidence customers have in a specific brand than the brand(s) of the competition (Lassar et al., 1995). Pope (1993) explained this with the following example: Packard Bell is only bought at price discount due to a low brand equity. In contrary, Compaq and IBM are able to ask a price premium due to the higher brand equity these brands have. Customers that are willing to pay a higher price premium perceive a higher value for the mentioned brands.
2.1.2 Influences on willingness to pay a price premium
13
product category is perceived as utilitarian or hedonic (Sethuraman & Cole, 1999; Dhar & Wertenboch, 2000) and congruity between the perception of the store format & the product category (Lee and Hyman, 2008).
This study focuses on the price image, perception of the store format, perception of the product category and the congruity between the perception of the store format and product category. In the next sections the variables that have been studied in this study will be discussed.
2.2 Price image of different store formats
Grocery stores can be divided into discounters and service supermarkets. Discounters characterize themselves as stores that focus on low prices, serving mainly private labels and serving a smaller assortment (Agarwal, 2003). This store format offers low prices, makes limited use of promotional activities and can be described as a simple, ‘no-frills’ format (M+M Planet Retail, 2005). Service supermarkets offer in general more NBs, have a broader/wider assortment and focus more on services. In the Netherlands, the grocery store Albert Heijn is an example of a service supermarket. Lidl and Aldi can be described as discounters. Furthermore, there is a distinction within the discounter format. Firstly, there are the soft discounters. These discounters, such as Dirk and Nettorama, offer to a smaller extent PLs in their assortment (less than half of the assortment is PL) than hard discounters. Nevertheless, hard and soft discounters offer significantly more PLs than service supermarkets. Furthermore, soft discounters have a larger assortment than hard discounters. Soft discounters have between 1400 and 7000 SKUs within their stores and have a size around 1500 square meters (Ter Braak, Deleersnyder, Geyskens, Dekimpe, 2013). Secondly, the other store format is called the hard discounter. The hard discounter has an assortment with mainly private labels and offers to a smaller extent NBs (Agarwal, 2003). Hard discounters have around 1400 SKUs in their assortment and have stores around 1000 square meters big (Ter Braak et al., 2013). Hard discounters are also more price-aggressive than soft discounters (Denstadli, Lines and Gronhaug, 2005). This study focuses on the difference between hard discounters and service supermarkets.
In order to examine the difference in WTPPP between stores, it is of importance to know which variables have an influence on store choice. These variables might also have an influence on the difference in WTPPP between stores. Pan & Zinkhan (2006) stated that price, assortment, employees and atmosphere affect store choice among others. Furthermore, the whole shopping experience can play a role in affecting the prices that customers are willing to pay for specific products and brands. Furthermore, components of shopping experience, such as physical environment and staff of a store have an influence on price perceptions (Berman and Evans, 1997). Brown and Oxenfeldt (1972) found for example, that cleanliness also has an impact on shopping experience and therefore price perceptions.
14
Prior studies found that the objective quality gap between private labels and NBs is eroding (Nielsen, 2014; Apelbaum, Gerstner & Naik, 2003). This finding also holds for the perceived quality (Steenkamp, 2010). Hence, the quality gap between NBs and private label gets smaller and is therefore of less influence in affecting the difference in WTPPP within store. In addition, the quality gap gets smaller and therefore the difference in WTPPP for NBs between store formats also gets smaller.
As mentioned before, price is an important factor with regard to store format choice (Pan & Zinkhan, 2006). Customers expect namely certain prices because of the price images of the store formats. Moreover, it might be assumed that the price at discounters is lower than at a service supermarket, because of the price image. The price image of a store can be defined as the perception of the expensiveness of the store (Lourenco, Gijsbrechts and Paap, 2015). A similar definition is stated by Hamilton and Chernev (2013), they defined price image as a concept that reflects the customer’s judgment about the average price level, with reference to the competitors of the store. Lourenco et al., (2015) found that price image is formed mainly by product categories that are purchased often and that have a small price range. In contrary, infrequently bought categories and categories without frequent price discounts, have a smaller impact on the shaping of the price image (Lourenco et al., 2015). Moreover, other papers found that price discounts, and price store format (for example Every Day Low Prices) do have an impact on the forming of the overall price image (Roggeveen et al. 2014; White and Yuan 2012). Besides this, assortment (Desai and Talukdar, 2003), advertising (Desmet and Le Nagard, 2005), interior (Baker et al., 1994) and exterior architecture (Zielke and Toporowski, 2012) are also found to be of importance in affecting the store price image. In other words, Desai and Talukdar (2003) found that the advertising spend of a retailer has a significant impact on lowering the price image of a store (Desmet and Le Nagard, 2005). With regard to the assortment, the amount of national brands sold increases the price image significantly (Desai and Talukdar, 2003). Also the interior and exterior of the store can affect the price store image. The layout (grid), color (brown/white) and organization of merchandise (messy) are of importance in lowering the store price image (Baker et al., 1994).
Prior studies also studied the consequences of high or low price images. Consumer beliefs (store image) and consumer behavior is affected by price image (Hamilton & Cherney, 2013). Zielke (2006) studied five price image dimensions and the influence on shopping intentions for several stores. With regard to supermarkets, the author found that a higher feeling of value for money (dimension of price image) increases shopping intentions. Value for money can be defined as a trade-off between what a customer receives and the costs (Zeithaml, 1988). Other studies have proven that price image is affecting customer’s price satisfaction (Lombart and Louis, 2014).
15
The prices of discounters are lower than those of service supermarkets and therefore the reference price level is lower. Thus, it can be expected that customers expect a lower price for a NB in a discount context.
In addition, Deleersnyder et al., (2007) found that a larger price gap of the NB between stores, results in more sales (for the NB and discounter). According to the authors, a NB is more successful in a discounter when it is priced lower at a discounter than at a service supermarket. Deleersnyder et al., (2007) studied also the optimal within store price gap (price gap between NB and PL at the discounter). The authors found that a larger NB-PL price gap improves the category sales for the discounter and the sales of the NB. In other words, a higher priced NB results in more category sales. The arguments of these authors were as follows: a larger price gap between NB and PL, may improve the attractiveness of the assortment and the NB may signal additional benefits. Another argument is that a smaller price gap leads to indifference in making a choice between the products/brands (Chiang & Padmanabhan, 1999). A larger price gap within the store has therefore a more positive impact on category sales.
Therefore, the following hypothesis is formulated:
16
2.3 Utilitarian / Hedonic Perception of Store Format
2.3.1 Hedonic Perception of Store FormatLee and Hyman (2008) classified stores into hedonic and utilitarian stores. Hedonic stores focus more on atmosphere, store layout and the needs of customers than utilitarian stores. Hedonic stores can be associated with premium quality products, higher prices and higher level of service degree (Lee and Hyman, 2008; Ligas & Chaudhuri, 2012). Furthermore, perceiving a store as hedonic relates to the variety of the assortment (Holbrook & Hirschman, 1982). An assortment with more choice is associated with hedonic stores. In line with Lee & Hyman (2008), it has been proven that hedonic stores, such as department stores, have a more enjoyable atmosphere (Donovan, Rossiter, Marcoolyn & Nesdale, 1994). In general, service supermarkets offer a deeper and broader assortment than discount stores. The following terms are also characteristic of a hedonic store: high-quality/status oriented shopping motivations, more services, more helpful staff on the floor (Baker, Levy, and Grewel 1992; Roy 1994; Vrechopoulous et al. 2004; Wakefield and Baker 1998; Wakefield and Barnes, 1996).
Is there a difference regarding the people that buy hedonic products? Wakefield and Inman (2003) found that people that buy hedonic products are less price sensitive. In general, service supermarkets sell more hedonic products than discount supermarkets. Therefore, it is expected that customers are less sensitive with regard to the prices within a service supermarket context than in the environment of a discount store. It may be assumed that customers have a lower WTPPP for NBs when they are less price sensitive.
The hedonic versus utilitarian perception a shopper has of the store may have a significant effect on the willingness to pay for national brands. Sethuraman & Cole (1999) found that product categories that are perceived as hedonic are able to attract a higher price premium. It may be assumed that a store that is perceived more as hedonic, is also able to attract higher prices than a store that is more perceived as utilitarian. In other words, hedonic stores are more in a position to ask higher prices than discount stores and actually benefit from a higher price. Therefore, the following hypothesis is formulated:
17 2.3.2 Utilitarian Perception of Store Format
Utilitarian stores can be associated with low prices, low level of services and a simple design (Lee and Hyman, 2008). Other characteristics of utilitarian stores are: customers have a value/convenience orientation, indifferent atmosphere, low number of staff available at the floor, utilitarian environment and a stony image (Baker, Levy, and Grewel 1992; Roy 1994; Vrechopoulous et al. 2004; Wakefield and Baker 1998; Wakefield and Barnes, 1996). Therefore, the following hypothesis is formulated:
H2b: The more a consumer sees a supermarket as utilitarian, the lower his/her WTPPP for NBs in that supermarket.
Based on the discussion above, hard-discounters may be perceived as utilitarian stores and service supermarkets as hedonic stores. It should be taken into account that the classification is not mutually exclusive. Hence, a supermarket can be perceived for a certain degree as hedonic and utilitarian.
2.4 Utilitarian / Hedonic Perception of Product Category
2.4.1 Hedonic Perception of Product Category
Several studies found that category characteristics play a role in affecting price premium (Dhar and Wertenbroch 2000). Hedonic categories and products obtain a higher level of fun, pleasure and excitement (Holbrook and Hirschman, 1982). Furthermore, hedonic products can be defined as products that deliver an emotional and multisensory experience (Holbrook and Hirschman, 1982). The hedonic shopping value that a customer attracts is an outcome related to non-task oriented activities and is more reflected in the shopping experience itself (Babin and Attaway, 2000). Examples of these product categories in grocery stores are wine and chocolate (Dhar & Wertenboch, 2000). Product categories that are perceived as hedonic, are able to attract a higher price premium (Sethuraman & Cole, 1999; Dhar & Wertenboch, 2000). So, measuring to which extent a category is perceived as hedonic may help managers with their pricing decisions (Voss, Spangenberg & Grohmann, 2003). Therefore, it is relevant to investigate if customers are willing to pay a higher price when the NB is part of a hedonic category.
18
Perceiving a category as hedonic, has also an influence on the effectiveness of sales promotions. It has been proven that the extent to which a product category is perceived as hedonic, influences the effectiveness of sales promotions (Chandon, Wansink & Laurent, 2000). These authors found that sales promotions are more valued as an utilitarian benefit (saving money) than as hedonic (being a thrifty shopper for example). Other authors also found that hedonic shopping value can be related to loyalty and repatronage intentions (Stoel et al., 2004). Hence, most studies have proven that perceiving a category as hedonic results in positive retail outcomes.
It should be noted that customers can perceive a category as hedonic and utilitarian at the same time. So, a category cannot be perceived as 100% hedonic (Voss, Spangenberg & Grohmann, 2003). To conclude, Lee & Hyman (2008) stated that a hedonic/utilitarian nature may have an effect on consumer choices. Moreover, product categories that are perceived as hedonic, are able to attract a higher price premium (Sethuraman & Cole, 1999; Dhar & Wertenboch, 2000). Sethuraman and Gielens (2014) stated that more research is required regarding product category characteristics and their influence on WTP. Therefore, the following hypotheses are formulated:
H3: The more a consumer sees a product category as hedonic, the higher his/her WTPPP for NBs in that category.
2.4.2 Utilitarian Perception of Product Category
19
Prior research studied the effects of product characteristics (hedonic/utilitarian) on service purchases (Chen, Kalra & Sun, 2009). These authors found that people purchase hedonic services more than utilitarian services. This might have to do with the self-gratification customers achieve while buying hedonic products (Khan, Dhar, and Wertenbroch 2005). The feelings of guilt might be an outcome of buying hedonic products, but people try to cope with these feeling by searching for reasons that justify their decisions (Khan, Dhar, and Wertenbroch 2005). The customer may justify the purchase by telling themselves that the product/service serves as a reward (Strahilevitz and Myers, 1988). Nevertheless, customers value hedonic products more and feel a pain of loss when they do not buy a hedonic service (Chen, Kalra & Sun, 2009). Therefore, it is expected that customers are willing to pay a lower price premium for utilitarian products. This expectation is in line with prior research (Sethuraman, 2003; Dhar & Wertenboch, 2000). Therefore, the following hypothesis is formulated:
H4: The more a consumer sees a product category as utilitarian, the lower his/her WTPPP for NBs in that category.
2.5 Store-to-Product Category (SPC) Congruity
Keller (1993) defines congruity as follows: “congruity is the extent to which a brand association shares content and meaning with another brand association”. Brands can be associated with companies, places of origin, channels of distribution and celebrities among others. In the context of this study, the NB is linked to a new channel of distribution; the hard discounter. Listing a NB at a discounter may result in incongruity; a NB does not fit in the category of a discounter. In that case, the image of the discounter may improve and the image of the NB may deteriorate (Grewal, Krishnan, Baker & Borin, 1998). Moreover, the discounter offers more variety in the assortment and differentiates itself (Deleersnyder et al. 2007). In other words, it is expected that the NB does not fit with a typical image of a discounter. This misfit may have to do with perceiving the store format and product category differently in terms of hedonism and utilitarianism (Lee and Hyman, 2008).
20
According to Dawes (1994) people have an intrinsic motivator to keep everything mentally structured and congruent. Thus, if the product category is perceived differently than the store format, this may result in incongruity.
Prior research also found that information that is inconsistent with the expectations of the customers, result in more elaborate processing and more rationalizing (Dijksterhuis & Van Knippenberg, 1995). In addition, Alden et al., (2000) found that inconsistencies in information may generate emotional reactions such as surprises and these emotional reactions result in more elaboration. Processing the information more properly might mean that customers behave more rational and absorb relevant information more properly (Fennis & Stroebe, 2015). A (rational) step-by-step process is a typical characteristic of a highly involved customer (Lee and Lou, 1995). Highly involved customers might be more aware about the incongruity between the store and the product category in terms of hedonism and utility. Perceiving a higher incongruity may lower the WTPPP. Another study stated that congruity can be related to the ease in which a stimulus is perceived and processed (Mantokanis, Whittlesea and Yoon, 2008). When the stimulus (NB) is processed with ease, it may result in positive emotions and this may help to evaluate the stimulus in a positive way. Furthermore, an increase in exposing the stimulus may increase liking. When a stimulus is exposed frequently it will result in a representation in the brain (Fennis & Stroebe, 2015). Another encounter with the stimulus will result in an increased ease of processing and this results in the end in a more positive evaluation (Winkelman et al., 2003). In this case, a NB within a hedonic category at a utilitarian store may not be processed with ease. In other words, the processing fluency is lacking and therefore it can be expected that customers would like to pay less for a NB in an incongruent context. As mentioned before, people try to resolve inconsistencies.
For stores that are more perceived as hedonic than utilitarian, the congruity is higher when the NB is part of a hedonic product category rather than an utilitarian product category. Therefore, the following hypothesis is formulated:
H5a: The higher the hedonic store-to-product category congruity, the higher his/her WTPPP for NBs in that category and supermarket.
For stores that are more perceived as utilitarian than hedonic, the congruity is higher when the NB is part of an utilitarian product category rather than a hedonic product category. The hypothesis is as follows:
21
2.6 Overview hypotheses
In this section an overview is given of the formulated hypotheses.
H1 The lower the consumer perceives the price image of the store, the lower his/her WTPPP for NBs in that supermarket.
H2a The more a consumer sees a supermarket as hedonic, the higher his/her WTPPP for NBs in that supermarket. H2b The more a consumer sees a supermarket as utilitarian, the lower his/her WTPPP for NBs in that
supermarket.
H3 The more a consumer sees a product category as hedonic, the higher his/her WTPPP for NBs in that category.
H4 The more a consumer sees a product category as utilitarian, the lower his/her WTPPP for NBs in that category.
H5a The higher the hedonic store-to-product category congruity, the higher his/her WTPPP for NBs in that
category and supermarket.
H5b The higher the utilitarian store-to-product category congruity, the higher his/her WTPPP for NBs in that
category and supermarket. table 2-1: overview hypotheses
2.7 Conceptual model
The conceptual model in figure 2.1 is based on the theoretical discussion in this chapter. The extent to which a store is perceived as utilitarian or hedonic (perception store format), store price image, utilitarian product category and hedonic product category (perception product category) are independent variables. As mentioned before, hard-discounters may be perceived as utilitarian stores and service supermarkets as hedonic stores.
Figure 2.1: Conceptual model
Perception Product Category
WTPPP for NB
Store Price Image
H1 H5a/b
H2a/b H3 H4
22
3 Research Design
This chapter starts with describing the data collection. After this, the constructs of the variables of the conceptual model will be discussed. Next, an overview is given of the constructs, operationalization of the constructs and the data sources. Lastly, the procedure of the analysis will be briefly discussed.
3.1 Data collection
The focus of this study is to examine the differences in WTPPP for NBs between different store formats. Furthermore, the price image of the store, the hedonic/utilitarian perception of a product category and the hedonic or utilitarian perception of the store are included in the study. The research type is quantitative and the data is collected with an online survey. The survey is created with an online survey tool called Qualtrics. The survey can be found in appendix A. In order to get enough respondents, various ways of distribution have been used. Moreover, the questionnaire was sent via Facebook, given to several people in a physical form and posted on several online fora. This study focused on the Dutch population, because it is proven that hedonic shopping value interacts with differences regarding culture. Moreover, interaction between hedonic shopping value and culture influences consumer behavior (Richard and Habibi, 2016). Even though, within a country the cultures can be really diverse, this study is limited to the Dutch nationality in order to keep the potential influence of culture small. The stated distribution methods make it possible to get a varied sample of respondents that differ in their age, income and education.
3.2 Measures
The variables of the conceptual model will be discussed in this section. 3.2.1 Store Price Image
Hamilton and Chernev (2013) made a comparison between direct measures and indirect measures with regard to price image. The direct measure means that customers are asked to evaluate the overall store price image (Hamilton and Chernev, 2013). An easy example of a direct question to measure store price image is the following: “How would you rate the prices at this store?” (Hamilton and Chernev, 2013). On the other hand, the indirect approach measures the consequences of the level of the price image. An example is the following: “How much would you expect this basket of
goods to cost at this store?” (Buyukkurt, 1986). Hamilton and Chernev (2013) mentioned that both
23
Similar as, Lourenco et al., (2015), this study uses the direct approach and uses the following question to measure store price image: “How do you rate the prices of products in this store?”. Moreover, the authors used an ordinal scale that ranged from 1 (excellent) to 9 (very bad) (Lourenco et al., 2015). In line with these authors, an ordinal scale with 9 options is chosen to make sure that there is a high reliability and validity (Lozano, García-Cueto and Muñiz, 2008).
3.2.2 Hedonic and Utilitarian Perception of Store Format and Product Category
In order to measure to which extent a product category and a store is perceived as hedonic or utilitarian, a pre-test has been conducted. As mentioned before, these dimensions are not mutually exclusive. Furthermore, a category cannot be perceived as 100% hedonic (Voss, Spangenberg & Grohmann, 2003). Previously mentioned is the example of milk, that can score high on both dimensions (Sloot et al., 2005). A pre-test has been done in order to know on which characteristic a product category scores high. Moreover, seeing a product category as more hedonic or utilitarian might differ per consumer. The selected product groups of the study of Sloot et al., (2005) have been chosen to include in the pre-test, because it has been proven that these categories score high on one or both dimensions. The selection of Sloot et al., (2005) is based on a study done by 40 food experts, that rated the product categories on a scale with the following range: from 1 (not hedonic) and 7 (very hedonic). The product categories are as followed: eggs, margarine, milk, detergent, beer, chips, cigarettes and cola.
Prior research differs in the constructs that have been used in order to measure hedonism and utilitarianism. Several studies used a multi-item construct in order to measure the hedonic or utilitarian perception (Spangenberg and Grohman, 2003; Lee and Hyman, 2008). Spangenberg and Grohman (2003) used for example, a 10 item construct and measured dimensions such as fun, excitement, joy, effectiveness, helpfulness and utilitarianism. In order to keep the pre-test parsimonious (Okada, 2015), the single-item construct of Okada (2005) has been used. Okada (2005) used the following construct: “How do you rate this product category?” Okada (2005) used an ordinal scale ranging from 1=not at all hedonic and 7=extremely hedonic. The utilitarian construct is measured in the same way (utilitarian level: 1 = not utilitarian, 7 = extremely utilitarian). Sloot et al., (2005) used an explanation of both dimensions in their survey, in order to make sure that the respondents understood the different dimensions. In line with Sloot’s study, this study used several definitions to explain the differences. The following definitions were used: Utilitarian products can be
defined as fundamentals, necessities and essentials (Barbin, Darden, and Griffin 1994; Mano and Oliver 1993). Hedonic categories and products obtain a higher level of fun, pleasure and excitement (Holbrook and Hirschman, 1982).
24
The findings of the pre-test can be seen in figure 3-1. Coffee is found to score high on both dimensions. Furthermore, milk, detergent and butter can be classified as utilitarian as expected. Beer and chips are typically hedonic. This study continues with the product categories coffee, milk, detergent, beer and chips.
After the pre-test, an online survey has been distributed and the questions related to the hedonic/utilitarian perceptions will be asked again.
3.2.3Store-to-Product Category (SPC) – Congruity
As previously mentioned, Lee & Hyman (2008) used the same construct to measure the level of hedonism/utilitarianism for stores and product group characteristics. Furthermore, Lee & Hyman (2008) calculated the congruity between the constructs. This study creates the congruity variable by creating interaction variables (hedonic store*hedonic product category and utilitarian store*utilitarian product category).
3.2.4 Willingness to Pay a Price Premium
The WTPPP is in line with price image measured with a single-item construct. The construct consists of the same scale that Steenkamp et al., (2010) used in their study. The authors asked directly how much the customer is willing to pay more for a NB compared to a PL. Furthermore, they used percentages to measure the extra amount of money customers would like to pay extra for a NB.
1 2 3 4 5 6 7 1 2 3 4 5 6 7 H e d o n ic Pe rc e p tion Utilitarian Perception
Utilitarian / Hedonic Perception of Product Categories
Chips Beer Coffee Milk Detergent Butter
25
Moreover, these authors used the following levels: 0% (nothing), 10% more, 20% more, 30% more, 40% more, 50% more, 75% more, 100% more (twice as much), more than 100% (more than twice as much). The “more than 100%” category has been recoded in their study to 125% (Steenkamp et al., 2010).
A prior study used the following question to measure WTPPP: “What price premium (average) would
you pay to deal with this rep firm with similar products? (%)” (Palmatier, Scheer, and Steenkamp,
2007). This question was adapted and used in the study of Steenkamp et al., (2010). This study uses the same measuring question as Steenkamp et al., (2010).
3.2.5 Overview constructs
In the table below an overview is given of the constructs and the related scales.
Construct Operationalization Data Source
Store price image “How do you rate the prices of products in this store?” Likert scale: 1 =very bad, 9 = excellent
Lourenco, Gijsbrechts and Paap (2015)
Perception store format – Utilitarian “How do you rate this store format?”
Likert scale: 1 = “not at all utilitarian”, 7 = “extremely utilitarian”
Okada (2005)
Perception store format – Hedonic “How do you rate this store format?”
Likert scale: 1 = “not at all hedonic”, 7 = “extremely hedonic”
Okada (2005)
Perception product category - Utilitarian “How do you rate this product category?”
Likert scale: 1 = “not at all utilitarian”, 7 = “extremely utilitarian”
Okada (2005)
Perception product category - Hedonic “How do you rate this product category?”
Likert scale: 1 = “not at all hedonic”, 7 = “extremely hedonic”
Okada (2005)
Willingness to pay a price premium (WTPPP)
“In the category X, how much more are you willing to pay for a brand compared to a shop’s own label?”
0% (nothing), 10% more, 20% more, 30% more, 40% more, 50% more, 75% more, 100% more (twice as much), 125% more.
Steenkamp et al., (2010)
Control variables
Gender What is your gender? 0 = female, 1 = male
Age What is your age? (any number can be filled in)
Education What is your highest level of education? (high school, MBO, HBO, University)
Income What is your monthly gross income?
< €1.000 €1.000-€2.000 €2.001- €3.000 €3.001-€4.000 €4.001-€5.000 > €5.0001
Family size What is the size of your household? Count only the persons that live in the household for 4 days a week (include yourself).
GfK/TNS
26
3.3 Analysis Procedure
First of all, descriptive statistics will be used in order to provide an overview of the sample. Secondly, a Chow-test has to be performed in order to examine if it is possible to pool the data of the different product categories. If this is not the case, multiple models will be created (one for each category) and the interpretation of the results should be done more carefully.
27
4 Results
First the descriptive statistics of the dataset are given. Secondly, the procedure of the chow-test and outcome has been described. Thirdly, a linear latent class analysis has been performed and the results are stated.
4.1 Descriptive statistics
In total 244 participants filled in the survey. Among those 244 participants, 40 did not complete the survey. These participants were excluded for further analysis. Moreover, a pre-analysis has been done in order to find outliers and missing values. Besides the 40 uncompleted surveys, no other missing values were found. The format of the questionnaire required an answer in order to finish it. This is the reason of the lack of missing values. Three other respondents were deleted, because they showed up as significant outliers. One respondent was below 18 years old, the other two outliers perceived beer as extremely utilitarian, which made these respondents outliers.
Furthermore, Fout! Verwijzingsbron niet gevonden.Fout! Verwijzingsbron niet gevonden. gives an overview of the descriptive statistics. Comparing the demographics of the sample to that of the actual supermarket shopping population in the Netherlands results in seeing that males are overrepresented and the age category 18-34 years as well. The supermarket shopping population data is based on data of EFMI (2014). A weighting variable is computed in order to make sure that the sample is representative for the supermarket shopping population. One weighting variable has been created that ensures that extra weight is given to females and underrepresented age groups. This resulted in proportions within the sample that are not completely similar to the Dutch shopping population as determined by EFMI. Nevertheless, this makes sure that the age groups with people of 55+ and 35-54 are not overweighted.
Gender Before weighting EFMI After weighting
Male 53% 29% 40%
Female 47% 71% 60%
Age Before weighting EFMI After weighting
18-34 80% 24% 65%
35-54 11% 41% 18%
55+ 9% 35% 18%
28
figure 4-1: box plot ofprice image Albert Heijn (1=very bad; 9=very good) figure 4-2: box plot ofprice image Aldi (1=very bad; 9=very good)
4.1.1 Store price image
As mentioned in chapter 3, participants were asked to rate the prices of stores. This question relates to the price image of the store formats. Hamilton and Chernev (2013) defined price image as a concept that reflects the customer’s judgment about the average price level, with reference to the competitors of the store. The results in the graphs below state that the prices of the Aldi are experienced better than the ones of Albert Heijn. This means that the Aldi has a better price image (mean 6,9) than Albert Heijn (mean 5,7).
4.1.2 Utilitarian / Hedonic Perception of Store Format
In the theoretical framework, it has been argued that Aldi would be perceived as an utilitarian store and Albert Heijn as a more hedonic store. The results in the figure below indicate that the Aldi scores higher on utility and Albert Heijn higher on hedonism. In order to examine if the perception of the Aldi and Albert Heijn differ significantly, a t-test has been performed. The t-test stated that the two stores differ significantly from each other (p<0,05).
figure 4-3: perception of store format
2,4 5,4 5,4 2,7 1 2 3 4 5 6 7
Aldi Albert Heijn
Store
How do you rate this store format?
(1 = not al all hedonic 7 = extremely hedonic)
29
4.1.3 Utilitarian / Hedonic Perception of Product Categories
In this study five product categories have been examined in terms of hedonism and utility. The scores on both dimensions per product category are shown in figure 4-44. The product categories milk and detergent score higher on the utilitarian dimension. In contrary, beer and chips are perceived as hedonic product categories. The mentioned categories meet the expectations mentioned in chapter 3. Coffee is an interesting category, because this category scores high on both dimensions. More variation within the dimensions make it easier to see if these dimensions moderate the studied relation or affect the dependent variable directly.
4.1.4 Willingness to Pay a Price Premium
In figure 4-5 (see next graph) can be seen that the participants in the study are willing to pay a higher price for national brands compared to private labels in the Aldi than in the Albert Heijn. This finding holds for every product category. This finding does not meet priori expectations. It was expected that customers would like to pay less for national brands in discount stores, because they refer to lower reference prices and buy within a store that has a cheap store price image. This finding will be discussed more extensively in chapter 5. Furthermore, the distribution of those variables looked normal. 1 2 3 4 5 6 7 1 2 3 4 5 6 7 H e d o n ic Pe rc e p tion Utilitarian Perception
Utilitarian / Hedonic Perception of Product Categories
BeerMilk Coffee
Detergent Chips
Figure 4-1: Perception Product Categories
30 figure 4-5: willingness to pay a price premium for a national brand
4.2 Pooling of the Product Categories
This study contains multiple entities (product categories) and therefore it is of importance to make a decision about the treatment of those entities in the model (Leeflang, Wieringa, Bijmolt & Pauwels, 2014). This means that a model can be specified as an unit-by-unit model (one model per product category), pooled model (the same model for all product categories) and a partially pooled model (some parameters are pooled and some are specific per product category) (Leeflang et al., 2014). The most important benefit of pooling is that by combining the data from cross-sections, there are more data points available so that the coefficients can be estimated with greater statistical efficiency. This means that there is a smaller variance of the estimates (Leeflang et al., 2014). In order to test if the product categories can be pooled, a CHOW test has to be conducted.
Before the Chow test, an one-way ANOVA has been performed in order to examine significant differences between the product categories regarding the WTPPP. This has been done for the WTPPP for product categories within the store Aldi and Albert Heijn separately. The ANOVA test stated that every product category scored significantly different of each other regarding the WTPPP. The only exceptions were coffee versus chips and milk versus detergent. These exceptions hold for both stores. The significant different product categories mean that there is heterogeneity in variances. Therefore, it might be expected that a fully pooled model is not suitable. In order to fully discard a fully pooled model a Chow-test has been performed.
4.2.1 Pooling of the product categories related to the Aldi
As mentioned in the previous section, a Chow-test has been performed in order to test if the data can be pooled or not (Leeflang et al., 2014). A pooled model means that all parameters are fixed for the different product categories in this case. A partially pooled model means that the slope parameters of the independent variables are fixed and the constants of the product categories are unique (Leeflang et al., 2014).
32% 11% 16% 41% 26% 37% 18% 19% 46% 30% 0% 5% 10% 15% 20% 25% 30% 35% 40% 45% 50%
Coffee Milk Detergent Beer Chips
In the category X, how much more are you willing to pay for a brand compared to a shop’s own label?
31 The formula of the Chow-test is as follows:
The terms are as follows: F = F-statistic
SSR = sum of squares residuals df = degrees of freedom
First of all the sum of squares residuals were calculated for the unit-by-unit model and the pooled model. These can be found in table 4-2. Secondly, the degrees of freedom had to be calculated for the pooled models and the unpooled models. These were calculated as follows:
Df pooled= NT - #estimated parameters= (5*204)–9 = 1011
Df un-pooled= N*(T -#estimated pars per product category)= 5*(204-9) = 975
Where N stands for the cross sections (product categories in this case) and T for the number of observations (Leeflang et al., 2014).
Model Residual Sum of Squares
Pooled Model Aldi 31.240
Unit-by-Unit Model 28.928 (sum of the unpooled models)
Coffee 8.901 Milk 3.965 Detergent 4.401 Beer 7.127 Chips 4.534 OLDSV-Model 30.187
table 4-2: residual sum of squares per model
Filling in the values in the Chow-test formula lead to the following result: F(36,1011) ={(31.240-28.928)/(1011-975)}/{(28.928/975)}= 2,16
32
These results suggest that the unit-to-model should be used. It might be the case that some parameters can be pooled (partially pooled model), instead of all parameters (fully pooled model). Therefore, the partially pooled model is compared with the unit-by-unit model. In order to compare these models, five dummy variables were created in order to have an unique intercept for the product categories. Moreover, in the partially model that is created, the parameters of the independent variables are pooled (OLSDV).
This lead to the following formula and result:
F(32,1007) ={(30.187-28.928)/(1007-975)}/{(28.928/975)}= 1,33
This output was found to be not significant (p>0,05; p = 0,14), based on the PQRS program that has been used. Therefore a partially pooled model is allowed.
4.2.2 Pooling of the product categories related to the Albert Heijn
In order to be parsimonious, the results of the Chow-test with respect to the Albert Heijn are briefly described. The sum of squares residuals can be found in table 4-3. Comparing the pooled model to the unit-by-unit model lead to the conclusion that pooling is not allowed. Furthermore, the partially pooled model was also not allowed due to the results of the F-statistics. Therefore, a unit-by-unit model is used for both chains in order to make useful comparisons. All F-statistics and p-values can be found in Appendix B.
Model Residual Sum of Squares
Pooled Model Albert Heijn 20.625
Unit-by-Unit Model 17.499 (sum of the unpooled models)
Coffee 5.562 Milk 1.622 Detergent 2.195 Beer 5.060 Chips 3.060 OLDSV-Model 19.402
table 4-3: residual sum of squares
4.3 Violations of assumptions
33
These two variables were added to the linear unit-by-unit models and these were checked on significance. Both variables were not significant in all the unit-by-unit models and therefore H0, the model has the right functional form, cannot be rejected (p>0,05).
A wrong estimate of variance can be caused by autocorrelation, heteroscedasticity, non-normality and multicollinearity (Leeflang et al., 2014). Multicollinearity has been checked by examining the correlations and the VIF scores. All models had scores above the VIF cut-off point of 5. The interactions (hedonic congruity and utilitarian congruity) caused the multicollinearity. It has been tried to resolve this issue by splitting up the sample in four parts. This is based on the values of the hedonic product category and utilitarian product category (cut-off was the median for low values (0) and high values (1)). After estimating the models for the selected respondents, the p-values and parameters changed for several variables. Therefore, it has been decided that the study uses a model without the interactions and splits the sample in four in order to test H5.
Another important problem that has to be solved is non-normality if this is the case. For several models it has been checked if the unstandardized residuals, tested with the Kolmogorov-Smirnov test, resulted in significant results (p<0,01). All the models that have been tested were significant and therefore several dummies were created that should account for outliers. Nevertheless, the Kolmogorov-Smirnov test continued with stating that the models were significant (p<0,01). Bootstrapping resulted in the same conclusion (non-normality), but the p-values of the variables did not differ that much. Autocorrelation is only relevant when the data is time dependent. Therefore, this problem does not apply to this study.
4.4 Testing the hypotheses
34
The ** indicates that the variable was significant (p<0,05 and * means p<0,10).
Coffee Milk Detergent Beer Chips
Hypotheses Aldi AH Aldi AH Aldi AH Aldi AH Aldi AH
H1: The lower the consumer perceives the price image of the store, the lower his/her WTPPP for NBs in that supermarket.
**
**
**
**
H2a: The more a consumer sees a supermarket as hedonic, the higher his/her WTPPP for NBs in that supermarket.
**
** ** ** ** **
**
H2b: The more a consumer sees a supermarket as utilitarian, the lower his/her WTPPP for NBs in that supermarket.
**
**
** **
**
H3: The more a consumer sees a product category as hedonic, the higher his/her WTPPP for NBs in that category.
** ** ** ** ** ** ** ** ** **
H4: The more a consumer sees a product category as utilitarian, the lower his/her WTPPP for NBs in that category.
H5a: The higher the hedonic store-to-product category congruity, the higher his/her WTPPP for NBs in that category and supermarket.
H5b: The higher the utilitarian store-to-product category congruity, the higher his/her WTPPP for NBs in that category and supermarket.
table 4-4: hypotheses (* = p<0,10; ** = p<0,05)
Striking is not only that several hypotheses are rejected, but one hypothesis (H2a) also showed the opposite result. In other words, the effect is the opposite from expected. Although, the problems that can cause wrong parameters and p-values are dealt with (as described in the previous section). Nevertheless, these results will be used for the discussion.
Hypotheses Aldi
35
As mentioned before, the sample has been split in order to examine if there are any significant interaction effects. It was hypothesized (H5b) that there is congruity between the utilitarian store perception and the utilitarian product category perception. For coffee, milk and beer this hypothesis is confirmed. Moreover, respondents that rated these product categories high on utility, are willing to pay a higher price premium when their perception of seeing the store as utilitarian increases. See Appendix I for an overview of the rejected/accepted hypotheses for the models that have been estimated on high/low values for the moderators.
Hypotheses Albert Heijn
For all the product categories with respect to the Albert Heijn, H3 is also confirmed. Besides this, for all product categories (except coffee), H1 is significant. This means that the price image positively influences the WTPPP for national brands at the Albert Heijn. H2a is partially confirmed (only for the product categories milk and detergent). Thus, the more a consumer perceives the Albert Heijn as hedonic, the more he/she is willing to pay a price premium for a national brand within the milk or detergent category. H2b is only confirmed for detergent.
In Appendix I can be seen that there is no clear interaction effect between the moderators and related independent variables. Therefore, H5a and H5b are not confirmed.
4.5 Latent class linear regression analysis
The previous section about pooling shows that the data is heterogeneous. Moreover, heterogeneity among the data might mean that latent classes can be identified. Latent classes are discrete segments that differ in their preferences (Eggers et al., 2015). With a latent class analysis all the respondents are separated in different segments with a certain probability. In other words, it is not totally sure to which segment a customer belongs in latent class analysis. In order to identify the optimal number of segments, information criteria, log-likelihood-based measures and the classification error can be used. Furthermore, it is also important to choose a number of segments that is useful for managerial purposes such as targeting. A model with few segments might be more convenient. On the other hand, more segments might identify niche markets and might provide a more detailed description of the segments.
36
The variables of the conceptual model are used as predictors and the demographics (age, gender, income, education and household size) are used as covariates. Covariates help to lower the classification error and are relevant to make useful latent classes. Due to time constraints and space, only the product category coffee within the Aldi is described extensively. The other estimates and profiles of the other Aldi product categories can be found in Appendix C. Furthermore, the latent classes of the Albert Heijn can be found in Appendix D . The results with respect to the Albert Heijn are placed in the appendix because this study focuses on the introduction of national brands at discounters. Another goal of this study, is to compare the differences between store formats in terms of WTPPP. Therefore a comparison has been made (see the previous section).
Product category coffee within the Aldi
First of all the number of segments has to be identified. table 4-5 shows that the BIC and CAIC criteria are the lowest for the 4 and 5 class model. Looking at the AIC and AIC3, the 7 class models have the lowest penalized log likelihood, but looks like it continuous with decreasing. Therefore, the BIC and CAIC criteria has been chosen to select the number of classes in this case. The classification error is not higher for 5 classes. In order to select 4 or 5 segments, the parameters of both models have been compared. Both models show that all variables are signifciant. Looking at the R^2, the model with 5 classes, explaines to a small extent more of the variance. In order to keep make it easier to describe the different classes, 4 classes have been selected.
LL BIC(LL) AIC(LL) AIC3(LL) CAIC(LL) Npar Class.Err. R² Model1 1-Class Regression -1055,4493 2152,9922 2124,8985 2131,8985 2159,9922 7 0,0000 0,0826
Model2 2-Class Regression -944,1919 2008,6513 1928,3838 1948,3838 2028,6513 20 0,0908 0,7962
Model3 3-Class Regression -900,9684 2000,3783 1867,9368 1900,9368 2033,3783 33 0,1498 0,7308
Model4 4-Class Regression -818,1105 1912,8365 1728,2211 1774,2211 1958,8365 46 0,1021 0,8469
Model5 5-Class Regression -773,0951 1900,9795 1664,1901 1723,1901 1959,9795 59 0,1047 0,8745
Model6 6-Class Regression -754,3424 1941,6482 1652,6849 1724,6849 2013,6482 72 0,0833 0,9046 Model7 7-Class Regression -704,2596 1919,6565 1578,5192 1663,5192 2004,6565 85 0,0836 0,8846
table 4-5: criteria classes
Overview of the segments
An overview of the segment is given in order to interpret the differences better. Furthermore, in table 4-7 is indicated how important the classes consider the studied attributes.
