0
Consumer Complaint Behaviour and Out-of-Stock:
How do consumers respond to a Promotional Out-of-Stock?
Martijn Jakobs
Master Thesis
1
Consumer Complaint Behaviour and Out-of-Stock:
How do consumers respond to a Promotional Out-of-Stock?
ABSTRACT
Using a sample of Dutch consumers this study empirically investigates how consumers respond to a promotional out-of-stock (POOS) and what the antecedents are that explain these responses. The construct of Consumer Complaint Behaviour (CCB), consisting of Voice and Private CCB, is applied to measure POOS responses. It is found that the larger the Hedonic Level of the product, the larger the likelihood of consumers to express their feelings to the store
employee/manager. Contrary to Hedonic Level, a higher Brand Equity causes consumers to complain more in their private environment. In addition it is found that the following antecedents; Regular Price, Discount %, Store Type, Shopping Trip Costs, Folder Read
Up-Front, and Education explain Voice or Private CCB.
2 Table of Contents:
1 Introduction ... 5
1.1 Promotions ... 5
1.2 Out-of-stock ... 7
1.3 Contribution and relevance ... 8
2 Literature review ... 10
2.1 Consumer OOS reactions ... 10
2.2 Antecedents ... 12
2.2.1 Product-related Antecedents ... 12
2.2.2 Store-related Antecedents ... 13
2.2.3 Situation-related Antecedents ... 14
2.2.4 Consumer-related Antecedents ... 14
3 Conceptual model and Hypotheses ... 16
3.1 Hedonic Level hypotheses ... 16
3.2 Brand Equity hypotheses ... 17
3.3 Interaction effect hypotheses ... 18
3.4 Product, Store, Situation, and Consumer related antecedents ... 19
3.4.1 Product-related antecedents ... 19 3.4.2 Store-related antecedents ... 20 3.4.3 Situation-related antecedents ... 21 3.4.4 Consumer-related antecedents ... 21 4 Research methodology ... 23 4.1 Data collection ... 23 4.2 Dependent variable ... 24
4.3 Main independent variables ... 25
4.4 Other IV’s ... 26
4.4.1 Product related questions ... 26
3
4.4.3 Situation related questions ... 27
4.4.4 Consumer related questions ... 27
5 Descriptive statistics ... 28
5.1 Sample ... 28
5.2 Socio demographics ... 28
5.3 Store visits ... 29
5.4 Representativeness of the research ... 29
6 Empirical results ... 31
6.1 Reliability check ... 31
6.2 Correlation analysis CCB ... 33
6.3 Regression Analysis ... 35
6.3.1 Voice CCB regression analysis ... 36
6.3.2 Private CCB regression analysis ... 37
6.3.3 Voice CCB regression analysis with interaction effect ... 38
6.3.4 Private CCB regression analysis with interaction effect ... 39
6.4 Hypothesized effects ... 40
7 Discussion ... 41
7.1 Antecedents ... 41
7.1.1 Product related antecedents ... 41
7.1.2 Store related antecedents ... 42
7.1.3 Situation related antecedents ... 42
7.1.4 Consumer related antecedents ... 42
8 Managerial implications ... 44
8.1 Implications for retailers ... 44
8.2 Implications for manufacturers ... 45
9 Limitations and further research ... 46
10 Acknowledgements ... 47
4
11.1 Appendix 1.0: Questionnaire ... 49
11.2 Appendix 2.0: Hedonic level average ... 53
11.3 Appendix 3.0: Product overview ... 54
11.4 Appendix 4.0: Consistency check ... 55
11.4.1 Voice CCB1 ... 55 11.4.2 Private CCB1 ... 55 11.4.3 Voice CCB2 ... 55 11.4.4 Private CCB2 ... 56 11.4.5 Brand Loyalty ... 56 11.4.6 Involvement ... 56 11.4.7 Brand Equity ... 57
11.4.8 Deal value perception ... 57
11.4.9 Store Loyalty ... 57
11.5 Appendix 5.1: Intercorrelations, variable abbreviations ... 58
11.6 Appendix 5.2: Intercorrelations Table* ... 59
11.7 Appendix 6.1: Voice CCB Regression ... 60
11.8 Appendix 6.2: Private CCB Regression ... 61
11.9 Appendix 6.3: Voice CCB and interaction effect Regression ... 62
11.10 Appendix 6.4: Private CCB and interaction effect Regression ... 63
5 1 Introduction
Over the past couple of years the Dutch consumer has been confronted with a major promotional price war in supermarkets and drug-stores that started in 2003 (Van Heerde, Gijsbrechts and Pauwels 2008). Promotions are of course a major tool in communicating the value of your retail environment towards consumers and have therefore been, and still are, extensively used jointly by retailers and manufacturers to attract as many consumers as possible. The price-war has made consumers more price conscious to weekly prices, and store price image resulting in the redistribution of consumer shopping across stores (Van Heerde, Gijsbrechts and Pauwels 2008). Promotional products thus became more important to retailers, manufacturers and consumers. A side effect of a promotion is the distorted view and sales figures generated in retail stores, making it harder to predict the needed inventories and allocated shelf space to certain products. This automatically leads to more out-of-stocks (OOS).
1.1 Promotions
6 users or change purchasing behaviour including increasing store traffic (Jobber 2004). These promotions might include, but are not limited to price promotions.
Every promotion varies in importance to consumers in terms of the characteristics of: (1) Promotion (large discount, presence of features, buy one get one (BOGO), X discount %), (2) The brand (e.g. unit share, price, and advertising), (3) The category (e.g. penetration, distribution, and concentration), and (4) Store (type of store, demographics of the market, and competitor’s density) (Ailawadi, Harlam, César & Trounce 2006).
Since 2008 consumers have become more sensitive to the value of products and specific promotions most likely caused by the economic recession. This is evidenced by the increasing interest in supermarket promotions and the increase in the number of different supermarket formulas visited per month. This suggests that consumers have started to actively purchase promotional products at different supermarket formulas and thus redistribute their purchases (ConsumentenTrends 2010). The increased sensitivity of consumers also affects business of both the retailer and the manufacturer.
A substantial amount of empirical literature determined that price promotions result in a substantial sales increase at the brand level (Blattberg et al. 1995, Van Heerde 1999). The increase in sales could be caused by within category brand switching (Gupta 1988) or category-expansion (Van Heerde 1999). These short-term effects can also have a negative long-term effect, such as post-promotion dips caused by consumers’ stock-piling large quantities. (Van Heerde et al. 2000). A positive effect for brand manufacturers is that consumers can only spend their Euro once, so the consumer will not and is not able to purchase a competing product. In addition it is found that consumers get accustomed to consume products they stockpile (Ailawadi et al. 2007). According to Nijs et al. (2001) the main short- and long-term effect of price promotions on business is to defend the current market share in that particular category. Large global players with sufficient funds can thus simply buy and sustain their current positions without losing any market share to new entrants (Nijs et al. 2001).
7 al. 2000; Burnham et al. 2003). Because consumers see switching stores or a brand as a cost, and might only be willing to switch with the help of an incentive such as a promotion, the switch can be seen as a very deliberate and well-thought of decision and must be very hard for a consumer to apply. It must be said however that consumers evaluate gains and losses in relative rather than in absolute terms, resulting in large variations in the rate at which money is exchanged for other things (Kahneman and Tversky 1984). In other words, there is a large difference in the evaluation of gains and losses between different products and services.
What would happen if a consumer deliberately made a choice to go to a supermarket to purchase a specific promotion, and finds the promotion to be out-of-stock (POOS)
1.2 Out-of-stock
OOS is a well-known phenomenon; a substantial amount of literature has been dedicated to this subject (Schary and Christopher 1979; Thayer 1989; Emmelhainz et al. 1991; Sloot et al. 2005; Musalem et al. 2010). Although new technology allows us to forecast sales, orders new products at certain critical stock levels, and retail environments are able to restock relatively fast, OOS are still common today. A product availability research amongst supermarkets in the U.S. for instance found that in 8 categories 8.2% of the times products were OOS on a typical afternoon (Anderson Consulting 1996). In some categories this percentage was even higher e.g. yoghurt (11.1%), bottled water (10.7%), and chilled juice (10.0%).
High pressure has been put on inventory managers to keep inventory as low as possible. Although this seems a viable and easy way to save money for retailers, working with minimum inventory levels increases the probability of OOS. The costs of OOS are usually not considered when inventory costs are calculated. The direct impact of OOS is a loss of revenue to the supplier of the store, but also to the store itself (Schary and Christopher 1979). Thayer (1989) claims that the increased pressure on inventory and from financial managers to keep stock as low as possible causes the current level of OOS.
8 Most of the OOS literature is empirical in nature, and more recent literature has made a distinction between the OOS direct impact (Category sales & profit) and indirect impact (Customer satisfaction, Store loyalty and retail image).
The direct impact of this troublesome phenomenon has been researched extensively over the past couple of years, and all authors agree on at least one important aspect; OOS is bad for business. It is bad for the retailer who can face losses up to 14% of consumers intending to buy the missing product (Emmelhainz et al. 1991), but in most cases it is even worse for manufacturers. Emmelhainz et al. (1991) showed that a single OOS can lead to a loss of more than 50% of consumers to competitors. Besides the loss of sales on the particular product, it also has a negative effect on other product categories (Campo et al., 2000). This means lower sales for the manufacturer and possibly increased sales for the retailer on other brands from the same category (Peckham 1963; Walter and Grabner 1975; Zinszer and Lesser 1981; Emmelhainz et al. 1991). This leads to a distorted sales figure and severe forecasting problems for retailers (Fitzsimons, 2000).
Limited research has been dedicated to the indirect impact of OOS. In addition it is unknown how consumers respond to a promotional OOS (Hereafter POOS). Therefore the aim of this paper is to empirically research the impact of POOS.
The main research question is:
1. How do consumers respond to a promotional out-of-stock and what are the antecedents partly explaining these responses?
1.3 Contribution and relevance
9 This research is just as valuable for retailers as it is for manufacturers because they often share the costs of promotions, either by lowering prices, and thereby reducing net margins, or by losing customers to other retail stores or brands (Campo et al. 2003). In addition POOS are the number one irritation amongst Dutch consumers at this very moment (ConsumentenTrends 2010). In addition manufacturers and retailers annually spend billions of Euro’s on marketing programs and a large part of consumer marketing is directed to satisfy consumer needs. POOS’s endanger this marketing objective.
To my knowledge this is the first time that a combination of both POOS and CCB is employed to consumer behavioural research in the Netherlands.
10 2 Literature review
In this section the relevant theoretical background is presented. It consists of two sections. Section one discusses regular Consumer OOS reactions and Consumer Complaint Behaviour. Section two discusses the antecedents that influence POOS.
2.1 Consumer OOS reactions
Besides the effects for businesses mentioned earlier, OOS also has an effect on consumers and their behaviour. A substantial number of literature researched OOS reactions in different situations, either through a field experiment (Emmelhainz et al. 1991; Verbeke et al. 1998), a survey (Walter and Grabner 1975; Campo et al. 2000; Sloot et al. 2005), a Quasi-experiment (Schary and Christopher 1979; Zinn and Liu 2001), or a laboratory experiment (Fitzsimons 2000). These studies also varied by providing a true stock-out situation or by a hypothetical situation. All researchers found a large number of behavioural consumers responses that could be grouped into six main behavioural responses: (1) Store Switch: visiting another store at the same day, (2) Item Switch: change the format or variety of the brand the consumer intended to buy, (3) Postpone Purchase: postpone the buy at that moment in time to the next store visit, (4) Cancel Purchase: completely drop the intended purchase from the list, or postpone the purchase for a longer period, (5) Category Switch: change the intended item for another product and another product category (Substitute items), and (6) Brand Switch: purchase a different brand but the same product (Sloot et al. 2005). Although there are six different behavioural consumer responses, only four of them are considered and researched as being significantly important; Store Switch, Item Switch, Brand Switch, and Postpone Purchase (Sloot et al. 2005). Research furthermore shows that item switching is the most common consumer response followed by the switching of the product’s package (size) (Campo et al. 2000).
11 between CCB and the first construct is that CCB is an effect of a dissatisfying situation that results in some kind of complaint behaviour from how one feels and what one does with that feeling.
According to Singh (1988) CCB consists of three parts: (1) Voice CCB, which is directed towards the consumer’s informal relationships or external social circle, but are involved in the exchange that caused the particular dissatisfaction (e.g. the seller), (2) Third Party CCB, are also informal relationships but do not have anything to do with the exchange that caused the dissatisfaction (e.g. Newspaper, legal agencies), (3) Private CCB, are relationships in the consumer’s social circle and are not involved in the transaction/exchange that lead to the dissatisfied experience (e.g. The consumer self, their friends and relatives) (Singh 1988). Broadbridge and Marshall (1995) define the construct as follows; (1) Do nothing, (2) Take Private action by switching brand or suppliers, boycotting the product/service, or warning family and friends, (3) Take public action by seeking direct redress from the retailer or manufacturer, bringing legal action, complaining to the media or registering a complaint with a consumer association. Broadbridge and Marshall (1995) construct differs from Singh (1988) in that they combine Voice and Third Party CCB into Public Action, and take ‘Do nothing’ from Voice CCB and make it a separate variable.
In a regular OOS situation consumers can respond by using either or both of the constructs presented above. In a POOS situation however consumers can always decide to switch stores (In the sense of different retail brands), but they will not always find the same promotion they intended to purchase. Therefore it is likely that consumers will behave towards the store or product in terms of CCB. CCB constructs are commonly used in situations where consumers are confronted with products that do not comply with what they requested or paid for. The use of the CCB construct as presented in this research is fairly new. Broadbridge and Marshall (1995), in a research about electronic appliances, found that the smaller and the less expensive goods purchased by consumers are, the less they will complain. Although their research focussed on electronic appliances, 92% of the respondents said to perceive these products as “essential rather than luxury items”. As groceries can also be considered to be essential items, the application of the CCB construct for the presented research is justified. This claim is supported by Day and Landon (1977).
12 dimensions. This allows the presented research to apply only two of the three dimensions as described above (Singh 1988). Therefore only Voice and Private CCB are used in this research. CCB is a formative construct which means that the latent construct is a combination of different indicators defining the construct without assumptions of intercorrelations between these items (Coltman et al. 2007). In the presented research CCB is formed by Voice and Private CCB. Voice and Private CCB are however formed independently and are a construct in itself, formed by three questions.
The focus of consumer complaints is an important matter in research; complaints can focus on the product (functional characteristics) or on non-functional characteristics of the promotional mix (E.g. the retail environment, the manufacturer, communication material and price promotion) (Jacoby and Jaccard 1981). This paper focuses solely on the non-functional characteristic of the product, and in particular on price promotion.
2.2 Antecedents
As discussed above prior research on OOS focussed solely on the type of consumer response, so the actual behaviour of consumers when they are confronted with an OOS situation. Although the nature of the POOS situation is different from prior OOS literature, the presented research does focus on the same antecedent clusters. The antecedent clusters are: (1) Product-related, (2) Store-related, (3) Situation-related, (4) Consumer-related (Campo et al. 2000; Zinn and Liu 2001; Sloot et al. 2005). To be able to distinguish between the four types of antecedents it is vital to determine the characteristics of the antecedents first.
2.2.1 Product-related Antecedents
Product related antecedents are variables related to a specific product category including the brand for which the POOS appears (Sloot et al. 2005).
Brand or item loyalty is an antecedent extensively used in prior research and found to have a significant impact on consumer behaviour in an OOS situation. It is defined as the consumer’s general tendency to be loyal to a certain brand or item, demonstrated by the consumer’s intention to acquire the brand/item as their primary choice (Oliver 1997). The higher the consumer’s Brand/Item Loyalty, the less likely it is that he or she will acquire a different brand. Consumers are more likely to go to another store and purchase the item there (Campo et al. 2000; Sloot et al. 2005).
13 products e.g. Coca-Cola, if the Coca-Cola Brand is on a promotion and the regular taste is POOS, a consumer might purchase a substitute item like Coca-Cola light. Heineken has only one flavour in different sizes and packages. In general only the crate of Heineken is on a promotion, which means that at the time of a POOS consumers do not have the ability to purchase another Heineken taste. (Assuming that only the crates or on a promotion). Consumers do have a choice with Coca-Cola or Lays chips because these brands have multiple substitute products which are all part of the promotion.
2.2.2 Store-related Antecedents
Store-related antecedents are variables which are related to the store but could also be related to the chain of stores (retail chain) in which the OOS occurs (Sloot et al. 2005).
According to Campo et al. (2000) loyalty is an important variable influencing the decision of the consumer at the time of OOS. This is furthermore supported by Emmelhainz et al. (1991), and Corstjens and Corstjens (1995). In their research they found that loyalty to either the store or the missing brand item leads to a substantial decrease of store and item switching. If acceptable alternative are present consumers are more willing to switch to another product or package size. In addition there is a significant difference between the mechanisms resulting in package size switching and item switching. The ConsumentenTrends 2010 research found an increase in the frequency of supermarket visits by Dutch consumers. Asking consumers how many supermarkets they visit provides a good indication of Store Loyalty. Store Loyalty is an important variable and although research by Sloot et al. (2005) found weak significant evidence for OOS, it is unclear what the influence of Store Loyalty on CCB will be at the time of a POOS.
14 Table 01: Supermarket formulas ConsumentenTrends 2010
Orientation Description Example
Full-Service Large assortment, high service level, high price
Albert Heijn, Plus, Super de Boer
Local supermarket / Convenience store
Decent service, close-by, high prices Coop, Golff, Spar
Value-for-money Medium service, medium price, C1000, Dekamarkt, Em-Té, Jan Linders, MCD, Poiesz, Sanders, Vomar
Service Discount* Above average service with attractive consumer prices
Deen, Hoogvliet, Jumbo, Supercoop
Brand Discount* Average service, attractive consumer prices
Boni, Bas van der Heijden, Digros, Dirk van den Broek, Nettorama
Hard discount Rock bottom price, less to no service Aldi, Lidl
*Service and Brand discount together are also referred to as the quality discount segment
2.2.3 Situation-related Antecedents
Situation related antecedents relate to a specific shopping trip in which the OOS situation occurs (Sloot et al. 2005). In addition to the regular factors being researched (E.g. type of promotion and OOS), it is also important to consider other factors influencing the impact of POOS on consumers. Consumers are also influenced by specific characteristics of their shopping trip. Campo et al. (2000) researched numerous antecedents influencing consumers during their trips such as: (1) Buying urgency; if consumers require the product immediately they will not postpone their purchase, and (2) Shopping trip type; if consumers have decided to do their monthly/weekly shopping trip, they might devote a lot of time to the process. On the other hand if they have decided to do daily shopping trips they will not devote a lot of time to their shopping process. Campo et al. (2000) found both antecedents to be significant, but not of high importance to consumers.
2.2.4 Consumer-related Antecedents
15 complaints. Household size can also have an influence on consumer behaviour (Inman et al. 2009)
All problems identified in the POOS section of this research can differ in magnitude. They can vary between product, consumer, and situation factors. A large number of consumer perceptions regarding their purchase and post-purchase behaviour have developed in recent years. An important product consumer classification that has evolved from sociological research is the level of involvement, which was introduced into the field of marketing by Krugmann (1966). Since then many types of involvement have been identified and confusion arose among researchers (Zaichkowsky 1985). According to the literature consumers can be involved with advertisements (Krugmann 1966), products (Howard and Sheth 1969), or with purchase decisions (Clarke and Belk 1978), another construct explains situational and enduring involvement. Situational involvement considers a specific situation in which involvement plays a role (e.g. a certain purchase situation). Enduring involvement, as the name says, has to do with a continuous level of involvement towards specific objects (Houstan and Rothschild 1978; Andrews, Durvasula and Akhter 1990; Day, Stafford and Camacho 1995). Most researchers define involvement as the state of a consumer: “A motivational state influenced by a person’s perception of the object’s relevance based on inherent needs, values, and interests, influenced by factors such as the characteristics of the person, object/stimulus and the situation” (Zaichkowsky 1985; Day, Stafford and Camacho 1995). An important distinction has been made between high and low level of involvement. Under high involvement products have greater personal relevance and consequences or elicit more personal connections than low involvement products (Petty, Cacioppo, and Schumann 1983).
16 3 Conceptual model and Hypotheses
The objective of this study is (1) to identify the types of CCB expressed by consumers at the time of a POOS, and (2) to examine the relationship between CCB and related antecedents (Product, Store, Situation, and Consumer). The conceptual model is displayed in Figure 1. The model’s main focus is on the relationship between the Hedonic Level and CCB, the relationship between Brand Equity and CCB, and the relationship between Hedonic Level and Brand Equity on CCB. In addition to the man moderating values (Hedonic Level and Brand Equity) antecedents are employed that are of importance to the CCB construct.
3.1 Hedonic Level hypotheses
17 manage to transform a purely Utilitarian product into a product that also offers Hedonic benefits to consumers, e.g. regular soap. The Utilitarian benefit to clean hands is transformed into a product offering Utilitarian and Hedonic benefits by the addition of scents and colours. The main difference between both types is that Hedonic products are purchased more often on the basis of emotions, whereas Utilitarian products, which in essence perform a practical and functional task, are purchased on rational motives (Sloot et al. 2005). Sloot et al. (2005) also found that OOS for Utilitarian products resulted more in Brand Switching. The main explanation to switch brands for a Utilitarian product, for instance a detergent, is that consumers are unwilling to postpone their laundry, indicating a lower loyalty towards Utilitarian products. An OOS of a Hedonic product results in more store switching, indicating that consumers are loyal to their product and are willing to travel to acquire it (Sloot et al. 2005). The behaviour of consumers at the time of an OOS is fairly well researched; however it is unclear how consumers respond to a POOS.
When a POOS occurs consumers cannot simply visit another supermarket to acquire their favourite product under the same promotional conditions. It is therefore to be expected that the higher the Hedonic level of a product the more likely consumers will engage in either Voice or Private CCB.
This leads to the following hypotheses:
H1a: In case of a Promotional Out-Of-Stock the Hedonic Level of a product increases the probability of Voice CCB
H1b: In case of a Promotional Out-Of-Stock the Hedonic Level of a product increases the probability of Private CCB
3.2 Brand Equity hypotheses
18 equity (Leone et al. 2006). One major commonality between large numbers of literature discussing Brand Equity is that Brand Equity is the incremental value of a product due to the brand name (Yoo et al. 2000). Customer Based Brand Equity (CBBE) is defined as: “Consumers’ different response between a focal brand and an unbranded product when both have the same level of marketing stimuli and product attributes. The difference in consumer response may be attributed to the brand name and demonstrates the effects of the long-term marketing invested into the brand” (Yoo and Donthu 2001). Another useful definition and distinction has been provided by Chandon et al. (2000). They provide the term ‘Brand Equity’ with a high and a low distinction. The less favourable consumers react, the lower the Brand Equity. The more favourably people react towards the brand, the higher the Brand Equity (Keller 2002).
Thus consumers have a strong dedication to acquire high Brand Equity products. It is therefore expected that higher Brand Equity POOS’s increase the likelihood of Voice and Private CCB.
This leads to the following hypotheses:
H2a: In case of a Promotional Out-Of-Stock the higher the Brand Equity level of a product the higher the probability of Voice CCB
H2b: In case of a Promotional Out-Of-Stock the higher the Brand Equity level of a product the higher the probability of Private CCB
3.3 Interaction effect hypotheses
19 Brand Equity and Hedonic Level, the more favourable consumers respond to a product. Therefore the following hypotheses have been developed.
H3a: In case of a Promotional Out-Of-Stock the Hedonic Level of a product increases the negative effect of Brand Equity on Voice CCB probability
H3b: In case of a Promotional Out-Of-Stock the Hedonic Level of a product increases the negative effect of Brand Equity on Private CCB probability
3.4 Product, Store, Situation, and Consumer related antecedents
In this section the effects of each antecedent on Voice and Private CCB in case of a POOS will be discussed.
3.4.1 Product-related antecedents
Brand Loyalty is one of the antecedents researched because it shows the loyalty of a consumer to a specific brand, which differs between consumers. In line with prior research a POOS of a product with a high Brand Loyalty (assuming the product is not on promotion in a different store) increases the likelihood of Voice or Private CCB (Oliver 1997; Campo et al. 2000; Sloot et al. 2005).
Type of Product: There is limited empirical evidence for the difference between Food and Non-food products. Day and Ash (1979) found that consumers are more likely to complain to the store or store manager (Voice CCB) for durable goods. As Food is a non-durable good, it is expected that Food products result in a higher likelihood of Voice or Private CCB then 8on-Food products.
Purchase Frequency: The number of times the respondent purchases the product per month. The more often the consumer purchases the product, the more disappointed the respondent will be when the product is POOS (Bawa & Shoemaker 1987; Sloot et al. 2005). Therefore it is expected that the higher the Purchase Frequency the higher the likelihood of Private or Voice CCB.
Brand Preference: Brand Preference refers to the preference consumers have towards the brand (Campo et al. 2000; Sloot et al. 2005). It is expected that Brand Preference will result in a higher likelihood of Voice or Private CCB.
20 product (Coca-Cola light instead of Regular) within the same brand. It is therefore expected that the substitutability of products decrease the likelihood of Voice or Private CCB.
Discount %, and Price: A higher discount, in either Euro or Percentages makes a deal more interesting (Ailawadi, Harlam, César & Trounce 2006). The more interesting a deal is the more dissatisfied a consumer will be if it is POOS. Therefore the higher the numbers are for all three variables, the more likely the consumer will engage in Private or Voice CCB.
3.4.2 Store-related antecedents
Store-loyalty was found to help consumers come back to the store, or at least stay with the store at times of an OOS (Emmelhainz et al 1991; Corstjens and Corstjens 1995). But what are the effects of Store Loyalty when a POOS occurs. Jap et al. (2000) and Burnham et al. (2003) found that high Store Loyalty leads to high Switching Costs. Therefore it is to be expected that the higher the store-loyalty, the lower the likelihood of Private or Voice CCB.
# Supermarket Visits: As identified in the literature consumers are visiting more supermarkets (Consumententrends 2010). This could be an indication of cherry picking, which means consumers are actively seeking for promotions and go to different stores to purchase these promotions. Assuming the particular product is not on promotion in another store the likelihood of disappointment is higher. Therefore it is to be expected that the higher the number of supermarket visits the higher the likelihood of Voice or Private CCB.
# Different Store Visits: This antecedent also measures whether consumers are loyal to a specific store, the more stores they visit, and the less loyal they are. Therefore it is expected that the higher the number of store visits, the higher the likelihood of Voice or Private CCB.
21 3.4.3 Situation-related antecedents
Shopping Trip Costs: Campo et al. (2000) found a significant difference between a minor or major shopping trip. They state that consumers in a major shopping trip have dedicated that particular time to their shopping. Due to their low shopping frequency they do not want to travel again to acquire a product that is POOS. Consumers on a minor shopping trip are frequent shoppers and if faced with a POOS will try to acquire the product during their next visit. Consumers however did not find Shopping Trip Type an important variable (Campo et al. 2000). In addition Sloot et al. (2005) did not find any significant relationship with the type of shopping trip and OOS. Shopping Trip Costs have been selected to measure this variable because a large or small shopping trip is a relative figure; costs are very absolute and measured in Euros which are the same for everyone. A research by Verbeke et al. (1998) showed that consumers who only spend small amounts of money are willing to postpone the purchase, and just come back at a later date. Based on the literature it is to be expected that the larger the Shopping Trip Costs the higher the likelihood of Voice or Private CCB will be.
Folder Usage: If a consumer reads the promotional folder up-front he or she might have planned, or at least was aware of the promotion. If this product is POOS the expectation is that the consumer will be more likely to engage in Voice or Private CCB.
Purchase decision: The decision to purchase the product on promotion in-store, also known as an impulse purchase (Narasimhan et al. 1996; Beatty and Ferrel 1998; Sloot et al. 2005). Sharma et al. (2010) found that the larger the impulsivity of the purchase the larger the complaint behaviour. Therefore if the purchase decision is made In-Store the consumer will be more likely to engage in Voice or Private CCB.
#Promoted products bought: If consumers have purchased a large number of products on promotion, they might not think it is important that one of the products is POOS, because they were able to purchase the majority of promotional products. Although no literature supports either a positive or negative relationship, it is expected that the more products on promotion the consumer has purchased, the less likely they will engage in either Voice or Private CCB.
3.4.4 Consumer-related antecedents
22 found to be willing to engage more in CCB than other identified groups (Fails and Francis, 1996). This means that the higher the income, the higher the educational background, and the lower the age, the higher the likelihood of either Voice or Private CCB will be.
Household size has been added to these three because one could imagine that a large household needs more groceries, and if every other variable stays equal they have less net income to spend, and thus might need to rely more on promotions. Therefore the larger the household the more likely the consumer will engage in either Voice or Private CCB.
Product involvement: When consumers are more involved into a particular product, service or even with a consumption situation the more likely it is that he or she is willing to devote time, effort, and money to complain to the retailer (Lau & Ng 2001, Sharma et al. 2010). The opposite occurs when consumer involvement is low (Chebat et al. 2005). It is therefore expected that the higher the consumer’s product involvement the more likely it is that he or she will engage in either Voice or Private CCB.
Deal-value perception: The deeper the promotion, the higher the perceived deal-value. (Darke and Chung 2005), this implies that when this deal would be POOS consumers will be more disappointed resulting in a higher likelihood of either Voice or Private CCB.
For clarification purposes the expectations explained above are summed in the Table 02. There are two possibilities for expectations displayed in the table: (1) +, means that if the variable increases it will increase the probability of CCB, (2) -, means that if the variable increases it will decrease the probability of CCB.
Table 02: Expected relationship between independent variables and CCB Product related antecedents Store related antecedents Variable Relationship with CCB
(Voice and Private)
Variable Relationship with CCB (Voice and Private)
Brand Loyalty + Store Loyalty - Type of product + # Supermarket Visits + Purchase Frequency + # Different Store Visits + Brand Preference + Store-Type VfM - Substitutability - Store-Type Service + Discount Percentage +
Regular Price +
Situation related antecedents Consumer related antecedents Variable Relationship with CCB
(Voice and Private)
Variable Relationship with CCB (Voice and Private)
23 4 Research methodology
4.1 Data collection
Dutch supermarkets were the scene of the data collection process. A structured questionnaire was used to interview a random sample of respondents at the exit of the supermarket (exit interviews). A questionnaire is a proper marketing management research tool in order to provide this research with information regarding CCB and its antecedents. This research used a hypothetical research setting in which consumers were asked to imagine they were confronted with a POOS with one of the products they purchased a few moments ago. Hypothetical out of stocks have been used by researchers before (Campo et al., 2000; Sloot et al., 2005) and there has been some debate on the reliability of a hypothetical situation. The drawbacks are that consumers are not always able to accurately recall such a situation, and that consumers sometimes try to answer in a socially responsible way (Campo et al., 2000; Sloot et al., 2005). The positive side of a hypothetical situation is that it allows the researcher to increase the scope of the research, and thereby reach more respondents, adding up to the reliability of the research. Furthermore this approach has been used by a large number of researchers and shown to be an appropriate research approach (Singh and Wilkes 1996; Campo et al. 2000; Kim et al., 2003; Sloot et al. 2005). Other settings (E.g. a lab setting) have their own drawbacks that could harm the outcome of this research. The questionnaire can be found in the appendix (Appendix 1.0)
To be certain that the questionnaire worked as it was meant to work Pre-test has been used. The environment of this test should come as close to the real survey as possible (Malhotra 2007). The Pre-test outcome showed that minor changes had to be made to make the questionnaire easier understandable for respondents. The 1-7 Likert –scale questions were all positioned in the centre of the questionnaire without any ‘regular’ questions in between. In the pre-test it was sometimes difficult for respondents to go back to the Likert 1-7 rating when a regular question was asked in between. This change increased the natural flow of the questionnaire. Also it was decided that specific product/promotion characteristics would be measured by the researcher afterwards, thereby shortening the time needed to perform the questionnaire. The pre-test also provided indications that consumers did not want to complain when they had only been faced with one POOS.
24 shopping baskets were also visually inspected in order to increase the likelihood of respondents to participate. The main advantage exit interview is that consumers can most accurately recall the purchase situation.
Data collection took place near three different supermarkets, Albert Heijn, Super de Boer and C1000. It was not possible to target consumers visiting hard discounts because in general these supermarkets do not sell ‘branded’ products, and are therefore difficult to compare to consumers who purchased ‘branded’ products. The consumers were monitored during four consecutive days, (Wednesday, Thursday, Friday, Saturday), and during two weeks. Another advantage of the Thursday, Friday, and Saturday is the higher likelihood of POOS. These days are generally the busiest days in supermarkets (Consumententrends 2010), thereby putting more pressure on the supermarket employees to keep all products in the shelves.
4.2 Dependent variable
On the basis of prior research, two pillars of one construct have been defined and will be used as the dependent variables of this research. CCB consists of Voice CCB and Private CCB. After the particular promotional product purchased by the consumer was mentioned, and a number of general questions concerning the variables were asked, the consumer was asked how he/she would react when that particular product was POOS. The respondent could provide answers to six main questions on a 7-point Likert scale. The following possible complaint behaviours had to be graded by the respondents. ‘1’ is highly unlikely they would respond in such manner, and ‘7’ is highly likely they would respond in such manner. As discussed before Voice and Private CCB consist of two sets of three different questions. These six questions are based on the research of Singh (1988). Singh (1988) tested a substantial amount of questions on their fit with CCB. Similar questions like the first three questions (1, 2, and 3) were found to best represent Voice CCB. The other three questions were found to best represent Private CCB (Singh, 1988). Singh’s (1988) research tested the questions in four general store environments including a supermarket environment. Voice CCB consists of: (1) I don’t care, (2) Complain to the supermarket manager, (3) Ask a supermarket employee why the promotion is not in stock. Private CCB consists of: (1) Decide not to use this supermarket again, (2) Speak to your friends and relatives about the POOS, (3) Convince your friends not to use the supermarket again.
25 For instance: “Could you imagine that another consumer/customer...followed by question 4, 5, 6. To emphasize that the question was not personal to the respondent it had specifically been brought to the attention that all three questions related to other consumers.
Because the pre-test provided indications that some consumers did not found the POOS incident to be worth complaining about, supported by Hirschman (1970). In his research he found indications that consumers sometimes do not want to complain because they are too loyal to the organization, even if they want to complain. Furthermore some consumers think their complaints will not matter and find it ‘fruitless’ to complain (Hirschman 1970). Therefore the same questions have been asked with a minor change in that the POOS has occurred for the 10th time, putting the respondent into the position where he/she thinks; “that’s enough”.
4.3 Main independent variables
In the general model three variables are hypothesized. After the data collection finished, the researcher consulted six Hedonic/Utilitarian experts to rate the collected products on their Hedonic Level from 1-7, where ‘1’ = not hedonic, and ‘7’ is completely Hedonic. The same products were also measured on their Utilitarian level from 1-7, where ‘1’ =not Utilitarian, and ‘7’ is completely Utilitarian. In a correlation analysis the two variables showed a very strong significant negative correlation (r= -.923; p=.000) indicating that the two variables are almost the exact opposite. This is according to prior research. In general the more Hedonic, the less Utilitarian a product is and vice-versa (Sloot et al. 2005). Sloot et al. (2005) summed the Hedonic and Utilitarian levels and created a new variable which is the Hedonic Level of the selected products. The presented research decided to use the same approach. The individual Hedonic and Utilitarian scores of each product are given in the appendix (Appendix 2.0).
The other dependent variable is Brand Equity, which will be measured by three main questions (Chandon et al., 2000, Sloot et al., 2005), (1) Perceived product price level, (2) Perceived product quality, (3) Preferences for the brand. All three questions will be measured on a seven- point Likert scale.
26 4.4 Other IV’s
4.4.1 Product related questions
Brand Loyalty: Overall item loyalty was measured by a scale used by Campo et al. (2000), and based on Baumgartner and Steenkamp (1996). Three main questions were asked: (1) “I think of myself as a loyal buyer of detergents (category / item)” (2), “I would rather stick with a brand I usually buy than try something I am not sure of”, (3) “I like to switch between different brands of…. (Category/item)”. All three questions were measured by a seven-point Likert scale.
Type of Product: The respondents were asked which products they had bought on a promotion; one of these products was the topic of discussion during the survey. Later these products were classified as 8on-Food or Food and changed into a dummy variable (0/1). An overview of the products can be found in the appendix (Appendix 3.0).
Purchase Frequency: Was measured by asking how often the respondent purchases this product per month on average.
Brand Preference: Was measured by asking the respondent if the acquired product also is the preferred brand, no/yes, will be recorded by dummy variables (0/1).
Substitutability: Was measured afterwards by the researcher, the number of substitutes was measured from 0 up to 5 or more.
Discount %: Was recorded by the researcher in percentages.
Regular Price: Was also recorded by the researcher to determine the initial price (Ailawadi et al. 2006) in real-euro value.
4.4.2 Store related questions
Store Loyalty: Similar to the Item loyalty measure general Store Loyalty was measured using three basic questions in the loyalty context. They are: (1) “I think of myself as a loyal customer to my supermarket”, (2) “I would rather stay with the supermarket I usually visit, than trying a different store I’m not very sure of”, (3) “I like to switch between different supermarkets”. The first two questions need a high grade to show Store Loyalty, the third question needed a low grade to show Store Loyalty (Campo et al., 2000) all were measured on a seven-point Likert scale.
27 Store-Type Visit (E.g. Full Service or VfM): Were measured and collected by the researcher at the different supermarkets (Consumententrends 2010), and replaced by a dummy variable (0/1).
4.4.3 Situation related questions
Shopping Trip Costs: Total cost of the shopping trip in Euros (Sloot et al. 2005).
Folder Usage: The respondent was asked when and if he/she has read the promotional folder with answer possibilities: (1) Did 8ot Read (2) Up-Front (3) In-Store, replaced by two dummy variables: 8ot Read (0) Read Up-Front (1), and 8ot Read (0) and Read In-Store (1). Purchase decision: The respondent was asked if he or she had planned to purchase the particular promotion: (1) Up-front or (2) In-Store, replaced with a dummy variable (0/1). #Promoted products bought: Was measured by asking the respondent for the number of products including the subjected products they had purchased on a promotion.
4.4.4 Consumer related questions
The demographic variables were classified according to the latest research of the CBL and EFMI in gender (Male, female = 0/1), Education (None, Mavo/VMBO, Havo, Vwo, MBO, HBO, University), Age (18-34, 35-54, 55 and over), Income (Ranging from less than €500, to €3.500 or more per month) and Household Size (1, 2, 3 or more). Besides these ‘regular’ variables, Product Involvement was researched based on three questions derived from literature (Kim et al. 2009) and based on the product discussed at that moment: (1) “I find the brand to be of importance”, (2) “I find the right taste/function important”, (3) “This product is valuable for me”. All were measured on a seven-point Likert scale.
28 5 Descriptive statistics
This chapter will provide an overview of the results of the field research conducted as described in the previous chapter. The chapter will start with the sample and explanation of the number of respondents, supermarkets used and socio demographic figures. Then the representativeness of the presented research will be proven in comparison to the CBL – EFMI research (ConsumentenTrends 2010) and Sloot et al. (2005). Thereafter the most important parts of CCB will be presented, followed by the antecedents as described above. All data has been processed with the help of SPSS 17.0.
5.1 Sample
300+ people were asked to participate in this research of which 104 were willing to cooperate. Main reasons for not cooperating were: A lack of time, no promotional products purchased, no brand promotional products purchased. The sample (n=104) is divided amongst the three supermarket formulas described earlier, in which the C1000 has the largest share of 43% (Figure 02).
5.2 Socio demographics
At first it was expected that the majority of respondents is female, as they are the main consumers visiting supermarkets (Figure 03). Representativeness of all socio demographics will be discussed in the next paragraph where they are compared to the most recent CBL and
EFMI consumer trends research
(ConsumentenTrends 2010) and to Sloot et al. (2005).
29 The third socio demographic variable
measured is Education. The large majority of respondents have either attended or completed an MBO Education followed by a large distance of respondents who have attended or completed an HBO Education. Only a small proportion of the sample went to a University (Figure 05).
5.3 Store visits
Respondents generally visit 2-3 stores per month with an overall percentage of 74%; some respondents even visit 5 different supermarket formulas per month (Figure 06).
The majority of respondents visit a supermarket 2-3 times a week (Figure 07); another group visit the supermarket on average 4 times a week. 1% even goes to the supermarket 7 times as week, which is even more than the supermarkets are generally opened in the area of research. Although this seems not valid, it is not unthinkable
that consumers visit a supermarket twice a day, e.g. a mix between a hard discount and a full-service supermarket.
5.4 Representativeness of the research
30 provide a reasonable comparison with the ConsumentenTrends 2010, Sloot & Verhoef (2005), and Singh (1988) research. The store visit average of 2.7 times a week is almost exactly the same as the ConsumentenTrends which positively influences the representativeness of this research.
To my knowledge the construct of CCB is new to POOS situations; therefore no other referencing material is available to check the validity of this research.
Table 03: Comparing figures Variables
Consumenten Trends 2010, n=2583
Sloot & Verhoef 2005, n=749 Singh (1988), n=176 Presented Research, n=104 Gender frequency Male 29% 23% 27% 20% Female 71% 77% 73% 80% Age frequency 18-34 years 23% 32% 52% 24% 35-54 years 42% 40% 48% 50% 55 years and older 35% 28% 26%
Household size frequency
1 person
69% 59% N/A 43% 2 persons
3 or more persons 31% 39% N/A 57%
Store visit average
Frequency 2,8 times a week N/A N/A 2,7 times a week
31 6 Empirical results
This chapter will illustrate the empirical results, and will start with the correlation analyses. Before the correlation analysis, a validity check will be performed to be certain that the questions presented in the methodology represent the constructs they mean to represent. This will be determined by the internal consistency reliability (Malhotra 2007). Hereafter correlations will be made to check if there is an indication that there is a relationship between the CCB construct and the variables. Thereafter a regression analysis will be performed in order to check multiple relationships between the multiple variables.
Before one continues with the correlation analysis, it has to be noted that variables that will be researched during the correlation analysis will not necessarily be researched in the regression analysis. It is even possible that variables that do not show significant results during the correlation analysis do show a significant result during the regression analysis, and vice versa (Malhotra 2007).
Different kinds of approaches are applicable to the presented research, and as the conceptual model shows there are a large number of variables. The conceptual model also shows there are two main independent variables, Hedonic Level and Brand Equity influencing the main dependent variables Voice and Private CCB. Therefore these relationships have been hypothesized, also the interaction effect of the two independent variables on each other have been hypothesized. The antecedents will be explorative researched for clarification and structural purposes, and are not inside the hypothesized effects. Although the other variables do not have stated hypotheses, they will be analyzed by the above mentioned statistical procedures to check for their influence on CCB.
6.1 Validity check
A number of questions were reversely displayed in this research, which means that some questions were positively formulated, and others were negatively. Therefore these questions first needed to be transformed to prevent the calculations to be affected. The questions needed to change are: “I like switching between different brands”, “I regularly switch between supermarkets”, and the questions of Voice CCB1 and Voice CCB2; “I don’t really care that much”. After the transformation they can be checked for their reliability to measure their construct.
32 if there is enough internal consistency between the
questions (Malhotra 2007). It must be noted that the Cronbach’s alpha only shows internal consistency between the questions and is not a measure of validity.
The validity has however been shown and proven in the literature review and methodology. The classifications are shown in Table 04 (George & Mallery 2003).
Although the CCB construct itself is Formative in nature because it is formed (in this research) by Voice and Private CCB, the separate constructs that form CCB (Voice and Private) are reflective which means that the questions used for the constructs share the same theme and are interchangeable (Coltman et al. 2007).
In order to test the internal consistency it is recommended that the Cronbach’s alpha test is repeated with fewer questions some questions to check if two questions form a better construct then the initial three. This is possible because the inclusion or exclusion of one or more questions from the construct does not change the content validity under reflective conditions (Coltman et al. 2007). If the Cronbach’s alpha figure improves it could mean that only two questions provide a better pair, and could therefore better predict the variable. This calculation has been performed at all the constructed variables and showed that all current questions fit best together. An overview of the calculations can be found in the appendix (Appendix 4.0)
Voice CCB1: Shows a Cronbach’s alpha of .615, which is according to the above shown classification questionable. Although the figure is nearing the line of being poor it is still within the margin of acceptance.
Private CCB1: The set of questions of Private CCB1 show a slightly larger internal consistency of .653, which is still questionable but not unacceptable.
Voice CCB2: The second Voice CCB questions show a larger internal consistency, again still questionable but not unacceptable with a Cronbach’s alpha of .642.
Private CCB2: The second set of questions for Private CCB show a larger internal consistency which can be classified as good, with a Cronbach’s alpha of .805.
Brand Loyalty: Brand Loyalty is next with also three questions trying to explain for the construct. The Brand Loyalty questions show an acceptable level of internal consistency with a Cronbach’s alpha of .763.
Product Involvement: Product Involvement is the next variable with three questions, and shows an questionable but acceptable level of internal consistency of .615,
Table 04: Cronbach’s alpha and classification of internal consistency Cronbach’s Alpha Classification
α ≥ .9 Excellent / Perfect .9 > α ≥ .8 Good
.8 > α ≥ .7 Acceptable
.7 > α ≥ .6 Questionable but not unacceptable .6 > α ≥ .5 Poor
33 Brand Equity: Brand Equity measured with the respondents shows an acceptable level of internal consistency of .752 between the three questions.
Deal-Value Perception: Deal-Value Perception shows a good level of internal consistency with a Cronbach’s alpha of .811 between the three questions.
Store Loyalty: Store Loyalty also shows an acceptable level of internal consistency with a Cronbach’s alpha of .751 between the three questions.
Now it has been determined that the three questions have enough internal consistency in general, the three questions per variable can be transformed into a single variable. The option is two-fold, the results can be summed and the summed results could be used for further research. The advantage here is that the summed results show larger numbers and could therefore be more easily interpretable. The other option is to take the average of the summed results, the advantage here has to do with missing variables, if one variable is missing the average would not change and therefore calculations would not severely be damaged in comparison with the summed-only option. The data has already been checked for missing variables, and for the information to be easily interpretable the variable will be used on a scale from 1-7 (Malhotra 2007).
6.2 Correlation analysis CCB
It has to be noted that before the correlation analysis was conducted with Voice and Private CCB, the variable intercorrelations have been researched. Some of the variables were left out to increase the reliability of the data set, and thereby the correlation and regression analysis, the overall regression table has been put in the Appendix (Appendix 5.1 & 5.2) the variables that have been left out are:
Product-related: Brand Loyalty, Type of Product, Purchase Frequency, Brand Preference. Store-related: # Different Stores Visited.
Situation-related: Purchase Decision, Folder Usage In-Store. Consumer-Related: Product Involvement, Deal-Value Perception.
The selections have been based on their intercorrelations with other variables that are of greater importance to this research.
34 The Table below (Table 05 next page) shows the independent variables as above together with Voice CCB and Private CCB. The first column shows the variable, the second column shows the expected direction explained by the ‘E’ as displayed in Table 02. The third column shows the correlation, the fourth shows the significance, and the fifth and last column shows if there is enough correlation between the variables to make statements about them, columns 6,7,8, are similar to 3,4,5, but now for Private CCB (Table 05 next page). Before one has a look at the results it has to be noted that correlations do not show causality, e.g. the correlation between sales of sun cream and nice weather is obvious, and in general nice weather increases the likelihood of sun cream sales, the opposite is however not true, the sales of sun-cream does not increase the likelihood of nice weather, although correlations at the time of nice weather and high sun-cream sales will indicate this. The correlation analysis will have the same structure as the variables were introduced in this research, and start with the overall correlation table displayed below (Table 05). The variables have different colours with a purpose: Blank = no relationship, Green = a positive relationship, Red = a negative relationship.
Table 05: CCB correlations variables, n=104
Hypothesized variables Voice CCB Private CCB
E Corr. Sig. Relation Corr. Sig. Relation
Hedonic Level + ,245** ,006 Positive ,069 ,243 None Brand Equity + ,232** ,009 Positive ,225* ,011 Positive BExHEDLVL + ,280** ,002 Positive ,220* ,013 Positive
Product related antecedents
Substitutability - ,116 ,121 None ,080 ,211 None Discount % + ,246** ,006 Positive ,106 ,141 None Regular Price + ,290** ,001 Positive ,159 ,054 None
Store related antecedents
Store Loyalty - ,098 ,162 None -,046 ,320 None # Supermarket Visits + ,062 ,264 None ,080 ,211 None
Service Store-Type + ,069 ,242 None -,192* ,025 Negative
VfM Store-Type + -,069 ,242 None ,192* ,025 Negative
Situation related antecedents
Shopping Trip Costs + ,061 ,270 None ,162* ,050 Positive
8o Folder Usage - ,217* ,013 Positive -,028 ,390 8one Up-front Folder Usage + ,243** ,006 Positive -,051 ,304 8one In-store Folder Usage + -,119 ,115 8one ,159 ,054 8one
#Promoted products bought - ,115 ,123 None ,121 ,110 None
Consumer related antecedents
Female Gender - -,186* ,030 Negative -,155 ,058 None
Male Gender + ,186* ,030 Positive ,155 ,058 None
Education + ,207* ,018 Positive ,015 ,438 None Age - -,053 ,296 None ,052 ,299 None Income + -,053 ,298 None ,198* ,022 Positive Household size + -,207* ,017 Negative ,085 ,195 None
* = Correlation is significant at the 0.05 level (1-tailed) Significant
35 Hypothesized effects: From the correlation table (Table 05) it is clear that all three variables have a positive significant relationship with Voice CCB. Only Hedonic Level does not show a significant relationship with Private CCB. The variables that do show a significant relationship all behave as expected. And show a relatively large correlation with the dependent variables, as was hypothesized.
Product related antecedents: The table above shows that only Regular Price, and Discount % are positively significant to Voice CCB, and that Regular Price is close to be significant to Private CCB as well, (p:.054)
Store related antecedents: In the table (Table 05) one can see that Service Store-Type shows a negative significant correlation and VfM Store-Type shows a positive significant relationship with Private CCB. This was however not as expected, which means that when consumers are visiting a Service supermarket, they are less likely to engage in Private CCB then when they are visiting a VfM supermarket.
Situation related antecedents: Folder Usage Up-front and 8o Folder Usage both show a positive significant relationship with Voice CCB. Shopping Trip Costs shows a positive significant relationship with Private CCB.
Consumer related antecedents: From the table it is clear that Male Gender and Education are the only positively significant variables related to Voice CCB, and that Female Gender and Household size show a negative significant relationship to Voice CCB. Income is the only variable showing a positive significant relationship with Private CCB.
6.3 Regression Analysis
36 importance of the variable compared to the unstandardized beta coefficient. Also the VIF will be measured, which is the Variance Inflation Factor that explains the amount of multicollinearity. In general rules of thumb are used for the VIF variable, some say it cannot be 4 or higher, others say it cannot be 10 or higher (O’Brien 2007; Huizingh, 2008) column 7. Because for the regression analysis the variables are all measured together, and only a limited number will prove to be significant, the order of analysis will be based on the types of CCB and start with Voice CCB, then Private CCB. The complete regression tables can be found in the appendix (Appendix 6.1, 2, 3, 4)
6.3.1 Voice CCB regression analysis
The regression analysis has been performed with only those variables that do not show too large intercorrelations. The table below shows the outcomes of the regression analysis, and that five variables are positively significant to Voice CCB. (Table 07), the overall regression table is in the appendix (Appendix 6.1)
Table 07: Voice CCB regression results, n=104
Unstandardized Coefficients Standardized Coefficients Collinearity Statistics B Beta t Sig. Tol VIF (Constant) -1,459 -1,169 ,246
Hypothesized variables
Hedoniclevel* ,185 ,189 1,812 ,073* ,732 1,366
Brand Equity ,058 ,049 ,485 ,629 ,782 1,278
Product related antecedents
Substitutability ,100 ,102 1,067 ,289 ,874 1,144 Regular price** ,058 ,276 2,616 ,010** ,718 1,393
Discount %*** ,030 ,292 2,795 ,006*** ,727 1,375
Store related antecedents
Store loyalty ,086 ,086 ,857 ,394 ,793 1,260 # Of supermarket visits ,058 ,052 ,489 ,626 ,711 1,406 Store-type visit -,076 -,030 -,296 ,768 ,752 1,330
Situation related antecedents
Shopping trip costs ,005 ,086 ,840 ,403 ,759 1,318 Folder usage up-front** ,570 ,209 2,156 ,034** ,847 1,181
# Of products bought on promo -,010 -,019 -,189 ,851 ,774 1,292
Consumer related antecedents
Gender -,200 -,065 -,620 ,537 ,721 1,387 Education* ,192 ,193 1,764 ,081* ,665 1,503
Age group ,223 ,128 1,211 ,229 ,716 1,397 Income group -,083 -,090 -,783 ,436 ,607 1,648 Household size Group -,148 -,076 -,659 ,512 ,600 1,665
Measure Cumber
R2 ,308
Adjusted R2 ,180 ANOVA F Sig. ,005***
* = Is significant at 90% level (0,05-0,10) (1-tailed) Significant
