Broadening the relationship between consumer and brand

Drivers of cross-buying in the fast moving consumer goods

market

S.W.M. Bot

Studentnumber 1333771

University of Groningen

2

Management Summary

In the cluttered landscape in retailing and brands today, manufacturers and retailers are searching for ways to grow their business. Cross-selling is one way of expanding sales and a growing interest for this strategy can be seen in the marketing field and in literature. Most research for drivers and consequences of cross-buying is conducted in a contractual setting, like financial services or catalogue retailing. Because cross-selling is an interesting strategy for the fast moving consumer goods industry too, the goal of this thesis is to identify the drivers of consumer cross-buying intentions for this industry.

Based on prior research several drivers have been identified and can be classified into attitudinal variables, product- and category characteristics, consumer characteristics, marketing efforts and company- and market related factors. A consumer survey has been designed to collect data on these drivers and ask consumers how they would react on a cross-selling proposition. The variables included in the regression model are satisfaction, brand awareness/associations, perceived quality, loyalty, hedonic and utilitarian attitude. And in the second part of the survey respondents were asked if consumers perceived fit/similarity between the promoted products and liked the promotion. And their cross-buying intentions were measured.

A multiple regression model has been created to assess the sign, magnitude and significance of the relationship between the independent variables and consumer cross-buying intentions. The results show that the appreciation of the promotion and the perceived fit or similarity are the most important influencers of cross-buying intentions. An unexpected negative influence of satisfaction was found. A utilitarian attitude has a positive effect and indicates that congruence between product and promotion is important.

Another important finding is that other attitudinal variables showed no significant results and are therefore not important in the cross-buying decision. So it can be concluded that prior relationship experience is no important driver in consumer cross-buying intentions. In the fast moving consumer goods industry, characterized by low-involvement and low perceived risk, the fit between promoted products is important in consumer cross-buying intentions.

Most important implication for marketers and practitioners is that designing cross-promotions should be considered carefully and tested thoroughly. Further research is needed for forms of promotions, what kind of products to promote, what categories to focus on and what type of promotion should be chosen.

3

Preface

Writing this preface, means that my last project as master marketing student, my master thesis, finally has come to an end. This thesis challenged me to show the knowledge and experience I gathered during my study. Though it was sometimes hard, it made me realize what I learned in the past years. Writing a thesis is also a personal challenge. I got to know myself quit well, my strengths, my weaknesses, the things I dislike, things I would definitely do different and most important all the things that make me happy.

In my opinion, being a student is more than following courses, writing papers and taking exams. It‟s also about meeting inspiring people, developing skills, finding ambitions and dreams and having lots of fun. Luckily I had the opportunity to do this during my time at the marketing association (MARUG). This period inspired me to choose marketing management as my specialization and I had the chance to see the ambitious work the marketing department in Groningen is doing and experienced the master as an intensive, but grateful course. My internship at Beiersdorf showed me marketing in practice and will be my guide in starting my career in marketing.

Nothing rests than thanking all the people who made me the person I am today. All the people I met during the past years who inspired and motivated me. All the people who made my time at university unforgettable.

Not to forget my classmates from the master marketing management, for a truly fantastic year!

I want to thank my supervisor Peter Verhoef for his constructive feedback, support, patience and trust in me. Also my second supervisor Janny Hoekstra for her warm support and feedback.

Thanks to all my dear friends for listening to all my thesis issues, helping with my thesis thoughts, talking about my thesis and sometimes for just not talking about my thesis!

Last, but definitely not least, I want to thank my mum and dad for their relentless support during this process. Without them it would never have been possible to finish it! Thank you so much, words cannot express!

Saskia Bot

4

Content

6

1.5 Scope of this study 10

2. Theoretical Framework 11

2.1 Cross-Selling and Cross-Buying 11

2.2 Drivers of Cross-Buying 13

2.2.1. Attitudinal Drivers 13

2.2.2. Product- and Category Characteristics 17

2.2.3. Marketing Efforts and Promotions 18

2.2.4. Consumer Characteristics 19

2.2.5. Past Relationship Variables 20

2.2.6. Market- and Company Related Factors 20

3. Research Design 22

3.1 Data-Collection 22

3.1.1. Sample 22

3.1.2. Data Collection Method 22

3.1.3. Survey Development and Stimuli Selection 23

3.1.4. Measurement and Items 24

3.2 Data-Analysis 27

3.2.1. Multiple Regression Analysis 27

5

4.3 Multiple Regression Analysis 36

5. Discussion and Implications 38

6. Limitations and Future Research 42

References 44

Appendices

Appendix 1 – Survey 48

Appendix 2 – Factor Analysis 53

Appendix 3 – Data Examination 55

Appendix 4 – Quadratic Effect Satisfaction 57

Appendix 5 – Interaction Effects 58

Appendix 6 – Control Variables 59

Appendix 7 – Final Regression Model 60

Appendix 8 – Correlation Matrix 61

Appendix 9 – Multicollinearity Diagnostics 62

Appendix 10 – Testing Assumptions 63

6

Introduction

1.1 Introduction

The modern consumer entering a drugstore or supermarket is dealing with choosing among many product categories, many products, many brands, innovations, brand- and line-extensions. As Laforet and Saunders (2007) state, the retailing landscape turned into a cluttered landscape. In their analysis of grocery supplier brands between 1994 and 2004, they found that brand structures are getting more and more complex, companies are mixing and matching brand names in their fight for differentiation of products, competition over segments and attract consumers. For consumers it‟s hard to make a choice, for manufacturers and brands even harder to make consumers choose them. The fast moving consumer goods (FMCG) industry is facing severe competition, advertising spendings are high and promotional pressure is high. For example, in the Dutch grocery channel in 2009, 50% of sales come from promotions (GfK, 2009).

The development Laforet and Saunders (2007) described led to a wealth of research for relational marketing. Marketers hope to decline the customer defection through deepening the relationship with their customers. Especially brand equity and customer loyalty gained lot‟s of interest from academics and practitioners, not quite surprising, as loyalty is believed to be one of the most important factors explaining consumer choice. A base of loyal consumers has shown to result in high market share and thus return on investment (Reicheld and Sasser, 1990). Building a strong brand, or brand equity, in a market with such severe competition is a goal of many organizations. This generates greater market share (Park & Srinivasan, 1994), more efficient product line extensions (Keller and Aaker, 1992) and more responsive advertising and promotions (Keller, 1998).

7

e.a. 2008) and increase customer share (Verhoef e.a., 2001). Cross-selling is an opportunity for firms to increase revenue from existing customers. In literature, the strong relationship between loyalty and cross-buying is well-known, except for the direction of this relationship (Reinartz e.a., 2008).

Especially in the services industry like the financial services and retail banking, lot of research has been done to identify drivers and consequences of cross-buying (Verhoef e.a., 2001, Ngobo, 2004), with promising results. The more products a consumer buys , the longer he or she is likely to stay with the firm (Reinartz and Kumar, 2000). Customers buying from multiple categories tend to have longer profitable lifetime durations (Reinartz and Kumar, 2003), it is also a driver of customer lifetime value (Venkatesan and Kumar, 2004) and stimulates multi-channel shopping behavior (Kumar and Venkatesan, 2005). Ngobo (2004) concludes that cross-selling is an important tool and that managers should very well understand the determinants of cross-buying behavior, before setting up cross-selling activities.

In the marketing field, cross-selling and cross-buying also gained interest with practitioners. In the FMCG industry, a growth of cross-selling initiatives can be seen. For example Kumar e.a. (2008) mention FMCG giant Procter & Gamble launching direct marketing initiatives to cross-sell several of it‟s food brands. Also Amazon is an example of a retailer trying to cross-promote their broad base of categories (Kumar e.a., 2008). Unilever reported an increase in dollar share and sales of Dove personal wash products by cross-selling the Dove master brand, as found in a study dating 20041. Not only an increase in share and sales, also an increase of 33% of cross-category behavior during promotional periods and the likelihood that consumers continue buying from multiple categories after the promotional period has ended.

Figure 1.1 Example of Cross-Promotion Unilever Dove

With their master brands, firms are trying to build a relationship with their customers, cross-selling different categories to the customer. Firms are increasingly trying to leverage their brand value to maximize revenues and profits. Research results indicate that these strategies are successful. But the impact of cross-selling can be greatly improved if firms identify and target the right customers and products for selling. This can be achieved by identifying the drivers of cross-buying, which can be used for classifying customers (Kumar e.a., 2008).

8

In the recent literature several variables have been identified as drivers of cross-buy in the catalogue retailing sector (Kumar e.a., 2008) and financial services sector (Verhoef e.a., 2001; Ngobo, 2004). Studies by Ngobo (2004) and Verhoef e.a. (2001) identified customer‟s attitude, socio-demographics and marketing effort by the firm as drivers of cross-buying. Kumar e.a. (2008) also found marketing effort and customer characteristics to be drivers of cross-buying, but also product characteristics influencing the cross-buying behavior. Attitudinal variables like satisfaction, perceived quality and loyalty are found to be related to consumer cross-buying decisions (Verhoef e.a., 2001; Ngobo, 2004; Reinartz e.a., 2008). Loyalty is another variable said to be related to cross-buying (Reinartz e.a., 2008). Though attitudinal variables also reflect the past relationship between a customer and a brand, other relational variables have been topic in research. Like Ngobo (2004) using past usage in his model, based on the work of Oliver (1999), saying that the more a customer uses the services of the retailer the more likely s/he will accept cross-buying from that same retailer because of learning and inertial effects. Verhoef e.a. (2001) and Kumar e.a. (2008) found a significant positive effect of marketing effort on cross-buying in a direct mailing context. Product characteristics have been studied by Kumar e.a. (2008), but also in the multi-category modeling literature. Wherein products being substitutes or complementary influence the cross-buying behavior (Seetharaman e.a, 2005; Manchanda e.a., 1999; Russel and Kamakura, 1997). The similarity of categories is an often used topic in a cross-buying context. Originating from literature on brand extensions (Aaker and Keller, 1990), the similarity or fit between categories makes customers more willing in accepting new categories (Ngobo, 2004; Liu and Wu, 2008). Found in the same multi-category literature, household traits are an important determinant in household choices. Household preferences (example: a preference for private labels) and household sensitivity for the marketing mix (example: proneness to promotions). Socio-demographics are also said to account for the differences in cross-buying, like age (Ngobo, 2004; Verhoef e.a., 2001; Liu and Wu, 2008), gender and income (Li e.a., 2005).

9

1.2 Problem Statement

In the section above, the background of the problem was described. For this thesis the problem statement will be:

“What are the drivers of cross-buying in the fast moving consumer goods market?”

To answer the problem statement, the following research questions will be addressed.

- What is cross-buying?

- What are the consequences for firms and brands of cross-buying? - What is the effect of marketing effort on cross-buying?

- Which attitudinal variables have influence on the cross-buying behavior? - What is the effect of product-category characteristics on cross-buying?

- What is the influence of social-demographics and household traits on cross-buying? 1.3 Structure

The next chapter gives the theoretical framework and contains an overview of the academic literature concerning cross-buying. In the third chapter a conceptual model and hypotheses will be formulated. In the fourth chapter the research method, data collection and data analysis will be described. The fifth chapter describes the results from the collected data, followed by the conclusions and recommendations in the sixth chapter.

1.4 Relevance

10

1.5 Scope of this study

11

2. Theoretical Framework

2.1 Cross-Selling and Cross-Buying

Cross-selling is the practice of promoting additional services or products to existing customers in addition to the ones the customer currently has (Butera, 2000). Especially in services industries multi-service providers persist in cross-selling their services, like insurances and retail banking. The interest in this particular growth strategy, lead to a wealth of academic research in this area (Verhoef e.a., 2001, Reinartz and Kumar, 2003, Ngobo, 2004) The first results from this research stream, estimated the positive effect of cross-selling on business results.

As a consequence, cross-selling also gained popularity in the more product driven consumer goods market. Many consumer goods manufacturers are trying out cross-selling activities, to increase their revenue and expand the relationship with their existing customers (Kumar e.a. 2008). Manufacturers like Unilever and Procter & Gamble employed initiatives to cross-sell their food and household brands and catalogue retailers try to cross-sell their categories by sending direct mailings. The first results show promising results, with the example of beauty master brand Dove in a study dating 2004 reporting an increase in share and sales by cross-selling their personal wash products (Kumar e.a., 2008).

With cross-selling activities, companies try to encourage customers to cross-buy from their different brands, products and categories. Cross-buying is defined as buying additional categories from a firm or brand (Kumar and Reinartz, 2006). Much theory in cross-buying literature is based on the brand extension literature.

Where consumers engage in buying additional categories from a firm or brand, the construct of cross-buying is closely related to brand extension literature, the concept of consumers buying completely new introduced categories of the firm or brand (Aaker and Keller, 1990). The difference with cross-buying is, consumer‟s might be familiar or aware of the product or category. For brand extensions, the category is completely new and consumer‟s have no prior knowledge. Therefore it is reasonable that antecedents might turn out different.

12

Though cross-buying also means that consumers stay with a brand or firm, cross-buying is about buying additional categories and as Verhoef (2003) states, this is a different decision. Not only because of the multiple category component, where different categories might mean different consumer needs. Consumers can also demonstrate behavioral loyal buying behavior out of convenience or inertia (Kumar and Reinartz, 2006). In cross-buying behavior consumers have to drop or add products and therefore cross-buying is a much deeper active engagement with the brand or firm (Verhoef, 2003).

Summarizing, the concept of cross-buying is related to loyalty literature and brand extension literature, but it can be said that the consumer decision to engage in a relationship with a brand or retailer, customer retention, is different from the incremental decision to add or drop existing products (Verhoef, 2003). Therefore, we adopt the view by Bolton e.a. (2004) and expect the antecedents for cross-buying behavior to be different from the antecedents of customer retention.

The concept of cross-buying is measured in more ways, extracted from the contract based services industry, and might therefore not all be applicable on the consumer goods industry. Nevertheless, the measures are discussed here, for better understanding the concept of cross-buying. Verhoef (2003) conceptualizes cross-buying as customer share development, according to Peppers and Rogers (1999), the ratio of a customers purchases of a particular category of products or services from supplier X to the customers total purchases of that category of products or services from all suppliers. This customer share can be increased by cross-selling other products from one category, but also from multiple categories from the same firm. Share-of-wallet (Sharp & Sharp, 1997; Mundt e.a., 2006), can be compared to customer share, and is an often used metric. But for this thesis it is important to focus on constructs where consumers buy from more categories and customer share does not necessarily mean multiple category share.

13 2.2 Drivers of Cross-Buying

Prior research focusing on the drivers of buying is limited, the foundation of modeling cross-buying determinants is the loyalty and brand extension research (Ngobo, 2004). In the introduction the drivers of cross-buying found in literature have been reviewed. Classified in attitudinal drivers, product- and category characteristics, marketing efforts, consumer characteristics, past relationship variables and other variables. In this section the variables will be discussed more extensively and an overview is presented in figure 2.1.

Figure 2.1 Conceptual Model

2.2.1. Attitudinal Drivers

14

are considered in cross-buying literature (Verhoef e.a ,2001; Ngobo, 2004). Like perceived quality and attitudinal loyalty (Ngobo, 2004) and satisfaction (Verhoef e.a., 2001). Furthermore, in the consumer attitude research, the interest has grown for difference in utilitarian and hedonic dimensions of attitude and the impact on consumer reactions to promotions (Voss e.a., 2003). In this thesis the focus will be on the attitude towards the brand.

Customer based brand equity is one of the most used measures for capturing consumers unique, favorable and strong associations in brand management. Higher brand equity leads to a higher response to advertising and promotions (Keller, 1993) and new brand extensions will be accepted faster with higher equity brands (Aaker and Keller, 1990; Lassar e.a., 1995). Regarding the relevance for buying, the different dimensions of brand equity and their relation to cross-buying will be discussed in more detail. Yoo and Donthu (2001) developed a multidimensional scale for measuring customer based brand equity that defines three dimensions of customer based brand equity, brand awareness/associations, perceived quality and loyalty. These dimensions will be used further in this thesis.

Brand Awareness

Brand awareness is related to the strength of the brand node or trace in memory, as reflected by consumers ability to identify the brand under different conditions (Rossiter and Percy, 1987). In a low involvement environment, like the FMCG sector, awareness is an important element in brand equity, because in these decision settings consumers lack motivation and are willing to choose more familiar brands (Keller, 1993). Aaker (1991) defines brand associations as “anything linked in memory to a brand”. In the former literature on cross-buying, the dimension of awareness/associations has not been specifically considered. Due to the fact that awareness is an important indicator of the success of brand extensions (Aaker and Keller, 1990) the first hypothesis is formulated:

H1) Brand Awareness and Associations are positively related to cross-buying intentions.

Perceived Quality

15

perceptions also improve the success of sampling activities (Erdem, 1998). Because high quality perceptions show positive effects in brand extension evaluation and umbrella brand literature, the second hypothesis will be:

H2) Perceived Quality is positively related to cross-buying intentions.

Attitudinal Loyalty

Aaker (1991) defines brand loyalty as “the attachment that a customer has to a brand.” A more loyal attitude towards the brand, creates higher brand equity and therefore loyalty is an important dimension of customer based brand equity (Lassar e.a., 1995). As discussed in the previous section, loyalty is associated with cross-buying, but it is important to understand the difference between behavioral and attitudinal loyalty (Kumar and Reinartz, 2006). The focus in this research is on attitudinal loyalty toward the brand, referring to the perceptions and attitudes of the consumer (Kumar and Reinartz, 2006). Reinartz e.a. (2008) examined the direction of the relationship between cross-buying and loyalty. They state that loyalty is not a consequence of cross-buying, but an antecedent of cross-buying. Implying that loyal customers are more intended to cross-buying. Verhoef (2003) focused on the effect of affective part of commitment, based on attitudinal loyalty, also concluding that higher affective commitment has a positive effect on customer share development. We adopt this view and the hypothesis is:

H3) Attitudinal loyalty is positively related to cross-buying intentions.

Satisfaction

Satisfaction has been topic in a lot of consumer research regarding relational marketing, repurchase intentions, loyalty and also in cross-buying context. As Verhoef e.a. (2001) find, the effect of satisfaction on cross-buying depends on the length of the relationship a consumer has with the firm or brand. Satisfaction has a positive effect on cross-buying when the relationship is longer. Ngobo (2004) finds the same results, that experiences in the former relationship are not that important for customers in enhancing the relationship. On the other hand, Liu and Wu (2008), state that the effect of satisfaction and trust on cross-buying depends on the category similarity and complexity. Contrary are the findings by Li e.a. (2005) in the financial services that overall satisfaction with the firm increases the ability to cross-sell. Furthermore, Bolton and Lemon (1999) find a positive effect of satisfaction on usage of service. In this thesis satisfaction is interpreted as the overall satisfaction a consumer currently holds toward the brand. Because cross-buying is related to repurchasing, the fourth hypothesis is:

16

Hedonic and Utilitarian Dimensions of Consumer Attitude

In consumer attitude research, the distinction between hedonic and utilitarian attitude towards products/brands gained more interest by scholars. Especially the relationship with effectiveness of cross-promotions is of interest in this cross-buying context. Thus measures of these two dimensions may be input for sales cross-promotion decisions (Voss e.a., 2003, Chandon e.a., 2000). Voss e.a. (2003) developed and validated a multidimensional scale, measuring the hedonic and utilitarian dimensions of consumer attitude toward product categories and different brands within categories. With on the one side a hedonic dimension, resulting from sensations derived from the experience of using products, and a utilitarian dimension derived from functions performed by products shown in table 2.1.

Utilitarian Dimension Hedonic Dimension

Effective Fun

Helpful Exciting

Functional Delightful Necessary Thrilling

Practical Enjoyable

Table 2.1 Utilitarian and Hedonic Dimensions of Consumer Attitude

Voss e.a. (2003) even found that classifying brands on this scale is not straightforward. They found that different brands from the same product category, may score different on this scale. For example, a brand adopting an experiential positioning strategy, may be positioned higher on hedonic dimensions than competing brands (Voss e.a., 2003). Relevant for product marketers, is the underlining of the authors, that even though a product group or category is viewed as more utilitarian, the brand itself can have such a positioning strategy, that consumers feel a more hedonic attitude towards the brand. Therefore in this thesis, the attitude towards the brand is used.

This attitude measure is useful, considering the impact on success of sales promotions (Chandon e.a., 2000), arguing that sales promotions offer consumers more benefits. Ranging from utilitarian benefits to hedonic benefits. As Chandon e.a. (2000) state, the congruence between the nature of the product and benefits of the promotion is an important indicator for the effectiveness of a sales promotion. Price cuts (this week 10% off) are seen as much more utilitarian compared with a other-product promotion (buy one, get another product for free) which scores higher on the hedonic dimension. Interesting, because Liao (2006) also concludes that other-product promotions work better for products/categories that can be seen as more hedonic. Considering these conclusions, it seems that cross-promotions work better for brands who are positioned higher on the hedonic dimension, or for which the consumer holds a hedonic attitude.

17

H5a) Hedonic Attitude toward the brand is positively related to cross-buying intentions. H5b) Utilitarian Attitude toward the brand is negatively related to cross-buying intentions.

2.2.2. Product- and Category Characteristics

Category similarity (fit)

Another important factor in the brand extension literature is the concept of fit between currently bought categories and new categories. Aaker and Keller (1990) suggest that the fit between the currently bought categories and additional categories is important for cross-category purchases. Perceived fit or category similarity will facilitate the transfer of attributes from the core categories to the extension categories (Liu and Wu, 2008). This similarity or fit wil make customers transfer their positive attitudes to the new business and accept these new categories faster (Aaker and Keller, 1990). Also Ngobo (2004) finds that image conflicts influence the consumer to perform additional purchases. Cross-buying intentions by customers seem to be higher if the focal cross-selling product is from a similar category and the customer perceives high fit.

The concept of perceived fit received attention in more studies. Aaker and Keller (1990) conceptualized the construct of fit in three bases: complement, substitute and transfer. According to Martin and Stewart (2001) transfer belongs to a more feature-based measure. Like Aaker and Keller (1990) they too conceptualized a more usage-based measure of fit. In the multi-category modeling literature (Russel and Kamakura, 1997; Manchanda e.a., 1999) complementarity and substitutability are constructs influencing the cross-category consumer behavior.

Taking a closer look at the multi-category models, the most used example is a household deciding to whether buy or not-buy cake mix and cake frost. The decision to buy cake mix may not be independent of the households decision to buy cake frost. These decisions are related across these categories because products may be complementary (cake mix and frosting) or substitutes (coke and orange juice) in fulfilling the households needs (Seetharaman e.a., 2005). Complementarity results show that pricing and promotion changes in one category affect purchase incidence in related product categories, this gives managers and retailers some control over consumer buying behavior (Manchanda e.a., 1999). Like, if two products are perceived as complementary, a price-off promotion should only cover one of the products, demand for the other product will increase either way.

18

explaining the evaluations of brand extensions. As they conclude, “transfer” alone is sufficient for measuring fit. That is the measure used in this thesis. Because the positive effects for similarity and/or fit found by Ngobo (2004), Liu and Wu (2008) and Aaker and Keller (1990) the sixth hypothesis is:

H6) Perceived fit/similarity is positively related to cross-buying intentions.

Category of first purchase and Complexity

The first category a consumer purchases as driver of cross-buying is not found much literature, but Kumar e.a. (2008) found that category of first purchase is a significant driver of cross-buy in the retailing sector. Also Ngobo (2004) compared differences between a financial services provider and a catalogue retailer, concluded that type of service has an effect on cross-buying intentions. Perhaps the explanation is in the fact that complexity of the product category has a moderating effect on the influence of satisfaction and trust on cross-buying (Liu and Wu, 2008). Category complexity is defined to which level the clients perceive complexity and lack an understanding of the products‟ attributes and find it hard to assess product quality (Devlin, 1998). In which sense consumers perceive complexity in the FMCG industry is not clear, because the research of Liu and Wu (2008) was conducted in the financial services sector. But it could be assumed that lesser known categories are faced as complex, like men buying facial moisturizer. Category of first purchase and complexity will be used in selecting the stimuli for the consumer survey.

2.2.3. Marketing Efforts

Kumar e.a. (2008) conducted research in the drivers and consequences of cross-buying in retailing. Direct mailings were send to customers of a large catalogue retailer pertaining previous bought categories and cross-promoting mailings pertaining categories never bought before. Both efforts showed significant effects on cross-buy. Wherein the increase of cross-buy due to the number of direct mailings is up to a threshold. The cross-promotional mailings show a strong positive effect on cross-buy. This is in line with the findings by Verhoef e.a. (2001) that direct mailings show a positive effect on purchasing new services.

19

Defined by Percy and Elliot (2005), a sales promotion is any direct purchasing incentive, reward, or promise that is offered for making a specific purchase or taking a specific purchase-related action. Examples are promotions with providing free samples (Erdem, 1998), other-product promotions (buy one, get another for free) (Liao, 2006) or money saving incentives for trying new products (buy one, get a 50% discount on other product) (Chandon e.a., 2000).

The finding of Erdem (1998) is that high perceived quality brands can increase their sales, and thus trial, with their existing customers by providing free samples. Liao (2006) finds that consumers tend to appreciate same-product promotions more than different-product promotions, but the product category has a moderating role in this process. Other product promotions are more appreciated for a more hedonic category, like shopping goods. Chandon e.a. (2000) find the same, that promotion benefits and the category benefits should be congruent for more effective promotions. So it can be concluded that cross-promotions should be carefully considered, because consumers tend to not appreciate all combinations of products. Because of the positive effects of marketing efforts and direct mailings in the services industry and catalogue retailing sector, a positive effect of (cross-)promotions is expected and the seventh hypothesis is therefore:

H7) Promotions are positively related to cross-buying intentions.

2.2.4. Consumer Characteristics

Consumer characteristics, or the so called consumer heterogeneity, play an important role in cross-buying literature. Overall, a distinction can be made between observable factors and unobservable factors.

Observable factors are socio-demographics. Kumar e.a. (2008) find that age of the head of the household and household income are significant predictors of cross-buy. This is consistent with Verhoef e.a. (2001) and Ngobo (2004) showing significant effect of age on cross-buying. Liu and Wu (2008) do find the same result, differences between older and younger people. Household income was only significant in some categories and gender did not show any significant results. Contrary, Li e.a. (2005) find that demographic variables like gender and education do have an impact on cross-buying. Mittal and Kamakura (2001) studied the effect of customer characteristics on satisfaction, repurchase intent and repurchase behavior and find significant effects for gender, education level, age, number of children in household and area of residence.

20

2003). Making it possible for marketers to estimate the preference of households in new categories, score potential target groups for new products (Singh e.a., 2003) or coordinating marketing activities, like promotions, across categories (Russel and Kamakura, 1997). In this thesis we limit ourselves to observable factors and will adopt the method of Ngobo (2004) to use this variables as control variables.

2.2.5. Past Relationship variables

Length of relationship and past usage seem to influence cross-buying intentions and the effect of drivers on cross-buying. Most research in cross-buying found a moderating effect for length of relationship and past usage, especially for the attitudinal variables. The effect of satisfaction differs between customers with short and long relationhips (Verhoef e.a., 2001). Longer relationships will only result in more cross-buying in highly similar categories (Liu and Wu, 2008). Also past relationship contributes to higher cross-buying intentions, because the longer the customer is with the firm, the more he will accept from the firm (Ngobo, 2004). The more a customer uses the services of a retailer, the more will be accepted from that firm because of learning (Oliver, 1999) and inertia effects (Rust e.a., 2000). Ngobo (2004) used past usage as a control variable and Verhoef e.a. (2001) found that past usage, the type of insurance product and number of services, explained some of the variance in their model. Past usage can be measured as frequency of buying and the share of purchases (Ngobo, 2004). As Ngobo (2004) did, past usage variables will be used as control variables in this thesis.

2.2.6 Market- and Company related factors

21

Variety seeking behavior is a development encouraged by the market situation. Consumer desire for variety may be due to different preferences within households or different household situations (Kotler, 1997). The buying situations characterized by low involvement and high competition, like the FMCG market, even stimulate variety seeking behavior (Nijssen, 1997). In first instance, this might be interesting for manufacturers as consumers seem to be prone to trying. But as Nijssen (1997) finds, most of the time this variety seeking behavior fragments a market instead of expanding it.

Cross-selling success also depends on some company related factors, as mentioned by Ngobo (2004). Like salesforce training, knowledge transfer between departments, teamwork, incentives and the support for extra promotional campaigns (Nijssen, 1997).

22

3. Research Design

3.1 Data Collection

3.1.1. Sample

For exploring consumers cross-buying intentions in our selected category, the personal care category, the target population are Dutch female and male shoppers aged 18 years and older, responsible for their own shopping in this category. For this research a nonprobability sampling technique, called convenience sampling, is used. A less expensive and time consuming technique, with highly accessible sampling units. Despite these advantages, the potential danger of selection bias, can create unrepresentative samples of the population. Nevertheless, it‟s a form of sampling often used, even in larger surveys (Malhotra, 2004). Next to the initial pool of respondents selected by the researcher, snowball sampling is used to gather additional respondents. This form of sampling asks the initial group of respondents to identify and forward the questionnaire to other people belonging to the target population (Malthotra, 2004). Though almost every adult in the Netherlands is a shopper in the personal care category, the objective is to create a sample representing the Dutch population aged 18 and older.

The method of analysis, requires at least 15 observations for every predictor for adequate predictive power and to assure the generalizability of the results (Hair e.a., 2006). So a multiple regression model with 8 independent variables, requires a minimal sample size of 120 respondents.

3.1.2. Data collection method

The survey method selected for gathering data is an internet survey. This method has the advantage that it is low cost and convenient for respondents, as they can complete the questionnaire in their own time and place (Malhotra, 2004). Besides that, internet surveys have a quick response time and the software tools designed nowadays give the researcher possibility to create a professional web-survey. However, there are limitations to electronic surveys. According to Malhotra (2004), internet or e-mail users are not representative of the general population.

23

3.1.3. Survey development and Stimuli Selection

For constructing the survey, choices have to be made concerning the stimuli. In the fast moving consumer goods market, different categories can be identified, food and non-food. As mentioned in the previous section, the selected category is the personal care category. This category is characterized by a large amount of brands, large and broad labels like NIVEA and Dove and smaller brands focusing on special consumer needs, like soap free showergel. And also a growing number of private labels can be seen, like Albert Heijn and Etos. Furthermore this category is interesting, because clear subcategories can be identified, shown in figure 3.1. In some subcategories the penetration is high, like deodorant and showerproducts. But in some subcategories, like facial care, penetration is only 45% and consumers are not familiar with the category. Moreover, the personal care category is a category both female and male shoppers are participating in. Meaning we can use both men and women for the data-collection. The personal car market is also characterized by a high distribution level and lot‟s of different channels, so products are available to most consumers (NCV annual report 2009).

93,8 76,6 40,6 78,3 74,6 64,4 45,2 58,3 34,3 25,6 33 30,3 29 15,4 0 20 40 60 80 100

Total Personal Care Dental Care Shaving Products Hair Care Shower/Bath/Soap Deodorant Facial Care Men Women

Figure 3.1 Gender of buyers and penetration of the personal care category in the Netherlands (NCV annual report 2008)

Because the definition of cross-buying is purchasing additional categories from a brand already bought, the starting point will be the showergel brand used by the respondent. In this manner, consumers are encouraged answer the questions elaborating from their actual buying behavior. The penetration in the showergel category is almost 75% for women and 30% for men, as can be seen in figure 3.1, so most respondents will be familiar with buying and using this product.

24

category like deodorant, to be highly associated to the basic personal care category, like showerproducts. Therefore the perceived fit by consumers is assumed to be high between these products. For facial moisturizers, we expect the fit to be lower. In this thesis deodorant and facial moisturizer are selected.

The survey will start with asking the attitudinal variables, cross-buying intentions without marketing effort, current usage of other subcategories from the current brand, past usage of the brand and length of relationship about the current used brand showergel. Next the consumers are confronted with hypothetical price-off cross-promotions. Differences are created between the product offered next to the showergel, a deodorant versus a facial moisturizer. The price-off promotion is a discount of 25% on the deodorant or facial moisturizer.

After confrontation with the promotion, respondents will be asked to answer some questions about the appreciation of the promotion, the perceived fit between the showergel and promoted product and cross-buying intentions. Concluding the survey with asking the respondents to fill in their socio-demographics, like age, gender, income and educational level. The complete survey can be found in appendix 1.

3.1.4. Measurement and Items

The items measuring cross-buying intentions are adapted from the research by Ngobo (2004) in which consumers are also confronted with a hypothetical promotion. The cross-buying intentions are used because we use a hypothetical situation. Consumer intentions are an often used measure in consumer research, because it is a relative easy manner for companies to gather information on future buying behavior. Though intentions are not a perfect predictor of actual behavior, they show to be a significant predictor of actual behavior (Blackwell, Miniard and Engel, 2006). Consumers are asked if they would engage in buying the promoted products, measured with four items all measured on a 7-point likert scale (Table 3.1.)

All attitudinal variables, too, are measured on a 7-point likert scale. The items measuring satisfaction are adapted from the work of Verhoef e.a. (2001) and Ngobo (2004), assessing the overall satisfaction of the consumer toward the brand. Awareness/Assocations, Perceived quality and Attitudinal Loyalty items are based on the multidimensional scale developed by Yoo and Donthu (2001). The hedonic and utilitarian attitude items are adapted from the HED/UT scale developed by Voss e.a. (2003). The item measurement can be found in table 3.1.

25

assessed a more overall measure of fit, asking the consumer if they perceive fit. In this thesis, the items are based on both forms of perceived fit to make sure the construct of fit is measured properly. These items are measured in questions 10 and 11 of the survey and the items can be found in table 3.1. Furthermore, the promotion is evaluated by the consumers, asking them if they appreciate the promotion offered, this measure is adapted from Liao (2006).

The past usage variables and consumer characteristics are used as control variables. Past usage is described in most cross-buying context as number of services used (Verhoef e.a., 2001; Ngobo, 2004) and share of wallet (Ngobo, 2004), we based the control variable in this research on this work. Respondents are asked for how long they have been buying this brand, what percentage of their total spending in the personal care category is spend on this brand and how many subcategories they currently use. Consumers were not asked how many other categories they use from their current showergel brand, because the cluttered retail landscape, low-involvement setting and multiple brand usage, makes it rather difficult to assess the reliability of these self-reported answers. Product awareness of their current brand is asked to assess if consumers are aware of the products offered by the brand.

Consumer characteristics, age and gender are asked on a respectively ratio and nominal level. For the level of education, the basic educational levels in the Netherlands are used. The income is based on the net average wage in the Netherlands 20102, asking the respondents if they earn below the average wage, the average wage, 50% above the average wage or twice the average wage. Table 3.1 contains a summary of the used constructs and items.

26

Table 3.1 Constructs, Definitions and Item Measurement

Definition Item Measurement Survey Questions

Construct

Satisfaction The emotional state that occurs as a result of a customer's interactions w ith the firm over time. (Anderson, Fornell and Lehmann, 1994)

Overall satisfaction w ith the current brand. (Ngobo, 2004; Verhoef e.a., 2001)

Overall I am satisfied w ith this brand.

Past Usage Reflection of the past usage a consumer has w ith the focal product and brand.

Share-of-Wallet (SOW), length of relationship, number of categories used. (Ngobo, 2004; Verhoef e.a., 2001)

For how long have you been using this show ergel brand? What percentage of your total spendings on personal care are spend on this brand?

How many subcategories do you use?

Aw areness/Associations The strength of the brand node or trace in memory. (Rossiter and Percy, 1987) Anything linked in memory to a brand. (Aaker, 1991)

Ability to recognize the brand, picture the logo, name associations, imagining the brand and being aw are of the brand. (Yoo and Donthu, 2000)

I can recognize this brand among

competing brands.

I am aw are of the brand. Some characteristics of this brand come

to my mind quickly. I can quickly recall the the logo or symbol

of the brand.

I have difficulty in imagining X in my mind.

Perceived Quality The consumers judgment about a product‟s overall excellence or superiority. (Zeithaml, 1988; Yoo and Donthu, 2000)

Judgment of quality and trust. (Yoo and Donthu, 2000)

My brand is of high quality. This brand offers products I can trust.

This brand offers products of high quality.

Loyalty The attachment that a customer has to a brand. (Aaker, 1991; Yoo and Donthu, 2000)

Being loyal, the brand being first choice and w hat reaction in an out-of-stock situation (Yoo and Donthu, 2000)

I consider myself to be loyal to this brand. This brand w ould be my first choice. I w ould not buy other brands if my brand

is available at the store.

Utilitarian Attitude An attitude based on functional, problem solving benefits using the product. (Voss e.a., 2003; Keller, 1993)

Rating on utilitarian benefits: effective, helpful, functional, necessary and practical. (Voss e.a., 2003)

Rating brand on follow ing items: Effective Helpful Funtional Necessary

Practical Hedonic Attitude An attitude based on sensory

benefits, experiencing the product. (Voss e.a., 2003; Keller, 1993)

Rating on hedonic benefits: fun, exciting, delightful, thrilling, enjoyable (Voss e.a., 2003)

Rating brand on follow ing items: Fun Exciting Delightful Thrilling

Enjoyable

Fit The consumer view s the new item to be consistent w ith the parent brand. (Tauber, 1988) The ability of the manufacturer to produce the new product. (Keller and Aaker, 1990)

Assessing if characteristics of products are similar (Martha and Stew art, 2001) and if the manufacturer is capable of producing the new product (Keller and Aaker, 1990)

The characteristics of facial moisturizer/deodorant are similar to show ergel. I think my show ergel brand is capable of manufacturing facial

moisturizer/deodorant.

Promotion Attractiveness The evaluation of promotion, positive or negative attitude tow ards the promotion. (Liao, 2006)

Based on scale for promotion preference by Liao (2006)

I can't apprecciate that my brand offers me this promotion. (reversed)

Cross-Buying Intentions Consumers intention to buy additional categories from a brand currently used. (Verhoef, 2003; Ngobo, 2004)

If consumers w ould consider the promotion and actually buy the promotion (Ngobo, 2004)

27 3.2 Data Analysis

3.2.1. Multiple Regression Analysis

Several variables have been identified in the previous chapter from literature which are found to have a relationship with cross-buying. The purpose of this thesis is exploring what the drivers are of consumers cross-buying intentions in the fast moving consumer goods industry and verifying this theoretical framework with statistical analysis. Multiple regression analysis is a multivariate statistical technique used to analyze the relationship between a single dependent variable and several independent variables (Hair e.a., 2006). Multiple regression gives the possibility to create a model with predicting the independent variables. When the researcher wants to create a model with maximal prediction power by the independent variables, the purpose is the prediction of the dependent variable. But multiple regression can also be used for objectively assessing the degree and character of the relationship between dependent and independent variables, this method of regression is also called explanation. It assesses the magnitude, sign and statistical significance of the variables. (Hair e.a., 2006). In this thesis explanation is a more appropriate approach. Multiple regression is a form of linear modeling and the relationship between the dependent variable and the independent variables can be noted in the following equation. Wherein the dependent variable is noted as Υ, the independent variables are noted as V, the regression coefficients as and the error term as .

For specifying the relationship, the selection of the variables is an important step and up to the researchers judgment which variables to include in the model. Selection should be based on theoretical or conceptual ground. The theoretical framework reviewed in the second chapter explored the variables associated with cross-buying in the literature and are summarized in the conceptual model. Those variables will be included in the multiple regression analysis. So the multiple regression equation for this research, with the independent variables selected from theory, is:

28

Some of the independent variables in the data-set are categorical. Because regression analysis requires metric variables, dummy variables need to be created. Dummy variables can act as replacement independent variables. Hair e.a. (2006) define a dummy variable as independent variable used to account for the effect that different levels of a nonmetric variable have in predicting the dependent variable. The dummies are created for gender, income, education and length of relationship according to the indicator coding method discussed in Hair e.a. (2006).

3.2.2. Plan of Analysis

29

4. Results

4.1 General results

The online survey had a response of 161 respondents. A first assessment of the data collected showed that 29 respondents failed to complete the questionnaire. Concentrated at the beginning of the survey, with 16 missing values for the second question. Probably due to technical problems or the respondent finding the software difficult. Furthermore, 9 respondents left the program at question number 8, probably finding it hard to answer the questions about awareness and quality. At last 3 respondents did not fill in question 10 and 11, that might be due to not fully understanding the question. From the respondents filling in the complete questionnaire, 2 respondents are found with non-consistent data. All questions wherein the respondent is asked to rate on a 7-point likert scale is answered with a zero, also for the reversed questions. With these control questions, we can assume that these respondents did not fill in the questionnaire profound. 2 more respondents are aged 10 year, that doesn‟t meet the required age of our sample, consumers aged 18 year or older. The minimum number is 120 respondents for a multiple regression analysis with 8 independent variables, so decided is to delete the cases discussed above. That results in a data set of 128 respondents, who completed the questionnaire.

4.1.1. Socio-demographic profile

Table 4.1 shows the demographic profile of the sample and will be discussed in this section. In the next section the brand usage of the consumers in our sample will be discussed.

Percentage of sample Demographic profile Gender Female 52,4 Male 47,6 Age (mean 28,77)

Less than 30 years old 78,1

Between 30 and 39 years old 14,1

Between 40 and 49 years old 3,1

50 years and older 4,7

Income Less than €1.500,- 25,8 Between €1.500,- and €2.500,- 53,1 Between €2.500,- and €3.000 14,1 €3.000,- and more 7 Educational Level

Middelbare School (VMBO) 0,8

Middelbare School (HAVO) 1,6

Middelbare School (VWO/Gymnasium) 3,1

Middelbaar Beroeps Onderwijs (MBO) 5,5

Hoger Beroeps Onderwijs (HBO) 25

30

Table 4.1 Demographic Profile Sample

In the Netherland the population consists of 49,5% men and 50,5% women3. This is comparable to our sample. The age of our sample is remarkable younger than the population. In our sample 92,2% is aged between 18 and 40, in the population this is only 25,3%. This is due to the distribution of the survey. The average income in our sample shows similarities with the population. The higher incomes are the same, but in our sample the group presenting the average income is much higher than in the Netherlands. The educational level of the respondents is high compared to the average in the Netherlands. In our sample 79,1% is educated on the highest levels, in the Netherlands this percentage is 27,1%. This explains the high percentage of the sample having an average income. This is, like age, due to the distribution of the survey and contains therefore a high number of (ex-)students. However, in literature, students samples are an often used method and we consider students or academic educated people to be consumers of the personal care category too. The results show respondents all buy one or more categories from the personal category.

4.1.2 Exploring Brand Usage

The brands used by the consumers in the sample are shown in figure 4.1. It can be seen that the larger brands performing on the market are also the largest in our data. NIVEA and Dove are market leaders, with respectively 26,6% and 20,3% in market share. These market shares do reflect the sum of female and male products. Respondents had the option to answer NIVEA for Men or Dove Men + Care, the subbrands targeting men. In the FMCG sector many brands are creating a special product line for men. It could be that male buyers are not aware of the difference between the family brand and the special sub-line for men. Because that‟s not the focus in this research, the female and male brand-lines are summed. Other brands that play an important role in the showerproduct market are FA, Palmolive and AXE. Interesting to notice that the private labels show a share of 8,6%.

Furthermore the length of the relationship consumers have with their chosen brand is asked. From the respondents 68% answered they have been buying the brand for more than a year, indicating medium loyal buying behavior. The respondents show high satisfaction with their current brand, with a mean of 5,6, measured on a 7-point likert scale. Interesting to note that consumers rating themselves on the loyalty scale (7-point likert) resulted in a mean of only 3,35.

31

Figure 4.1 Current Brand Usage

Asking the respondents how much of their total spendings in the personal care category are spend on their current showergel brand, in other words the share-of-wallet, shows a mean of 16,22%. Because not all showergel brands manufacture products in all subcategories, it‟s not possible to assess the impact of this share-of-wallet, so it‟s deleted from the analysis. Interesting, that the perception of consumers is that they spend nearly 1/5 of their total spendings on one brand.

The last questions show interesting results. In the previous chapter, the penetration numbers shown. Showing that not all categories are that familiar to consumers. In the data collected for this research, consumers show a broad category usage (table 4.2), with over 63,3% using 6 or 7 categories. This could be due to the sample drawn.

Percentage of Respondents Number of Categories Used Aware of number of Categories of

Focal Brand 1 3,1 18,8 2 2,3 26,6 3 5,5 15,6 4 7 12,5 5 18,8 11,7 6 25 5,5 7 38,3 9,4

Table 4.2 Category Usage and Awareness

When asking consumers in how many subcategories their focal brand is participating in, 45,4% answers that next to their showergel the brand has one other product and only 15% thinks their brand runs products in all subcategories. If we divide the brands chosen by the respondents in broad brands participating in almost al categories (NIVEA, Dove) and small brands (FA,

32

Palmolive), we can see that consumers are actually aware of the products from their brand (table 4.3). From the consumers using a broad brand, more than 50% knows that their brand is participating in 5 or more subcategories. Where from the respondents using smaller brands 27,9% thinks their brand only produces showergel and 41,2% thinks they manufacture products in one other category. Asking the respondents if they would intent to buy other subcategories from their currently used showergel brand on a 7-point likert scale, the mean is 4,2.

Broad Brands Small Brands Percentage of Respondents Number of Categories 1 8,3 27,9 2 10 41,2 3 16,7 14,7 4 13,3 11,8 5 23,3 1,5 6 10 1,5 7 18,3 1,5

33 4.2 Measurement Validation

4.2.1. Factor Analysis

Because the constructs in the model are measured with more items, a factor analysis is performed to define the underlying structure of the predictor variables and combine the original items to reduce the number of variables (Hair e.a., 2006). An eigenvalue cut-off of 1 was used and the scree plot was interpreted, for determining the number of components. The results of the rotated factor analysis show nine components (appendix 2), wherein we can see clear factors for the attitudinal constructs. Except for satisfaction, which appears in the awareness/associations component. Based on the theoretical framework is decided to retain satisfaction in the further analysis as single construct. Furthermore, the evaluation of the promotion shows cross-loadings, but based on the literature and a better interpretable result, these two variables are combined in the ninth component. The perceived fit constructs are combined in two components, one for deodorant and one for facial moisturizer, this is due to the high correlation of these two constructs. For the further analysis, the perceived fit of both components is combined, because this gives the best interpretable results.

Reliability is the extent to which a scale produces consistent results if repeated measurements are made (Malthotra, 2004). In the data-collection, several constructs were measured with more than one item and the factor analysis revealed clear factors . To make sure the constructs are correct, a internal consistency reliability test is done, with using the coefficient alpha of cronbach‟s alpha. This coefficient ranges between 0 and 1, with values above 0,6 or 0,7 indicating internal reliability (Hair e.a., 2006). For the constructs and the items measured the coefficient has been calculated, table 4.4 shows that all constructs are internally reliable and can be used for further multivariate analysis.

Construct No. Of Items Cronbach's Alpha Awareness/Associations 5 0,827 Quality 3 0,941 Loyalty 3 0,899 Utilitarian Attitude 5 0,74 Hedonic Attitude 5 0,859 Fit 4 0,753 Cross-Buying Intentions 4 0,835

34

4.2.2. Data examination

As discussed in the research design, a multiple regression analysis will be performed and in this section the variables to be included in the regression model will be examined. First will be checked if any outliers or influencers are present in the data. Next the assumptions for performing multiple regression analysis are assessed, a linear relationship between the predictors and the dependent variable, normality of the error term distribution, constant variance of the error terms and independence of the error terms (Hair e.a., 2006). If needed, possible actions, like transformation of data or omitting outliers will be performed.

First the variables are checked for possible multicollinearity. Therefore the correlation matrix and the Variance Inflation Factor (VIF) and Tolerance values are assessed (Appendix 8 and 9). In the correlation matrix can be seen that correlation exists between the dependent variable and the independent variables. Also correlation is detected between the independent variables. However, the tolerance and VIF values are respectively above 0,10 and below 10, and therefore multicollinearity seems to be no problem in the data-set (Hair e.a., 2006).

In the first regression model created, including all variables, two remarkable results show up. The partial correlation plots and the residual examination show that three cases, case 20, 38 and 86, seem to be outliers on more predictors (Appendix 3). Decided is to perform a new regression analysis with deletion of the three outlier cases. Model fit is assessed with the multiple R and R². The multiple R is the correlation coefficient for the regression between variables and the dependent variable. The R² indicates the percentage of total variation of the dependent variable explained by the model (Hair e.a., 2006). The R in our second model without outliers shows a large improvement for the null model, from 0,594 to 0,693. Also the R² shows an improvement from 0,353 to 0,480. Furthermore, the plots show no outliers anymore, presented in appendix 7. Therefore we will continue the data-analysis without these cases.

35

Another explanation for the negative sign for satisfaction could be the presence of an interaction or moderator effect. A moderator effect is when an independent variable changes the form of the relationship between another independent variable and the dependent variable (Hair e.a., 2006). Therefore the possible interaction with the other attitudinal variables is included in a new regression performed. The attitudinal variables considered to show interaction with satisfaction are perceived quality, awareness/associations and loyalty. Assessing the significance of an interaction term is accomplished by evaluating incremental R², not the significance of individual coefficients, due to high multicollinearity (Hair e.a., 2006). The results of the regression analysis show no significant improvement of R² in the null model (appendix 5), so no significant interaction term is found.

4.2.3. Control Variables

Covariates or control variables are variables that are not part of the research design, yet need to be accounted for in the analysis (Hair e.a., 2006). Based on the literature some variables have been identified to influence cross-buying intentions, consumer characteristics, like age, gender, income and education and past usage, measured as the length of the relationship. These variables will be controlled for in the multiple regression analysis. They are entered first in to the model, to remove the variation associated with the covariates. In appendix 6 we can see that age, gender, income, education and length of relationship don‟t have a significant effect on the dependent variable. Therefore we will limit the regression analysis to the main effects, to assure the model is interpretable.

36

4.3 Multiple Regression Analysis

A forward regression has been performed, in this section the results will be described and are summarized in table 4.5. The first variable entered into the model is perceived FIT or similarity, because it shows the highest correlation with the dependent variable. In the first model can be seen that FIT explains almost 50% of the variance in the dependent variable. The coefficient used is the standardized beta-coefficient, because this gives the relative magnitude of the variable (Hair e.a., 2006). The β for the first model is 0,639 and shows high significance (p<0,0001). If consumers perceive high fit between the focal product and the promoted product, cross-buying intentions will rise. Next the partial correlations and significance of the other variables are checked. Promotion appreciation shows the highest partial correlation and is therefore the second variable included.

The second model shows significant improvement (R²=0,506, p<0,05) and a significant positive relationship (β = 0,179, p<0,05). So a positive attitude towards the promotion and high perceived fit increase cross-buying intentions. In the third model SATISFACTION is included and a significant model improvement is shown (R²=0,521, p<0,05). However, SATISFACTION shows a negative beta-coefficient (β = -,132) and significance (p<0,05). In model 4 the UTILITARIAN ATTITUDE variable is included, with significant improvement of the model (R²=0,544) and the positive coefficient (β = 0,158) is significant. The utilitarian attitude measures if consumers perceive the brand used as functional, the positive coefficient, shows that the more functional a products is, the higher the cross-buying intentions.

The other variables reflecting the past relationship experience by the consumer AWARENESS/ASSOCIATIONS, QUALITY, LOYALTY and HEDONIC ATTITUDE are entered last in the regression analysis. They show no significant partial correlation with the dependent variables and show no significant improvement of the model or show any significant coefficient.

37 Independent Variables Model I Model II Model III Model

IV Model V Model VI Model VII Model VIII Constant 0,304 -0,425 0,469 -0,568 -0,638 -0,652 -0,624 -0,626 Fit 0,693*** 0,611*** 0,652*** 0,633*** 0,629*** 0,626*** 0,627*** 0,628*** Promotion Appreciation 0,179* 0,174* 0,179* 0,179* 0,179* 0,184* 0,183* Satisfaction -0,132* -0,167* -0,181* -0,183* -0,186* -0,186* Utilitarian Attitude 0,158* 0,145* 0,142* 0,139 0,139 Awareness/Association s 0,042 0,040 0,036 0,033 Perceived Quality 0,010 0,004 0,002 Loyalty 0,023 0,022 Hedonic Attitude 0,009 R 0,693*** 0,711* 0,722* 0,738* 0,739 0,739 0,739 0,739 R² 0,480 0,506 0,521 0,544 0,546 0,546 0,546 0,546 F - 6,221 4,008 6,063 0,333 0,014 0,094 0,012 * p < 0,05 ** p < 0,01 *** p < 0,001