Master Thesis
“Private label copycatting and the effect on national brand
evaluation and willingness to buy.”
Name:
Florian Daniël Pruim
Studentnr: 104234702
Study:
MSc Business Administration
Track:
Marketing
Institution: University of Amsterdam
Supervisor: Dr. K. Venetis
Date:
24/06/2016
Statement of originality
This document is written by student Florian Daniël Pruim who declares to take full responsibility for the contents of this document.
I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the text and its references have been used in creating it.
The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.
Acknowledgement
This master thesis is written as a final project of the MSc Business Administration, specialization in Marketing at the University of Amsterdam. There are several people I would like to thank for their help during the process of research and writing my thesis.
First of all, it would like to thank my thesis supervisor, Dr. Karin Venetis for her help and critical feedback, which kept me on the right track and allowed me to keep an eye on the final goal. By providing additional and important insights, I was able to improve the thesis and to be self-critical.
Additionally, I would like to thank my family, friends and my girlfriend who greatly supported me throughout the whole process of writing the thesis with their moral support.
Finally, I would like to thank the participants who filled in my survey, as without their help, this project could not be finished.
Table of contents
Statement of originality 2 Acknowledgement 3 Abstract 6 1. Introduction 7 2. Literature Review: 9 2.1 Private label brand 9 2.1.1 Private label growth 9 2.1.2 Quality perception 11 2.1.3 PLB evaluation 11 2.1.4 PLB willingness to buy 13 2.2 National Brand 14 2.2.1 NB brand equity 14 2.2.2 NB willingness to buy 15 2.3 Presence of NB 17 2.4 National brand type 21 3. Conceptual model 23 4. Methodology 24 4.1 Survey and population 24 4.2 Brand selection 24 4.3 Research design 25 4.3 Variables and measurement 27 4.4 Pre-test 28 5. Analysis 28 5.1 Pre-test 28 5.1.1 Outcomes of the pre-test 28 5.2 Experiment 30 5.2.1 Response analysis 30 5.2.2 Group differences 31 5.2.3 Brand broadness 35 5.2.4 Scale reliabilities 35 5.2.5 Distribution 36 5.3 Hypotheses testing 37 6. Discussion 44 6.1 Private label brand 45 6.1.1 PLB evaluation 45 6.1.2 PLB Willingness to buy 47 6.2 National brand 49 6.2.1 Effect on the NB 49 6.2.2 NB presence 51 6.2.3 NB type 53 6.3 Implications 54 6.3.1 Theoretical implications 54 6.3.2 Managerial implications 55 7. Conclusion 56 7.1 Summary 56 7.2 Limitations 578. References 59
9. Appendices 62
Appendix I: (Dutch) Pre-test 62
Appendix II: (Dutch) Final Experiment 67
Abstract
Private label brands (PLB) are growing rapidly and in order to signal high quality, manufactures are relying more on copycatting the national brand (NB) product label. The prevailing belief is that the more similar the PLBs are to the NB, the more positive their evaluation is. However, it is found that this effect depends on the evaluation mode of the consumer. Whenever the NB is present, the copycatting practices would be highlighted, resulting in a less favorable evaluation for high degree copycats. This study aims at investigating this relationship and to explore whether there is also an effect of copycatting on the NB. Two different product categories (i.e. chocolate and olive oil) where used in order to compare the difference between a broad and a narrow NB. The conceptual model is tested through statistical data obtained from an online survey. The analysis showed that in the olive oil category, the higher degree copycat was indeed evaluated more positively. However, no effect was found on the willingness to buy the PLB as well as on the evaluation and the willingness to buy the NB. In contrast to previous studies, no moderating effect was found for the presence of the NB. A comparison between the two different NB types showed that there was no difference in effect on the evaluation and willingness to buy. Contrary to previous findings, the results of this study imply that the copycatting practices of the PLB are not too much of a treat for the NB. This has implications for both marketing theory and practice, which will be discussed.
1. Introduction
Brands that are manufactured by one company and sold under another retailers’ brand, also known as private label brands (PLB), are experiencing great successes in the recent history. Over the past five years, the growth of private label brand sales has been exponential and there is still a huge growth potential since sales are only ten percent of all grocery value (Euromonitor International, 2013). As a group, PLB have higher market share than the top national brands in over a quarter of product categories (Quelch & Harding, 1996 in Batra & Sinha, 2000) and as the worldwide share of PLB increases, the importance of PLB-related research increases as well (Hyman, Kopf & Lee, 2010).
The rapid growth of PLB can be explained by several factors, including price advantage, increased quality and better packaging. Most importantly, it is found that the primary determinant of the success is the ability of offering an acceptable level of perceived quality (DelVecchio, 2001). At the same time, it is found that consumers with low product category knowledge are likely to rely on the commonly held belief that PLB are inferior in quality relative to national brands as a heuristic in making a purchase decision (DelVecchio, 2001).
In order to signal to the consumer that PLB products are of similar quality, manufactures of PLB rely more and more on copycatting the national brand product label. In a survey in the US, Scott-Morten & Zettelmeyer (2004) found that half of the PLB products imitated the NB in size, shape and color. Likewise, in a survey among 100 brand managers in the UK, over a half had their brand copied by a supermarkets and more than eighty percent lost sales to imitators (Collins-Dodd & Zaichkowsky, 1999).
The reasoning behind the copycatting practice is that larger similarity leads to a higher likelihood of brand confusion which might transfer the positive associations of the NB to the PLB. Several studies found that this could harm the NB, since they incest in building a strong brand, but the PLB takes advantage of this (Loken, Ross, and Hinkle 1986; Warlop and Alba
2004 in Van Horen & Pieters, 2012b). Additionally, consumers could may make inferences and attributions of similarity of use, content and other product characteristics, which could be in favor of the PLB (Kapferer, 1995).
However, it is found that when the NB is present, highly similar copycat PLBs are evaluated less favorable then subtle copycats (Van Horen & Pieters, 2012b). The reasoning behind that is that consumers could feel resistance to blatant copycatting products, because they realize the PLB tries to leverage the NB reputation trough imitation. Additionally, it could be that deliberate copycatting highlights the qualities of the NB, causing the copycat to pale in comparison with the NB, which could lead to the consumer valuating the PLB more negatively (Van Horen & Pieters, 2012b).
Although it is known what the effect on the PLB is, it remains unclear whether this effect also works the other way around, and that besides that the PLB is evaluated more negatively, the NB is evaluated more positively. It could for instance be the case that when consumers feel resistance to blatant copycatting, they shy away from the PLB and try to
‘defend’ the NB and so evaluate the NB more positively. Additionally, theory remains silent on
what this means for both NB and PLB product preference. It could be that the negative PLB evaluation leads to a switch to the NB, but it could also be the case that consumers evaluate the PLB less favorable, but still buy their products because of a price advantage.
To overcome this gap in the literature, the research question for this study will be: “Whether and how does the degree of copycat of private label brands influence product
preference of national brands and how is this effect moderated by the presence of the national brand?”
The remainder of this study will firstly describe the existing literature about PLB and the effect of copycatting, to identify what is already known and which parts needs further examination. After this a conceptual framework will be established together with corresponding hypotheses. Thirdly, the sample will be described together with the methods used for this study. Subsequently, a statistical analysis on the obtained data will be performed, as well as a discussion about the implication of the results. Finally, a conclusion will be provided including recommendations for future research.
2. Literature Review:
2.1 Private label brand
2.1.1 Private label growthBrands owned by a retailer or supplier, also known as private label, are quite common in retailing since the economic recession of the late 1980s and have gained a sizeable share of the global retail sales since then (Ailawadi, Pauwels, & Steenkamp, 2008). All developed countries have witnessed a steady increase in the share of PLB with nowadays an overall penetration of over 40 percent in some European countries with peaks up to 70 percent in some product categories (Boyle & Lathrop, 2013). A part of the recent success of PLB is because of the global economic crisis. Research found out that the PLB share increases during economic downturns and that consumers tend to switch more extensively to PLB during bad times that they switch back to NB after a subsequent recovery (Lamey, Deleersnyder, Dekrimpe & Steemkamp, 2007).
However, the success of PLB depends to a large extend to the category it belongs to. First of all, in making a buying decision consumers will take into account the risk of making a
wrong brand choice. Whenever this risk is high, for example with durables, an established brand that has proven itself over a longer period of time, can serve as a way to reduce this risk (Batra & Sinha, 2000). At the same time, the nature of the product features makes a difference in the choice for a PLB of a national brand. It is found that in hedonic product categories, in which people first have to experience the product before they can evaluate it, consumer tends to prefer a national brand over an PLB (Batra & Sinha, 2000). So both come down to the risk reduction because of the stronger credibility of a NB then a PLB. This study focuses on the FMCG, in which the risk of making a wrong purchase decision is very little and consumers are more likely to try out different brands. Therefore, it isn’t very surprising that the share of PLB in this sector is relatively high compared to other sectors.
This success is partly due to the fact that private label products aren’t perceived as low cost alternatives to national brands anymore, but they are of increasing quality and fulfill consumer needs across a variety of price points (Nielsen, 2014). Besides this, PLB’s are typically 15%-40% cheaper than national brands, which make them highly attractive to price conscious consumers (Ashley, 1998). In a study where participants had to rate both national brand and store brand products after a blind test, it turned out that the store brand was preferred above the established national brand, which is consistent with the trend of increasing product quality of private label (De Wulf, Odekerker-Schröder, Goedetier & Van Ossel, 2005).
On the other hand, from the retailer perspective, offering more PLB’s is interesting since it leads to higher margins, higher store loyalty and better bargaining positions with manufacturers. Additionally, it allows the retailer to differentiate from competitors (Ailawadi & Keller, 2004). These benefits lead to retailers introducing more and more PLB for a larger share of consumers that are willing to buy these products.
2.1.2 Quality perception
Although consumers recognize the increased PLB product quality, it isn’t perceived as good as NB yet. In a consumer survey research by Nielsen (2014) on PLB, 29% of global consumers stated that the quality equals that of NB. This implies that the majority of consumers still views the PLB quality as lower. Further, it is found that consumers with low product category knowledge are likely to rely on the commonly held belief that PLB are inferior in quality relative to national brands as a heuristic in making a purchase decision (DelVecchio, 2001).
By the frequent introduction of new and improved products, NB manufacturers try to retain their position of being of better quality. PLB manufacturers namely often lack the technical and financial resources to introduce these products themselves, so they are stuck with trying to copy the innovations that NBs introduce (Steenkamp et al. 2010). Additionally, NB try to increase the quality gap by distinctive packaging to be recognized in the large number of competing stock keeping units (SKUs) in a retail store and given the limited time consumers spend on each purchase decision (Kapferer, 1996). Visual uniqueness will namely increase the speed of brand recognition and the accuracy with which it can be differentiated from competitors (Warlop & Alba, 2004).
2.1.3 PLB evaluation
In order to be competitive, PLB manufacturers rely more and more on copycatting the national brand product by minimizing the feature differentiation in terms of packaging, size and typeface. This is a huge difference with the functional oriented packaging at the beginning of PLB. By having minimum differentiated products, retailers try to signal to their customers that the quality of their PLB is similar to the national brand (Choi & Coughlan, 2006). However, an important precondition for this copycatting strategy to be successful is to make a connection or relation to the NB. In that case, the (positive) NB association is activated and the transfer of
knowledge and affect can eventually take place which can lead to a more favorable evaluation of the PLB (Fazio, 1986 in Van Horen & Pieters, 2012a). Additionally, increasing visual similarity could lead to brand confusion, whereas the product differences are concealed and the market transparency reduces (Kapferer, 1995). When the consumer mixes up the PLB and the NB products because of the similarity, the positive association of the latter can be transformed to the PLB, resulting in a more favorable evaluation of the PLB.
On top of this, consumers could use price, packaging or brand names as cues to asses a products quality thereby reducing uncertainty of making a wrong choice (Kirmani & Rao, 2000). So, when consumers are in a situation where information is lacking and a known brand is not present, they can use similarity to a brand to overcome the gap and to reduce the risk. The similarity in packaging design will generate a positive feeling of familiarity and will be used to infer similarity of quality, performance and reliability (Van Horen & Pieters, 2013).
Interestingly, Warlop & Alba (2004) found out that consumers do understand that the copycat brand tries to suggest similarity of quality with the NB, but that skepticism is not the modal response. Additionally, it turned out that visually salient features which are not necessarily relevant for the product, are of great influence in making a purchase decision (Warlop & Alba, 2004). So this would imply that a salient package, which is highly similar to the NB would lead to a more favorable attitude towards the PLB. It could namely on the one hand lead to the transformation of positive associations, while on the other hand consumers seem not to be skeptical about the persuasion tactics that are used. This leads to the following hypothesis:
H1: A higher degree of copycat from the PLB will result in a more favorable attitude towards this PLB
2.1.4 PLB willingness to buy
On the one hand, brand confusion because of similarity in trade dress could lead to the consumer accidentally picking the PLB over the intended NB. In such situation the consumer makes a subconscious mistake which leads to the PLB product at least be tried once. Whenever the consumer thinks the quality turns out to be sufficient, this could lead to a conscious switch to the PLB. The decision heuristic could in that case become ‘buy the brand I bought last time, since it was satisfactory’ (Hoyer & Brown, 1990).
In the case of FMCG, purchase decisions are often made with low involvement, because these products are bought frequently. This implicates that the consumer spends little time and cognitive effort and will rely on heuristics in choosing brands. The major goal in repetitive and relatively unimportant decisions is to make a satisfactory choice while minimizing effort rather than making the ‘optimal’ choice (Hoyer, 1984). Consumers may because of the low involvement, wrongly believe that NB and PLB products are made by the same company or by companies that have a business connection, because they look so similar (Loken, Ross & Hinkle, 1986). In that case the consumer observes the difference in products, but generalizes the origin of the products. It could also be the case that because consumers exercise a low degree of care when evaluating the product, this causes them to be less likely to notice differences between the brands, which leads to the conclusion that they are of similar origin (Howard, Kerin & Gengler, 2000).
Additionally, it could be that consumers not only generalize the origin of the product but use the similarity between the exterior features to infer similarity of product quality. This could be due to the heuristic rule that “things that look the same, must work the same” (Eagly & Chaiken, 1993 in Warlop & Alba, 2004). Given the fact that most copycats are priced below the imitated NB, this would result in the consumer willing to buy the PLB rather than the NB
since it offers better value for money (Collins-Dodd & Zaichkowsky, 1999). This leads to the following hypothesis:
H2: A higher degree of copycat from PLB will result in an increase of willingness to buy the PLB
2.2 National Brand
2.2.1 NB brand equityBased on the above, it can be concluded that by being more similar, the PLB takes advantage of the established NB brand equity and a part of the brand equity is transferred to the PLB (Van Horen & Pieters, 2012b). However, it is unclear whether this goes at the cost of NB brand equity or that the effect is unilateral.
On the one hand, the exposure to a copycat brand can reduce the accuracy and speed of the recall of first-user brand information. As Morrin & Jacoby (2000) demonstrate, the usage of brand diluting logos tends to reduce brand exclusive recall levels. In their study, a single exposure to diluting brand stimuli is found to have already a damaging effect on the speed and accuracy of brand information. This would of course be unfavorable for the NB, who want their brand information to be recalled accurately and quickly in order to maintain their image.
Secondly, when the consumer wrongly beliefs that because of the similarity in trade dress, the products are made by the same company (Loken, Ross & Hinkle, 1986), the PLB product may be seen as a brand extension. When the PLB information is inconsistent with the NB information, this could cause confusion by consumers. As Loken & John (1993) argue, beliefs about a parent brand are diluted mostly, when extensions are consistent with some attributes from the parent brand, but not all. When the extension product is at par with the most attributes, but is inconsistent in some important other attributes, this could harm the parent
would suggest that a PLB that looks very similar to the NB and might be at par with some product attributes, but is inconsistent with others (e.g. the design of the products looks similar, but the quality is less), could dilute the beliefs about the NB.
In order to protect the NB, trademark laws exist to penalize copycat brands that are too similar to the leader brand and take unfair advantage of the marketing efforts of the NB. The reasoning behind this is that trademark dilution leads to the NB being less able to identify and distinguish itself from competitors (Morrin & Jacoby, 2000). When the PLB creates the same associations because of similarity in trade dress, this could of course be harmful for the NB, since being less distinctive could in the end lead to a less unique and favorable evaluation of the NB.
On the other hand, it could be the case that blatant copycats highlight the superiority of the NB product and at the same time shows the weaknesses of the PLB (Van Horen & Pieters, 2012b). In that case, the PLB pales in comparison with the NB and the consumer could become aware of the differences between the products rather than the similarities. When the consumer focusses on the product attributes in which the NB is superior, this could lead to a more favorable evaluation of the NB. Since the discussed effects of inconsistent information and highlighting the superiority are contrary, they might rule each other out. This leads to the following hypothesis:
H3: A higher degree of copycat from the PLB will have no effect on the evaluation of the NB.
2.2.2 NB willingness to buy
As discussed earlier, purchase decisions in the FMCG sector are often made with low involvement and on the base of heuristics (Hoyer, 1984). An important cue in making the purchase decision is brand awareness. Consumers often don’t buy the products that are of the
advertising, strict quality controls and extrinsic cue effects from NB have led to high awareness and signaling a quality reassurance (Baltas, 1997). It is found that consumers rely so much on brands, that they prefer established brands over alternatives even when the established brand is inferior on other product characteristics (Muthukrishnan, Wathieu, & Xu, 2009). Additionally, consumers with low product category knowledge are likely to rely on the commonly held belief that PLB are inferior in quality relative to national brands as a heuristic in making a purchase decision (DelVecchio, 2001). This would result in the consumer picking the NB over the PLB.
However, when the PLB looks very much alike, it can benefit from the brand awareness created by the NB. As discussed earlier, by brand confusion, consumers might mix up the products and wrongly recognize the PLB as the NB (Kapferer, 1995). The PLB then makes use of the awareness created by the NB and can in that manner attract the consumer.
Secondly, the WTP a price premium for a NB depends on the perception of a quality gap between the NB and the PLB and the involvement in the product category (Steenkamp et al., 2010). When the packaging of the PLB is similar to the NB, stimulus generalization is likely, where similarity in exterior are used to infer similarity in product quality. Whenever the consumer asses the products as of equal quality, this would imply that he isn’t willing to pay a price premium for the NB anymore. In that case the consumer would pick the PLB over the NB since it offers better value for money.
Additionally, it turned out that that consumers who are highly involved in a product category, are more concerned with the product they buy. Quality is then more important together with the functional, social and psychological outcomes of the product (Steenkamp et al., 2010). Since purchases in the FMCG are often made with low involvement, these outcomes are less important and people are more likely to try out other (for example cheaper) products. This would imply that when the PLB looks similar, this would have a negative effect on the willingness to buy the NB. This leads to the following hypothesis:
H4: A higher degree of copycat from the PLB will result in a decrease of willingness to buy the NB
2.3 Presence of NB
The presence of the NB can have an important impact on both the effect of the degree of copycat on the PLB as on the NB. Van Horen & Pieters (2012b) namely found that the presence of the NB negatively influences the evaluation of a highly similar copycat PLB. This is due to the fact that the evaluation of a brand depends on the evaluation mode of the consumer, which could be comparative or non-comparative.
Non-comparative evaluations are made without an explicit comparison, so that there are no direct references and the evaluation is made in isolation (“How do you like this brand?”). This would be when the PLB is evaluated without the NB product being present. In that case, by similarity in trade dress, the positive NB information is interpreted and included in the representation of the PLB. The (positive) NB association is activated and the transfer of knowledge and affect can eventually take place which can lead to a more favorable evaluation of the PLB (Van Horen & Pieters, 2012b).
On the other hand, comparative evaluations are made with the presence of an explicit comparison, so that the evaluation will be guided by this benchmark (“How do you like this brand, compared to the other brand?”). This would be when both the PLB and the NB are present. It turned out that under the comparative evaluation method, highly similar PLB are evaluated more negatively than moderate or low copycat PLBs. This is partly due to the fact that comparison will lead to the PLB looking less attractive next to the high standard of the NB. The copycat PLB highlights the positive features of the NB, which makes the PLB pale and makes the NB looking more attractive (Van Horen & Pieters, 2012b). Additionally, consumers are more likely to become aware of the similarities between the PLB and the NB and that their
judgment might be biased because of the positive feelings generated by the NB. Consumers might realize that the similar PLB tries to leverage the positive image of the NB to their own brand and in that manner try to persuade the consumer. This could result in an opposite reaction in the form of resistance against the PLB (Van Horen & Pieters, 2012b).
In addition, it turned out that if people are aware that contextual information influences their judgment, they will consult their beliefs in judging the appropriateness of the influence (Petty, Brinol, Tormala, & Wegener, 2007). So when the consumer judges the imitation strategy as inappropriate, they tend to correct for the positive feelings induced through similarity (Van Horen & Pieters, 2012a). Ultimately the presence of the NB could lead to consumers evaluating the highly similar PLB in a more negative manner (Van Horen & Pieters, 2012b). This leads to the following hypothesis:
H5a: NB presence will have a negative moderating role, such that a higher degree of copycat from the PLB will result in a less favorable evaluation of this PLB.
When the consumer becomes aware of the persuasion tactics used by the PLB, this could not only lead to a more negative evaluation of the PLB (Van Horen & Pieters, 2012b), but could lead to such resistance that the consumer isn’t willing to buy the PLB product anymore. In that case, the unethical persuasion tactics lead to the consumer shying away from the PLB product, and trying to ‘defend’ the NB by buying their products. However, given the fact that the purchases in the FMCG are made with low involvement (Hoyer, 1984), it might be that consumers don’t react that drastically.
It could namely also be the case that on the one hand consumers have a less favorable attitude towards the PLB, but on the other hand are still willing to buy the product because of the price advantage. With the national brand being present, the consumer is namely better
capable to compare prices and might try out the cheaper brand. At the same time, it would enable the consumer to closely compare the products, which might be in favor of the NB since it could highlight the NB’s superiority and let the PLB pale. Hence, a negative moderating role is expected for the NB, since it could on the one hand lead to the consumer feeling such resistance due to the persuasion tactics and on the other hand observing the superiority of the NB, which both would lead to a negative effect on the willingness to buy the PLB.
This leads to the following hypothesis:
H5b: NB presence will have a negative moderating role, such that a higher degree of copycat from the PLB will have no effect on the willingness to buy this PLB.
As previously discussed, the presence of the NB has a negative moderating role on the evaluation of the PLB, however the question is what the effect is on the evaluation of the NB itself. On the one hand, comparative evaluation could lead to consumers shying away from the PLB and feel sympathy for the NB which is under attack. Close comparison could lead to the consumer observing the superiority of the NB, what would result in a more favorable evaluation, when the NB is present.
However, it could also be the case that the presence of the NB causes more confusion since there is a direct comparison. When the competing products propagate different brand information, this would lead to inconsistency in the consumer’s brand representation. This could especially be damaging when the consumer uses the similarity in trade dress to infer similarity of origin. When the consumer expects both the products to be produced by the NB company, but the information of the PLB is different from the NB, this could lead to confusion. For example, when the consumer thinks that the cheaper PLB is also manufactured by the NB
company, this could lead to the NB associated with ‘cheap’ as well, which could damage their quality image (Aaker, 1990).
The above would imply that the NB presence would have a negative moderating role when the consumer infers similarity of origin (Loken et al. 1986), which would lead to confusion because of inconsistent brand information. This confusion would in the end then lead to a less positive evaluation of the NB. However, the NB would have a positive moderating role, when the consumer feels sympathy for the NB being under attack and when comparison highlights the superiority of the NB. Since these effect are contrary, no moderating role is expected for the NB presence. This leads to the following hypothesis:
H5c: NB presence will have a neutral moderating role, such that a higher degree of copycat from the PLB will have a no effect on the evaluation of the NB.
For the NB, the question remains whether consumers consider the PLB copycatting practices as such unfair that they shy away from the PLB products, or that the price advantage is to important. According to Steenkamp et al. (2010), the NB should not heavily rely on price promotions to reduce the PLB price advantage, since that forces consumers to primarily focus on price. This would make consumers more price sensitive and causes the willingness to pay a price premium for the NB to decrease. Rather than focusing on price promotions, NB should focus on advertising and differentiation, since this would result in the consumer to perceive the products as different rather than similar. The presence of the NB would in that case highlight both the unethical practices and the inferiority of the PLB.
In line with the reasoning that the presence of the NB would have a negative moderating role on the willingness to buy the PLB, a positive moderating role is expected for the NB. The presence of the NB would namely enable the consumer to observe the blatant copycatting,
which could both lead to resistance to the PLB and highlight the superiority of the NB. Both would result in a positive effect on the willingness to buy the NB. This results in the following hypothesis:
H5d: NB presence has a positive moderating role, such that a higher degree of copycat from the PLB will have no effect on the willingness to buy the NB.
2.4 National brand type
While there are both narrow (i.e. brands with products in only one product category) and broad (i.e. brands that have products in several product categories) NB types, this could influence the moderating role of the NB presence. For a brand that is only competing in one product category, there will be a strong category association, while this won’t be the case with a broad brand. Additionally, it is found by Meyvis & Janiszewski (2004) that broad brands tend to have more accessible benefit associations that narrow brands, since the strong category association cannot interfere with the benefit associations.
For this research this might imply that when consumers observe a PLB that copycats a narrow brand that is strongly connected to a category, this leads to more resistance then a PLB copycatting a broad brand. This could be because the associations between the category and the NB are so strong, that the consumer considers them always simultaneously. In that case, it could be that once there is a new brand that tries to make use of the NB brand equity by copycatting, this brand might feel like an ‘intruder’ which leads to high resistance. For the broad NB, this link between the category and the brand would be less strongly, such that there are also different and additional brand associations. A PLB that uses copycatting then leads to less resistance, since it interferes less with the category associations. This would imply that the when the NB is present and the brand type is narrow, a higher degree of copycat from the PLB will have a
positive effect on both the evaluation and the willingness to buy the NB. This would be due to the strong resistance against the unethical copycatting practices. This leads to the following hypotheses:
H6a: NB type will have a positive moderating role, such that a narrower NB type will lead to a positive effect on the evaluation of the NB, with a higher degree of copycat from the PLB.
H6b: NB type will have a positive moderating role, such that a narrower NB type will lead to a positive effect on the willingness to buy the NB, with a higher degree of copycat from the PLB.
3. Conceptual model
Figure 1 summarizes the theoretical framework and the hypotheses that have been derived.
Degree of copycat PLB Willingness to buy private label brand Evaluation of national brand Willingness to buy national brand National brand type H1 H2 H3 H5a H5d H6a H6b Figure 1: conceptual model H5b Presence of national brand Evaluation of private label H4 H5c
4. Methodology
This section discusses the methods being used to conduct the study. First, the main characteristics of the sample will be discussed. Secondly, the research design and the explanation of the treatments will be discussed. Thirdly, the variables that are used in the study will be discussed together with their measurements. Finally, the design of the pre-test will be discussed.
4.1 Survey and population
This study was done by conducting an online survey using Qualtrics and was distributed via social media and e-mail. All questionnaires were filled in online. The main study involved 295 participants, from which 212 fully completed the experiment (response rate of 72%). The sample was drawn from a large population, including different kinds of participants regardless of age, gender or level of education. From the respondents, 58% were female and 42% were male. Incomplete questionnaires were completed from the analysis, except for demographic information.
Respondents were collected in the Netherlands and to overcome any language biases, the survey was offered in Dutch. Respondents were randomly confronted with one of the four treatments and did not know about the treatments that others received. The questions to be answered were identical.
4.2 Brand selection
In accordance to the study by Van Horen & Pieters (2012b), Bertolli is chosen as the broad national brand for this study, since it has the highest market share, a strong reputation and a distinctive name. Additionally, Milka is chosen as the narrow national brand, as this brand is
used by Van Horen & Pieters (2012a) as well. In this manner, the results of this study are more easily comparable with those previous ones.
Both the high-similarity brand label and the low-similarity brand label of Bertolli could be used from the study by Van Horen & Pieters (2012b). For the Milka brand, only a high-similarity brand label was present (Van Horen & Pieters, 2012a), such that a low-high-similarity brand label is created.
4.3 Research design
The online controlled experiment had a 2 (similarity: low, high) * 2 (national brand: absent, present) * 2 (broadness: narrow, broad) mixed design. Four different treatments were formed in order to test the hypotheses.
The introduction was the same for all groups and indicated that this research was about the introduction of new PLB’s in the supermarket. Respondents were instructed to form their opinion about this brand based on the brand information that was provided and the brand label that they got to see. Every respondent was exposed to both questions about the chocolate and the olive oil categories and related brands. However, it was ensured that none of the respondents had two times a highly similar copycat brand in combination with a national brand being present, as this could reveal the goal of this research. Additionally, some fillers that had nothing to do with the research where included to hide the real purpose. Presentation order was counterbalanced in order to rule out bias arising from order.
After the brief introduction, participants were told that a well-known Dutch supermarket chain would introduce a new PLB in the specific product category (either olive oil or chocolate). The name of the brand was indicated and it was told that this brand would be next to the NB in the store, but for a lower price. The respondents were told to imagine themselves being in the supermarket and looking at the product with the label that was displayed. In the NB present
conditions, on the next page, the NB label appeared, together with the information that this was the NB where the PLB would lay next to. In the NB absent conditions, the NB label was not displayed and only after filling out the questions about the PLB, the NB name was revealed. In this manner, the influence of the presence of the NB could be measured.
The four different treatments will be indicated based on either high (H) or low (L) similarity, and whether the NB is present (P) or absent (A). Even though in the experiment presentation order was counterbalanced, in the labels for the treatments, the olive oil conditions will be indicated first.
Respondents of the first treatment where exposed to the Bellamia (H) label, without Bertolli (A) and the Choco (L) label together with Milka (P). The second treatment included the labels of Santini (L) in combination with Bertolli (P) and the label of Lecha (H), without Milka (A). The third treatment was the same as the first, however now Bellamia (H) was together with Bertolli (P) and Choco (L) was showed without Milka (A). Finally, the last treatment was the same as the second treatment, but now Santini (L) was presented without Bertolli (A) and Lecha (H) was presented together with Milka (P).
Treatment 1: HA | LP Treatment 2: LP | HA Treatment 3: HP | LA Treatment 4: LA | HP
A visualization is given in the table below.
Tabel 1. Visualization of treatments
Degree of Copycat NATIONAL BRAND LOW HIGH PRESENT LP HP ABSENT LA HA
4.3 Variables and measurement
The dependent variables in this study are the evaluation and the willingness to buy both the PLB and the NB. In order to measure the evaluation of the brands, the same scales are used as by Van Horen & Pieters (2012b). The usage of scales derived from previous studies is done in order to maximize the reliability and internal validity of this study. Just like the study by Van Horen & Pieters (2012b), four seven-point scale differentials were used to measure the evaluation of the brands (bad-good, unattractive-attractive, uninteresting-interesting, negative-positive).
For willingness to buy, an addition to the study by Van Horen en Pieters (2012b) for measuring the success of a brand was used. To measure this variable, three seven-point Likert-scale statements where used (totally disagree – totally agree). The statements were, respectively: (1) It is likely that I would buy brand X, (2) Buying brand X is a good decision and (3) I would try out brand X. For brand evaluation and willingness to buy, the same scales where used for both the PLB and the NB, in order to maximize internal validity.
In the experiment, there is controlled for involvement, buying behavior and price sensitivity of both the olive oil and chocolate categories by the usage of five different seven-point Likert-scale statements (totally disagree – totally agree). The statements were, respectively: When I buy olive oil/chocolate: (1) the taste matters, (2) the brand matters, (3) the price is important, (4) All brands look the same, (5) When buying olive oil/chocolate, making the right choice is important.
Additionally, the familiarity with the NB (i.e. Bertolli and Milka) is checked with one seven-point scale question (totally unfamiliar/totally familiar) and the perceived broadness of the brand with another seven-point scale question (small/broad).
Finally, the demographic characteristics of the respondents were collected with respect to age, gender and level of education.
4.4 Pre-test
In order to test whether participants indeed perceive the labels to differ in similarity but to be equally attractive, and to test if Bertolli is seen as a broad brand an Milka as a narrow, a pre-test was established. The pre-pre-test involved 18 Dutch participants (12 men and 6 woman) who were between 16 and 66 years old (mean = 23, 06, std = 14,83).
For the Bertolli brand manipulation, both the low similarity (i.e. Santini) and high similarity (i.e. Bellamia) copycat labels from the study by Van Horen en Pieters (2012b) where used. This was done, since these labels where already found to be different in terms of similarity, but not in attractiveness. For the Milka brand manipulation, the high similarity brand label (i.e. Lecha) was used from the study by Van Horen en Pieters (2012a) and the low similarity brand label (i.e. Choco) was self-created.
5. Analysis
5.1 Pre-test
5.1.1 Outcomes of the pre-test
Two paired t-tests were conducted to compare the level of similarity between the high and low similarity copycat labels. There was a significant difference in the similarity scores for Bellamia (M=5.78, SD=1.17) and Santini (M=3.22, SD=1.31); (t (17) = 5.848, p<.000), as well as for the labels of Lecha (M=5.17, SD=1.47) and Choco (M=3.22, SD=1.59); (t (17) = -4,357, p<.000). These results imply that the high similarity copycat labels (i.e. Bellamia and Lecha) are indeed perceived as more similar to the NB (i.e. Bertolli and Milka) then the low similarity copycat labels (i.e. Santini and Choco).
Additionally, two paired t-test were conducted to compare the level of attractiveness between the high and low similarity copycat labels. There was no significant difference in attractiveness for Bellamia (M=4.67, SD=1.37) and Santini (M=4.06, SD=1.30); (t (17) =1.775, p=.094). The same applies to Lecha (M=4.17, SD=1.72) and Choco (M=4.28, SD=1.36), were there was neither a significant difference in attractiveness; (t (17) =.220, p = 0.828). This implies that although the level of similarity is significant different for the labels, they are perceived as equally attractive, which is a necessary precondition for the labels to be used in the final experiment.
In order to check the broadness of the Bertolli and Milka brand in comparison with each other, four other brands that are available in the supermarket (e.g. Unox, Robijn, Calve and Dubbelfriss) where included in the pre-test. The broadness of a brand was measured by asking both the main product association with the brand and the number of remainder product associations. The rank and corresponding number of associations is displayed in the table below, together with the significance of the mean differences.
Tabel 2. Number of associations and mean differences
Rank Brand Mean of
associations Std. dev 1 2 3 4 5 1 Unox 2.11 1.13 - - - - - 2 Bertolli 2.0 1.71 .828 - - - - 3 Robijn 1.72 1.60 .401 .491 - - - 4 Milka 1.56 1.20 .213 .392 .708 - - 5 Dubbelfriss 1.33 1.78 .044* .199 .287 .655 - 6 Calve 1.28 .96 .031* .033* .289 .439 .902
* difference is significant at the 0.05 level (2-tailed)
From a repeated measures ANOVA, it turned out that there was a significant difference between the broadness of the brands (F (5,13) = 3,33, p=.037). However, looking at the pairwise comparison it became clear that there is a non-significant difference between the means of associations of Bertolli and Milka (p=.392). Nevertheless, it was decided to use both Milka and Bertolli in the main research, since both studies by Van Horen en Pieters (2012b, 2012a) used
these brands as well and it was excepted that the difference in broadness between the brands would be revealed in a direct comparison in the main study.
5.2 Experiment
5.2.1 Response analysisThe main study involved 295 participants, from which 212 fully completed the experiment (response rate of 72%). Another 6 participants were excluded because of missing data, leaving 206 respondents for the analysis. Respondents who only didn’t fill in either age, gender or level of education were decided to be retained, as these characteristics are not part of the main analysis.
First all the descriptive statistics were analyzed. The study involved 52% woman and 48% man, ranging from 16 till 73 years old, with a mean of 27,5 years. Eight respondents didn’t fill in their age and one did not fill in gender. The biggest part of this group (76,7%) was higher educated (HBO/WO). All respondents filled in their level of education. The age distribution and of the total sample is displayed below:
Tabel 3. Age distribution of the sample
The respondents were randomly divided into one of the four different treatments. In order to compare the different treatments in terms of age, gender and educational distribution, the
Age Frequency Percent Cum. Percent
Under 18 11 5.6% 5.6% 18-25 135 67.7% 73.3% 26-34 14 7.5% 80.8% 35-49 18 9.1% 89.9% 50-64 16 8.1% 98% 65 and older 4 2% 100% Total 197 100%
Tabel 4. Demographic statistics of the sample
Treatment 1 Treatment 2 Treatment 3 Treatment 4 Freq. Percent Freq. Percent Freq. Percent Freq. Percent Gender: Male 19 38.8% 24 49% 26 42.6% 17 37% Female 30 61.2% 25 51% 35 57.4% 29 63% Age: under 18 2 4.2% 1 2.1% 5 8.2% 3 7.1% 18-25 32 66.6% 36 76.6% 42 68.8% 25 59.6% 26-34 4 8.4% 2 4.3% 5 8.2% 3 7.1% 35-49 6 12.6% 4 8.5% 3 5% 5 11.9% 50-64 3 6.3% 4 8.5% 5 8.2% 4 9.5% 65+ 1 2.1% 0 1 1.6% 2 4.8% Missing 1 3 4 Education: Elementary 0 0 1 1.6% 0 VMBO / MAVO 1 2% 1 2% 1 1.6% 3 6.5% MBO 4 8.2% 3 6% 4 6.6% 5 10.9% HAVO / VWO 4 8.2% 7 14% 9 14.8% 5 10.9% HBO / WO 40 81.6% 39 78% 46 75.4% 33 71.7% Total 49 50 61 46 5.2.2 Group differences
The distribution of the characteristics of the respondents among the different treatments will be compared in order to find out if there are any significant differences between the groups. This will be done, since group differences could influence the analysis. A Pearson Chi-Square test for independence was conducted in order to check whether there were differences in the distribution of the characteristics. The assumptions of the Chi-Square test for a larger then 2 by
2 table are that (1) no more than 20% of the contingency cells can have an expected value of lower then 5 (2) no cell has an expected count of 0 and (3) data groups must be independent.
Age
For age, the null hypothesis that ‘there is no significant difference between the age distribution among the four different treatments’ was tested with a significance level of 0.05. For this analysis, the age distribution in groups, as discussed in the previous chapter, was used instead of the original data of exact number. This was done since the age groups are more easily comparable then individual exact numbers. However, since all cells have an expected count of less than 5, the first assumption of the Chi-Square test was violated, resulting in the usage of the Likelihood ratio. The outcomes of the comparisons in terms of p-values are tabled below. Tabel 5. Chi-Square test age
Treatment 1 (P-value) Treatment 2 (p-value) Treatment 3 (p-value)
Treatment 1 -
Treatment 2 .540 -
Treatment 3 .458 .435 -
Treatment 4 .540 .456 .246
Based on these outcomes, it can be concluded that the null hypothesis will not be rejected and that there are no significant differences between the different treatments in terms of age distribution.
Gender
For gender, the null hypothesis that ‘there are no significant differences between the gender distribution among the four different treatments’ was tested with a significance level of 0.05. None of the assumptions where violated, such that the Chi-Square test could be used. The outcomes of the comparison in terms of p-values are displayed in the table below.
Tabel 6. Chi-square test gender
Treatment 1 (P-value) Treatment 2 (p-value) Treatment 3 (p-value)
Treatment 1 -
Treatment 2 .140 -
Treatment 3 .295 .906 -
Treatment 4 .417 .360 .142
Based on these outcomes, it can be concluded that the null hypothesis will not be rejected and that there are no significant differences between the different treatments in terms of gender distribution.
Level of education
For level of education, the null hypothesis that ‘there are no significant differences between the distribution of level of education among the four different treatments’ was tested with a significance level of 0.05. Just as with age, the first assumption of the Chi-Square test was violated, resulting in the usage of the Likelihood ratio. The outcomes of the comparisons in terms of p-values are tabled below.
Tabel 7. Chi-Square test level of education
Treatment 1 (P-value) Treatment 2 (p-value) Treatment 3 (p-value)
Treatment 1 -
Treatment 2 .663 -
Treatment 3 .375 .466 -
Treatment 4 .754 .084 .500
Based on these outcomes, it can be concluded that the null hypothesis will not be rejected and that there are no significant differences between the different treatments in terms of the distribution of level of education.
Overall, it can be concluded that there are no significant differences among the different treatments in any of the characteristics that where measured. This implies that the chance that any possible differences in the outcomes are indeed caused by the treatment, increases.
Buying behavior and price sensitivity
Next to the respondents’ characteristics, it was tested whether there were any differences among groups in terms of buying behavior and price sensitivity. The null hypothesis that ‘there are no significant differences in buying behavior and price sensitivity among the four different treatments’, was tested with a significance level of 0.05. The four labels measuring buying behavior were combined into one variable and the one label measuring price sensitivity was analyzed separately. An independent samples T-test was used in order to test the differences among groups. This was done for both buying behavior and price sensitivity of the olive oil (first significance number) and the chocolate category (second significance number) The outcomes are presented in the tables below.
Tabel 8. P-values of difference in buying behavior
Treatment 1 (P-value) Treatment 2 (p-value) Treatment 3 (p-value)
Treatment 1 -
Treatment 2 .452 | .988 -
Treatment 3 .163 | .585 .612 | .592 -
Treatment 4 .796 | .601 .621 | .608 .275 | .995
Tabel 9. P-values of difference in price sensitivity
Treatment 1 (P-value) Treatment 2 (p-value) Treatment 3 (p-value)
Treatment 1 -
Treatment 2 .161 | .254 -
Treatment 3 .892 | .595 .198 | .567 -
Treatment 4 .003** | .980 .172 | .270 .005** | .593
** significant at the 0.01 level (2-tailed)
As can be seen from the tables, there was no significant difference between any of the treatments in terms of buying behavior. However, for the price sensitivity in the olive oil category, there is a significant difference (at the 1% level) between the treatment 1 and treatment 4 and between treatment 3 and treatment 4. Respondents in treatment 4 turned out to
could imply that since they care less about price, they are more in favor of the NB. This will be taking into account in the discussion.
5.2.3 Brand broadness
In order to find out whether in the final experiment Bertolli is perceived as a broader (i.e. higher number of product associations) brand then Milka, a paired samples T-test was used. The results of this test are displayed in the table below.
Tabel 10. Paired samples t-test on NB broadness
Treatment M SD DF t-value Significance 1 Bertolli Milka 4.22 3.20 1.57 2.00 48 3.47 .001** 2 Bertolli 4.46 1.79 49 3.53 .001** Milka 3.30 1.92 3 Bertolli 4.33 1.45 60 2.70 .009** Milka 3.59 2.16 4 Bertolli 4.37 1.37 45 1.57 .123 Milka 3.85 1.87
** significant at the 0.01 level (2-tailed)
For every treatment, except for the fourth, Bertolli has significantly more product associations then Milka, which indicates that Bertolli can indeed be seen as a broader brand. Four the fourth treatment, there is a non-significant effect, however the mean association of Bertolli is higher than that of Milka. This indicates that the manipulation is successfully implemented and that conclusions can be drawn for the hypothesis about brand broadness.
5.2.4 Scale reliabilities
Scale reliabilities tests were performed in order to check the internal consistency of the different measures. The reliability was measured using Cronbach’s alfa for each measurement in the different samples. The scores for each measurement and corresponding treatment can be found in the table below. For every scale, Cronbach’s alfa’s was higher than 0.70, indicating that the
reliability of the scales was good and no modification was needed. Also, none of the items would substantially affect reliability if they were deleted.
Tabel 11. Cronbach’s alfa for the measures in each treatment
Treatment Cronbach’s alfa (olive oil) Cronbach’s alfa (chocolate) Brand evaluation PLB (4 items) 1 .919 .903 2 .950 .934 3 .921 .927 4 .944 .908 Willingness to buy PLB (3 items) 1 2 .841 .886 .855 .905 3 .833 .887 4 .802 .855 Brand evaluation NB (4 items) 1 .964 .892 2 .902 .945 3 .918 .936 4 .860 .901 Willingess to buy NB (3 items) 1 .905 .900 2 .794 .786 3 .734 .927 4 .885 .896 5.2.5 Distribution
An analysis was done in order to find out if the data was normally distributed. The skewness and the kurtosis of the variables are reported in the table below.
Tabel 12. Skewness and Kurtosis of each variable
Skewness Kurtosis
Evaluation PLB (olive oil) -.723 .485
Evaluation PLB (chocolate) -.010 -.704 WTB PLB (olive oil) -.956 .771 WTB PLB (chocolate -.140 -.911 Evaluation NB (Bertolli) -.819 2.323 Evaluation NB (Milka) -1.233 2.218 WTB Bertolli -1.019 1.346 WTB Milka -1.592 3.424
From the output it becomes clear that for the variables measuring the PLB, the values for the skewness and kurtosis are between -1 and 1. It is commonly accepted that variables with such values are accepted to be normally distributed. For the variables measuring the NB, the values for the skewness and kurtosis are outside this range, indicating that these data are not normally distributed. This could very well be because respondents already formed an opinion about these national brands that is either more positive or more negative, rather than skewed around the middle. However, since the dependent variables in this study are measured on an ordinal level, nonparametric tests will be used. For these tests normality is not required, such that this distribution does not cause any problems.
5.3 Hypotheses testing
H1: A higher degree of copycat from the PLB will result in a more favorable attitude towards this PLB
To test this hypothesis, a separate Kruskall-Wallis H test was used for both the evaluation of the olive oil category and the chocolate category. This nonparametric test is used instead of the One-Way ANOVA since the evaluation of the PLB is measured at the ordinal level. For the olive oil category (i.e. Santini and Bellamia), there was a significant effect of degree of copycat on the PLB evaluation χ2(3) = 14.777, p = .002. A Bonferroni adjustment is made for the level alfa for the post-hoc testing. Since there are four pair-wise comparisons, the level of alfa for the post-hoc testing is adjusted to .0125 (0.05 divided by 4). Mann-Whitney U post-hoc tests revealed that there was a significant better evaluation of the higher degree copycat (i.e. Bellamia) than for the low degree copycat (i.e. Santini). For the first treatment, where Bellamia was tested in absence of Bertolli, there was a significant higher score then treatment 2 (i.e. Santini with presence of Bertolli) (M=5.22 > M=4.5), U= 829.5, p=.005 and then treatment 4 (i.e. Santini in absence of Bertolli) (M=5.22 > M=4.49) U = 663, p = .001. For the third
treatment including Bellamia, no significant difference was found with second (p =.088) and fourth treatment (p =.021), although the mean score is higher (M=4.9). However, this could be due to the negative moderating role of the NB presence.
For the chocolate category (i.e. Choco and Lecha), there was a non-significant effect of degree of copycat on the PLB evaluation, χ2(3) = 3.232, p = .357. This indicates that for this category, a higher degree of copycat from PLB has no influence on the evaluation of this brand. Interestingly, the mean scores of the low degree copycat PLBs (i.e. treatment 1: M=5.22 and treatment 3: M=4.90) are higher than their high degree counterparts (i.e. treatment 2: M=4.5 and treatment 4: M=4.49). This indicates that for this category, an opposite effect occurred, although this effect is non-significant. These results imply that the hypothesis is partially supported, but since there is no unanimous result, the hypothesis cannot be accepted.
H2: A higher degree of copycat from PLB will result in an increase of willingness to buy the PLB
To test this hypothesis, a separate Kruskall-Wallis H test was used for both the willingness to buy the PLB of the olive oil category and the chocolate category. For the olive oil category, there was a non-significant effect of degree of copycat on willingness to buy the PLB, χ2(3) = 2.794, p = .424. However, the mean score of treatment 1 (M=4.88) is slightly higher than the mean scores of treatment 2 (M=4.83) and treatment 4 (M=4.59), which direction is in favor of the second hypothesis. The score for third treatment (M=4.81) however, is only higher than the fourth treatment.
Additionally, for the chocolate category, there was also a non-significant, χ2(3) = .490, p = .921. The mean scores of treatment 1 (M=3.96) and treatment 3 (4.0) are higher than that of treatment 2 (M=3.95) and treatment 4 (3.80), which direction is against the second
hypothesis. This implies that there is no effect of the degree of copycat from the PLB on the willingness to buy this PLB, which means that hypothesis 2 must be rejected.
H3: A higher degree of copycat from the PLB will have no effect on the evaluation of the NB.
To test this hypothesis, a separate Kruskall-Wallis H test was used for both the evaluation of the NB of the olive oil category and the chocolate category. For the olive oil category, there was a non-significant effect of the degree of copycat from the PLB on the evaluation of the NB, χ2(3) = 5.057, p = .168. Additionally, for the chocolate category, there was a non-significant effect as well χ2(3) = 3.531, p = .317. This implies that the third hypothesis is supported and that there is indeed no effect of the degree of copycat from the PLB on the evaluation of the related NB.
H4: A higher degree of copycat from the PLB will result in a decrease of willingness to buy the NB
To test this hypothesis, a separate Kruskall-Wallis H test was used for both the willingness to buy the NB of the olive oil category and the chocolate category. For the olive oil category, there was a non-significant effect of the degree of copycat from the PLB on the willingness to buy the NB, χ2(3) = 4.054, p = .256. However, the mean score of treatment 1 (M=5.04) is slightly lower than the mean scores of treatment 2 (M=5.53) and treatment 4 (M=5.20), which direction is in favor of the fourth hypothesis. The score for third treatment (M=5.50) however, is only lower than the second treatment.
Additionally, for the chocolate category, there was a non-significant effect as well, χ2(3) = 3.936, p = .268. The mean scores of the high degree copycat treatments 2 (M=5.64) and 4 (M=5.55) are higher than the mean score of treatment 3 (M=5.53), which direction is against
the fourth hypothesis. However, the mean score of the low degree copycat treatment 1 (M=5.83) is higher than all other treatments, which direction is favor of the the hypothesis. However, the non-significant effects imply that the fourth hypothesis is rejected and that there is no effect of the degree of copycat from the PLB on the willingness to buy the NB.
H5a: NB presence will have a negative moderating role, such that a higher degree of copycat from the PLB will result in a less favorable evaluation of this PLB.
To test this hypothesis, several independent Mann-Whitney U tests are performed for both the olive oil and the chocolate category. For this hypothesis, the difference between the evaluation score of the PLBs with and without the presence of the NB are compared. For the high degree copycats in the olive oil category, the mean score of the NB absent condition was indeed higher (M=5.22) than the mean score in the NB present condition (M=4.90). Although the difference supports the hypothesis, it turned out to be non-significant (U=1251, p = .141). For the low degree copycats it was expected that there would not be a lot of difference between the NB absent and NB present conditions, which was indeed the case. The mean of the NB present (M=4.50) and the NB absent (M=4.49) did not significantly differ from each other (U=1125, p = .857).
For the chocolate category, the same holds for the high degree copycat labels. The mean score for the condition with the NB being absent (M=3.79) was higher than the score where the NB was present (M=3.51). However, the Mann-Whitney U test showed that this difference was non-significant, (U=1039, p =.412). For the low degree copycats it was interesting to see that the mean score for the NB absent (M=4.02) was also higher than for the NB present (M=3.66). But the difference between the two turned out to be non-significant as well, (U=1276, p =.187).
This implies that there is no effect of the presence of the national brand on the evaluation of the PLB and that hypothesis 5a is rejected.
H5b: NB presence will have a negative moderating role, such that a higher degree of copycat from the PLB will have no effect on the willingness to buy this PLB.
As there was a non-significant effect found of the degree of copycat from the PLB on the willingness to buy this PLB (i.e. hypothesis 2), a negative moderating role could only be shown by a negative influence on the willingness to buy. The mean score of the high degree copycat treatment where the NB was present (M=4.81) was indeed lower than when the NB was absent (M=4.88). However, a Mann-Whitney U test showed, that this difference was non-significant (U=1473, p =.894). For the low degree copycats, an opposite effect was found, where the condition with the mean score for the treatment with the NB being present (M=4.83) was higher than the absent condition (M=4.59). Nevertheless, as expected this difference was non-significant (U=1008, p=.294).
The same holds for the chocolate category, where the mean score of the high degree copycat together with the NB is lower (M=3.80) then the score without the NB (M=3.95). Once again, this difference turned out to be non-significant (U=1104, p =.735). For the low degree copycats, the mean scores where almost identical for the NB present (M=3.96) and the NB absent condition (M=4.0). As expected, this difference was non-significant (U=1449, p=.784). This implies that there is no prove for a moderating role of NB presence and that hypothesis 5b needs to be rejected.
H5c: NB presence will have a neutral moderating role, such that a higher degree of copycat from the PLB will have no effect on the evaluation of the NB.
To test this hypothesis, several Mann-Whitney U tests were conducted. For the olive oil category, there was a non-significant difference between the high degree copycat treatments (U=1327, p =.309), as well as for the low degree copycat treatments (U=944, p=.128).
The same holds for the chocolate category, where both the high degree copycat treatments (U=1030, p=.412) and the low degree copycat treatments (U=1453, p =.798) do not significantly differ from each other. This implies that there is no moderating effect of the NB presence on the evaluation of the NB, such that hypothesis 5c can be accepted.
H5d: NB presence has a positive moderating role, such that a higher degree of copycat from the PLB will have no effect on the willingness to buy the NB.
As there was a non-significant effect found of the degree of copycat from the PLB on the willingness to buy the NB (i.e. hypothesis 4), a positive moderating role could only be shown by a positive influence on the willingness to buy. Several Mann-Whitney U tests were conducted. For the olive oil category, there was a non-significant difference between the high degree copycat treatments (U=1304, p =.246), as well as for the low degree copycat treatments (U=930, p=.101). However, although the effect was non-significant, mean scores do indicate that the willingness to buy the NB is slightly higher when the high degree copycat is combined with the presence of the NB (M=5.50) rather than with the absence (M=5.04).
The same holds for the chocolate category, where both the high degree copycat treatments (U=1109, p=.757) and the low degree copycat treatments (U=1264, p =.159) do not significantly differ from each other. The mean scores show a different tendency than in the
olive oil category, since the mean for the high degree copycat in combination with the NB (M=5.55) is lower than the for the one without the NB (M=5.64). Since no significant effect was found, this implies that there is no support for a moderating effect of the NB presence on the willingness to buy the NB. This leads to the rejection of hypothesis 5d.
H6a: NB type will have a positive moderating role, such that a narrower NB type will lead to a positive effect on the evaluation of the NB, with a higher degree of copycat from the PLB.
In order to test this hypothesis, the difference between the evaluation of the narrow NB (i.e. Milka) and the broad NB (i.e. Bertolli) is checked. This is done by comparing the evaluation of the NBs in the treatments of high degree copycats in combination with the presence of the NB. A Mann-Whitney U test was performed and showed that there was a non-significant difference (U=1364, p=.802) between the evaluations of the narrow and broad NBs. The mean scores even indicate that the evaluation of the narrow brand (M=5.52) is slightly below that of the broad brand (M=5.66), which is opposite to what was expected. However, the non-significant score implies that there is no moderating role of NB type and that a narrower NB does not have a positive effect on the NB evaluation. This leads to the rejection of hypothesis 6a.
H6b: NB type will have a positive moderating role, such that a narrower NB type will lead to a positive effect on the willingness to buy the NB, with a higher degree of copycat from the PLB.
In order to test this hypothesis, the same analysis is performed as was done for hypothesis 6a, except now the willingness to buy the NB was used as the dependent variable. From a Mann-Whitney U test it became clear that there was a non-significant (U=1213, p=.222) difference between the willingness to buy the narrow and the broad NBs. The mean scores do indicate that
