Buscar en
Journal of Innovation & Knowledge
Toda la web
Inicio Journal of Innovation & Knowledge Premium Private Label strategies: social networks and traditional perspectives
Journal Information
Vol. 6. Issue 2.
Pages 78-91 (April - June 2021)
Share
Share
Download PDF
More article options
Visits
518
Vol. 6. Issue 2.
Pages 78-91 (April - June 2021)
Regular article
DOI: 10.1016/j.jik.2020.06.003
Open Access
Premium Private Label strategies: social networks and traditional perspectives
Visits
...
Eloy Gil-Cordero
Corresponding author
egcordero@us.es

Corresponding author.
, Francisco Javier Rondan-Cataluña, Manuel Rey-Moreno
University of Seville, Avda. Ramón y Cajal, 1, 41018 Sevilla, Spain
Article information
Abstract
Full Text
Bibliography
Download PDF
Statistics
Tables (13)
Table 1. Influence of the establishment's image on perceived quality
Table 2. Description of the model, social network feelings, Lidl.
Table 3. Fit of the model, social network feelings, LIDL.
Table 4. Parameters of the ARIMA model, social network, Lidl.
Table 5. Description of the model, social network feelings, Carrefour
Table 6. Fit of the model, social network feelings, Carrefour.
Table 7. Fit of the model, social network feelings, Carrefour.
Table 8. Statistics of the model, social network, DIA.
Table 9. Verification of the propositions.
Table A1. Description of the ANOVA model, perceived quality.
Table A2. Description of the ANOVA model, brand value.
Table A3. Description of the ANOVA model, incomes intervals.
Table A4. T test, Lidl, Carrefour and DIA.
Show moreShow less
Abstract

The main aim of this work is to analyze Premium Private Labels (PPL) in the food sector. To do so, we have obtained data from both on-line and off-line channels, analyzing social networks, specifically Twitter, and from carrying out personal surveys of customers in the sales point to check their opinions about the perceived quality, brand value, and other PPL-related variables. We analyzed a total of 7 Twitter accounts of the main supermarket and hypermarket chains over a year and did 600 customer surveys in points of sale. Among the main results stands out the fact that the perceived quality of the PPL compared with the brand leader varies among the main distributors. There are also significant differences between the brand value of the 3 PP L analyzed. On the other hand, no relation was detected between the income levels and the brand equity of PPL buyers.

Keywords:
Premium Private Labels
social network sites
retailing
perceived value
JEL classification:
M31
M19.
Full Text
Introduction

The study of the strategies which the major retailers are carrying out in their establishments is necessary to understand the boom of Premium Private Labels (PPL) (Choi, 2016). For Ailawadi and Keller (2004), this new context of growing sales and greater acceptation of PPL is a consequence of a well-designed retailer strategy. The offer of a portfolio of own brands includes, generally, one that is generic, or economical, one that is standard and one that is Premium. These occupy different positions in the minds of the consumer (Geyskens, Gielens, & Gijsbrechts, 2010; Kumar & Steenkamp, 2007). The PPL is determinant for retailers, given that it is leading the consumer to see them as a separate group with respect to the rest (Nenycz-Thiel & Romaniuk, 2016).

In spite of the growing academic research on PPL, there is scant analysis of this phenomenon from the brand value perspective (Cuneo, Lopez, & Yagüe, 2012). To understand if the value of the distributor brands depends on the category into which they are grouped (Huang & Huddleston, 2009), it is fundamental to comprehend the Premium phenomenon within the distributor brand (Ohlson, 2019). The behavior of consumers has been investigated in the literature in many areas (Keller, 1993), but not from the perspective of PPL. This indicates the relevance of this research.

Some authors (Keller, Dekimpe, & Geyskens, 2016; Puelles, 1995) establish that the main retail leaders are developing PPL in innovative products with a better quality-price relationship than the manufacturers' brands. From this standpoint, PPL are in the upper extreme of the market and their unique characteristics in terms of flavor, origin and/or ingredients enable retailers to compete with higher quality national brands (Ter Braak, Geyskens, & Dekimpe, 2014). These products, according to Puelles (1995), are aimed at a public of different sociodemographic characteristics but who have a good purchasing power.

In its origins, one of the first authors to write about PPL together with Bazoche (Bazoche, Giraud-Héraud, & Soler, 2005) was Kumar (Kumar, 2007). The latter author stated that retail companies were looking for ways to compete against manufacturer brands inside their stores. Consequently, they began to search for solutions, so they introduced variations of the distributor brands in their shelves, but with a superior quality. In this sense, Dobson, & Chakraborty (Dobson & Chakraborty, 2009) established that the market share of PPL grew faster than the market share of PL. This fact was a crucial factor for the implementation by other retailers of this type of brand within their stores. PPLs can be defined as those products branded as private labels that are positioned at the top end, with respect to quality, within the retail chains. These products have unique characteristics in terms of flavour, origin and / or ingredients that allow retailers to compete with the highest quality manufacturer brands (Bazoche et al., 2005). In this sense, the positioning of the PPL is very different from the positioning of the standard PL and is thus affirmed by some authors (Ter Braak, et al. 2014). While the standard PL tends to imitate manufacturer brands (Geyskens, et al. 2010), selling at a 20-30% lower price (Steenkamp, ​​et al. 2010), PPLs are positioned at the upper end of the market, and offer quality similar or superior to the manufacturer brand (Geyskens, et al. 2010). Also, their prices are very similar to the price of manufacturers' brands, being sometimes higher (Lewin, Gomez-Arias, & Bello-Acebron, 2008). Notorious examples are Loblaw's President's Choice in the United States, and Tesco's Finest in the United Kingdom (Ter Braak, et al. 2014). Many relevant retailers sell PPL such as Carrefour (Selección Carrefour), Eroski (Eroski SELEQTIA), DIA (Delicious), or LIDL (Deluxe) in Spain.

In this article we analyze the sector of distributor brands in Spain, which, along with Switzerland, has led the market share of distributor brands and their Premium variant during recent years with shares over 40% of total sales (Arias Fernández, Pineda, & Florencio, 2015). The main aim of this work is to analyze the PPL in food retailers in the off-line and on-line channels, centering on the constructs of perceived quality, perceived value, brand value, awareness value and brand image. We will carry out a novel contribution with regard to the extant research, through incorporating the study of PPL in the social networks environment, which has become an accessible and multitudinous form of interacting, achieving and sharing information on brands and products (Barreto, 2014).

With the emergence of these means of communication, brands are faced with the challenge of finding new ways, tones and contents which capture the attention of the users who browse on social networks (Bigné & Hyder, 2012; Kuo & Yen, 2009). In this context, it is an imperative for any retailer to understand, assume and evolve in a similar way to how the virtual social networks and their tools develop (Lee, Park, & Han, 2008). Brands now comprehend not only how their commercial message must be, but also how to deal with more elusive and restrictive users who base a great part of their decisions on impulses generated through the networks (Curras, Ruiz, & Sanz, 2011).

The current article is structured as follows. First, we describe the theoretical antecedents of the study's topic and establish the research propositions. Then, we present the methodology applied and, finally, the analysis of the empirical data, setting forth, at the end, the conclusions and limitations for future research.

Theoretical review and formulating of propositions

Lamey, Deleersnyder, Steenkamp, and Dekimpe (2012)) point out that consumers are more receptive to distributor brands and PPL in times of economic difficulties. Therefore, the periods of monetary contraction are a great opportunity for retailers. The value proposals of distributor brands cover a broad spectrum, from Premium to low cost (Nenycz-Thiel & Romaniuk, 2016). The different positioning of distributor brands hence have a different value proposal to deal with the distinct profiles of consumers (Geyskens, Keller, Dekimpe, & de Jong, 2018; Kumar, 2019). To know the consumers who are more inclined to their purchase is important to have advantages in categories in which the distributor brands and the PPL have a greater power (Ailawadi et al., 2010).

Perceived quality and value of premium private labels

Few studies have tried to characterize distributor brand shoppers based on their sensitivity to the product's perceived quality (Ohno & Takemura, 2019). This concept was defined by Olshavsky (1985), who considers a product's perceived quality as a relatively global value judgment generated in the relations between customers and suppliers and in which both parts are active. Ailawadi, Neslin, and Gedenk (2001)) qualify the distributor brand customer as not being sensitive to the service's perceived quality. Other studies share this statement, such as Ailawadi and Keller (2004) and Ailawadi, Pauwels, and Steenkamp (2008)). There are also conflicting positions, such as Nenycz-Thiel and Romaniuk (2016) who consider the perceived quality of PPL to be perfectly comparable with manufacturers' brands. In this sense, many authors have measured this variable, establishing that this perceived quality can be affected by different qualities, such as the packaging (Immonen, 2010), the product itself (Kim, Lee, Lee, & Taylor, 2019) or the retailer's own image (Martinelli & Vergura, 2020).

Other authors, such as Geyskens et al. (2010) establish that the perceived quality of PPL and their brand value generally cannibalize other products within the same establishment, acting with the same pattern in different retail sales points. This means that if the profile of customers of different chains varies from one to another, their brand value perceptions will also be different. From the previous literature we propose the following:P1a

The perceived quality of PPL with respect to the brand leader varies among the main distributors.

P1b

There are significant differences between the brand values for distinct PPL.

Many distributors currently have sophisticated distributor brand strategies which range from options which compete in prices (generic brands) to others that do so due to being the best products in the market (Premium brands), aiming them at another market segment (Reimann & Wagner, 2016). The distributors therefore handle a portfolio of distributor brands with the aim of reaching different segments whose main differentiation lies in their purchasing power (Geyskens et al., 2010; Kumar & Steenkamp, 2007).

Some authors establish that PPL cause a positive impact on customer loyalty through a causal relation with the perceived value (Martinelli & Vergura, 2014). There are various examples in the literature of the impact of incorporating these Premium brands into the range and providing the distributor with the possibility of increasing its margins, enhancing its own range with exclusive products (Ailawadi & Keller, 2004; Garretson, Fisher, & Burton, 2002; Pauwels & Srinivasan, 2004; Sprott & Shimp, 2004). These exclusive products can be identified by a high income customer segment, as the prices are similar to those of the manufacturer's brand (Dunne & Narasimhan, 1999; Pınar & Girard, 2020).

IRI (2014) establishes that the distributors are moving ever more further away from large discounts and promotions, increasing prices to satisfy the demand of the consumers who want better quality Premium products. In this sense, the distributors are taking into account that there is a segment of their clientele that seeks superior quality above price and therefore, the different distributors are adapting to this reality (Geyskens et al., 2010; Loebnitz, Zielke, & Grunert, 2020).

In the literature, the brand equity of the distributor brand has been dealt with as a determinant factor when purchased by different segments within the retailers' portfolios (Ailawadi et al., 2001). Defining it as a concern about paying low prices, subject to some restriction of quality (Lichtenstein, Netemeyer, & Burton, 1990). The customers' brand equity of their brands is a characteristic which retailers must know to determine what strategy to follow in their establishments regarding these brands (Pillai & Kumar, 2012; Ruiz-Real, Gázquez-Abad, Esteban-Millat, & Martínez-López, 2018). Retailers must continue working to achieve consumer learning with tools such as PPL in their establishments (Beldona & Wysong, 2007), or the delivery of samples of products to improve the valuing of their brands.

Baltas, Paraskevas, and Argouslidis (2007)) show that consumers' preferences for distributor brands, including their Premium version, stem from a broad process of evaluation in which the quality, time, gender, age, etc., along with the awareness value have a main role. In this way, the manufacturers do not assume that the distributor brands are only cheap alternatives (Olsen & Sallis, 2010), bought by those who cannot pay for the national brands, as with the emergence of the PPL the role given to them by the manufacturers has changed (Beneke, 2013). It is important to highlight the extraordinary change in both quality and perceptions of standard and premium private labels currently (Bronnenberg, Dubé, & Sanders, 2020; Abril & Rodriguez-Cánovas, 2016). Now, they are seen as attractive alternatives to products of national brands. According to this viewpoint, this work will try to note if the public who buys PPL corresponds to a middle/high socioeconomic segment. From the aforesaid, we propose the following: P2a

There is a positive relation between the perceived value of PPL and the income level.

P2b

There is no relation between the income level and the perceived value of the buyers of PPL.

Image and frequency of visits

The distributor brand strategies developed throughout recent years have meant the coinciding of the brand denomination with that of the chain brand-name (Nenycz-Thiel & Romaniuk, 2012; Tandon & Verma, 2020). Or using it as a support of an own brand to obtain credibility and a guarantee for this brand. This shows that the interrelation between the chain-label and its distributor brands has been taken advantage of to improve the value of the latter, both in their standard and Premium facets (Beristain, 2009).

The image of the retail chain is a multidimensional aggregate of associations linked to the label in the consumer's mind. Not in vain are those brand-chain-linked associations nodes of information bound to that of the brand, which can host not only perceptions about its attributes, but also the fruits which derive from them (Martinelli, De Canio, & Marchi, 2019). That is to say, the benefits and attitudes which the consumer associates with the brand (Nabec, 2002). From this, we deduce that the brand's image will be influenced to a greater or lesser extent by its corporate conduct. This behavior is manifested in the firm's social (Jones, Comfort, & Hillier, 2007) and strategic area (Fombrun & Shanley, 1990; Higgins & Bannister, 1992).

On the other hand, it is to be highlighted that the management of the brand identity has become more sophisticated with the passing of time (Floor, 2006). Hence we will need to analyze this statement based on the quality of the establishment to verify if what other researchers propose about the matter is also the case for PPL (Beristain & Zorrilla, 2011; Schnittka, 2015). On this basis, we formulate the following:P3

The image of the establishment influences the PPL's perceived quality.

Social networks and premium private labels

According to the aforementioned, to ponder the perceived quality of retail establishments' PPL and compare them with the corporate image of chains can turn out to be interesting (Blanco, García, & Martín, 2011). Social networks play increasingly more an important role in the development of the strategies of the large distributor firms, being the means by which customers comment freely about the aspects which each brand is projecting at the consumer. Therefore, social networks should be considered a fundamental communication tool within organizations (Sokolova & Kefi, 2020).

Social networks are considered a suitable context for learning (Dabbagh & Kitsantas, 2012), being a key to inquire into the process by which people learn in these new environments and what their motivations are to do so. The use of Internet as a factor of digital inclusion has been explored from different perspectives in the scientific literature. The motivations of use being the most common (Lin & Lu, 2011; Seidman, 2013; Tosun, 2012; Yang & Brown, 2013). More specifically, there are investigations centered on the motivations of use of Facebook and Twitter (Arteaga, Cortijo, & Javed, 2014; Dogruer, Menevis, & Eyyam, 2011; Seidman, 2013; Tosun, 2012; Yang & Brown, 2013).

Due to all this, social networks can be understood as spaces of information exchange and the generation of relations which are becoming more relevant (Hsiao, Wang, Wang, & Kao, 2019). As they make possible the interaction of people interested in common topics who can share or exchange information, knowledge, experiences, interests and/or needs. The potential benefits which stem from the use of these networks are not only for the users who participate in them, but also for the firms (Brown, Mawson, & Rowe, 2019). These organizations, through the comments made on these virtual platforms, can know the tastes, desires and needs of the people who are on them, their uses, consumption behaviors and background, as well as their levels of satisfaction and dissatisfaction (Neelotpaul, 2010). This is why we analyzed the feelings of the retail brand's users in the social networks to know if its brand strategy is perceived in the same way by the consumers and establish if the Premium products have a place in their establishments.

Social networks have become indispensable for consumers (Ferreira, Santos, & Malta, 2020). Thanks to them they are now not limited to getting opinions and recommendations only through the people who they know. They also are in touch with comments made by unknown people about the product or brand (Barreto, 2014). An essential point for firms to take into account when they participate in social networks, and which many articles underscore, is the handling by firms of the e-WOM about their brands (Alvandi, Fazli, & Amiri, 2014). The success of e-WOM is determined by the satisfying of the consumers' social and functional needs, as they seek both useful information about the product (functional needs) and relations with other members through these platforms (social needs) (Chih, Wang, Hsu, & Huang, 2013). Either as on-line or off-line shoppers, people more and more turn to Internet and social networks as a source of information to advise them on their purchase (Rondán-Cataluña, Arenas-Gaitán, & Ramírez-Correa, 2015), especially concerning products and services about which they have little information.

Currently, Internet users have access to virtual information in their mobile phone, being able to consult personal comments about any product through recommendations shared or conversations held with other users, before, during and after the shopping process (Court, Gordon, & Perrey, 2012). This is why brands are banking on channel 2.0 promotions.

The virtual contents generated by users are an important source for the analysis of feelings (Phan et al. 2020). Those most used in the research on this discipline have been reviews of products and services, microblogging (especially Twitter), publications in blogs and news in webs (Boyd & Ellison, 2008; Medhat, Hassan, & Korashy, 2014; Tsytsarau & Palpanas, 2012; Vinodhini & Chandrasekaran, 2012). Various researchers state in their investigations that Twitter is appropriate for the analysis of feelings and that it is possible to link feelings and attitudes (Pak & Paroubek, 2010; Thelwall, Buckley, & Paltoglou, 2011; Jiang, Yu, Zhou, Liu, & Zhao, 2011). Therefore, the content generated by the users of social media based on their feelings is a source of notable influence in the shopping decisions of other users (Cantallops & Salvi, 2014; Trusov, Bucklin, & Pauwels, 2009). Therefore, electronic word of mouth affects corporate brand image and consequently it would also modify the perceived image of all private labels including PPL. This relationship would be more intense when the name of retailer brand is included in the name of the PPL.

In this sense, consumers increasingly turn to social network sites to obtain information to take decisions (Kozinets, 2002). Therefore, social networks have transformed the shopping experience of consumers, influencing purchasing intentions (Pookulangara & Koesler, 2011). In this sense, the E-WOM is the fundamental factor on which researchers base to verify the influence that social networks exert on purchasing decisions (Bataineh, 2015; López & Sicilia, 2014; Yoon, 2012). Since, according to the Signaling Theory, the buyer and the seller have access to different information on many occasions. Hence, the buyer tries to interpret the communication that the seller wants to provide complementing it with social networks where the user's previous experience is essential (Connelly, Certo, Ireland, & Reutzel, 2011; Mumi, Obal, & Yang, 2019). Finally, all of the above is supported by the Herding Theory, according to this theory users seek to follow the group's behavior and therefore they get through social networks (Baddeley, Pillas, Christopoulos, Schultz, & Tobler, 2007).

Therefore, social networks are of great importance both for the users and for the retailer firms which commercialize distributor brands (Borgaonkar, 2020). Although they have different aims, they comprise a common initiative, which is to maintain existing social links or form new connections between users and retailers (Boyd & Ellison, 2008; Ellison, Steinfield, & Lampe, 2007). From all of this, we conclude that brands frequently use social networks as one more instrument of their communication strategy for both manufacturers' brands and distributor brands. And, thereafter, foster their variant of PPL, increase the trust that the consumers have in them and influence their shopping intentions (Hajli, 2014). Based on this discussion, we make the following proposition:P4

The brand image of retail establishments in social networks influences the perceived image of PPL in these networks.

Loyalty and value of PPL

There have been various investigations developed about PPL which are based on the study of factors such as price, perceived quality, extent of the range, etc. (Partal & Benito, 2014; Geyskens 2010; Kethüda, 2020). Nevertheless, there is a gap with respect to the average visits which take place in the retail establishments by consumers who purchase PPL.

Average visits are defined as the frequency with which a consumer goes to the establishment with the intention of buying a specific product (Martínez-Ruiz, Jiménez-Zarco, & Izquierdo-Yusta, 2010). Given that the buying of both Premium and standard distributor brands generally entails higher levels of functional risk than the those of manufacturers' brands (Richardson, Jain, & Dick, 1996; Zielke & Dobbelstein, 2007), consumers tend to prefer the latter for the categories of products with a high social or functional risk. Although with each purchase and visit to the shop they become more familiar with the distributor brand's labels (Fischer, Völckner, & Sattler, 2010; Semeijn, Van Riel, & Ambrosini, 2004). The aforementioned is our basis for formulating the following proposition: P5

The average number of visits to the establishment influences the value of its PPL

MethodologyMeasuring image in Twitter

The ways in which people speak and write facilitate the comprehension of their emotional and cognitive facet. During the last three decades, researchers have furnished proofs which indicate that emotional states toward products or brands can be deduced by, for example, the words that they use (Stiles, 1992).

What has been commented upon before can be linked to 2.0 technology and to its dynamic area, social networks. According to Rissoan (2013), social networks are a means of communication which favor social interactions and in which technology and the creation of content intervene. The latter implies that the potential consumer becomes the author of the information, publicly or privately, and without any geographical limit. The new information and communication technologies enable the creating and publishing of this content easily, quickly and in significant volumes, making social interaction possible as a result of the reactions of each human being (Rissoan, 2013).

It is important to know what consumers say about this technology and how it influences them. With this aim, we will use a method for the study of the different emotional and cognitive states. Tools such as Sentiment140 and LIWC develop a text analysis application. Sentiment140's algorithms are designed to analyze the written text on a word basis, to calculate the percentages of words in the text which coincide with the dimensions of the language. These algorithms generate the outcome as a result of identifying the negative, neutral or positive feelings which users of the social platform Twitter have about a specific product or a brand. The Sentiment140 method is a web application which classifies the tweets according to their polarity. The evaluation is carried out using the remote monitoring approach proposed by Go, Bhayani, and Huang (2009)). This is based on monitored learning procedures and is tackled by using positive, negative and neutral feelings.

The messages of Twitter have a maximum length of 140 characters and thanks to Twitter's API millions of tweets can be collected to know users' feelings. The approach of Sentiment140 enables the use of different classifiers of automatic learning and characteristic extractors. The classifiers of automatic learning are Naive Bayes, maximum entropy (MaxEnt), and Support Vector Machines (SVM). Not only do they use emotions, as if we analyze them there is a negative impact on the precision of the MAXENT and SVM classifiers. The difference lies in the mathematical models and the selection of weight of the MaxEnt and SVM function. Therefore, and excluding emoticons, the classifier can learn from other characteristics (for example, Unigrams and Bigramas) present in the tweet.

With the help of the tool described, we proceeded to compile a database of the feelings which Twitter users transmit about the perceived image, both of the PPL and the global image of the establishment from 01/01/2015 till 31/12/2015, regarding some of the main retailers of the food sector in Spain (Nielsen, 2015) according to sales volume, such as Mercadona, Lidl, Carrefour and DIA.

Consistent with the retailer's concern for endowing their own brands with a better quality (Hoch & Banerji, 1993), the Premium Private Labels of the distributors selected were incorporated into the market segmentation of the distributor brands (Dunne & Narasimhan, 1999). Thus, the three distributor brands analyzed are Delicious (DIA), Deluxe (Lidl) and Carrefour Selección (Carrefour). These criteria were established based on the sales volume of the main brand or brands of the distributor and the sales volume of its main PPL (Nielsen 2015).

We hence compiled all the tweets of the monthly consumers of these retailers to check their feelings when buying products of the PPL and of the standard distributor brand. Along with this, and due to everything that has been described, we established a follow- up of the number of followers during the year 2015. Once we knew the place of these brands in the consumer's minds and what they comment about them on social networks, we undertook the adding of information through the compiling of data via personal surveys.

A questionnaire was worked out (see Table A1) with a view to analyzing the propositions formulated. The different questions were adapted from scales validated by different authors (Yoo & Donthu, 2001; Delgado-Ballester, Hernandez-Espallardo, & Rodriguez-Orejuela, 2014). The scales corresponding to the image of the establishment, functional risk, financial risk, social risk, psychological risk and awareness value were validated by Delgado-Ballester et al. (2014) and those corresponding to the brand value by Yoo and Donthu (2001). A pre-test of 10 surveys was carried out for each of the 3 retail distributors (DIA, Carrefour, Lidl) selecting users of different ages and genders to test the questionnaire.

Table A1.

Description of the ANOVA model, perceived quality.

NMeanTypical deviationTypical error95% of the confidence interval for the mean
Lower Limit  Upper Limit 
25CARREFOUR  200  3.010  .9078  .0642  2.883  3.137 
DIA  200  2.990  .9563  .0676  2.857  3.123 
LIDL  200  2.725  1.0559  .0747  2.578  2.872 
Total  600  2.908  .9823  .0401  2.830  2.987 
26CARREFOUR  200  2.765  .8266  .0584  2.650  2.880 
DIA  200  2.780  .8976  .0635  2.655  2.905 
LIDL  200  2.570  .9746  .0689  2.434  2.706 
Total  600  2.705  .9052  .0370  2.632  2.778 
27CARREFOUR  200  2.845  .8393  .0594  2.728  2.962 
DIA  200  2.815  .8744  .0618  2.693  2.937 
LIDL  200  2.580  .9687  .0685  2.445  2.715 
Total  600  2.747  .9021  .0368  2.674  2.819 
28CARREFOUR  200  2.555  .7549  .0534  2.450  2.660 
DIA  200  2.505  .7894  .0558  2.395  2.615 
LIDL  200  2.245  .8768  .0620  2.123  2.367 
Total  600  2.435  .8187  .0334  2.369  2.501 
29CARREFOUR  200  2.505  .7159  .0506  2.405  2.605 
DIA  200  2.420  .7459  .0527  2.316  2.524 
LIDL  200  2.185  .8628  .0610  2.065  2.305 
Total  600  2.370  .7879  .0322  2.307  2.433 
30CARREFOUR  200  2.605  .7759  .0549  2.497  2.713 
DIA  200  2.725  .9401  .0665  2.594  2.856 
LIDL  200  2.405  .9028  .0638  2.279  2.531 
Total  600  2.578  .8842  .0361  2.507  2.649 
31CARREFOUR  200  2.450  1.1242  .0795  2.293  2.607 
DIA  200  2.350  1.0925  .0772  2.198  2.502 
LIDL  200  2.260  1.1528  .0815  2.099  2.421 
Total  600  2.353  1.1242  .0459  2.263  2.443 
32CARREFOUR  200  2.120  .8829  .0624  1.997  2.243 
DIA  200  1.970  .7888  .0558  1.860  2.080 
LIDL  200  1.910  .8865  .0627  1.786  2.034 
Total  600  2.000  .8571  .0350  1.931  2.069 
33CARREFOUR  200  2.220  .9466  .0669  2.088  2.352 
DIA  200  1.980  .7955  .0562  1.869  2.091 
LIDL  200  2.035  .9425  .0666  1.904  2.166 
Total  600  2.078  .9020  .0368  2.006  2.151 
34CARREFOUR  200  1.895  .7791  .0551  1.786  2.004 
DIA  200  1.800  .7229  .0511  1.699  1.901 
LIDL  200  1.675  .7363  .0521  1.572  1.778 
Total  600  1.790  .7507  .0306  1.730  1.850 
35CARREFOUR  200  2.030  .8849  .0626  1.907  2.153 
DIA  200  1.870  .7655  .0541  1.763  1.977 
LIDL  200  1.760  .8402  .0594  1.643  1.877 
Total  600  1.887  .8377  .0342  1.820  1.954 
36CARREFOUR  200  2.425  .9997  .0707  2.286  2.564 
DIA  200  2.425  1.0343  .0731  2.281  2.569 
LIDL  200  2.155  1.0375  .0734  2.010  2.300 
Total  600  2.335  1.0302  .0421  2.252  2.418 

A non-random sampling method was done, the population studied being customers over 18 years old of the nine establishments analyzed: three of DIA, three of Carrefour and three of Lidl, all in the south of Spain. The interviewers were trained in different sessions in the technique of administering the questionnaire to the individual respondents. The interviewing took place in July 2016. This resulted in a total of 614 questionnaires, 14 of which had to be removed due to being incorrectly filled out. The division of the valid questionnaires between establishments was equitable; that is to say, 200 respondents customers of DIA, 200 customers of Lidl and 200 of Carrefour.

The sample we worked with was therefore made up of 600 people of which 61.2% were women and 38.8 % men. The ages were distributed as follows: 20% between 18 and 35 years old, 33.5% between 36 and 45, 26% between 46 and 55, 14.8% between 56 and 65, and 5.7% over 65. With respect to their marital status, 51.8% were married, 30.3 % single, 11.5% were divorced and 6.3% were widows or widowers.

We thus had a study based on personal surveys carried out directly with the end customer in the sales point. From this, we could establish comparisons with what consumers voiced in Twitter, as this is a social network where firms officially project their brand image and in which consumers have primary information about them.

Different statistical techniques have been applied in this study with the aid of version 23 of the IBMSPSS software, such as temporal series analysis, ANOVA, the DMS, Bonferroni and Tamhane tests, the Kolmogorov-Smirnov statistic, the Mann-Whitney procedure to verify if the two independent samples have been extracted from populations with the same variance and, finally, the T test.

Results

First we analyzed the reliability of the scales used in each of the constructs as well as their dimensions. According to George and Mallery (1995), reliability is related to the fact of the measuring instrument producing the same results each time it is run with the same person in the same circumstances. The instruments used in social sciences are considered reliable if, irrespective of who runs them and the way in which this is done, similar results are obtained. In this research we used Cronbach's alpha, an indicator widely used for this analysis.

According to George and Mallery (1995), a Cronbach's alpha above 0.7 can be considered an acceptable level. In our study we have obtained for the brand equity scale a Cronbach's alpha of 0.847, for the brand value 0.94, for the functional risk 0.942, for the financial risk 0.987, for the social risk 0.917, for the psychological risk 0.842 and, lastly, for the perceived quality scale 0.909. It can therefore be concluded that all the scales used are reliable.

In the first analysis we took as a source the data obtained from the surveys. To check Proposition 1a (the perceived quality of PPL with respect to the brand leader varies among the main distributors), we applied an ANOVA analysis (see Table A1) using as dependent variables the questions of the perceived quality scale about the manufacturer's brand and the three PPL analyzed (Deluxe, Carrefour Selección and Delicious).

According to the ANOVA, there are significant differences in all the variables except in number 31. That is to say, buying the PPL instead of the brand leader can negatively affect what others think of me. This fits in social risk and, within that, in the perceived quality with respect to the 3 chains. These results are also confirmed with the non-parametric Kruskal-Wallis test for independent samples. On the other hand, according to the post hoc tests of Bonferroni and Tamhane, it is noted that the respondents who consume the PPL of Lidl (Deluxe) offer significantly lower average values for these variables than for Carrefour Selección and Delicious, except for variable 33 (buying the PPL instead of the brand leader can make others see me in a way I don't want) where Delicious shows the lowest results.

Deluxe is better valued than the Premium of Carrefour and DIA. Due to the values of the variables of the scale being lower and that the scale is reversed, we conclude that the perceived quality with respect to the brand leader of Deluxe is above Carrefour Selección and Delicious. There is therefore not enough evidence to reject Proposition P1a.

Taking into account the contributions of Aaker (1996) and of Keller (1993) about brand equity, we determine that to verify Proposition P1b (there are significant differences between the brand values for different PPL) an ANOVA is also applied (see Table A2). The result is that there are significant differences between the brand values for the 3 PP L.

Table A2.

Description of the ANOVA model, brand value.

NMeanTypical deviationTypical error95% of the confidence interval for the mean
Lower Limit  Upper Limit 
40CARREFOUR  200  2.535  0.9181  0.0649  2.407  2.663 
DIA  200  2.88  0.8942  0.0632  2.755  3.005 
LIDL  200  2.92  1.0484  0.0741  2.774  3.066 
Total  600  2.778  0.9699  0.0396  2.701  2.856 
41CARREFOUR  200  2.575  0.9481  0.067  2.443  2.707 
DIA  200  2.885  0.875  0.0619  2.763  3.007 
LIDL  200  2.94  1.0781  0.0762  2.79  3.09 
Total  600  2.8  0.9823  0.0401  2.721  2.879 
42CARREFOUR  200  2.52  0.9076  0.0642  2.393  2.647 
DIA  200  2.805  0.8429  0.0596  2.687  2.923 
LIDL  200  2.85  1.0502  0.0743  2.704  2.996 
Total  600  2.725  0.9474  0.0387  2.649  2.801 
43CARREFOUR  200  3.395  0.7082  0.0501  3.296  3.494 
DIA  200  3.45  0.7418  0.0525  3.347  3.553 
LIDL  200  3.7  0.8447  0.0597  3.582  3.818 
Total  600  3.515  0.7772  0.0317  3.453  3.577 
44CARREFOUR  200  3.605  0.6332  0.0448  3.517  3.693 
DIA  200  3.66  0.6758  0.0478  3.566  3.754 
LIDL  200  3.785  0.7886  0.0558  3.675  3.895 
Total  600  3.683  0.7051  0.0288  3.627  3.74 
45CARREFOUR  200  3.315  0.818  0.0578  3.201  3.429 
DIA  200  3.32  0.7619  0.0539  3.214  3.426 
LIDL  200  3.32  0.912  0.0645  3.193  3.447 
Total  600  3.318  0.8316  0.0339  3.252  3.385 
46CARREFOUR  200  3.86  0.65  0.046  3.769  3.951 
DIA  200  3.925  0.4476  0.0317  3.863  3.987 
LIDL  200  4.01  0.6097  0.0431  3.925  4.095 
Total  600  3.932  0.5781  0.0236  3.885  3.978 
47CARREFOUR  200  3.05  0.8129  0.0575  2.937  3.163 
DIA  200  3.175  0.8171  0.0578  3.061  3.289 
LIDL  200  3.24  0.9091  0.0643  3.113  3.367 
Total  600  3.155  0.8498  0.0347  3.087  3.223 
48CARREFOUR  200  3.39  0.9657  0.0683  3.255  3.525 
DIA  200  3.445  0.96  0.0679  3.311  3.579 
LIDL  200  3.54  0.9711  0.0687  3.405  3.675 
Total  600  3.458  0.966  0.0394  3.381  3.536 
49CARREFOUR  200  2.495  1.0174  0.0719  2.353  2.637 
DIA  200  2.5  1.0514  0.0743  2.353  2.647 
LIDL  200  2.465  1.0509  0.0743  2.318  2.612 
Total  600  2.487  1.0384  0.0424  2.403  2.57 
50CARREFOUR  200  2.68  0.775  0.0548  2.572  2.788 
DIA  200  2.94  0.8122  0.0574  2.827  3.053 
LIDL  200  2.975  0.9266  0.0655  2.846  3.104 
Total  600  2.865  0.8493  0.0347  2.797  2.933 
51CARREFOUR  200  2.7  0.8082  0.0572  2.587  2.813 
DIA  200  2.94  0.7997  0.0566  2.828  3.052 
LIDL  200  2.96  0.9014  0.0637  2.834  3.086 
Total  600  2.867  0.8446  0.0345  2.799  2.934 
52CARREFOUR  200  2.675  0.7826  0.0553  2.566  2.784 
DIA  200  2.93  0.8114  0.0574  2.817  3.043 
LIDL  200  2.93  0.8995  0.0636  2.805  3.055 
Total  600  2.845  0.8399  0.0343  2.778  2.912 

With the one-way ANOVA, the differences of the average values of the variables of the scale are significantly different for all the brand value scale items except 45 (I can recognize the PPL from among the competitors), 47 (some characteristics of the PPL come quickly to mind), 48 (I can quickly remember the symbol or logo of the PPL) and 49 (I have trouble imagining the PPL). The previous results are corroborated by the non-parametric Mann-Whitney test of independent samples.

According to the results of the DMS, Bonferroni and Tamhane tests, in items 40 to 52 (which correspond to brand value), there are significant differences between Carrefour Selección and the other two brands, that is to say, Deluxe (Lidl) and Delicious (DIA). This takes place except in the variables 43 (the quality of the PLP is extremely high), 44 (the likelihood of the PPL being useful is very high), 46 (I know the PPL), 47 (some characteristics of the PPL come quickly to mind), and in 45 and 49 there are not significant differences, Carrefour's PPL generally having lower scores for the latent variable of the brand value.

Regarding the perceived value (P2a “There is a positive relation between the perceived value of PPL and income levels”), linear and non-linear relations between the variables were sought. As the adjusted R2 is 0.006 P2a is rejected; that is to say, the relation between income levels and the perceived value of the PPL does not exist. Various non-linear adjustments were carried out (logarithmic, reverse, quadratic, cubic, composite, power, growth, exponential and logistical) and the results were confirmed in all of them.

As there was not a validation with the regression, we proceeded to carry out the ANOVA, dividing the incomes into intervals (see Table A3). To do so, we codified the income variables into 4 groups: less than 1,000 euros, between 1,001euros and 2,000 euros, from 2,001euros to 3,000 euros and over 3000 euros. With this variable there are partially significant results.

Table A3.

Description of the ANOVA model, incomes intervals.

NMeanTypical deviationTypical error95% of the confidence interval for the mean
Lower Limit  Upper Limit 
401,00  156  2.673  .8884  .0711  2.533  2.814 
2,00  181  2.663  .9558  .0710  2.523  2.803 
3,00  178  2.927  1.0030  .0752  2.779  3.075 
4,00  85  2.906  1.0308  .1118  2.684  3.128 
Total  600  2.778  .9699  .0396  2.701  2.856 
411,00  156  2.737  .9306  .0745  2.590  2.884 
2,00  181  2.718  .9793  .0728  2.575  2.862 
3,00  178  2.888  1.0244  .0768  2.736  3.039 
4,00  85  2.906  .9835  .1067  2.694  3.118 
Total  600  2.800  .9823  .0401  2.721  2.879 
421,00  156  2.564  .8284  .0663  2.433  2.695 
2,00  181  2.613  .9455  .0703  2.475  2.752 
3,00  178  2.876  .9894  .0742  2.730  3.023 
4,00  85  2.941  .9923  .1076  2.727  3.155 
Total  600  2.725  .9474  .0387  2.649  2.801 
431,00  156  3.519  .6765  .0542  3.412  3.626 
2,00  181  3.486  .7499  .0557  3.376  3.596 
3,00  178  3.573  .8355  .0626  3.449  3.697 
4,00  85  3.447  .8797  .0954  3.257  3.637 
Total  600  3.515  .7772  .0317  3.453  3.577 
441,00  156  3.686  .5881  .0471  3.593  3.779 
2,00  181  3.663  .6768  .0503  3.564  3.762 
3,00  178  3.725  .7721  .0579  3.611  3.839 
4,00  85  3.635  .8143  .0883  3.460  3.811 
Total  600  3.683  .7051  .0288  3.627  3.740 
451,00  156  3.417  .7180  .0575  3.303  3.530 
2,00  181  3.298  .8363  .0622  3.176  3.421 
3,00  178  3.343  .8704  .0652  3.214  3.471 
4,00  85  3.129  .9101  .0987  2.933  3.326 
Total  600  3.318  .8316  .0339  3.252  3.385 
461,00  156  3.859  .5730  .0459  3.768  3.950 
2,00  181  3.950  .5703  .0424  3.867  4.034 
3,00  178  3.994  .5261  .0394  3.917  4.072 
4,00  85  3.894  .6905  .0749  3.745  4.043 
Total  600  3.932  .5781  .0236  3.885  3.978 
471,00  156  3.128  .7929  .0635  3.003  3.254 
2,00  181  3.061  .8378  .0623  2.938  3.184 
3,00  178  3.270  .8735  .0655  3.140  3.399 
4,00  85  3.165  .9109  .0988  2.968  3.361 
Total  600  3.155  .8498  .0347  3.087  3.223 
481,00  156  3.186  .9073  .0726  3.042  3.329 
2,00  181  3.420  .9431  .0701  3.282  3.558 
3,00  178  3.708  .9292  .0696  3.570  3.845 
4,00  85  3.518  1.0647  .1155  3.288  3.747 
Total  600  3.458  .9660  .0394  3.381  3.536 
491,00  156  2.788  .9504  .0761  2.638  2.939 
2,00  181  2.530  1.0030  .0746  2.383  2.677 
3,00  178  2.225  1.0168  .0762  2.074  2.375 
4,00  85  2.388  1.1659  .1265  2.137  2.640 
Total  600  2.487  1.0384  .0424  2.403  2.570 
501,00  156  2.801  .7222  .0578  2.687  2.916 
2,00  181  2.851  .8332  .0619  2.729  2.973 
3,00  178  2.949  .9583  .0718  2.808  3.091 
4,00  85  2.835  .8570  .0930  2.650  3.020 
Total  600  2.865  .8493  .0347  2.797  2.933 
511,00  156  2.808  .6920  .0554  2.698  2.917 
2,00  181  2.845  .8289  .0616  2.724  2.967 
3,00  178  2.949  .9583  .0718  2.808  3.091 
4,00  85  2.847  .8797  .0954  2.657  3.037 
Total  600  2.867  .8446  .0345  2.799  2.934 
521,00  156  2.756  .6944  .0556  2.647  2.866 
2,00  181  2.812  .8152  .0606  2.693  2.932 
3,00  178  2.944  .9550  .0716  2.803  3.085 
4,00  85  2.871  .8700  .0944  2.683  3.058 
Total  600  2.845  .8399  .0343  2.778  2.912 

From the one-way ANOVA results, there are signs that the users who earn more value the PPL more than those who earn less for some of the scale's variables, specifically for variables 40 (I consider I'm loyal to the PPL), 42 (I'm not going to buy other brands if the PPL is available in the shop), 48 (I can quickly remember the symbol or logo of the PPL) and 49 (I have trouble imagining the PPL), but they are not determinant as the same does not take place for the rest of the items of the scale.

Subsequently, we are going to study the degree of income levels and the brand equity of the buyers of the PPL to determine if Proposition P2b (There is no relation between the income level and the awareness value of the buyers of PPL) is accepted, again through linear regression analysis. As the R2 is 0.007, P2b is rejected. Therefore, the linear relation between the income level and the perceived value of the PPL does not exist. As before, we carried out other non-linear adjustments, such as logarithmic, reverse, quadratic, cubic, composite, power, growth, exponential and logistical, and the same results were confirmed in all of them.

Recent works have studied perceived quality in different areas. However, very few works have been published relating this with PPL (Grunert & Aachmann, 2016). In Table 1 we investigate if the chain's image positively influences its PPL's perceived quality in order to verify P3. Linear regression analysis is again done and is shown in the following table.

Table 1.

Influence of the establishment's image on perceived quality

    Non-standardized coefficientsStandardized coefficients  Sig.  Collinearity statistics
Model    Standard error  Beta      Tolerance  VIF 
1(Constant)  4.333  .160    27.034  .000     
IMAGE  -.546  .044  -.455  -12.511  .000  1.000  1.000 

The adjusted linear correlation coefficient is 0.207, a low value. Furthermore, the image variable has a significant effect on the perceived quality, but the variable's coefficient in the regression equation is negative: -0.455. Hence there is a negative relation with a low correlation. Due to this, we determine that Proposition 3 is rejected.

Given the importance of social networks, as has been seen in the literature review, in Tables 2, 3 and 4 we study the retail establishments' brand image in social networks, checking if it influences these networks' perceived image of the PPL (Proposition P4). This is done for the three brands. SPSSIBM's expert modeler, has been applied. This aims to identify and automatically estimate the ARIMA model or the exponential smoothing model which fits best for one or more of the destination variables, thus removing the necessity of identifying an appropriate model via the trial and error system.

Table 2.

Description of the model, social network feelings, Lidl.

Model type 
ID of modelPOSITIVEFEELINGSLIDL  Model_1  ARIMA(0,0,0) 
NEGATIVEFEELINGSLIDL  Model_2  ARIMA(0,0,0) 
Table 3.

Fit of the model, social network feelings, LIDL.

Fit statisticMeanS.E.Min.Max.Percentile
10  25  50  75  90  95 
Stationary R-squared  .707  .355  .456  .958  .456  .456  .456  .707  .958  .958  .958 
R-squared  .707  .355  .456  .958  .456  .456  .456  .707  .958  .958  .958 
RMSE  .050  .021  .034  .065  .034  .034  .034  .050  .065  .065  .065 
MAPE  21.391  16.673  9.602  33.180  9.602  9.602  9.602  21.391  33.180  33.180  33.180 
MaxAPE  93.567  86.891  32.125  155.008  32.125  32.125  32.125  93.567  155.008  155.008  155.008 
MAE  .033  .019  .020  .047  .020  .020  .020  .033  .047  .047  .047 
MaxAE  .089  .048  .055  .124  .055  .055  .055  .089  .124  .124  .124 
Normalized BIC  -5.465  .568  -5.866  -5.063  -5.866  -5.866  -5.866  -5.465  -5.063  -5.063  -5.063 
Table 4.

Parameters of the ARIMA model, social network, Lidl.

Estimation  SE  Sig. 
POSITIVEFEELINGSLIDL-Model_1  POSITIVEFEELINGSLIDL  Without transformation  Constant.508  .019  26.069  .000 
NEGATIVEFEELINGSLIDL-Model_2NEGATIVEFEELINGSLIDL  Without transformation  Constant.357  .075  4.788  .002 
POSITIVEFEELINGSDELUXELidlWithout transformationDelay     
Numerator  Delay 0  -.449  .134  -3.355  ,012 
NEGATIVEFEELINGSDELUXELidlWithout transformationNumerator  Delay 0  -1.070  .261  -4.091  ,005 
Difference     

Firstly, we analyze the influence of Lidl's brand image on its PPL, again from the data collected in Twitter. It is determined in this study that the greater the positive feeling of the Deluxe brand, the lower the negative feeling about Lidl. Also, the greater the negative feeling of Deluxe, the less negative feelings about Lidl. This was established by the predictor of Model 2 (negative feelings), which we have ascertained with an ARIMA model.

In Tables 5, 6 and 7 we analyze these same factors of Tables 2,3 and 4 for Carrefour and Carrefour Selección.

Table 5.

Description of the model, social network feelings, Carrefour

Model type 
ID of modelPOSITIVEFEELINGSCARREFOUR  Model_1  Winters additive 
NEGATIVEFEELINGSARREFOUR  Model_2  ARIMA(0,0,0) 
Table 6.

Fit of the model, social network feelings, Carrefour.

Fit statisticMeanSEMin.Max.Percentile
10  25  50  75  90  95 
Stationary R-squared  .684  .684  .684  .684  .684  .684  .684  .684  .684  .684 
R-squared  .842  .223  .684  1.000  .684  .684  .684  .842  1.000  1.000  1.000 
RMSE  .036  .051  7.716E-15  .072  7.716E-15  7.716E-15  7.716E-15  .036  .072  .072  .072 
MAPE  19.015  26.891  5.321E-13  38.030  5.321E-13  5.321E-13  5.321E-13  19.015  38.030  38.030  38.030 
MaxAPE  52.590  74.373  6.366E-12  105.180  6.366E-12  6.366E-12  6.366E-12  52.590  105.180  105.180  105.180 
MAE  .025  .035  1.938E-15  .049  1.938E-15  1.938E-15  1.938E-15  .025  .049  .049  .049 
MaxAE  .062  .088  2.315E-14  .125  2.315E-14  2.315E-14  2.315E-14  .062  .125  .125  .125 
Normalized BIC  -34.450  42.313  -64.370  -4.530  -64.370  -64.370  -64.370  -34.450  -4.530  -4.530  -4.530 
Table 7.

Fit of the model, social network feelings, Carrefour.

Estimation  SE  Sig. 
NEGATIVEFEELINGSCARREFOUR-Model_2NEGATIVEFEELINGSCARREFOUR  Without transformation  Constant.096  .030  3.235  .018 
POSITIVEFEELINGSCARREFOURSELECCIÓNWithout transformationNumeratorDelay 0  1.058  .486  2.174  ,073 
Delay 2  -1.279  .482  -2.651  ,038 
Difference     

It is noted that in the social networks, the greater the positive feelings about Carrefour selección, the lower are the negative feelings about Carrefour. This coincides with what we have investigated in Tables 2,3 and 4 for Lidl. But in this case Model 1 was calculated with the Winters additive model, the ARIMA model not having a predictor for positive feelings and studying negative feelings.

In Table 8 it is seen that in the relation between DIA and Delicious, both for Model 1 and for Model 2, and coinciding with positive and negative feelings in the social networks, the statistical system used suggests to us that it is advisable to study Model 1 with the Winter additive and Model 2 with the Winter multiplicative.

Table 8.

Statistics of the model, social network, DIA.

Statistics of the model   
ModelNumber of predictorsFit statistics of the modelLjung-Box Q(18)
Stationary R-squared  R-squared  Normalized BIC  Statistics  DF  Sig. 
POSITIVEFEELINGSDIA-Model_1  1.000  -59.331 
NEGATIVEFEELINGSDIA-Model_2  1.000  -62.828 

It is observed that there are not predictors for the Delicious brand. That is, there is no relation between the feelings of Delicious with DIA, this brand setting itself apart from the establishment's own image. One of the results which we can establish is the possibility that social networks users comment on Delicious without perceiving that this is DIA's PPL.

For the study of the average number of visits to the establishment in relation to its PPL, we have re-codified the variable into two groups: up to 2 visits per month to the retailer establishment or 3 times or more. To check P5 we have used various procedures. On the one hand, the Kolmogorov-Smirnov comparison. In this analysis, instead of studying the probabilities of different events, the values of the distribution functions are compared, both in the sample and what would technically be derived from the population which has been made explicit in the null hypothesis. On the other hand, we have checked with the Mann-Whitney procedure to verify if two independent samples have been extracted from populations with an equal variance, compared to the alternative hypothesis that they have been extracted from populations with different variations. Finally, we have centered on the T test (see Table A4).

Table A4.

T test, Lidl, Carrefour and DIA.

    N LIDL  Mean LIDL  Standard deviation LIDL  Standard error average LIDL  N Carrefour  Mean Carrefour  Standard deviation Carrefour  Standard error average Carrefour  N DIA  Mean DIA  Standard deviation DIA  Standard error average DIA 
4078  2.603  1.0236  0.1159  97  2.196  0.7989  0.0811  57  2.509  0.8045  0.1066 
122  3.123  1.017  0.0921  103  2.854  0.9119  0.0899  143  3.028  0.8877  0.0742 
4178  2.654  1.0669  0.1208  97  2.268  0.8356  0.0848  57  2.491  0.7588  0.1005 
122  3.123  1.049  0.095  103  2.864  0.9605  0.0946  143  3.042  0.8711  0.0728 
4278  2.577  1.0385  0.1176  97  2.175  0.7639  0.0776  57  2.386  0.7009  0.0928 
122  3.025  1.0242  0.0927  103  2.845  0.9156  0.0902  143  2.972  0.8387  0.0701 
4378  3.526  0.785  0.0889  97  3.144  0.707  0.0718  57  3.211  0.818  0.1084 
122  3.811  0.8656  0.0784  103  3.631  0.6261  0.0617  143  3.545  0.6892  0.0576 
4478  3.641  0.7555  0.0855  97  3.392  0.7005  0.0711  57  3.316  0.7596  0.1006 
122  3.877  0.7985  0.0723  103  3.806  0.4862  0.0479  143  3.797  0.5881  0.0492 
4578  3.167  0.8283  0.0938  97  3.227  0.8721  0.0885  57  3.07  0.6777  0.0898 
122  3.418  0.9521  0.0862  103  3.398  0.7585  0.0747  143  3.42  0.7729  0.0646 
4678  3.833  0.6328  0.0717  97  3.753  0.7504  0.0762  57  3.737  0.5518  0.0731 
122  4.123  0.5689  0.0515  103  3.961  0.5225  0.0515  143  0.3753  0.0314 
4778  2.923  0.8021  0.0908  97  2.876  0.8324  0.0845  57  2.825  0.7349  0.0973 
122  3.443  0.9186  0.0832  103  3.214  0.7625  0.0751  143  3.315  0.8086  0.0676 
4878  3.192  0.9542  0.108  97  0.9354  0.095  57  3.035  0.9056  0.12 
122  3.762  0.9186  0.0832  103  3.757  0.8456  0.0833  143  3.608  0.9347  0.0782 
4978  2.756  1.0089  0.1142  97  2.814  1.0341  0.105  57  2.86  0.9899  0.1311 
122  2.279  1.0386  0.094  103  2.194  0.9081  0.0895  143  2.357  1.0441  0.0873 
5078  2.795  0.8583  0.0972  97  2.412  0.6732  0.0684  57  2.544  0.7336  0.0972 
122  3.09  0.9535  0.0863  103  2.932  0.7829  0.0771  143  3.098  0.79  0.0661 
5178  2.795  0.8274  0.0937  97  2.412  0.6885  0.0699  57  2.579  0.7058  0.0935 
122  3.066  0.9336  0.0845  103  2.971  0.822  0.081  143  3.084  0.7917  0.0662 
5278  2.769  0.8046  0.0911  97  2.423  0.6897  0.07  57  2.526  0.7098  0.094 
122  3.033  0.9442  0.0855  103  2.913  0.7934  0.0782  143  3.091  0.7953  0.0665 

From the T test results of independent samples, for DIA the differences of the mean values of the scale's variables are significantly different in all the variables. The previous results with the non-parametric Mann-Whitney test of independent samples are corroborated. It is shown that the people who go to DIA more frequently value the PPL more than those who visit less regularly.

For Carrefour, from the T test of independent samples it is noted that the differences of the mean scores of the scale are significantly distinct for all those of the scale except variable 45. As before, the results with the non-parametric Mann-Whitney test of independent samples are confirmed. This means that the people who visit Carrefour more often value the Carrefour Selección more than those who go to this hypermarket less frequently.

Lastly, for Lidl the same procedure is followed. From the T test results for independent samples, the differences of the scale's mean values are statistically distinct for all the scale's variables except number 45, which is on the limit of the parameter of acceptation. The previous results with the non-parametric Mann-Whitney test of independent samples are corroborated. It is shown that the customers who go more to Lidl value the Deluxe brand more than those who visit the establishment less.

To summarize, we set out the propositions accepted and rejected from the different data analyses carried out in Table 9.

Table 9.

Verification of the propositions.

PROPOSITIONS  RESULT 
P1a: The perceived quality of PPL with respect to the brand leader varies among the main distributors.  ACCEPTED 
P1b: There is a positive relation between the perceived value of PPL and income levels.  ACCEPTED 
P2a: There is a positive relation between the perceived value of PPL and income levels.  REJECTED 
P2b: There is no relation between the income level and the awareness value of the buyers of PPL.  ACCEPTED 
P3: The establishment's image influences the perceived quality of the PPL.  REJECTED 
P4: The brand image of retail establishments in social networks influences the perceived image of the PPL in these networks.  PARTIALLY ACCEPTED 
P5: The average number of visits to the establishment influences the value of its PPL.  ACCEPTED 
Discussion and conclusions

The study of PPL is increasingly more important, given the evolution and margins which they contribute to retailers (Ter Braak et al., 2014). In the first proposition (P1a) we have studied this kind of brand based on the perceived quality with respect to the manufacturer's brand leader within the retail establishment. The result of the research shows that the perceived quality of PPL compared to that of the brand leader varies among the main distributors. The proposition set out is therefore accepted. It is seen that the respondents who consume the Deluxe brand offer lower values for this variable than for Carrefour Selección and Delicious. While standard PPL tend to imitate mainstream-quality manufacturer brands in perceived quality (Geyskens et al., 2010), we can deduce that not all retailers carry out this type of strategies in the same way or at least the consumers do not perceive it the same. So, it is essential that retailers are aware that perceptions of the perceived quality of PPLs vary between stores in order to establish their future strategies.

The proposition P1b stable that there are significant differences in the brand value for the 3 Premium brands analyzed. Especially between the Carrefour Selección Premium and its competitors, Deluxe and Delicious, determining that the consumers of Carrefour Selección have a lower brand value than those of their two rivals analyzed. This may be determined by several factors, for example, the distributor's strategy would not have been positioned correctly with regards to its PPL. Subsequently, we suggest that each retailer should analyze its PPL brand value with those of its competitors to determine possible impacts within its positioning or its business strategy. Since this research has shown that there are some retailers that even having a higher perceived quality on their PPL with respect to their competitors, their PPL brand value may not be properly positioned.

In propositions P2a and P2b we analyze the income levels of the consumers related with the perceived level of PPL and the brand equity of the Premium distributor buyers. A previous investigation (Beneke, Greene, Lok, & Mallett, 2012) showed that on average consumers do not consider PPL as financially risky, or that their perception depends on the consumer's income level. However, the same has never been demonstrated for the perceived value and the brand equity. In this study the respondents with a greater income level did not show significant differences in their value perceptions with respect to those with lower incomes. This indicates that the level of incomes does not influence the perceived value of PPL. Hence, customers who buy PPL have an awareness value and a perceived value of these brands which does not depend on the different socioeconomic levels. In this sense, although many people think that PPLs are exclusive products for affluent customers, we recommend that managers do not focus their efforts on the consumer's income level to explain the brand value and the perceived value of these brands. Because in this article we have shown that the income levels of PPL customers do not influence these variables.

Although there is broad literature related to a product's brand value and its perceived quality (Kamakura & Russell, 1993; Park & Srinivasan, 1994; Rangaswamy, Raymond, & Terence, 1993), little is directly linked with the distributor chain owner of a PPL. Although other authors have centered on this topic (Bigné, Borredá, & Miquel, 2013) for the distributor brand, investigations have not been done for PPL. In this article the coefficient of the variable in the regression line is negative, specifically -0.455, thus rejecting proposition P3. This is why it can be concluded that the establishment's image does not positively influence the perceived quality of PPL. We therefore think that an establishment with a good image is not a sign that its PPL has more quality. This can be due to many customers not having a clear idea that the PPL analyzed are really distributor brands and think that they are manufacturer brands.

Including information obtained directly from social networks is a key part of this article's originality. Although some authors study brand influences in social networks (Zhao, Li, & Jin, 2016), there is a gap concerning the distributor brand and even more PPL. We have determined that brand image in retail establishments affects the brand image of their PPL – this is the case for Lidl and Carrefour but not for DIA. This is why we accept P4 partially, concluding that for Lidl and Carrefour the greater the positive feeling about their PPL, the lower the negative feelings about their labels are. The non-fulfillment in the case of DIA and its PPL Delicious concerning this criterion may be due to different factors. One of them may be the strategy of the distributor itself, given that in its establishments there is a tendency to show this brand in one or a few gondola ends, thus they do not often have the contact with other products which could facilitate their comparison. This phenomenon does not tend to happen in the other distributors studied. On the other hand, it may be that the customers of DIA have not been able to mentally separate the discount store from one they go to with higher range products, DIA having a hard job linking its PPL to its label. Therefore, we recommend that discount retailers take into account the above in order to correctly establish their strategy with respect to their PPL brand. Due to the fact that the positioning of a discount store could be conflicting with a PPL.

According to the aforementioned, social networks have become a fundamental instrument in retailers' communication strategies (Zhang, Guo, Hu, & Liu, 2016), being a tool which can cause feelings for or against their label. This is why it is necessary to center efforts on which strategies of retailers with their products in the on-line channel are parallel to the off-line world. This is the reason why it is recommended to take care of what is shown in the on-line channel about distributor brands and PPL. Given that different opinions of consumers in this medium are a source of a very direct influence on the behavior of future shoppers (Kang & Johnson, 2013). It should be especially emphasized that DIA, with its PPL Delicious, must showcase its label to shoppers in a way that is strategically determined for its establishments.

Finally, there are various authors who try to explain the relations between distributor brands and PPL with distinct variables as the epicenter of their research (Keller et al., 2016; Reimann & Wagner, 2016). None, however, has centered specifically on explaining if there is some relation between the average number of visits to the retail establishment and its PPL's value. In this case, we have determined that P5 is accepted given that the 3 retail establishments studied all coincide with the same common denominator. That is, the consumer who visits each establishment more yields a more positive value for the establishment's PPL. This is to say, the customers who visit a store more frequently have a better perception of its PPL's value. Therefore, PPL managers should be aware that average visits are an important component of brand value. Therefore, we recommend that managers who wish to improve the perception of value of their PPL should encourage their customers to visit their establishments more frequently, by means of loyalty tools.

For future research, we propose to extend this research with the extension of the sample in other European countries using the same brands. Subsequently, it would be quite interesting to investigate if this behaviour is verified in the same way in other continents and other retailers. In addition, we will consider introducing other variables to extend the studied model.

References
[Aaker, 1996]
D.A. Aaker.
Measuring brand equity across products and markets.
California Management Review, 38 (1996), pp. 102-120
[Abril and Rodriguez-Cánovas, 2016]
C. Abril, B. Rodriguez-Cánovas.
Marketing mix effects on private labels brand equity.
European journal of management and business economics, 25 (2016), pp. 168-175
[Ailawadi et al., 2001]
K.L. Ailawadi, S.A. Neslin, K. Gedenk.
Pursuing the value-conscious consumer: store brands versus national brand promotions.
Journal of Marketing, 65 (2001), pp. 71-89
[Ailawadi et al., 2008]
K.L. Ailawadi, K. Pauwels, J.-B.E.M Steenkamp.
Private label use and store loyalty.
Journal of Marketing, 72 (2008), pp. 19-30
[Ailawadi et al., 2010]
K.L. Ailawadi, E.T. Bradlow, M. Draganska, V. Nijs, R.P. Rooderkerk, K. Sudhir, J. Zhang.
Empirical models of manufacturer-retailer interaction: A review and agenda for future research.
Marketing Letters, 21 (2010), pp. 273-285
[Ailawadi and Keller, 2004]
K.L. Ailawadi, K. Keller.
Understanding retail branding: Conceptual insights y research priorities.
Journal of Retailing, 80 (2004), pp. 331-342
[Alvandi et al., 2014]
M. Alvandi, S. Fazli, M. Amiri.
Investigating the effect of interactive media on consumer-firm relationship.
International Journal Of Academic Research, 6 (2014),
[Arias Fernández et al., 2015]
E.J. Arias Fernández, J.M. Pineda, B.P. Florencio.
Factores determinantes en la elección de las marcas de distribuidor por parte de los consumidores.
Gestão e TI, 5 (2015),
[Arteaga et al., 2014]
R. Arteaga, V. Cortijo, U. Javed.
Students’ perceptions of Facebook for academic purposes.
Computers & Education, 70 (2014), pp. 138-149
[Baddeley et al., 2007]
M. Baddeley, D. Pillas, Y. Christopoulos, W. Schultz, P. Tobler.
Herding and social pressure in trading tasks: a behavioural analysis.
(2007),
[Baltas et al., 2007]
G. Baltas, C. Paraskevas, P.C Argouslidis.
Consumer characteristics and demand for store brands.
International Journal of Retail & Distribution Management, 35 (2007), pp. 328-341
[Barreto, 2014]
A.M. Barreto.
The word-of-mouth phenomenon in the social media era.
International Journal of Market Research, 56 (2014), pp. 631-654
[Bataineh, 2015]
A.Q. Bataineh.
The impact of perceived e-WOM on purchase intention: The mediating role of corporate image.
International Journal of Marketing Studies, 7 (2015), pp. 126
[Bazoche et al., 2005]
P. Bazoche, E. Giraud-Héraud, L.G. Soler.
Premium private labels, supply contracts, market segmentation, and spot prices.
Journal of Agricultural & Food Industrial Organization, 3 (2005),
[Beldona and Wysong, 2007]
S. Beldona, S. Wysong.
Putting the "brand" back into store brands: an exploratory examination of store brands and brand personality.
Journal of Product and Brand Management, 16 (2007), pp. 226-235
[Beneke, 2013]
J. Beneke.
A closer inspection of the impact of perceived risk on purchase intention of premium private label brands: The effect of age, gender, income and racial group.
Journal of Business and Retail Management Research, 7 (2013),
[Beneke et al., 2012]
J. Beneke, A. Greene, I. Lok, K. Mallett.
La influencia de la percepción del riesgo en la compra intención - el caso de la prima de comestibles marcas privadas en Sudáfrica.
Diario de Gestión de Producto y Marca, 21 (2012), pp. 4-14
[Beristain and Zorrilla, 2011]
J.J. Beristain, P. Zorrilla.
The relationship between store image and store brand equity: A conceptual framework and evidence from hypermarkets.
Journal of Retailing and Consumer Services, 18 (2011), pp. 562-574
[Beristain, 2009]
J.J. Beristain.
El valor de las marcas de distribuidor y sus antecedentes: el caso de las marcas de cadena.
Revista de Dirección y Administración de Empresas, 16 (2009), pp. 123-149
[Bigné and Hyder, 2012]
E. Bigné, A. Hyder.
Does Web Site Engagement Lead To Making A Purchase? Summer Educators’ Conference, American Marketing Association, Chicago.
(2012),
[Bigné et al., 2013]
E. Bigné, A. Borredá, M.J. Miquel.
El valor del establecimiento y su relación con la imagen de marca privada: efecto moderador del conocimiento de la marca privada como oferta propia del establecimiento.
Revista Europea de Dirección y Economía de la Empresa, 22 (2013), pp. 1-10
[Blanco et al., 2011]
E.F. Blanco, D.A. García, I.M. Martín.
Las estrategias publicitarias de las marcas en el contexto de crisis.
adComunica, 1 (2011), pp. 119-138
[Borgaonkar, 2020]
V.H. Borgaonkar.
Impact of Social media Influencers on the Cosmetic Industry and the Buying Behaviour of Consumers.
Purakala with ISSN 0971-2143 is an UGC CARE Journal, 31 (2020), pp. 591-600
[Boyd and Ellison, 2008]
D.M. Boyd, N.B. Ellison.
Social network sites: Definition, history, and scholarship.
Journal of Computer-Mediated Communication, 13 (2008),
[Bronnenberg et al., 2020]
B.J. Bronnenberg, J.P. Dubé, R.E. Sanders.
Consumer misinformation and the brand premium: a private label blind taste test.
Marketing Science, 39 (2020), pp. 382-406
[Brown et al., 2019]
R. Brown, S. Mawson, A. Rowe.
Start-ups, entrepreneurial networks and equity crowdfunding: a processual perspective.
Industrial Marketing Management, 80 (2019), pp. 115-125
[Cantallops and Salvi, 2014]
A.S. Cantallops, F. Salvi.
New consumer behavior: A review of research on eWOM and hotels.
International Journal of Hospitality Management, 36 (2014), pp. 41-51
[Chih et al., 2013]
W.H. Chih, K.Y. Wang, L.C. Hsu, S.C. Huang.
Investigating electronic word-of-mouth effects on online discussion forums: the role of perceived positive electronic word-of-mouth review credibility.
Cyberpsychology, Behavior, and Social Networking, 16 (2013), pp. 658-668
[Choi, 2016]
S.C. Choi.
Defensive strategy against a private label: Building brand premium for retailer cooperation.
Journal of Retailing and Consumer Services, 34 (2016), pp. 335-339
[Connelly et al., 2011]
B.L. Connelly, S.T. Certo, R.D. Ireland, C.R. Reutzel.
Signaling theory: A review and assessment.
Journal of management, 37 (2011), pp. 39-67
[Court et al., 2012]
D. Court, J. Gordon, J. Perrey.
Measuring marketing’s worth.
McKinsey Quarterly, 3 (2012), pp. 113-118
[Cuneo et al., 2012]
A. Cuneo, P. Lopez, M.J. Yagüe.
Measuring private labels brand equity: a consumer perspective.
European Journal of Marketing, 46 (2012), pp. 952-964
[Curras et al., 2011]
R. Curras, C. Ruiz, S. Sanz.
“Antecedentes y consecuencias de la actitud hacia el uso de las redes sociales”, XXIII Congreso Nacional de Marketing.
(2011),
[Dabbagh and Kitsantas, 2012]
N. Dabbagh, A. Kitsantas.
Personal Learning Environments, social media, and self-regulated learning: A natural formula for connecting formal and informal learning.
The Internet and Higher Education, 15 (2012), pp. 3-8
[Delgado-Ballester et al., 2014]
E. Delgado-Ballester, M. Hernandez-Espallardo, A. Rodriguez-Orejuela.
Store image influences in consumers’ perceptions of store brands: the moderating role of value consciousness.
European Journal of Marketing, 48 (2014), pp. 1850-1869
[Dobson and Chakraborty, 2009]
P. Dobson, R. Chakraborty.
Private labels and branded goods: consumers’‘horrors’ and ‘heroes’.
Private Labels, Brands and Competition Policy, Oxford University Press, (2009), pp. 99-124
[Dogruer et al., 2011]
N. Dogruer, I. Menevis, R. Eyyam.
What is the motivation for using Facebook?.
Procedia Social and Behavioral Sciences, 15 (2011), pp. 2642-2646
[Dunne and Narasimhan, 1999]
D. Dunne, C. Narasimhan.
The new appeal of store brands.
Harvard Business Review, 77 (1999), pp. 41-48
[Ellison et al., 2007]
N.B. Ellison, C. Steinfield, C. Lampe.
The benefits of Facebook “Friends”: Social capital and college students’ use of online social network sites.
Journal of Computer Mediated Communication, 12 (2007),
[Ferreira et al., 2020]
J. Ferreira, V. Santos, P. Malta.
Potential and Advantages of Social Shopping in Portugal.
In Marketing and Smart Technologies, Springer, (2020), pp. 125-136
[Fischer et al., 2010]
M. Fischer, F. Völckner, H. Sattler.
How important are brands? A cross-category, cross-country study.
Journal of Marketing Research, 47 (2010), pp. 823-839
[Floor, 2006]
K. Floor.
Branding a store: How to build successful retail brands in a changing marketplace.
Kogan Page Publishers, (2006),
[Fombrun and Shanley, 1990]
C. Fombrun, M. Shanley.
What’s in a name? Reputation building and corporate strategy.
Academy of Management Journal, 33 (1990), pp. 233-258
[Garretson et al., 2002]
J.A. Garretson, D. Fisher, S. Burton.
Antecedents of private label attitude y national brand promotion attitude: Similarities y differences.
Journal of Retailing, 78 (2002), pp. 91-101
[George and Mallery, 1995]
D. George, P. Mallery.
SSPS/PC. step by step: A simple guide and reference.
Wadsworth Publishing Company, (1995),
[Geyskens et al., 2010]
I. Geyskens, K. Gielens, E. Gijsbrechts.
Proliferating private-label portfolios: How introducing economy and premium private labels influences brand choice.
Journal of Marketing Research, 47 (2010), pp. 791-807
[Geyskens et al., 2018]
I. Geyskens, K.O. Keller, M.G. Dekimpe, K. de Jong.
How to brand your private labels.
Business horizons, 61 (2018), pp. 487-496
[Go et al., 2009]
A. Go, R. Bhayani, L. Huang.
Twitter sentiment classification using distant supervision.
CS224N Project Report, Stanford, 1 (2009), pp. 12
[Grunert and Aachmann, 2016]
K.G. Grunert, K. Aachmann.
Consumer reactions to the use of EU quality labels on food products: A review of the literature.
Food Control, 59 (2016), pp. 178-187
[Hajli, 2014]
M.N. Hajli.
A study of the impact of social media on consumers.
International Journal of Market Research, 56 (2014), pp. 387-404
[Higgins and Bannister, 1992]
R.B. Higgins, B.D. Bannister.
How corporate communication of strategy affects share price.
Long Range Planning, 25 (1992), pp. 27-35
[Hoch and Banerji, 1993]
S.J. Hoch, S. Banerji.
When do private labels succeed?.
Sloan Management Review, 34 (1993), pp. 57
[Hsiao et al., 2019]
S.H. Hsiao, Y.Y. Wang, T. Wang, T.W. Kao.
How social media shapes the fashion industry: The spillover effects between private labels and national brands. Industrial Marketing Management.
(2019),
[Huang and Huddleston, 2009]
Y. Huang, P. Huddleston.
Retailer premium own-brands: creating customer loyalty through own-brand products advantage.
International Journal of Retail & Distribution Management, 37 (2009), pp. 975-992
[Immonen, 2010]
L. Immonen.
Package cues and their influence on the perception of premium quality of premium private label products.
(2010),
[IRI, 2014]
IRI.
La marca de distribuidor vive su primera desaceleración en Europa desde el inicio de la recesión económica. Insights and Publications. IRI.
(2014),
[Jiang et al., 2011]
L. Jiang, M. Yu, M. Zhou, X. Liu, T. Zhao.
Target-dependent twitter sentiment classification.
(2011), pp. 151-160
[Jones et al., 2007]
P. Jones, D. Comfort, D. Hillier.
What’s in store? Retail marketing and corporate social responsibility.
Marketing Intelligence & Planning, 25 (2007), pp. 17-30
[Kamakura and Russell, 1993]
W.A. Kamakura, G.J. Russell.
Measuring brand value with scanner data.
International Journal of Research in Marketing, 10 (1993), pp. 9-21
[Kang and Johnson, 2013]
J.Y. Kang, K. Johnson.
How does social commerce work for apparel shopping? Apparel social e-shopping with social network storefronts.
Journal of Customer Behaviour, 12 (2013), pp. 53-72
[Keller, 1993]
K.L. Keller.
Conceptualizing, measuring, and managing customer-based brand equity.
Journal of Marketing, 57 (1993), pp. 1-22
[Keller et al., 2016]
K.O. Keller, M.G. Dekimpe, I. Geyskens.
Let Your Banner Wave? Antecedents and Performance Implications of Retailers’ Private-Label Branding Strategies.
Journal of Marketing, 80 (2016), pp. 1-15
[Kethüda, 2020]
Ö Kethüda.
Evaluating the Success of the Economic and Premium Private Labels in Retailers Positioned at the Opposite Ends of the Price-Quality Axis.
In Improving Marketing Strategies for Private Label Products, IGI Global, (2020), pp. 83-109
[Kim et al., 2019]
S. Kim, S. Lee, J.H. Lee, C.R. Taylor.
Can premium private labels compete with luxury brands: the impact of advertising on perceived luxuriousness.
International Journal of Advertising, (2019), pp. 1-22
[Kozinets, 2002]
R.V. Kozinets.
The field behind the screen: Using netnography for marketing research in online communities.
Journal of marketing research, 39 (2002), pp. 61-72
[Kumar, 2007]
N. Kumar.
Private label strategy: How to meet the store brand challenge.
Harvard Business Review Press, (2007),
[Kumar, 2019]
S. Kumar.
Consumer Preference Towards Private Label Brands with Reference to Retail Apparel in India.
Indian Journal of Marketing, 49 (2019), pp. 49-66
[Kuo and Yen, 2009]
D. Kuo, S. Yen.
Towards an understanding of the behavioral intention to use 3G mobile value-added services.
Computers in Human Behavior, 25 (2009), pp. 103-110
[Lamey et al., 2012]
L. Lamey, B. Deleersnyder, J.-B.E. Steenkamp, M.G. Dekimpe.
The effect of business-cycle fluctuations on private-label share: what has marketing conduct got to do with it?.
Journal of Marketing, 76 (2012), pp. 1-19
[Lee et al., 2008]
J. Lee, D.H. Park, I. Han.
The effect of negative online consumer reviews on product attitude: An information processing view.
Electronic commerce research and applications, 7 (2008), pp. 341-352
[Lewin et al., 2008]
J. Lewin, J.T. Gomez-Arias, L. Bello-Acebron.
Why do leading brand manufacturers supply private labels?.
Journal of Business & Industrial Marketing., (2008),
[Lichtenstein et al., 1990]
D.R. Lichtenstein, R.G. Netemeyer, S. Burton.
Distinguishing coupon proneness from value consciousness: An acquisition-transaction utility theory perspective.
Journal of Marketing, 54 (1990), pp. 54-67
[Lin and Lu, 2011]
K.Y. Lin, H.P. Lu.
Why people use social networking sites: An empirical study integrating network externalities and motivation theory.
Computers in Human Behavior, 27 (2011), pp. 1152-1161
[Loebnitz et al., 2020]
N. Loebnitz, S. Zielke, K.G. Grunert.
Consumers’ brand decision: a matter of social risk.
International Journal of Retail & Distribution Management., (2020),
[López and Sicilia, 2014]
M. López, M. Sicilia.
Determinants of E-WOM influence: the role of consumers’ internet experience.
Journal of theoretical and applied electronic commerce research, 9 (2014), pp. 28-43
[Martinelli and Vergura, 2014]
E. Martinelli, D.T. Vergura.
Evolving the Private Label Role in the Retailer-Customer Relationship: Antecedents and Impact of Premium. Handbook of Research on Retailer-Consumer Relationship Development.
(2014), pp. 86
[Martinelli and Vergura, 2020]
E. Martinelli, D.T. Vergura.
Satisfying Customers Through Premium Private Labels: Identifying Drivers of Customer Loyalty to the Retailer.
In Handbook of Research on Retailing Techniques for Optimal Consumer Engagement and Experiences, IGI Global, (2020), pp. 33-56
[Martinelli et al., 2019]
E. Martinelli, F. De Canio, G. Marchi.
Premium private labels (PPLs): From food products to concept stores.
In Case Studies in Food Retailing and Distribution, Woodhead Publishing, (2019), pp. 267-279
[Martínez-Ruiz et al., 2010]
M.P. Martínez-Ruiz, A.I. Jiménez-Zarco, A. Izquierdo-Yusta.
Customer satisfaction’s key factors in Spanish grocery stores: Evidence from hypermarkets and supermarkets.
Journal of Retailing and Consumer Services, 17 (2010), pp. 278-285
[Medhat et al., 2014]
W. Medhat, A. Hassan, H. Korashy.
Sentiment analysis algorithms and applications: A survey.
Ain Shams Engineering Journal, 5 (2014), pp. 1093-1113
[Mumi et al., 2019]
A. Mumi, M. Obal, Y. Yang.
Investigating social media as a firm’s signaling strategy through an IPO.
Small Business Economics, 53 (2019), pp. 631-645
[Nabec, 2002]
L. Nabec.
L’évolution des représentations de la marque et de l’enseigne dans l’esprit du consommateur lors de la formation de nouveaux couples marque-enseigne. Paris Dauphine University.
(2002),
[Neelotpaul, 2010]
B. Neelotpaul.
A study on interactivity and online branding.
Advances in Management, 3 (2010), pp. 13-16
[Nenycz-Thiel and Romaniuk, 2012]
M. Nenycz-Thiel, J. Romaniuk.
Value-for-money perceptions of supermarket and private labels.
Australasian Marketing Journal (AMJ), 20 (2012), pp. 171-177
[Nenycz-Thiel and Romaniuk, 2016]
M. Nenycz-Thiel, J. Romaniuk.
Understanding premium private labels: A consumer categorisation approach.
Journal of Retailing and Consumer Services, 29 (2016), pp. 22-30
[Nielsen, 2015]
Nielsen.
Nielsen 360 2015 edition available from: Nielsen.
(2015),
[Ohlson, 2019]
J. Ohlson.
Premium Private Label Market Growthon the expense of National Brands: A study on the Swedish premium private label market shares and its characteristics.
(2019),
[Ohno and Takemura, 2019]
T. Ohno, M. Takemura.
Evolution of private label development: A japanese case.
International journal of multidisciplinarity in business and scienc, 5 (2019), pp. 53-62
[Olsen and Sallis, 2010]
N.V. Olsen, J.E. Sallis.
Processes and outcomes of distributor brand new product development.
International Journal of Retail & Distribution Management., (2010),
[Olshavsky, 1985]
R.W. Olshavsky.
Perceived quality in consumer decision making: an integrated theoretical perspective.
Perceived Quality, 4 (1985), pp. 3-29
[Pak and Paroubek, 2010]
A. Pak, P. Paroubek.
Twitter as a Corpus for Sentiment Analysis and Opinion Mining.
In LREc, 10 (2010), pp. 1320-1326
[Park and Srinivasan, 1994]
C.S. Park, V. Srinivasan.
A survey-based method for measuring and understanding brand equity and its extendibility.
Journal of Marketing Research, 31 (1994), pp. 271-288
[Partal and Benito, 2014]
M.M. Partal, O.G. Benito.
Efectos de la extensión de marca de distribuidor en el eje calidad-precio.
Universia Business Review, 1 (2014), pp. 18-39
[Pauwels and Srinivasan, 2004]
K. Pauwels, S. Srinivasan.
Who benefits from store brand entry?.
Marketing Science, 23 (2004), pp. 364-390
[Pillai and Kumar, 2012]
K.G. Pillai, V. Kumar.
Differential effects of value consciousness and coupon proneness on consumers’ persuasion knowledge of pricing tactics.
Journal of Retailing, 88 (2012), pp. 20-33
[Pınar and Girard, 2020]
M. Pınar, T. Girard.
Comparing Private Label Brand Equity Dimensions of the Same Store: Their Relationships, Similarities, and Differences. In Improving Marketing Strategies for Private Label Products.
IGI Global, (2020), pp. 61-82
[Pookulangara and Koesler, 2011]
S. Pookulangara, K. Koesler.
Cultural influence on consumers’ usage of social networks and its’ impact on online purchase intentions.
Journal of Retailing and Consumer Services, 18 (2011), pp. 348-354
[Puelles, 1995]
J.A. Puelles.
Análisis del Fenómeno de las Marcas de Distribución en España y de su Tratamiento Estratégico.
Información Comercial Española, 739 (1995), pp. 117-129
[Rangaswamy et al., 1993]
A. Rangaswamy, B. Raymond, A.O. Terence.
Brand equity and the extendibility of brand names.
International Journal of Research in Marketing, 10 (1993), pp. 61-75
[Reimann and Wagner, 2016]
O. Reimann, U. Wagner.
The Potential of Co-branding as a Branding Strategy for Premium Private Labels: A Theoretical Assessment.
In Advances in National Brand and Private Label Marketing, 12 (2016), pp. 91-94
[Richardson et al., 1996]
P.S. Richardson, A.K. Jain, A. Dick.
Household store brand proneness: a framework.
Journal of Retailing, 72 (1996), pp. 159-185
[Rissoan, 2013]
Ramón. Rissoan.
Redes Sociales. Comprender y Dominar las Nuevas Herramientas de Comunicación. Barcelona. Ediciones ENI.
(2013), pp. 12-79
[Rondán-Cataluña et al., 2015]
F.J. Rondán-Cataluña, J. Arenas-Gaitán, P. Ramírez-Correa.
Travel buying behavior in social network site users: to buy online vs. offline.
Journal of Theoretical and Applied Electronic Commerce Research, 10 (2015), pp. 49-62
[Ruiz-Real et al., 2018]
J.L. Ruiz-Real, J.C. Gázquez-Abad, I. Esteban-Millat, F.J. Martínez-López.
Betting exclusively by private labels: could it have negative consequences for retailers?.
Spanish Journal of Marketing-ESIC, 22 (2018), pp. 183-202
[Schnittka, 2015]
O. Schnittka.
Are they always promising? An empirical analysis of moderators influencing consumer preferences for economy and premium private labels.
Journal of Retailing and Consumer Services, 24 (2015), pp. 94-99
[Seidman, 2013]
G. Seidman.
Self-presentation and belonging on Facebook: How personality influences social media use and motivations.
Personality and Individual Differences, 54 (2013), pp. 402-407
[Semeijn et al., 2004]
J. Semeijn, A.C. Van Riel, A.B. Ambrosini.
Consumer evaluations of store brands: effects of store image and product attributes.
Journal of Retailing and Consumer Services, 11 (2004), pp. 247-258
[Sokolova and Kefi, 2020]
K. Sokolova, H. Kefi.
Instagram and YouTube bloggers promote it, why should I buy? How credibility and parasocial interaction influence purchase intentions.
Journal of Retailing and Consumer Services, 53 (2020),
[Sprott and Shimp, 2004]
D.E. Sprott, T.A. Shimp.
Using product sampling to augment the perceived quality of store brands.
Journal of Retailing, 80 (2004), pp. 305-315
[Stiles, 1992]
W.B. Stiles.
Describing talk: A taxonomy of verbal response modes.
Sage Publications, (1992),
[Tandon and Verma, 2020]
N. Tandon, J. Verma.
The Effects of Private Label Brand Strategies on Consumer Perception. In Improving Marketing Strategies for Private Label Products.
IGI Global, (2020), pp. 187-205
[Ter Braak et al., 2014]
A. Ter Braak, I. Geyskens, M.G. Dekimpe.
Taking private labels upmarket: Empirical generalizations on category drivers of premium private label introductions.
Journal of Retailing, 90 (2014), pp. 125-140
[Thelwall et al., 2011]
M. Thelwall, K. Buckley, G. Paltoglou.
Sentiment in Twitter events.
Journal of the American Society for Information Science and Technology, 62 (2011), pp. 406-418
[Tosun, 2012]
L.P. Tosun.
Motives for Facebook use and expressing ‘‘true self’’ on the Internet.
Computers in Human Behavior, 28 (2012), pp. 1510-1517
[Trusov et al., 2009]
M. Trusov, R.E. Bucklin, K. Pauwels.
Effects of words-of-mouth versus traditional marketing: findings from an Internet Social networking site.
Journal of Marketing., 73 (2009), pp. 90-102
[Tsytsarau and Palpanas, 2012]
M. Tsytsarau, T. Palpanas.
Survey on mining subjective data on the web.
Data Mining and Knowledge Discovery, 24 (2012), pp. 478-514
[Vinodhini and Chandrasekaran, 2012]
G. Vinodhini, R.M. Chandrasekaran.
Sentiment analysis and opinion mining: a survey.
International Journal of Advanced Research in Computer Science and Software Engineering, 2 (2012),
[Yang and Brown, 2013]
C.C Yang, B. Brown.
Motives for using Facebook, patterns of Facebook activities, and late adolescents. Social adjustment to college.
Youth Adolescence, 42 (2013), pp. 403-416
[Yoo and Donthu, 2001]
B. Yoo, N. Donthu.
Developing and validating a multidimensional consumer-based brand equity scale.
Journal of Business Research, 52 (2001), pp. 1-14
[Yoon, 2012]
S.J. Yoon.
A social network approach to the influences of shopping experiences on e-wom.
Journal of Electronic Commerce Research, 13 (2012), pp. 213
[Zhang et al., 2016]
M. Zhang, L. Guo, M. Hu, W. Liu.
Influence of customer engagement with company social networks on stickiness: Mediating effect of customer value creation.
International Journal of Information Management, 37 (2016), pp. 229-240
[Zhao et al., 2016]
Y. Zhao, S. Li, F. Jin.
Identification of influential nodes in social networks with community structure based on label propagation.
Neurocomputing, 210 (2016), pp. 34-44
[Zielke and Dobbelstein, 2007]
S. Zielke, T. Dobbelstein.
Customers’ willingness to purchase new store brands.
Journal of Product & Brand Management, 16 (2007), pp. 112-121
Copyright © 2020. Journal of Innovation & Knowledge
Article options
Tools