«of the AM’s Brand, Corporate Identity and Reputation SIG INSTRUCTIONS FOR THE SESSIONS Sessions chairs The main function of a session chair is to ...»
Hirschman, E.C. 1986. The Effect of Verbal and Pictorial Advertising Stimuli on Aesthetic, Utilitarian and Familiarity Perceptions. Journal of Advertising 15 (2), 27–34.
John, D.R., Loken, B., Kim, K. & Monga, A.B. 2006. Brand Concept Maps: A Methodology for Identifying Brand Associations Networks. Journal of Marketing Research 43 (4), 549– 563.
Kapferer, J.-N. & Laurent, G. 1986. Consumer Involvement Profiles: A New Practical Approach of Consumer Involvement. Journal of Advertising Research 25 (6), 48–56.
Keller, K.L. 1993. Conceptualizing, Measuring, and Managing Customer-Based Brand Equity. Journal of Marketing 57 (1), 1–22.
Koffka, K. 1922. Perception: An Introduction to the Gestalt-theories. Psychological Bulletin 19, 531–585.
Lee, D.H. 1990. Symbolic Interactionism: Some Implications for Consumer Self-Concept and Product Symbolism Research. Advances in Consumer Research 17, 386–393.
Lee, M. 1995. Effects of Schema Congruity and Involvement on Product Evaluations.
Advances in Consumer Research 22, 210–216.
Levy, S.J. 1959. Symbols for Sale. Harvard Business Review 37 (4), 117–124.
Low, G.S. & Lamb Jr., C.W. 2000. The measurement and dimensionality of brand associations. Journal of Product & Brand Management 9 (6), 350–370.
Luchs, M. & Swan, K.S. 2011. Perspective: The emergence of product design as a field of marketing inquiry. Journal of Product Innovation Management 28 (3), 327–345.
Noble, C.H. & Kumar, M. 2010. Exploring the Appeal of Product Design: A Grounded, Value-Based Model of Key Design Elements and Relationships. Journal of Product Innovation Management 27 (5), 640–657.
Petty, R.E. & Cacioppo, J.T. 1986. The Elaboration Likelihood Model of Persuasion.
Advances in Experimental Social Psychology 19, 123–205.
Pohl, R.F., Eisenhauer, M. & Hardt, O. 2003. SARA: A cognitive process model to stimulate the anchoring effect and hindsight bias. Memory 11 (4/5), 337-356.
Rafaeli, A. & Vilnai-Yavetz, I. 2004. Emotion as a Connection of Physical Artifacts and Organizations. Organization Science 15 (6), 671–686.
Ravasi, D. & Stigliani, I. 2012. Product Design: a Review and Research Agenda for Management Studies. International Journal of Management Reviews 14 (4), 464–488.
Rindova, V.P. & Petkova, A.P. 2007. When Is a New Thing a Good Thing? Technological Change, Product Form Design, and Perceptions of Value for Product Innovations.
Organization Science 18 (2), 217–232.
Rubera, G. 2014. Design Innovativeness and Product Sales' Evolution. Marketing Science, forthcoming.
Verganti, R. 2006. Innovating Through Design. Harvard Business Review 84 (12), 114–122.
Veryzer, R.W. 1993. Aesthetic Response and the Influence of Design Principles on Product Preferences. Advances in Consumer Research 20, 224–228.
Appendices Appendix 1: Preliminary studies and main study Preliminary studies and main study
Main study (N=490) Measurement of the influences of the stimuli connotations Appendix 2: Connotations scale Connotations scale Feminine: rough/gentle, hard/soft, angular/round and masculine/feminine Noble: ordinary/exclusive, primitive/elevated, cheap/expensive and common/noble Further items: unprofessional/professional, dishonest/honest, undynamic/dynamic, Uncharismatic/charismatic, incompetent/competent, distorted/undistorted, not chic/chic, not Classy/classy, unimaginative/imaginative, not fresh/fresh, artificial/natural, not catchy/catchy and unsuccessful/successful Appendix 3: Stimuli with PDC evaluations
f: feminine connotation scale n: noble connotation scale N=120 (For each product category: N=30) *** = p.001 (Significances of differences between connotated and default products)
Mixture modeling in branding research: Latent classes of brand equity and loyalty in an organizational context Juntunen, Mari Juntunen, Jouni Paananen, Mikko Purpose of Paper Researchers using structural equation modelling (SEM) frequently treat their data as if they were collected from a single population (Muthén 1989). This assumption of homogeneity is often unrealistic and may provide misleading results, as the data may contain unobserved heterogeneity (Jedidi et al. 1997). Mixture modeling, or finite mixture modeling (McLachlan and Peel 2000), refers to modeling with categorical latent variables that represent subpopulations where the population membership is not known but is revealed from the data (Muthen and Muthen 1998–2007; Van Horn et al. 2009). Respondents are not segmented a priori because their group membership is unknown (Bart et al. 2005); instead, the techniques have been developed to identify unobserved heterogeneity−or latent classes−from the data.
Finite mixture structural equation modeling (FMSEM) extends the traditional multi-group SEM by allowing to uncover unobservable customer segments and to estimate segmentspecific path coefficients in a research model simultaneously (Bart et al. 2005). In this study we use FMSEM to reveal latent classes of brand equity (BE) and loyalty in an organizational context.
Customer loyalty is one of the most important goals that brand managers want to achieve.
Loyalty is one of the central concepts in most BE models in different contexts (see, e.g.
Aaker 1991; King & Grace 2010, Faircloth 2005; French & Smith 2010; Kim et al. 2011). In an organizational context several researchers see loyalty−customers’ intentions to continue buying from the service provider, along with a deeply held commitment−as an outcome of BE (e.g. Keller 1993; van Riel et al. 2005; Taylor et al. 2004; Vogel et al. 2008). BE refers to “differential effect of brand knowledge on consumer response to the marketing of the brand” (Keller 1993: 1). Other central concepts of BE are brand image (BI) and brand awareness (BA). BI refers to the customer’s perceptions about a brand) (Keller 1993). BA refers to collective awareness: when a brand is known, each individual knows that it is known (Kapferer 2012: 11).
Davis et al. (2008) found that in the business services context both BA and BI have a positive influence on BE. Berry (2000) says that BI rather than BA is the major determinant of BE in the services context. Thus, it is possible that the antecedents of BE differ in different unobservable heterogeneous customer segments. The purpose of this study is 1) to reveal the number of latent classes, that is, the number of different models of BE and loyalty in our data, and 2) to discover the model of BE and loyalty for each latent class with the help of FMSEM analyses.
Methodology/approach In our research model (Figure 1) BE is a structure which consists of BA and BI. We assume that BE influences loyalty. The variable C indicates the existence and number of latent classes and the dashed arrows mean that the slope of the linear regressions of independent variables varies across these latent classes. As the approach is data-driven, no hypotheses are formulated. Measures are presented in Appendix.
Figure 1: Research model We chose business customers of a Finnish brewery Olvi plc. for our respondents. Olvi delivers products for its customers through a variety of business services (e.g. leasing and maintaining beer taps, logistics) and has no typical customer type: they vary from small privately owned restaurants to large restaurant chains. Our data was gathered from 645 restaurant customers through a Webropol survey in April 2013. A total of 173 answers (26.8 %) were received.
We began by analysing the data with SEM using the MPlus software package. As the model was unworkable (as was assumed in order to reveal unobserved heterogeneity), we continued our analyses with FMSEM to elicit the number of the latent classes and a SEM model for each class. The model was first estimated using one latent class, then two latent classes, and so on, until the model-fit information suggested that the previous model was better than the current model.
Findings Fit indices (Table 1) suggest the three latent class solution. Log-likelihood (LogLH) is lower in solutions with more classes, which indicates that the model works as it should. The Bayesian information criteria (BIC), the most reliable fit index with a small sample size (less than 500) (Tolvanen 2007), offers the lowest value for the solution with three latent classes.
Parametric bootstrapped likelihood ratio (PBLR) is reliable when it can be produced and offers a statistically significant p-value (0.000) with three latent classes solution. Other values are suitable for large sample sizes (Tolvanen 2007).
Table 1. Evaluation of latent classes
LC1 has 7 members (4.0 % of the population), LC2 has 37 members (21.4 %) and LC3 has 129 members (74.6 %). Average posteriori probabilities reveal that the probability of the members belonging to the suggested classes is very high (LC1 0.995; LC2 0.942; LC3 0.982).
In all SEM models BI has a positive relationship with BE and BE has a positive influence on loyalty (Figure 2). The relationship between BA and BE varies, thus we focus on it. In LC1 the relationship is positive: increase (decrease) in BA increases (decreases) loyalty. In LC2 the relationship is negative: increase (decrease) in BA diminishes (increases) loyalty. In LC3 BA has no relationship with BE.
Although estimates in the models vary, model structures are mathematically indifferent. This allows us to consider BA as an outcome of BE. This view might conflict with traditional brand research (Keller 1993), but can be accepted as the role of awareness needs attention (Berry 2000; Kapferer 2012). In LC1 increase (decrease) in BI or loyalty increases (decreases) BA; in LC2 increase (decrease) in BI or loyalty diminishes (increases) BA; and in LC3 change in BI or loyalty has no influence on BA. Additionally, BA, BI and loyalty can all be interpreted as dimensions of BE. In LC1 higher (lower) BA values indicate higher (lower) BE values; in LC2 higher (lower) BA values indicate lower (higher) BE values; and in LC3 only higher (lower) BI or loyalty values indicate higher (lower) BE values.
Figure 2: The SEM models of the three latent classes Theoretical implications Our results show that even a small data sample may contain latent classes that behave differently. We revealed three latent classes: brand advocates (LC1), brand rebels (LC2), and performance seekers (LC3). Brand advocates represent a traditional view. The higher they value any of the components of BE, the better they consider the BE being. The better known they consider the brand, or the better their image of the brand is, the more loyal they intend to be to the brand. This is in line with Keller (1993) and with our research model. Our finding is that when their brand image improves or loyalty intentions increase, they feel that the collective awareness of the brand increases.
Brand rebels behave in the opposite way. The higher they think the collective awareness of a brand is, the lower they consider the BE, and vice versa. When collective awareness increases, their loyalty intentions decrease; and when they think collective awareness decreases, their loyalty intentions increase. Thus this group of respondents might search for new brands and experiences. For performance seekers awareness of the brand has no influence; only their image on company’s actual performance influences their loyalty intentions.
Practical implications It is important for managers to notice that the antecedents of BE and loyalty may vary among customer segments. Different customers may appreciate different aspects: collective awareness might be extremely important for some of the customers but absolutely negative for others, while some care about how the company performs only. Thus, new insights and views are needed in order to find new kinds of customer segments and answer the contemporary customer needs in new ways.
Mixture modeling offers new, statistically acceptable ways for researchers and managers to approach the data sets when the original SEM model proves to be unworkable, as well as it might offer new insights to handle big data.
Limitations We focused on FMSEM, but a family of mixture modeling techniques is much wider and thus offers several other aspects for branding research. We used Keller’s (1993) view on BE; a number of other BE models might have been chosen. One of our latent classes (LC1) consists of 4 % of the respondents only, although its’ existence is both statistically and theoretically justified. Finally, our data is small and gathered in one country and context only, which limits the generalizability of the results, specifically in today’s global world where big data is emphasized.
Originality/value This study is one of the rare attempts to employ mixture modeling techniques in general and FMSEM in particular in branding research. Our study might be the first to reveal respondents’ unobserved heterogeneity in an organizational branding context.
Keywords Branding; Finite mixture structural equation modeling (FMSEM); Unobserved heterogeneity;
Latent class approach; Segmentation References Aaker, D.A. 1991. Managing Brand Equity. Capitalizing the Value of a Brand Name. New York: The Free Press.
Aaker, D.A. 1996. Measuring brand equity across products and markets. California management review 38 (3), 102–120.
Bart, Y., Venkatesh, S., Sultan, F. & Urban, G.L. 2005. Are the Drivers and Role of Online Trust the Same for All Web Sites and Consumers? A Large-Scale Exploratory Empirical Study. Journal of Marketing 69 (4), 133–152.
Berry, L.L. 2000. Cultivating service brand equity. Journal of the Academy of Marketing Science 28 (1), 128–137.
Davis, D.F. Golicic, S.L. & Marquardt, A.J. 2008. Branding a B2B service: Does a brand differentiate a logistics service provider? Industrial Marketing Management 37 (2), 218–227.
Faircloth, J.B. 2005. Factors influencing nonprofit resource provider support decisions:
applying the brand equity concept to nonprofits. Journal of Marketing Theory & Practice 13 (3), 1–15.
French, A. & Smith, G. 2010. Measuring political brand equity: a consumer oriented approach. European Journal of Marketing 44 (3/4), 460–477.
Jedidi, K., Jagpal, H.S & DeSarbo, W.S. 1997. Finite-Mixture Structural Equation Models for Response-Based Segmentation and Unobserved Heterogeneity. Marketing Science 16 (1), 39–59.