Hierbij verklaar ik, Duane de Freitas, dat ik deze scriptie zelf geschreven heb en dat ik de volledige verantwoordelijkheid op me neem voor de inhoud ervan.
Ik bevestig dat de tekst en het werk dat in deze scriptie gepresenteerd wordt origineel is en dat ik geen gebruik heb gemaakt van andere bronnen dan die welke in de tekst en in de referenties worden genoemd.
De Faculteit Economie en Bedrijfskunde is alleen verantwoordelijk voor de begeleiding tot het inleveren van de scriptie, niet voor de inhoud.
A Freudian Business:
Probing the Effectiveness of Free Association Techniques
When Eliciting Controversial Brand Associations
Duane J. de Freitas, BSc.
Supervisors: drs. J. Labadie & drs. R. Pruppers
Amsterdam Business School
To my late mother, my father and my brothers,
Abstract. Free association is a popular research technique in the field of branding, and forms the basis
for a plethora of branding research methods, such as the Brand Concept Map (BCM). Confrontational, explicit techniques like free association however, require introspective self-analysis and truthful re-porting, exposing the technique to social desirability bias. This study investigates the effectiveness and validity of free association in the case of ethnically controversial brands. More specifically, through comparison of free association results with results obtained through implicit measures, investigated is under what circumstances cross-method heterogeneity might arise. The results show that in some cases, indeed, respondents might not be forthcoming in listing all of their controversial associations during free association. Also, communication to consumers can expose a brand to, or shield it from ethnic controversy. The implications for (international) managers, marketers and branding research-ers, especially for brands adopted by controversial ethnic minorities, are discussed.
TABLE OF CONTENT
Ch. 1: Introduction 1
§1.1 Brands and Associations 3
§1.2 Scientific contribution 4 §1.3 Managerial contribution 5 §1.4 Outline 5 §1.5 Delimitations 6 §1.6 Methodology 6 Ch. 2: Theoretical Framework 7
§2.1a Brands as Associative Networks 7
§2.1b Branding and Deeper Psyche 9
§2.1c Brand Sensitivity for Prototype Association 13
§2.2a Mapping Brand Associations 14
§2.2b Free Association to Answer Our Call 17
§2.2c Free Association and Controversy 19
§2.3a Alternatives to Free Association 22
§2.3b Dawn of the Implicit Test 23
§2.3c Free Association versus Implicit Tests 24
Ch. 3: Hypotheses 27
§3.1 Nationalistic Orientation and Ethnic Distinctiveness: Brands With Passports 27 §3.2 Birth of Controversial Associations: the Mediating Role of Controversial
Stereotyping 29
§3.3 Social Desirability: the Case of Free Association 30
§3.4 Dissociative Referencing: the Moderating Role of Reference Bases 31
§3.5 Exploring Brand Nationalistic Orientation 32
§3.6 Dissociative Referencing: the Role of Reference Base 33
Ch. 4: Methodology & Results 34
§4.1 Pretest for Study 1 34
§4.1.1 Participants 34 §4.1.2 Independent variable 35 §4.1.3 Dependent variable 35 §4.1.4 Procedure 35 §4.1.5 Results 36 §4.1.6 Conclusion 38 §4.2 Study 1 38 §4.2.1 Participants 38 §4.2.2 Independent variable 39 §4.2.3 Dependent variable 39
§4.2.4 Covariates and Mediator 40
§4.2.5 Procedure 41
§4.3 Pretest for Study 2 50 §4.3.1 Participants 50 §4.3.2 Independent variable 50 §4.3.3 Dependent variable 50 §4.3.4 Procedure 51 §4.3.5 Results 51 §4.3.6 Conclusion 55 §4.4 Study 2 56 §4.4.1 Participants 56 §4.4.2 Materials 57 §4.4.3 Procedure 58 §4.4.4 Results 59 §4.4.5 Conclusion 65 Ch. 5: Discussion 66 §5.1 Managerial implications 72 References 74 Appendix 78
A Freudian Business
1 1. Introduction
Different times require different measures. As the world around us trans-forms, so too must the tools for navigating that world. For most of the Western civiliza-tion, most notably Western-Europe and the United States, the turn of the century marks a momentous transformation: through suc-cessive bolts of globalization, what were once relatively homogenous nations with regards to ethnicity, culture, language and religion transformed into nations with a colorful flux. Just as social stratification slowly but surely met it’s demise halfway through the previous century, geographical segmentation of different cultures becomes an idea of the past and diversity is the new black. More than ever do cultures, lan-guages and ethnicities mix in countries such as Germany, Belgium, France and the Unit-ed Kingdom. The city of Amsterdam is home to people from more than 180 differ-ent countries (OIS, 2014), exemplary of the immigration trend characterizing the Neth-erlands which sparked in 2010 as close to 150,000 immigrants entered the country that year alone (Verkade, 2011). One year later, the United States hit a demographic mile-stone as well, as – for the first time in the history of the nation – more than half of the children born were born out of immigrant parents (FoxNews, 2012). The inflow of immigrants in the United States is expected to grow to a point where no single group could be considered a majority in 2040: Caucasians, former majority, will then make up 49 per cent of the population which ef-fectively alters the nation from United States to United Minorities. Indeed, people increasingly cross borders and countries
slowly form new, more heterogeneous iden-tities built on the backgrounds and identi-ties of all those living there. Critically exam-ining navigational tools international man-agers use to understand and manage in-creasingly heterogeneous societies, is the essential contribution of this paper.
To the international manager or market-er, this change provides opportunities and threats, since his potential customers are more heterogeneous in appearance, culture, rituals and language: that is, the “American wife” is no longer Caucasian by default and could very well speak Spanish as a first lan-guage, the “Dutch young man” is not typi-cally Christian or atheist but could very well be Islamic. Put differently, the surge in immigration has bridged the gap between the ‘national’ manager and the ‘internation-al’ manager: previously, the latter could be distinguished from the former since inter-national managers tackle intercultural chal-lenges and the national manager would not. However, today managers in Western socie-ties are all ‘international’ managers, consid-ering the diverse influx of cultures they have to deal with. While this increase in diversity is welcomed and cherished by many, it has been met with evenly passion-ate resistance, sometimes degenerating into aggression, xenophobia, discrimination and hatred. In countries such as the ones men-tioned, the immigration has led to a surge of societal problems and sometimes put groups into confrontational setting. The growth of nationalistic political parties in France (Front National), Belgium (Vlaams
Belang), UK (British National Party), Greece
(Golden Dawn) and the Netherlands (Partij
Voor de Vrijheid) for example mirrors the
differ-2
ent cultures, built on very different norms and values as the host countries: clearly, the process of mixing two groups of citizens has profound consequences that continue to shake the societal foundations. Oftentimes, immigrant families are overrepresented in the lower classes of society, with poverty, lack of education, criminality and thus dis-crimination as a key consequence. Case in point: in 2007, 62 percent of Europeans con-sider it all-round disadvantageous with regard to discrimination to be a person of different ethnic origin than the rest of the population (Eurobarometer, 2007). On the other hand, oppositional tendencies by the majority group are quickly interpreted as xenophobic, boorish or racist, hindering open communication and fueling frustra-tion. In short, although people from differ-ent ethnic backgrounds now live closer to each other with evermore intertwined lives, the intergroup differences are increasingly underscored with minorities dreading dis-crimination and majorities avoiding accusa-tions of such discrimination. It should come as no surprise that in 2007 Europeans con-sidered ethnic background the most com-mon source for discrimination and believed that the prevalence has increased in the preceding five years (Eurobarometer, 2007).
This has an effect on the importance of brands and branding. Brands have always been somewhat of an extension of identity providing a sense of community to it’s adopters (O’Reilly, 2005). When identities are under pressure and intergroup conflict intensifies, brand positioning may have greater significance as brands develop into symbols of (ethnic) groups, putting the brand at the front line of intergroup con-flict, much more today than fifty years ago.
As it has been shown that image transfer from celebrity to brand can modify brand awareness and sales (Vandeberg, 2014), presumably, the same logic applies to trans-fer of brand image from the (stereo)typical user of a product to the brand. That is, if a car brand (e.g. Opel) is mostly adopted by a homogenous group (e.g. older men), then the constant pairing of older men and Opel in the eyes of the consumer might lead to considering Opel a “car brand for old guys”, with consequences for sales. When clothing brand Lonsdale was adopted by a relatively small, but recognizable and ho-mogenous group of Neo-Nazi’s, it enticed an iconic association to what is to many a dissociative reference group. Case in point: in the city of Groningen, The Netherlands, club owners jointly decided to disallow access to guests wearing the brand (NRC, 2005), signifying the power of user groups and stereotyping. The same rationale ap-plies for brands adopted by homosexuals, blacks, Muslims or any other minority, albe-it that the link to those groups are more controversial.
The surge in immigration briefly de-scribed in the opening paragraph, com-bined with the human tendency to use brands as an extension of our identities, has two important ramifications: (1) ethnic groups use brands to identify and display themselves and (2) to identify and recog-nize other groups, to which they do not (wish to) belong. Marketing towards His-panics in the US inadvertently becomes ‘marketing away from non-Hispanics’ if there are strong conflicts between Hispanics and non-Hispanics: it eventually turns branding into a zero-sum game in which success in one segment means losing
cus-A Freudian Business
3 tomers in opposing segments. This is
exem-plified by brands such as FUBU (For Us, By
Us) for African-Americans and Tommy
Hil-figer for Caucasians. The founders of both firms have been quoted stating that they feel their clothes should be worn only by respectively black and white people (Man-ual-Logan, 2012): by ‘choosing sides’ with the existing intergroup conflict in the back-ground, the brand simultaneously increases popularity amongst some users and infuri-ates others, especially those in the group not targeted.
The increase in diversity – and thus so-cial turmoil – warrants our attention as the tools and models the (international) manag-er utilizes to map the world might assume more homogenous, non-confrontational groups and therefore become obsolete. A popular research technique to uncover con-sumer branding under examination here is termed free association. Oftentimes complex branding tools, such as the devising of Brand Concept Maps (BCM) (Roedder John, Loken, Kim, & Basu Monga, 2006) are built on free association: this technique of un-covering what is associated to a brand, comes down to exposing a research partici-pant to the name or logo of the brand (e.g. Nike) and requiring the participant to freely report any- and everything that comes to mind (e.g. ‘sport’; ‘active’; ‘summer’; ‘sweat’ etc.). The response is then analyzed and interpreted. While free association may have functioned in the past, it is proposed here that the adoption of brands by ethnic groups is controversial to such a great ex-tent – due to all the consternation sur-rounding the immigration flow of the past decades – that it subverts the outcome of the test and possibly renders it unreliable.
In summary, the problem statement tackled in this thesis is that confrontational brand mapping techniques such as free associa-tion, that require voicing personal beliefs, carry the burden of social desirability bias more today than in the past, and are there-fore of lower validity and usefulness when controversial brands associated to ethnic minorities are concerned.
1.1 Controversial Brands and Associations.
Following Keller’s article in ’93, scientific institutions and businesses alike have surged to investigate brand associations (Gawronski, 2009). According to Keller (1993), who builds heavily on Human Asso-ciative Memory theory, brands are nected in consumer minds to other con-cepts, such as experiences, other brands, properties, and so on. A brand’s existence in consumer mind then, is basically a net-work of concepts and associations in the brain. Activation of the brand (e.g. “McDonalds”) automatically activates a variety of other concepts (e.g. “Nice food” or “Bad for you”). According to Keller, brands that establish strong, favorable and unique links in consumer minds, have higher brand equity and are preferred by consumers.
The discovery of associative networks has led to a plethora of research into meas-uring, changing and strengthening brand associations. So-called projective techniques for example, are techniques where the brand is displayed (activating the brand node) and the respondent is required to express his or her associations (Koll, Von Wallpach & Kreuzer, 2010). Sometimes this takes the form of asking respondents to verbally list everything that comes to mind
4
(i.e. free association), sometimes they are required to tell stories or make impression-istic posters. Whatever the exact form of response, the idea is that these projective techniques do not limit the scope of the as-sociations, nor the type of association (e.g. semantic, visual, olfactory etc.) and there-fore provide a rich snapshot into consumer associations.
There are obstacles to the validity of projective techniques however: the re-spondent might not be capable of accurately reporting his or her associations. Even for trained psychologists it can be hard to ana-lyze and verbalize what they feel and why: the degree of introspection needed for the task is so high that results become unrelia-ble, so more rigorous approaches to test associations were invented. More recently invented methodologies circumvent the respondent’s abilities – or lack thereof – altogether. Through so-called implicit tech-niques like the Implicit Association Test (IAT) or the Lexical Decision Task (LDT), respondents’ cognitive and affective associ-ations are laid bare regardless of the inabil-ity to express associations (Greenwald, McGhee & Schwartz, 1998). Implicit tests – which usually involve computer programs in which respondents have to react as fast and accurate as possible to brands and test stimuli in a seemingly unrelated task – eliminate the need for respondent self-anal-ysis: associations influence behavior and the tests are designed to register this influence, thereby shortcutting the respondent. A downside of implicit testing however, is that the nature of the procedure requires fixed preconceived concepts, which makes uncovering obscure associations intricate and projective techniques still necessary.
1.2 Theoretical contribution
In this thesis, the suitability of free associa-tion in the case of associaassocia-tions between brands and controversial ethnic groups is investigated. It is expected that while free association is effective and intuitively justi-fiable for non-controversial associations, it is less reliable when respondents have – in their view – shameful or unwanted associa-tions. Branding research techniques should be appropriately armed against social de-sirability bias now that the prevalence of intertwining ethnic groups and the subse-quent controversy surrounding discrimina-tion has increased. The comparison be-tween the projective techniques and implicit testing has not been investigated exten-sively. Despite revealed shortcomings of re-spondent self-analysis, many popular asso-ciative techniques still require an ‘elicitation phase’ of which free association forms the starting point (e.g. Roedder John et al., 2006; Till, Baack & Waterman, 2011; Schnittka, Sattler & Zenker, 2012). Moreover, develop-ers of brand mapping methods – a process of visualizing brand association networks through maps – claim to cover almost all associations through extensive elicitation processes, apparently assuming that what respondents mention accumulates to an exhaustive list of everything that actually came to mind (Teichert & Schöntag, 2010). Perhaps, the romantic view of the Freud-instigated technique has the illusory ap-pearance of scrupulous, mind-delving psy-cho-analysis. Or perhaps the influence of these problems is expected to be small and insignificant to the research or no alterna-tive for this method is present.
Whatever the source of the popularity may be, the applicability of the free
associ-A Freudian Business
5 ation technique, perfected by Sigmund
Freud, is questioned and scrutinized as both free association and implicit measures are administered in conjunction with contro-versial associations, to investigate cross-method homogeneity. Association tests are like tools in the brand mapping toolbox: advancing the knowledge of which tool to use – or more specifically, not to use – under what circumstances, lies at the heart of the scientific contribution of this thesis.
1.3 Managerial contribution
Managers of firms and marketers of brands operating in ethnically diverse fields, espe-cially in international settings where brands are positioned across different cultural con-texts, can use the insight from this research to adapt their approach to branding. While the findings in this writing definitely do not apply to all brands in all settings, its merit may prove more important today and in the future, considering the trend towards mul-ticultural societies. Next to that, managers or marketers trying to improve the position-ing of problem brands might be eluded by results of branding research executed through free association. In that case, un-derstanding the full range of shortcomings of the technique can aid in selecting proper alternatives if the context so dictates. In short, since the effectiveness of branding research is increased, managers and mar-keters with concerns in that area might gain advantages from the insights produced here, especially in international setting.
1.4 Outline
In order to examine the con- or divergence of results produced by different methods, the theoretical framework of this thesis is
split up in building blocks. The first build-ing block examines the concepts of associa-tive networks in the brain and the connec-tion to brands. Many neurobranding as-sessment techniques – if not all – are built on these theories. Building block two zooms in on the uncovering of associations itself. Next to explicit techniques, the free associ-ation technique is elaborated on: the idea behind it, it’s protocol, the range of brand mapping procedures it provides the basis for and it’s merits and drawbacks. Basically answering the question of what free associ-ation is, this block also examines the more recent popularity of measuring the order of mentioned associations, the time it takes to mention associations and the “clumps” of associations mentioned together. Finally, this block also relates to the concept of brands and controversial associations. Es-pecially the relationship between contro-versy, free association and international vs. domestically orientated brands is covered, since international and thus more diverse brand images might be less vulnerable to adoption by ethnic groups. Ergo, the social desirability threat might be mediated through this factor.
Building block three provides a brief analysis of alternatives to explicit testing and free association, thereby introducing implicit tests. This section answers the ques-tion of what implicit testing entails, provid-ing the basis for the methodology used in this thesis as well as the comparison to free association to bring any shortcomings of the latter to light.
1.5 Delimitations
Importantly, my aim is not to investigate alternatives to free association techniques or
6
to provide solutions for any encountered problems, such as low validity in controver-sial contexts or low reliability. Following that reasoning, the implicit measures are administered to highlight discrepancies be-tween the two methods and any found in-consistencies could – but do not necessarily – underline the weaknesses of free associa-tion. Although this might bolster the case for implicit measures, it is not meant as an argument for the widespread application of implicit measures as an alternative form of testing, for those methods have shortcom-ings of their own and are not under evalua-tion here. Likewise, I do not strive to im-prove free association, even though the recognition of proper contexts for free asso-ciation is attenuated to. Next to that, the use of brands is limited to those to be consid-ered controversial on the basis of the brand’s history or it’s adoption
(brand-spe-cific controversy) and not on the basis of
con-troversy surrounding the nature of the product that would extend throughout the entire industry (industry-specific controversy), as would be the case with for example con-dom brands. Finally, while controversy comes in many different forms across dif-ferent cultures, controversy surrounding negatively stereotyped ethnic groups – and their associations to brands – is central in this thesis, because of the increased preva-lence of multicultural societies. Other forms of controversy are ignored, and the implicit assumption is that any found results could be generalized to other domains of contro-versy.
1.6 Methodology
In an effort to investigate the formation and existence of controversial associations – and
its effect on free association – this research is split up in four parts. Two parts function as pretests to test stimuli, and two studies test the hypotheses. During study 1, which is based on questionnaires, respondents will be exposed to fictitious brand advertise-ments to test the effect of communicated cultural attributes on the formation of con-troversy. In order to test, select and examine these stimuli, study 1 is preceded by a pre-test. Pretest for Study 1. Study 2 is an exper-iment in which respondents will again re-spond to brand advertisements in a free association task and in an implicit task, only this time with real brands, that should bring about controversy. To test which brands, couple to which ethnic minorities, bring about the most controversy, Pretest for Study 2 is executed, preceding study 2.
In short, in four waves of research re-spondents will help answer (1) what ficti-tious brands serve best in (2) investigating how controversy is formed and how com-panies can influence this. After that, (3) what real brands are coupled to which eth-nic minorities and finally (4) how does con-troversy affect the effectiveness of free asso-ciations for real brands.
A Freudian Business
7
2. Theoretical Framework 2.1a Brands as Associative Networks
The success of a brand comes down to the workings in the grid of networks and asso-ciations in the human mind. Keller (1993) contributed to this insight by elaborating on brand equity and brand knowledge. Brand equity can best be defined as the extra value that consumers place on products of a given brand, as compared to comparable un-branded products. The extra value of the product is mostly translated to an actual premium price, which turns brand equity into a measurable construct, interpretable as the brand-induced elevated willingness-to-pay (Cheng & Chen, 2001). The idea of cre-ating a successful brand revolves around the concept of increasing willingness-to-pay, ergo increasing brand equity, and Kel-ler claims that this equity hinges on the lat-ter concept of brand knowledge. He dif-ferentiates brand knowledge into brand
image and awareness. The former relates to
the attributes ascribed, benefits experienced by and attitudes held towards the brand by consumers. Brand image can be based on functional, product related properties – such as hiking shoes that have the image of durable, reliable and sturdy material (e.g.
The North Face) – but it could also be more
obscure and relate to functional, non-product related attributes, such as a ‘joyful’ (e.g. Fanta) or ‘sexy’ (e.g. Bacardi) brand image. Finally, brand awareness indicates the ease in which a brand comes to mind. Part of awareness is based on recall: when confronted with some form of cue, such as a product category or a need for a certain product, consumers activate a series of brands in their mind that is considered for
purchase. Because of bounded rationality and satisficing strategies, consumers mostly consider only a handful of brands in a small set called the consideration set (Romaniuk & Gailard, 2007). Brands that are easily elic-ited, have a head start on their competition since these brands are actually considered for purchase. It has been shown that enter-ing this set, especially beenter-ing the first brand mentioned – the so-called Top-of-Mind (ToM) brand – can be enough to lead to a purchase transcending existing negative other beliefs and attitudes towards the brand, most notably when elaboration is low (Romaniuk & Sharp, 2004). Awareness is more than recall alone however: recogni-tion, the process of recognizing a brand amidst a series of stimuli such as other brands in the shopping isle, too leads to higher awareness. When consumers see the famous arch-shaped golden M on a red background, the brand ‘McDonalds’ is pre-sumably easily recognized, even when con-fronted with many other advertising and non-advertising stimuli when driving for example. Recall and recognition combined contribute to brand awareness, recall most-ly when consumers consider purchases without brand presence and recognition while shopping in stores or browsing on the web, and brand awareness as well as brand image contribute to brand knowledge: the Holy Grail for increasing brand equity, ac-cording to Keller.
The manifestation of brand knowledge in the consumer mind mirrors the manifes-tation of any other concept in the human mind as suggested by Human Associative Memory theory. Basically, brands are thought of as central points in the brain, or ‘nodes’, just as how any other concept is
8
represented in the brain as a node. The links between these nodes, or ‘ties’, form a net-work. When one node is ‘activated’, be it consciously or unconsciously, the activation automatically spreads through the ties to other nodes that also become activated, al-beit to a lesser degree (Carrol, 2007). This process is called spreading activation: one thinks of Toyota and through the links in the brain the nodes “Prius”, “modern”, “holiday” or “Japan” might very well be activated as well. The reverse is also true: reading the word “Japan” might activate “Toyota”, although associations are not symmetrical. That is, the strength of the concept X – concept Y association is not necessarily identical to the strength of the concept Y – concept X association. Romaniuk and Sharp (2004) provide the insightful example that after thinking of McDonalds, ‘salad’ might very well be acti-vated but when thinking about ‘salads’ the chances of an activated McDonalds-node are smaller. The entire network of all the nodes activated by activation of the brand node is what is considered brand knowledge, and Keller stipulates that busi-nesses should aim for strengthening the ties between the brand node and positive con-cepts and avoiding ties to negative concon-cepts (1993). The stronger the ties are to these positive nodes and the more unique these ties are to the brand – and not to competing brands or the entire industry – the more valuable the brand knowledge.
Brand association management is a tricky but rewarding business. In the past decennium, the emphasis in modifying brand knowledge has shifted towards awareness, more specifically towards recall rather than recognition. Romaniuk and
Sharp (2004) have shown that brand
sali-ence, the sensitivity of the brand node for
activation very similar to brand awareness can explain a large share of the variance in market share and is essential in the process of purchase deliberation. As explained with consideration sets, a consumer might very well know and recognize the Daihatsu brand, but if this node is not salient and thus not activated when evaluating car of-ferings, it’s positive properties are never engaged in consumer minds and it’s pur-chase is never evaluated. This also explains why it is up to five times as expensive to acquire a new customer than to retain the old one (Palumbo & Herbig, 2000): product usage by the consumer inevitably leads to many new ties, strengthens and enlarges the network and thus the brand knowledge, which increases the likelihood of a strong and established position in the considera-tion set, more so than any ad alone could achieve. The old mantra that bad publicity is better than no publicity makes sense when the power of consideration sets are considered. Taken to an extreme, strong brand loyalty is the ultimate outcome of strong brand association management: the propensity towards repeated purchases that consumers experience.
It is easy to underestimate the strong effects these associations can have on con-sumer behavior. Repeated pairing of brand and positive concepts can eventually ele-vate a brand to ‘favorite’ brand status. Re-search has shown that when that happens, and an individual is confronted with the image of the favorite brand, the reward cir-cuit in the brain is strongly activated enforc-ing the release of dopamine just as how this is activated when confronted with a favorite
A Freudian Business
9 person – such as a loved one – or a favorite
place, such as the home (Schaefer & Rotte, 2006). But also less successful yet positively viewed brands can elicit a positive feeling in consumers, very similar to other, more abstract concepts can. Brand association management assimilates brands into associ-ative networks and the brands are no longer just brands: they become a part of us, our identity and of our culture (O’Reilly, 2005).
2.1b Branding and Deeper Psyche
The preference for brands is more complex than the above however: Keller’s contribu-tion only superficially examines the psy-chology behind branding. Why consumers seek out some brands more than others, regardless of their awareness and positive image, remains unresolved in his contribu-tion. Jillz and Budweiser are both well-known, well-recalled, well-recognized brands that presumably transfer both posi-tive affect and posiposi-tive product information through highly successful marketing efforts. Yet men and women do not favor these brands indiscriminately: their self concepts steer them towards certain brand images and away from others. The associative net-work theories and priming alone fail to ex-plain this, since a more social constructivist view is required to explain the selectivity consumers have for brands. Combined with social psychology however, a powerful the-ory emerges that explains and predicts this behavior. That is, human self-identity can also be thought of as a node in the network of nodes in the mind (Willingham, 2007). The self is an important node and connect-ed to other nodes with ties just like brands. Humans are social animals, so their self-concept and the ties to nodes are not
formed in isolation: our identities are for a large part based upon the identities of the groups to which we belong (Brewer & Brown, 1998). These groups in turn also represent nodes (or groups of nodes) in the mind and associating ourselves with these groups eventually leads to assimilation of ties and nodes previously linked to the group. ‘Being’ a Boston Red Sox fan basi-cally comes down to belonging to the group of Boston Red Sox fans and steering our self concept in that direction. By mimicking the archetypical behavior of the typical repre-sentative of the group – termed the prototype – the self-concept becomes connected to the nodes also connected to the group. If those nodes involve clothing (e.g. Boston Red Sox fan outfit) or behaviors (e.g. watching Bos-ton Red Sox matches), than the individual is inclined towards these concepts. This ulti-mately also holds for brands adopted by the group. People constantly enter this process of matching their identity to that of the pro-totype, a process coined prototype matching (Edson Escalas & Bettman, 2003).
It is important to note the difference between prototype matching and basic
priming. In priming, positive affect is
trans-ferred from an object to a brand: Kim, Ling and Barghava (1998) have shown that this can be effective for any positive image, re-gardless of a link to the brand or even in-dustry. In a clever set-up, the authors suc-cessfully transferred positive affect from an image of a cat – which does not communi-cate any product-related information – to a pizza brand, leaving the respondents more positive towards the brand after pairing. Basic priming with celebrities follows the same logic. The positive affect brought about by the ‘familiar’ famous face of
celeb-10
rities (e.g. Michael Jackson) is transferred to the brand (e.g. Pepsi) through ambassador-ship, even in the absence of a logical link to the product.
However, in some cases celebrities are not picked at random but chosen for what they represent in the mind, that is, for what the celebrity-node is tied to in the brain. Marketers hope that associations tied to the celebrity in such case, eventually connect to the brand: the celebrity is displayed as a user or an advocate of the brand and image from the celebrity is transferred to the brand. On a neurological level, after repeat-ed pairing, a direct tie between the brand node (e.g. Gilette) and a property node (e.g. excellence) is established by circumventing the linking node of the athlete (e.g. Roger Federer), a process very similar to Pavlovian conditioning that is built on the presumption that neurons that fire together,
wire together: the basal premise of Hebbian
learning (Willingham, 2007). Other well-known association attempts, involve the sponsoring of popular sport teams (e.g. the Chicago Bulls and Budweiser) and the spon-soring of events (e.g. Olympic Games and
Coca-Cola).
Prototype matching runs on a deeper level as the concept, and thus the self-concept node, is involved in this type of branding. There are different groups that are involved in the process of prototype matching. Reference groups are groups to which we (think we) belong and consider ourselves members. From time to time, in-dividuals feel the need to reaffirm their membership of these reference groups, a natural tendency usually referred to as a need for self-verification. Aspiration groups on the other hand are groups to which the
in-dividual wishes to belong (Edson Escalas & Bettman, 2003). Sometimes the prototypes of these groups are actual individuals (e.g. Lionel Messi in soccer) but the prototype needn’t be an actual individual by neces-sity: rather, it is the abstract mental projec-tion (also called schema, Willingham, 2007) of the representative of this group. Associa-tions to this prototype the individual wish-es to connect to his self-concept, which is why celebrity endorsement is so powerful if the chosen celebrities are firmly associated with prototypes (Edson Escalas & Bettman, 2003). The cues mentioned before (as dis-cussed by Chen & Cheng, 2001) also play a role in activating the self-concept. Depend-ing on the situation, the individual might wish to display a certain identity, which is in fact a group of associations over the oth-er, following different urges for self-veri-fication. While attending a plenary lecture at college (cue), the student who speaks to fellow students (cue) might wish to bolster his image of intellectuality, that is, reaf-firming the tie between the self-node and the “intellectual”-node, creating a prece-dence towards the brands used by his pro-totype of “intellectuals”: next to changing his tone of voice and clothing for that situa-tion, he might opt for an Apple computer considering the popularity of the brand amongst higher education students. When that same student during the break of said lecture is hungry (cue) and decides to eat in the cafeteria (cue) the need for reaffirming his sense of health-consciousness might steer him away from brands known for their high-calorie or non-organic foods and towards brands associated with the proto-typical health-conscious individual. When going out with friends (cue) later that night
A Freudian Business
11
to a club (cue), the same individual might unconsciously reaffirm his ties to ‘hip indi-viduals’ and experience an inclination to-wards those brands he perceives as associ-ated with the prototypical hip person (e.g.
Bacardi). Identity then, comes down to a
series of memberships to reference and as-piration groups, that we constantly and actively, yet mostly unconsciously, wish to reaffirm by our behaviors. Branding, seen from a social constructivist perspective, is the act of altering the prototype in the minds of the consumer, by creating an association between brand-node and prototype-node and human social nature steering our be-havior does the rest. As said, the forces driving these behaviors can be strong, as exemplified by individuals who go to great
lengths to associate themselves with brands with for example tattoos of the brand (e.g.
Harley Davidson- or Apple-tattoos). When
brands are seen as the ideal way to establish an aspired connection to the self-concept and a group, very successful branding oc-curs. See Figure 1 for a visualization of the process of prototype matching.
In the aforementioned research by Schaefer and Rotte (2006) into activation of the reward circuit of the brain after confron-tation with favorite brands, brands typically labeled as favorite by the participants were brands such as Lambourghini and Ferrari. Most, if not all of the participants have nev-er actually came into contact with the brands by owning or driving a Lambourghini or Ferrari (with the exception of small mer-Figure 1: A visualization of the process of prototype matching, naturally steering the self-concept towards and away from attractive or unattractive prototypes. Firms influence this prototype in multiple ways, by repeatedly pairing the display of the brand to important nodes. NB.: This sche-matic representation is an example and not exhaustive.
12
chandizing products), nevertheless this brand is still somehow vital to the individ-ual’s self-concept. Apparently, the nodes of these car brands are associated to the pre-ferred prototypes (e.g. wealthy or success-ful prototypes) through the characteristics of the user group. What’s more, these brands hardly do any marketing as com-pared to other car manufacturers, which further exemplifies the power of aspiration groups, and the deeper psyche of branding.
The reverse is also true, however: while Edson Escalas and Bettman speak of refer-ence and aspiration groups as the two major players (2003), White (2006) suggests that there is an important third group.
Dissocia-tive reference groups are groups to which the
individual does not wish to belong or be associated. Ties to this group, i.e. ties to the node of the prototype of that group and it’s associations, are strongly avoided in much the same way as ties to aspiration groups are sought after. The concept of dissociative reference groups is relatively new in brand-ing (first research published on the matter in branding stems from 2006) but effective in explaining the avoidance consumers ex-perience towards certain brands. The force driving this avoidance is apt to be stronger than the force towards aspired groups. That is, in White’s research both males and fe-males were instructed to make a selection between a smaller and a larger steak for dinner. In one condition the smaller steak was called the ‘Chef’s cut’ whereas in the other it was called ‘Ladies’ cut’. In the Chef’s cut-condition, 47 percent of the males opted for the small steak and 53 per-cent for the larger one. In the Ladies’ cut-condition however, only 5,3 percent of the males opted for the smaller steak and
pref-erence for the larger steak grew to a whop-ping 94,7 percent. While females ex-perienced a small and insignificant prefer-ence for the Ladies’ cut between conditions, male respondents apparently experienced a much stronger aversion to and avoidance of the Ladies’ cut. The concept of asymmetric
dominance, that revolves around
asymmetri-cally designing offering favorability to em-phasize attractiveness of certain options over others, is well documented: usually pricing measures are used to influence con-sumer preference (see Ariely, 2013 for a classic demonstration on holiday prefer-ences). The promising news is that it ap-pears that dominated-choice effects are also attainable by less drastic measures as price cuts by simply altering the name of an al-ternative: after all, the best way to sell big steaks is ostensibly to unpleasantly term smaller ones, and utilize dissociative refer-encing as a tool.
White (2006) explains that the avoidance of associations to dissociative reference groups is strongest when choosing brands in public and higher for self-conscious indi-viduals. Cues play a role here as well: when out for “a drink with the guys” (cue), males presumably display a far greater aversion towards brands positioned as feminine. This has far-reaching consequences for the concept of brand loyalty as elaborated on by Palumbo and Herbig (2000). Individuals might indeed experience a strong motiva-tion for repeated purchases on the basis of brand awareness and image, but this could also be because they experience an aversion towards poorly positioned competing brands. In other words, the preference for McDonalds might indeed stem from suc-cessful branding, but it could also be that
A Freudian Business
13 individuals do not wish to be seen as
“peo-ple who go to KFC”. Especially if that group is homogeneous, easily recognizable and a dissociative reference group, for ho-mogeneity and distinctiveness provide a broader base for referencing (Lambert, Bar-ton, Lickel & Wells, 1998; Brewer & Brown, 1998).
The concept of prototype matching, dis-sociative and (asdis-sociative) referencing as well as marketing attempts to connect the brand to the prototype or to a celebrity con-nected to the prototype, is visualized in Figure 1. This visualization is by no means exhaustive: there are far more ways in which firms can influence prototype match-ing. The image simply displays the dis-cussed.
2.1c Brand Sensitivity for Prototype Association
An important question arises when consid-ering the previous: are some brands more susceptible for associations to a prototype than others? Some brands are loosely nected to a large group of unrelated con-cepts, which poses the problem of brand
dilution: Dutch brand V&D provides a clear
example of such an incoherent ‘all and nothing’ brand (Haveman & Labadie, 2003). Other brands however, especially those with clear, homogeneous and restricted target and user groups are successfully connected to less but more important and more coherent concepts, facilitating the pro-cess of prototype matching (e.g. Amstel, Dutch daughter brand of Heineken).
As said, brand association management is tricky considering user groups: brands trapped in unfavorable positions – such as
Opel (“old men”) and Bacardi Breezer
(“pro-miscuous teenage girls”) – all have or had a
hard time disentangling the connection to the adopting group (Haveman & Labadie, 2003), which might be why some firms take drastic measures to avoid this. Clothing brand Abercrombie & Fitch created a stir when it became known that the firm refuses to create plus-size clothing and destroys old or damaged clothing rather than donating it to charity, in an effort to avoid the display of obese or poor people in Abercrombie &
Fitch wear leading to dissociative references
(Zimmer, 2013). Likewise, it was rumored that the pricing strategy of clothing giant
Tommy Hilfiger was aimed at deterring
Afri-can-Americans – who tend to populate the poorer segments of the American society – from buying and wearing the clothing to allegedly conserve the image of a white and prestige brand (Manuel-Logan, 2012). Whether this is true remains to be seen and whether this is just is beyond the scope of this thesis. These examples are however, from a purely strategic marketing per-spective a logical and effective step: is not unthinkable that firms, today or tomorrow, secretly seek to avoid certain customers because of dissociative referencing, those who are overweight or poor for example, by restricting access to the brand.
The international, national, or even chauvinistic nature of the brand might play a role in this. Some brands actively strive for associations with a certain country. Con-sider the airline industry: Lufthansa, the German airline, altered it’s services, it’s presentation and allied with strategically sought out German suppliers, such as
Sennheiser (a well known German
audio-electronics manufacturer) for it’s produc-tion of onboard electronic equipment to strengthen it’s position and image as a
14
German firm. This is not just true for
Lufthansa: it seems that linking a brand to a
nation – and thus to the nation’s prototype – is popular for a multitude of firms in the airline industry, since many competitors follow the same strategy (e.g. Air France or
United Arabic Emirates) in designing their
services and appearance. This fact is further exemplified by the choice in brand names alone (e.g. British Airways; South Africa
Air-ways etc.). However, other competitors in
that same industry – such as Vueling or
RyanAir – do not aim for this nationalistic
connection. Since it is harder to tie Vueling to a certain group of consumers, the proto-type of the Vueling-flyer might be more in-coherent. The result could be that it’s per-ceived user groups might display higher variability, hindering judgment and thus dissociative referencing (Lambert et al., 1998). That is, if you are Dutch, and do not wish to be associated to Germans, you might avoid Lufthansa, but the avoidance of what group could lead to the avoidance of
Vueling? The positioning of the brand is less
specific which might entail weaker associa-tions and lower brand equity, but it might also encapsulate the brand from the risk of being associated to a group of people and hinder dissociative referencing. Obviously, the discrepancy in sensitivity for dissocia-tive referencing might also exist between two purely domestic brands that differ in coherence: for example, it might be easier to be deterred from the prototypical Amstel-drinker as compared to the prototypical
V&D-shopper. While the nationalistic
na-ture of brands is not core subject matter in this thesis, it’s possible effect on branding and branding research in the context of dis-sociative referencing is explored and it is
included as a variable in this research.
2.2a Mapping Brand Associations
Both research into branding from a more theoretical standpoint executed by scholars, as more practically orientated efforts as a part of marketing research by firms are common today. The aim is to uncover which associations we have, which we don’t have, which associations are strong or weak, how favorable and how unique these associations are and – most importantly – how all of this relates to consumer buying behavior measures. Branding networks and it’s elements of nodes and ties are not di-rectly observable however: while research-ers effectively and objectively monitor and manipulate what stimuli respondents are exposed to as well as the effect on their pur-chasing behavior, the mind itself is ob-served only indirectly. Some models cir-cumvent this problem by ignoring the in-fluence of the mind altogether – the so-called black box models – most notably in the time when behaviorism was popular in psychology. Other, more modern models however, have tried opening the black box through workarounds and logical leaps of mind. It is no wonder then that branding research methods are rooted in psychology and sociology, where the scientific conun-drum of investigating the unobservable, yet influential, human psyche has led to crea-tive solutions. Complex branding concepts such as consumer dissociative referencing therefore strongly relate to well-established psychological/sociological concepts such as the in-group/out-group dichotomy (Brewer & Brown, 2007).
Roedder John and his colleagues (2006) argue that one of the most powerful tools to
A Freudian Business
15 visualize inner brain associations through
branding, is by construing what is called a Brand Concept Map (BCM). This comes down to creating a visual schema – much like a spider web – where the brand and all its relevant associations are shown as con-nected nodes to the brand (see Roedder John et al., 2006, p. 1. for a powerful exam-ple). Not only does the BCM show associa-tions, it also displays many other properties of consumer psyche presumed important, such as the tie strength between nodes as the thickness of the lines, network density as a variable, cut points in order to group associations, importance of associations through the size of the node and many more (see Teichert & Schöntag, 2010 p. 377 for an extensive overview). The process of generating a BCM is intensive, and starts with a phase of consumer mapping, which basically entails uncovering and visualizing all the relevant associations – and links – the respondent has. According to Roedder John et al., this process should (1) start with an elicitation phase in which all relevant as-sociations are uncovered. After this, (2) the
linking phase initiates, to construe a
matrix-like map, by investigating which associa-tions link to each other. Finally, the con-sumer mapping procedure ends with (3) the
aggregating phase: the process of combining
individual respondent inputs into one ag-gregated brand map, by selecting the most important nodes and ties, usually indicated by frequency of mention and frequency of first-order mention. With this input the con-sumer mapping component is followed by the analytical component, which is used to analyze the network. Together, consumer mapping and analysis lead up to the BCM, that businesses and scientists alike use to
predict consumer behavior.
The techniques for eliciting associations – the starting point for the BCM – are con-troversial and have evolved over the years. The prime research methods to uncover associations fall within the realm of explicit
testing and self-rapports. Explicit tests are
usually survey-like tests or interviews that generally involve showing the respondent the brand and subsequently asking the re-spondent to indicate on a 1-5 or 1-7 Likert-scale basis to what extent he or she feels the brand corresponds to a given association. In other cases, more in-depth forms of explicit testing involve asking the respondent fol-low-up questions to reveal whether any associations are favorable and to what ex-tent they influence behavior. A researcher interested in Audi for example could show the Audi-logo or an Audi-ad and conse-quently ask ‘to what extent do you consider
Audi to be … adventurous?’, followed by
‘how important is being adventurous to you?’. Explicit tests are termed explicit since respondents are required to make thought processes explicit through verbalized out-put. The associations under investigation are part of a preconceived set of associa-tions, a list usually generated by the re-searcher or the interested firm (in this ex-ample Audi) and usually following or pre-ceding marketing efforts, such as a market-ing campaign or a change in brand logo. The produced data of research along these lines are easy to communicate and defend, which is possibly a driver behind it’s popu-larity. Since the answer space to explicit tests is restricted (e.g. Likert-scale or multi-ple choice), the research can be made highly specific, reliable and – presumably – useful for evaluating marketing efforts or
justify-16
ing budgeting decisions.
The shortcomings of explicit testing have gradually surfaced however. Firstly, since the answer options are restricted and the set of associations is preconceived, it is impossible to uncover any unsuspected associations. When the goal is to test for a specific association (e.g. only interested in adventurousness), this would not be prob-lematic, but in drawing up brand maps or during more general research into the posi-tioning of a given brand, explicit methods could fail to uncover that which is truly important. Next to that, probing questions that form the basis of the test are inherently suggestive and lead to framing (Vandeberg, 2014). That is, the consumer might not con-sider Audi adventurous at all, but – now that the researcher mentions it – it in fact is more or less adventurous. In a recent pub-lishing, Vandeberg insightfully lists these problematic properties of explicit tests (and others) with regard to the validity of the techniques. More than alternatives, the ex-plicit tests risk false positives, and Vandeberg catalogs eight major scientific pitfalls common to the technique, of which the three most relevant to this thesis will be discussed here.
Explicit testing requires conscious intro-spective access of the respondent into his or her mind. The basal premise is that re-spondents are capable of analyzing their own impulses and evenly capable of ver-balizing the associations they have. Intro-spective access however, is limited and dif-ficult even amongst those that have studied the matter, let alone for the regular re-spondent. The human mind is built on a multitude of biases blurring our introspec-tive insight and coloring our model of the
world, rendering the respondent input in-herently unreliable. Explicit tests are better suited for investigating what respondents think they have for associations, but not for what they actually have. Many cognitive psychologists – most notably Loftus (e.g. Loftus, 2003) – have exposed and examined the malleability of human memory: even the most powerful, seemingly objective processes, such as memory on the basis of eye-witnessing, can be molded through framing and serve to elude respondents. Asking consumers to analyze concepts much more complex and obscure than what they have seen with their own eyes is even more sensitive to invalid or unreliable an-swers. Secondly, explicit tests, and espe-cially self-rapports, hint at the goal of the research. When asked by (1) a marketing researcher of Sony to (2) view a Sony-com-mercial and to (3) state to what extent Sony could be considered ‘modern’ and (4) to what extent modernity is important, the respondent imaginably is aware of the con-cepts the researcher is interested in, espe-cially if respondent groups are comprised of scientific students, as is often the case. This knowledge potentially biases the results as respondents have the natural tendency to please researchers, by answering in the manner perceived as preferred or normal. Finally, most scholars recognize that the favorability of an association, if present, is of great importance (Schnittka, Sattler & Zenker, 2006). In explicit testing, probing favorability mostly takes the form of a pos-itive-negative dichotomy, which forces re-spondents to extremes and does not capture the richness of the favorability spectrum. The downside is a significant loss of valid-ity as many associations are not entirely
A Freudian Business
17 positive or negative, and fall in the grey
area between the extremes (Vandeberg, 2014).
2.2b Free Association to Answer Our Call
To circumvent the aforementioned method-ological issues, many researchers have di-verted to an elicitation technique called free
association. Rooted in Freud’s
psychody-namic school of thought in psychology, free association comes down to providing spondents a cue upon which they are re-quired to verbally list any- and everything that comes the mind. The answer space is in no way restricted (there are no ‘right’ or ‘wrongs’) and all the output produced by the respondent is gathered. The technique is credited to Freud who used it in his notori-ous analysis of the unconscinotori-ous (i.e. das Ich
und das Es). Freud reasoned that when
pa-tients are allowed to let their minds wander freely and censure-free report whatever comes to mind, a truer display of (re-pressed) urges and feelings emerge (Kelman, 1960). This concept has been gen-erally debunked (Brunel, Tietje & Green-wald, 2003), but free association is still con-sidered to be a powerful tool of unrestricted access to relevant associations, repressed or not, and seen as a silver bullet for uncover-ing ‘true’ associations.
The result of a free association session is usually a list of diverse, sometimes seem-ingly unrelated concepts mentioned, which serves as a basis for analysis. Because there are no probing questions other than the brand itself the effect of framing is smaller, as is the obviousness of the research intent. Since respondents have nothing to go on, the chances of a more accurate list of acti-vated associations is considered higher.
Moreover, due to unrestricted answer space, free association rose as the quintes-sential strategy for uncovering unexpected or unknown associations to the brand. Con-sider a fictional example: Image clothing manufacturer X has serious trouble selling clothing Y, even though several investments have been made to improve the quality and effort has been put in increasing efficiency and thus decreasing the price. For some reason, target group Z seems reluctant to buy clothing Y. Unique selling point to clothing Y is it’s image of hip and mo-dernity, following profound marketing ef-forts: a goal well achieved considering the convincing feedback on explicit tests sub-jecting respondents to questions such as ‘on a scale of 1 to 7, to what extent is clothing Y ‘hip’?’. Averaging a 5,5 score and even en-dorsed by ‘hip’ celebrities, it seems a mys-tery why clothing Y won’t sell. So why not? Possibly, the brand is associated to much more than hipness alone: it could be associated to a dissociative reference group outside of the firm’s control. Suppose that clothing Y – while targeting target group Z – has been adopted extensively by young consumers (group Z’). While trying to ap-pear hip on Instagram, the efforts of this user group backfires however and many other consumers – including targeted group Z – now consider the brand a brand for teenagers who want to be hip, but aren’t (i.e.
wannabees), a group which is dissociative to
most. Explicit tests might initially miss im-portant associations like this, since they require that someone within the marketing department suspects this association and the answer space does not allow for elabo-rating on other associations than hip and modernity. Free association in this case
18
could have signaled this issue both easier and sooner: when confronted with the brand image of clothing Y, some respond-ents might exclaim “wannabees; selfie;
Instagram” of which post-hoc analysis
re-veals that the brand is hip, but that user group Z’ could be considered wannabees, deterring target group Z: a new and power-ful insight retrieved.
Larger answer spaces inevitably incur higher complexity and more statistic obsta-cles when it comes to analysis. Free associ-ation techniques build heavily on Human
Associative Network theories, using the
con-cepts of nodes, ties and spreading activation as anchor points to logically distill ‘rules’ that must hold true in order to interpret responses. For example, spreading ac-tivation predicts that nodes that have the strongest connection to the node activated, will be activated first. This has led to the rule that the order of produced responses correspond to the strength of associations, making free association researchers assign special value to first mentioned associations for example. Next to that, the theory pos-tulates that activation travels from node to node, which means that the order in which associations are activated, and thus men-tioned, provides insight into the grouping of concepts in the brain. When the associa-tion ‘annoying’ is menassocia-tioned directly after “Drive-through’ when investigating
McDonalds brand, it is presumably different
from mentioning ‘annoying’ directly after the association ‘Kids’: the former might signal associations of long waiting times at the drive-through, whereas the latter might hint at negative experiences with children at the restaurant (Teichert & Schöntag, 2010). Moreover, reaction time in
associa-tions provides another cue to association strength: the associative model predicts that stronger activation results in more promi-nent ‘popping-up’ of concepts. Associations that surface after lengthy, elaborate pro-cessing (long reaction time) are therefore presumed to be weaker than easily ex-claimed concepts (short reaction time). There are many more important and pre-dictive variables that are entered in statisti-cal models and analyzed.
Some advocates of free association are particularly convinced of the power of the technique and the ease of administering the test. In the words of BCM-specialists Roedder John and colleagues:
Second, data used to identify salient asso-ciations should be based on consumer re-sponses to open-ended questions (e.g., “when you think of [brand], what comes to mind?”). Open-ended questions allow consumers to voice whatever brand asso-ciations are most accessible and important to them in their own words (Roedder John et al., 2006, p. 552).
Another form of free association is Zaltman’s Metaphor Elicitation Technique (ZMET) for example (Zalman & Coulter, 1995), which involves a very thorough analysis of respondents by instructing par-ticipants to send researchers up to 12 pho-tos that convey their thoughts and feelings on the brand, followed by a two-hour in-depth interview. Free association is not nec-essarily as extensive, but the essence re-mains the same: provide respondents with a small cue such as a brand name, logo or advertisement and they will respond with an exhaustive list of associations, ranging
A Freudian Business
19 from superficial to deep, from obvious to
obscure, which the capable researcher can then combine with HAM-theories to make sense of it all.
2.2c Free Association and Controversy
The power and utility of free association when it comes to easily elicited, easily ver-balized associations that are non-threaten-ing to the individual is not disputed here. It would seem reasonable to suspect that he who associates Nike with ‘sportive excel-lence’ would actually produce this response during a session. But let us revisit the ex-ample of retailer X, with clothing Y targeted at target group Z, that had trouble selling because of user group Z’. What if the disso-ciative reference group that adopted the brand, i.e. user group Z’, were not
wannabees but a group of blacks? Possible
problems of social desirability arise when respondents are requested to list associa-tions like that: a problem that is well recog-nized as a shortcoming of explicit testing in general. Controversy – or taboo – is related to shame and fear: when having an associa-tion is considered shameful or a respondent experiences fear of social disapproval for linking two concepts together. Even though free association is still an explicit test (it requires respondents to make their associa-tions explicit) it is somehow proposed free from social desirability concerns, even for controversial associations, considering the widespread neglect of this possibly influ-ential variable. More specifically, free asso-ciation in combination with controversial associations might lead respondents to (1) not mentioning associations, (2) covertly mentioning associations while avoiding the real association (e.g. stating ‘gangsters’
while meaning ‘blacks’), (3) reluctantly mentioning associations later than they sur-faced and many other possible sources of interference. The consequences of this are far reaching for – for example – the use of brand concept maps in which the elicitation phase of free association is followed by se-lecting ‘the 25 most important associations’. When neglect of associations altogether or reluctance in mentioning due to contro-versy is muddling the waters, the whole idea of selecting the most important associ-ations becomes unreliable, as is the act of measuring response time or listing men-tioning order.
The possibly interfering role of contro-versy in the examples above are evident in situations when the respondent is aware of an existing, but shame-inducing association. Even more likely is the situation in which the respondent has a negative association to the brand but is unaware of the priming power of this association. Therefore, even the most anti-racist might associate blacks with poverty and be primed to associate poverty to a brand widely adopted by blacks in a certain area. When subjected to free association this person is likely to state ‘poverty’ as a relevant association, but may never realize that this association is medi-ated through ‘blacks’ since he consciously rejects and suppresses the idea of being a racist. Pity the firm that on the basis of the results through free association invests heavily in enhancing this brand’s image to a more luxurious status, only to find that sales remain low among the targeted group. This shortcoming of free association is very similar to the shortcomings of limited intro-spective access mentioned by Vandeberg (2014). The morale of it all is that
