28
JOURNAL OF CUSTOMER BEHAVIOUR, 2013, Vol. 12, No. 1, pp.25-51 http://dx.doi.org/10.1362/147539213X13645550618489 ISSN1475-3928 print /ISSN1477-6421 online © Westburn Publishers Ltd. An associative learning account of branding effects of sponsorship Luke Lunhua Mao, University of Florida, USA* James J. Zhang, University of Georgia, USA Daniel P. Connaughton, University of Florida, USA Stephen Holland, University of Florida, USA John O. Spengler, University of Florida, USA Abstract For sponsors, the essence of commercial sponsorship is the right of being associated with the sponsored organisation, which can later be leveraged for branding purposes. The branding power of sponsorship relies on its associative power, and consumers learn sponsorship in two qualitatively distinctive ways: evaluative conditioning and predictive learning. These two processes can lead to different branding outcomes (e.g., a decrease in brand loyalty, but an increase in perceived quality). This study stresses the limitations of traditional theoretical accounts in explaining branding effects of sponsorship and proposes that associative learning would be the fundamental branding mechanism of sponsorship marketing. Specifically, this study examines the multivariate relationships between a set of learning variables (i.e., sport involvement, event involvement, event attitude, emotional experience, brand knowledge, and brand-event relatedness) and a set of branding effect variables (i.e., perceived quality, attitudinal loyalty, and behavioural intention). Through a series of canonical correlation analyses on a survey data, we found evidence to support the associative learning account of branding effects. Keywords Associative learning, Evaluative conditioning, Predictive learning, Sponsorship, Branding *Correspondence details and biographies for the authors are located at the end of the article. JOURNAL OF CUSTOMER BEHAVIOUR

JOURNAL OF CUSTOMER BEHAVIOUR...intangibles as meaningful words, nonsense syllables, music, pleasant scenes, and celebrity. Furthermore, the nature of the relationship between conditioned

  • Upload
    others

  • View
    7

  • Download
    0

Embed Size (px)

Citation preview

JOURNAL OF CUSTOMER BEHAVIOUR, 2013, Vol. 12, No. 1, pp.25-51http://dx.doi.org/10.1362/147539213X13645550618489ISSN1475-3928 print /ISSN1477-6421 online © Westburn Publishers Ltd.

An associative learning account of branding effects of sponsorship

Luke Lunhua Mao, University of Florida, USA*James J. Zhang, University of Georgia, USADaniel P. Connaughton, University of Florida, USAStephen Holland, University of Florida, USAJohn O. Spengler, University of Florida, USA

Abstract For sponsors, the essence of commercial sponsorship is the right of being associated with the sponsored organisation, which can later be leveraged for branding purposes. The branding power of sponsorship relies on its associative power, and consumers learn sponsorship in two qualitatively distinctive ways: evaluative conditioning and predictive learning. These two processes can lead to different branding outcomes (e.g., a decrease in brand loyalty, but an increase in perceived quality). This study stresses the limitations of traditional theoretical accounts in explaining branding effects of sponsorship and proposes that associative learning would be the fundamental branding mechanism of sponsorship marketing. Specifically, this study examines the multivariate relationships between a set of learning variables (i.e., sport involvement, event involvement, event attitude, emotional experience, brand knowledge, and brand-event relatedness) and a set of branding effect variables (i.e., perceived quality, attitudinal loyalty, and behavioural intention). Through a series of canonical correlation analyses on a survey data, we found evidence to support the associative learning account of branding effects.

Keywords Associative learning, Evaluative conditioning, Predictive learning, Sponsorship, Branding

*Correspondence details and biographies for the authors are located at the end of the article.

JOURNAL OF

CUSTOMERBEHAVIOUR

INTRODUCTION

Corporations spend millions of dollars acquiring sponsorship rights, and planning and implementing sponsorship marketing activities. In the United States alone, 96 corporations spent more than $15 million on sponsorship in 2006 (IEG, 2008). Collectively, twelve of The Olympic Partners (TOP) of the Beijing Olympic Games contributed $866 million in cash, goods, and/or services for the 2006 and 2008 Olympic Games (IOC, 2009). The average contribution for each TOP sponsor exceeded $60 million. Even in a time of economic downturn, eleven companies spent a total of $957 million to secure the TOP sponsorship for the 2009-2012 cycle, ending with the 2012 London Olympic Games (IOC, 2012). The average contribution for each TOP sponsor was about $87 million. In most cases, sponsors garner some rights of being associated with the sponsored organisation in return for their contributions. The prime objective of sponsorship marketing is to build and communicate an association to a sponsorship. It is an underlying assumption that the mere fact of “being associated” has commercial value and can be exploited for corporate or marketing objectives. For instance, as an official sponsor of the Beijing Organising Committee for the Olympic Games, the Chinese dairy product brand, Yili, was allowed to print certain proprietary symbols of the 2008 Olympic Games on the packages of its products along with its own brand name. It is believed that such an alliance would have an impact on consumers’ perceptions and attitudes towards its branded products and activate their choice decision towards the sponsor’s brand (Geylani, Inman & Hofstede, 2008; Simonin & Ruth, 1998). Although corporate marketing managers perceive sponsorship as an effective vehicle to enhance elements of brand equity, including brand awareness, brand associations, perceived quality, and brand loyalty (Cornwell, Roy & Steinard, 2001; Henseler, Wilson, Götz & Hautvast, 2007), the evaluation of its effectiveness remains a central topic in sponsorship research.

In this study, it is posited that associative learning (i.e., “the learning of the ways in which concepts are related”, Van Osselaer, 2008, p. 699) is the mechanism underlying the branding effects of sponsorship. Pairing a brand with a sport event may lead to more positive brand evaluation because the brand name itself may have acquired a positive halo as a result of associative learning. Alternatively, the branding effects may occur because the cue of association with an event provides diagnostic information about the function and quality of the product, brand reputation, and consumption experience as a result of predictive learning, which is a second way of associative learning and is explored in this paper. These two ways of associative learning are qualitatively distinct. There is a difference between choosing a product because one likes its search characteristics and choosing a product because its search characteristics allow one to predict a consumption experience (Van Osselaer, 2008). The objective of the present study was to extend the emerging body of research on the branding effects of sponsorship. This study included: (a) an analysis of two ways of learning sponsorship associations - evaluative conditioning and predictive learning (Van Osselaer, 2008; Van Osselaer & Janiszewski, 2001); and (b) demonstrating their impact on perceived product quality, attitudinal loyalty and behavioural intention toward the sponsoring brand.

Journal of Customer Behaviour, Volume 12 JCB26

Mao, Zhang, Connaughton, Holland & Spengler An associative learning account 27

REVIEW OF LITERATURE

Two ways of learning sponsorship associations

The theory of associative learning can be traced back to two distinct views of classical conditioning. In the early 20th century, Ivan Pavlov found that when a neutral stimulus (e.g., bell) was consistently followed by a biologically significant stimulus (e.g., food), dogs eventually started to salivate upon perceiving the initial neutral stimulus. This process, termed classical conditioning, was interpreted by two distinct perspectives: adaptive learning (i.e., learn to predict, Rescorla & Wagner, 1972) and associative learning (Staats & Staats, 1959). Although Pavlov originally was interested in understanding how animals predict their environment, the paradigm of associative learning gained popularity with human behaviour researchers over time. The shift from Pavlovian conditioning to associative learning has taken place in conjunction with a relaxation of assumptions underpinning the seminal Pavlovian paradigm. With human subjects, researchers are more interested in their cognitive and affective responses instead of automatic physiological responses; the conditioned stimulus has further generalized to a wide variety of instruments including such intangibles as meaningful words, nonsense syllables, music, pleasant scenes, and celebrity. Furthermore, the nature of the relationship between conditioned stimulus and unconditioned response is more associative than causally inferable.

Despite these fundamental differences, the term “classical conditioning” is often used instead of “associative learning” in marketing literature (e.g., Gorn, 1982; Speed & Thompson, 2000; Stuart, Shimp & Engle, 1987; Till, Stanley & Priluck, 2008). Moreover, inconsistent usage of the terminology can be found in the psychology and marketing literature. In this study, we adopted the definition of associative learning as “learning of the ways in which concepts are related” (Van Osselaer, 2008, p. 699). The two ways of associative learning that were explored in this study are referred to as “evaluative conditioning” and “predictive learning” to reflect the fact that the relatedness between concepts can be merely correlated or causally inferable.

Evaluative conditioning is a process in which pairing a neutral element (unconditioned stimulus) with a valenced element (conditioned stimulus) leads to a transfer of valence from one to the other element (Sweldens, Van Osselaer & Janiszewski, 2010; Van Osselaer, 2008). Evaluative conditioning is a form of associative learning that relies on an exemplar-based process (Baeyens, De Houwer, Vansteenwegen & Eelen, 1998). Van Osselaer (2008) proposes two types of processes through which consumers may learn brand associations - exemplar-based process and adaptive process. The exemplar-based process is relatively unfocused in which all stimulus elements are contextually encoded and get cross-referenced for later retrieval. The connection between a cue and an associated item is strengthened every time the cue is presented with the associated item. The exemplar-based process is consistent with Keller’s (1993) conceptualisation of brand knowledge and is the fundamental way of learning brand associations (Janiszewski & Van Osselaer, 2000). Most investigations into the branding effects of sponsorship begin with the assumption that a brand is a collection of associations and event sponsorship might be one of many vehicles that can create secondary associations (Keller, 1993). From an associative learning perspective, the learned association will be represented as a network of concept nodes in memory, which may later be activated. As demonstrated in previous studies, the branding effectiveness of sponsorship often relies on this

evaluative conditioning (Speed & Thompson, 2000). We propose that evaluative conditioning is the first way of learning sponsorship associations.

Consumers often predict consumption benefits (e.g., product quality, hedonic experience) based on the product cues (e.g., brand names, ingredients). Predictive learning, relying on adaptive process, occurs when consumers have motivations to predict consumption outcomes. Unlike the exemplar-based process, an adaptive process is error-driven and focuses on prediction with minimal error. Furthermore, co-occurrence does not guarantee the strengthening of associations. The updating of associations only takes place to the extent that the learning system is not already correctly predicting an outcome (Van Osselaer & Janiszewski, 2001). Sponsorship is believed to provide diagnostic information for consumers, such as scale of the company, product function, and product quality (Pracejus, 1998; Rao & Ruekert, 1994). We propose that predictive learning is the second way of learning sponsorship associations.

Branding effects of sponsorship

Corporations often invest in sponsorships to actualise their branding goals. The power of a brand lies with its brand equity. Branding effects can be conceptualised as the perceived enhancement of elements of brand equity. In Aaker’s (1991) seminal brand equity framework, brand equity comprises brand awareness, perceived quality, brand associations, and brand loyalty. Previous studies have predominately focused on the impact of sponsorship on brand awareness (i.e., brand recall and recognition, Pham, 1992; Tripodi, Hirons, Bednall & Sutherland, 2003; Wakefield, Becker-Olsen & Cornwell, 2007) and brand associations (i.e., brand image, Donahay & Rosenberger, 2007; Grohs, Wagner & Vsetecka, 2004; Gwinner & Eaton, 1999). The general finding is that sponsorship strategy can be employed to enhance brand awareness and association, but the effectiveness is subject to numerous moderators and is extremely difficult to evaluate in the field setting due to many confounding variables (Cornwell, Weeks & Roy, 2005; Pham, 1991). For example, Pham (1992) measured consumer recognition of embedded sponsorship stimuli in a controlled laboratory setting and found involvement with a soccer game had a curvilinear effect, where arousal in reaction to the game had a negative effect on the outcome variable. Tripodi et al. (2003) conducted a field study prior to the Sydney 2000 Olympics to evaluate the awareness of sponsorship. They considered three approaches to measure recall, event sponsorship prompt, brand sponsorship prompt, and category sponsorship prompt as well as one measure of recognition. Tripodi et al. (2003) found that these different approaches to evaluate sponsorship effectiveness yielded different estimates of sponsorship awareness. In terms of brand associations, much work has been conducted in understanding the mechanism of brand image (e.g., Donahay & Rosenberger, 2007; Grohs et al., 2004). Based on a survey of 160 Australian F1 motor sport fans, Donahay and Rosenberger (2007) found a functional-based sponsor relationship had a major influence on the efficiency of the image-transfer process.

In this study, we focused on the other two elements of brand equity - perceived quality and brand loyalty. Perceived quality is a consumer’s subjective judgment about a product’s overall excellence. It plays an important role in determining a consumer’s choice decisions. Whereas perceived quality is mainly determined by personal product experiences, unique needs, and consumption situations, it is also subject to the influence of other factors, such as marketing communication.

Journal of Customer Behaviour, Volume 12 JCB28

Mao, Zhang, Connaughton, Holland & Spengler An associative learning account 29

According to Pracejus’ (1998) inference-based model of building brand equity through sponsorship, consumers would draw different inferences about the sponsoring brand based on event associational factors. These inferences may include but not be limited to consumers’ beliefs in brand size, brand legitimacy, and sponsor’s function in event facilitation. Most consumers, through past patronage or other forms of exposure (word-of-mouth, television, free trial, etc.) to sponsoring brands, will develop some beliefs about the associations between the event and brand. When encountering a novel event-brand alliance in which the quality of the product is unobservable or the product is totally new to a segment of market, consumers may infer consumption benefits based on their beliefs about the links between events and brands. Associating a brand with a sport event therefore can serve as predictive cues about perceived product performance and consumption experience. Following this theoretical reasoning, it is speculated that some consumers chose a box of Yili dairy product with the logo of the Beijing Olympic Games on the package over its competitor, Mengniu, not only because their affect toward the event transferred to the brand, but also because the association may lend credibility to the product and provide cues about the unobservable product quality. It is likely that consumers adopt a predictive learning process when they learn sponsorship associations in predicting product quality.

According to a recent IEG survey, 65% of sponsors rate “increase brand loyalty” as an extremely important objective (IEG, 2011). Brand loyalty is a deeply held commitment to purchase a preferred product or service in the future. It generally comprises behavioural and attitudinal loyalty, suggesting that loyal customers hold favourable attitudes and thus routinely purchase a particular brand (Kaynak, Salman & Tatoglu, 2008). Speed and Thompson (2000) demonstrated that the liking of an event can lead to the liking of a brand and willingness to consider the branded product. Sirgy, Lee, Johar and Tidwell (2008) directly demonstrated the positive impact of sponsorship on brand loyalty. This study focuses on the attitudinal dimension of brand loyalty. As an attitudinal construct, brand loyalty may be enhanced through evaluative conditioning (Gorn, 1982; Stuart et al., 1987). In addition, behavioural intention to purchase a sponsoring brand has also been investigated. Purchase intention reflects a consumer’s motivation and conscious effort to actualise the purchasing behaviour, thus serving as a good proxy for actual purchase behaviour. Crompton (2004) suggested that purchase intention is perhaps the most useful indicator when assessing the impact of sponsorship on sales.

Common determinants of learning sponsorship associations

The two ways of learning sponsorship associations are different. Whereas predictive learning focuses on the predictive value of the association cue (i.e., experienced outcome - expected outcome, q-o), evaluative conditioning emphasises the valence of unconditioned stimulus to be paired with the conditioned stimulus (i.e., value of association, q). But in a single learning trial, both processes are determined by the salience of cue () and efficiency of learning (i.e., learning rate parameter ). In the context of sport sponsorship marketing, there are many possible associative cues, including but not limited to on-site signage, in-stadium advertising, title of proprietary area, website presence, publicity information, and co-branded products.

The first determinant of associative learning is the salience of cue. Although this parameter is often dichotomised for simplicity, it is essentially continous and a function of a consumer’s extrinsic involvement, defined as “allocation of attentional

capacity to a message source” (Greenwald & Leavitt, 1984, p. 591). As the amount of attention paid to sponsorship information influences the salience of an associative cue in one’s mind, it affects the efficiency of learning. In sponsorship research, it is believed that the level of extrinsic involvement affects the recall and recognition of sponsors (Pham, 1991). However, due to the difficulty of manipulating and measuring extrinsic involvement in field settings, the extent to which it influences the branding effects has not yet been empirically tested.

Traditionally, researchers believe that consumers’ involvement with sport sponsorship messages is rather low because in-stadium advertisements can hardly compete with the game for attention. When spectators’ involvement with the game itself is the focus of attention, the limited nature of attentional capacity makes it unlikely that involvement with any concurrent advertising message will exceed the pre-attention level (Greenwald & Leavitt, 1984; Pham, 1991). However, with the advancement of in-stadium technology (e.g., jumbotron and mega-auditory equipment) and innovation in sponsorship leveraging (e.g., interactive games), it is possible that the involvement of consumers with the sponsorship message may be elevated to higher levels - focal attention, comprehension or elaboration. In this study, we did not explicitly investigate the impact of the extrinsic involvement (i.e., salience of cue) on outcome variables. Instead, we assume attention to sponsorship information is a prerequisite for evaluating branding effects and hold it as a constant.

The second determinant is the learning rate parameter. It varies by individual and is a function of numerous variables including intrinsic involvement and knowledge. Intrinsic involvement is the psychological state of perceived relevance of the stimulus based on inherent needs, values, and interests (Zaichkowsky, 1986). Intrinsic involvement has often been used as a proxy for extrinsic involvement in sponsorship research. It is often referred to as involvement (Pham, 1992), sport involvement (Cornwell, Relyea, Irwin & Maignan, 2000), or event involvement (Martensen, Gronholdt, Bendtsen & Jensen, 2007) in sponsorship research. However, due to conceptual confusions about involvement, findings in previous studies have often been inconsistent. For instance, Pham (1992) reported a curvilinear effect of involvement with a soccer game on the recognition of sponsors. Cornwell et al. (2000) found that involvement with a particular sport does not directly affect recall and recognition of sponsors, but it has a positive influence on game attendance that in turn results in exposure to sponsor’s messages. Martensen et al. (2007) showed that participant’s involvement with an event was predictive of sponsorship value.

From a learning perspective, intrinsic involvement, which represents a motivational aspect of learning, influences the efficiency of learning sponsorship associations. Knowledge, on the other hand, represents the ability aspect of learning (Kyllonen, Tirre & Christal, 1991). Because relationships between a cue and an associated item are constructed by drawing on previously stored facts, the breadth of factual knowledge an individual brings to the learning situation might be expected, at least partly, to determine associative learning efficiency. Kyllonen et al. (1991) also suggest that with a richer knowledge base, a constructed relationship will more likely be a distinctive relationship, which will also contribute to memorability. In sponsorship contexts, researchers have mostly focused on consumer knowledge about the sponsor’s product, which is related to the familiarity, experience, expertise, and use of the sponsor’s branded product (Lacey, Close & Finney, 2010). Pham and Johar (2001) found that consumers’ familiarity with a brand name will significantly influence their ability to correctly identify event sponsors, and consumers tend to bias toward those prominent brand names. Lacey et al. (2010) found that consumers’

Journal of Customer Behaviour, Volume 12 JCB30

subjective knowledge about the sponsor’s product significantly influences brand commitment and purchase intentions. Roy and Cornwell (2004) suggested another element in sponsorship knowledge, knowledge about an event, which has been found to be influential in processing sponsorship messages. From an associative learning perspective, consumer knowledge influences the ability of mapping relationships between the cue and the associated item. Whereas, in evaluative conditioning models, knowledge always facilitates the learning process, in predictive learning models, certain existing knowledge can block learning. For instance, the prominence bias effect found in previous studies (Pham & Johar, 2001; Roy & Cornwell, 2003) may be attributable to existing associations between prominent brands with events blocking the learning of a new association.

The third determinant is the value of association. In sponsorship terms, the value of association is a function of numerous variables including, among others, event attitude and the emotional experience of the event, and relatedness between the sponsoring brand and the event. Event attitude and emotional experience represent the valence of the unconditioned stimulus. Event attitude and event emotions are two drivers of sponsorship value (Christensen, 2006; Hansen, Martensen & Christensen, 2005). Hansen et al. (2005) and Christensen (2006) postulated that attitudinal responses to a sponsorship message, as a central route, could influence consumers’ reactions to the sponsor. A more comprehensive theoretical framework was proposed by Speed and Thompson (2000), in which event attitude (defined as “personal liking for the event”), event status, attitude toward sponsor, sincerity of sponsor, ubiquity of sponsor, and sponsor-event fit were six determinants of sport sponsorship response. They found that event attitude was positively related to three response variables (attention to the sponsor, favourability toward the sponsor, and willingness to consider sponsor’s product).

Regarding emotional experience, Hansen et al. (2005) and Christensen (2006) postulated emotional response to a sponsorship message as a peripheral cue, which could influence consumers’ reactions to the sponsor. Pham (1992) examined

31Mao, Zhang, Connaughton, Holland & Spengler An associative learning account

FIGURE 1 Conceptual relationships between learning determinants and branding outcomes

1 Cue Salience

Extrinsic Involvement

2 Learning RateIntrinsic InvolvementConsumer Knowledge

3 Association ValueEvent AttitudeEmotional ExperienceEvent-Brand Relatedness

Branding Outcomes

• Perceived Quality• Brand Loyalty• Behavioural Intention

Evaluative Conditioning

Predictive Learning

the influence of two specific emotions, namely arousal and pleasure, on sponsor recognition, which revealed that arousal, in reaction to watching an edited session of a soccer game, had a negative effect on the recognition of embedded billboards brand names; however, an emotional experience of pleasure did not have the hypothesised positive effect on recognition.

According to the associative learning theory, repeated presentations of a brand with the same affective stimulus can forge an indirect mental connection between the brand and an affective reaction triggered by the event (Sweldens et al., 2010). As the affective response to the brand is assumed to be mediated by the Conditioned Stimulus-Unconditioned Stimulus (i.e., brand-event) association, it appears that the relatedness between brand-event plays a role in this process. Relatedness, also referred to as congruence, match-up, or fit, is regarded as an important contributing factor to the branding effects of sponsorship. Previous studies have shown that event-brand relatedness is associated with consumers’ ability to recognise and recall a sponsor (Speed & Thompson, 2000), positively perceived image transfer (Gwinner & Eaton, 1999), and consumer attitude toward sponsor and purchase intentions (Becker-Olsen, 2003). Figure 1 summarises the conceptual relationships of these three classes of learning determinants with branding outcomes through the two routes of learning, namely evaluative conditioning and predictive learning.

METHOD

Participants

Research participants (N = 1,512) were students in two Chinese universities. One university is located in a large metropolitan area in East China and the other is located in a medium-sized city in South China. Because both universities accept students from all over China, the respondents represented a wide range of Chinese geographical locations. Table 1 presents the demographic characteristics of those research participants in this study.

Journal of Customer Behaviour, Volume 12JCB32

TABLE 1 Descriptive statistics for the demographic variables (N = 1,512)

Variable Category N %Gender Male 589 40.4

Female 868 59.6Education Undergraduate

GraduateOther

797415300

52.827.419.8

Age 17 years or younger18-23 years old24-28 years old29 years or older

151331

11832

1.089.0

7.92.1

Resident Urban Setting Metropolitan cityMiddle-size citySmall town/Countryside

369406554

27.630.541.6

Other 4 0.3

Event and brand selection

To examine variability and improve generalisability of the results, two different sport events and two different product brands were purposefully selected as stimuli to measure the branding effects of sponsorship. The two sport events were Chinese Basketball Association (CBA) and Chinese Soccer League (CSL) games, which were representative of the primary sport events in China at the time of data collection. The two product brands were Li-Ning and Nongfu Spring. Li-Ning is a well-known sporting apparel and equipment brand, and Nongfu Spring is a very popular beverage brand in China. These brands were selected based on the following criteria: (a) athletic equipment and beverages are two product categories that are most active in sport sponsorship in China, (b) both brands are well known to Chinese consumers and have recently been actively involved in sport sponsorship in China, (c) both brands are not the foremost leaders in their respective product categories (chosen in an effort to control for the influence of market dominance), and (d) both brands were not actual sponsors of the events (with the intention to eliminate individual and event biases and set the stage for introducing the two brands as fictitious sponsors). The respondents were randomly assigned to receive a fictitious announcement of a sponsorship deal involving one brand and one event.

Measurement

A questionnaire was developed that contained two parts (Appendix A). The first part included the following associative learning factors: event involvement, emotional experiences, event attitude, brand knowledge, and event-brand relatedness. These items were adapted from original measures and each of them had different response formats in a Likert or semantic-differential type scale. To maintain the unique measurement features of each measure, its original format of scaling was adopted. Event involvement was defined as perceived relevance of an event to an individual, and was measured by Zaichkowsky’s (1994) Personal Involvement Inventory. A measure of 20 statements assessing personal feelings developed by Hansen et al. (2005) and Hansen, Halling and Christensen (2006) was chosen to assess emotional responses to the viewing of the event. Attitude toward the event was measured by three items that were adopted from Muehling and Laczniak (1988). Brand knowledge was operationalised as consumer’s use, familiarity, and experience of the sponsor’s product (Lacey et al., 2010). Respondents answered “yes”/“no” to the question “have you ever used/owned any (brand name) products?” Two follow-up questions, adopted from Bloch, Sherrell and Ridgway (1989), measured their familiarity and experience with the brand. The perceived event-brand relatedness was tested using two items from Gwinner and Eaton’s (1999) study, representing functional-based similarity and image-based similarity respectively. In an effort to reduce potential demand effects, the focal brand (either Li-Ning or Nongfu) followed by four filler brands (Nike, Lenovo, Adidas, Sony) were rated in terms of consumer knowledge and their relatedness with the focal event (either the CBA or CSL games).

The second part of the questionnaire measured the following branding effects factors: perceived quality, attitudinal loyalty, and behavioural intention. The measurement of attitudinal loyalty and perceived quality was based on a modified application of the brand equity measure developed by Yoo and Donthu (2001). Five items related to perceived quality and brand loyalty were adopted in the current study. An announcement that the focal brand (i.e., Li-Ning or Nongfu Spring) was recently decided as sponsor of the focal event, along with the brand logos, was presented on

Mao, Zhang, Connaughton, Holland & Spengler An associative learning account 33

the front page of the second questionnaire. Similar to previous sponsorship studies that adopted self-rated behavioural intention (e.g., Hansen et al., 2005), in this survey respondents were asked to indicate the extent to which their evaluation of the focal brand would change due to this particular sponsorship deal. Additionally, for sample description purposes, the demographic background section included four variables: gender, age, education level, and main residence, which was included at the end of the second questionnaire.

Procedures

Institutional approval was obtained to conduct the study. For one university, 10 classes on a weekday were randomly selected according to the school’s curriculum. For the other university, general physical education classes on all weekdays were selected. Instructors of these classes were contacted and asked to support and cooperate with the data collection process. With the support of the instructor, the first part of the questionnaire was administered at the beginning of the class, and the second part was administered during the last 15 minutes of the same class session. A class session typically lasted two to three hours. We purposely administered the two parts with some time lapse to reduce potential contextual effect that responses to earlier questions affect the responses to the later questions. Tourangeau, Rasinski, Bradburn and D’Andrade (1989) referred to this contextual carry-over effect as a common problem for attitude surveys that potentially hamper the internal validity of the study. Participation in the study was voluntary and a small notebook gift was used as an incentive to those who completed the questionnaire. Generally, it took an individual approximately 25 minutes to complete the two parts of the questionnaire. All responses were anonymous and confidentiality was ensured. A total of 1,512 respondents completed the survey.

Data analyses

The statistical package SAS 9.2 was used to conduct statistical analyses. Descriptive statistics for the demographics, associative learning variables, and branding effects variables were calculated. Cronbach’s alpha coefficients and item-to-total correlation coefficients were calculated as measures of internal consistency. To provide an overview of the relationship between associative learning variables and branding effects, a canonical correlation analysis (CCA) was conducted by PROC CANCORR procedure in SAS 9.2. CCA is a type of multivariate analytical technique that enables the assessment of relationships between multiple independent and multiple dependent variables (Hair, Anderson, Black & Tatham, 2009). This method is widely used when the research interest is not centred on a single dependent variable, but to explore the impact of different actions on a wide range of outcomes (Lattin, Carroll & Green, 2003). It can simultaneously identify multiple unique relationships, if they exist. As a multivariate technique, CCA can limit the probability of over-committing Type I errors because this method allows researchers to assess multiple inter-dependences in a single relationship framework. In many business studies, CCA was used to explore the interrelationships between two sets of variables, such as between innovation strategies and market orientation (Lukas & Ferrell, 2000), between customer service and market response (Pisharodi & Langley, 1991), and between adoption of outsourcing services and the characteristics of environments in which the services operate (Zeynep Aksin & Masini, 2008).

Journal of Customer Behaviour, Volume 12 JCB34

Mao, Zhang, Connaughton, Holland & Spengler An associative learning account 35

The primary goal of CCA is to find two linear combinations of the original variables - one combination from the set of predictor variables and one combination from the set of criterion variables - that exhibit the largest possible covariance, thus determining how the specific variables function in this multivariate relationship. After finding the first canonical function, CCA identifies the second pair (and subsequent pairs) of canonical variates by requiring that it be uncorrelated with the first. The number of functions that can be computed in a CCA usually equals the number of variables in the smaller of the two variable sets unless there is perfect multicollinearity (Lattin et al., 2003). The relative importance of a variable in each set of variables is indicated by the standardised canonical function coefficient (also called canonical weight) and the canonical structure loading (also called canonical loading) that are extracted for the variable (Stevens, 2009). CCA requires a very large sample size to achieve a stable solution. Stevens (2009) suggested that a sample of 1,000 or more subjects is advisable in order to increase the probability that the interpretation made in the given sample would hold up in another sample from the same population. In the current study, CCA was used to detect the interrelationships between a set of learning variables and a set of branding effects variables, and its adoption was consistent with the purpose of the study. As the object of analysis was sponsorship, the data were pooled in the canonical correlation analysis.

RESULTS

Preliminary analyses

Before conducting canonical correlation analysis, internal consistency of the measures for the proposed constructs was first assessed. Table 2 presents descriptive statistics, item-to-total correlation coefficients, and Cronbach alpha coefficient for each construct and its items. Except for brand knowledge, brand-event relatedness, and behavioural intention, the Cronbach alpha coefficients of all remaining constructs were greater than 0.7, the traditional recommended threshold for a strong reliability (Hair et al., 2009). Mean scores of sport involvement, event involvement, positive emotions, negative emotions, event attitude, attitudinal loyalty, and perceived quality were computed by averaging the scores of all reliable items within each construct, respectively. Furthermore, emotional experience, a derived variable aiming to measure the net emotional surplus, was calculated by subtracting the mean score of positive emotions from the mean score of negative emotions (i.e., [(Sum of Positive Emotions)/12 - (Sum of Negative Emotions)/7]). The brand knowledge construct included one dummy indicator. The Cronbach alpha coefficient for the remaining two variables was 0.62, lower than the traditional recommended threshold for reliability. Therefore, brand usage, brand familiarity, and brand experience items were treated as three separate constructs as they likely have measured three different aspects of brand knowledge. Likewise, the Cronbach alpha coefficient for event-brand relatedness was 0.51, lower than the traditional recommended threshold for reliability. Function-relatedness and image-relatedness were treated as two separate constructs as they likely have measured two different aspects of relatedness. As a result, there was a total of nine associative learning variables: event involvement, brand usage, brand familiarity, brand experience, event attitude, event emotional experience, function-relatedness, image-relatedness and sport involvement.

Journal of Customer Behaviour, Volume 12 JCB36

TABLE 2 Summary of descriptive statistics of items and internal consistency of constructs

Construct and Items Cronbach Mean SD Item-to-Total Correlation

Sport Involvement 0.70

How do you like sports in general? 3.60 0.85 0.58

How often do you actually take part in sports? 2.77 1.21 0.45

How often you watch sports on television? 2.98 0.83 0.50

Event Involvement 0.86

Important-Unimportant 3.74 2.15 0.57

Interesting-Boring 4.02 1.94 0.50

Relevant-Irrelevant 3.47 1.85 0.54

Exciting-Unexciting 3.45 1.77 0.66

Means a lot to me-Means nothing 4.19 1.90 0.52

Appealing-Unappealing 3.56 1.83 0.65

Festinating-Mundane 3.31 1.70 0.57

Valuable-Worthless 4.30 1.85 0.54

Involving-Uninvolving 3.78 1.90 0.54

Needed-Not needed 4.37 1.98 0.52

Brand Knowledge 0.62Have you ever used/owned any Li-Ning/Nongfu Spring

brand product? 70.49 - -

Familiar with Li-Ning/Nongfu Spring branded products 3.29 0.99 0.45Favourable experience with the Li-Ning/Nongfu Spring

branded products 3.31 0.71 0.45

Positive Emotions 0.92

Hope 1.22 0.87 0.58

Joy 1.24 0.88 0.70

Love 1.15 0.85 0.71

Enjoyment 1.12 0.87 0.73

Happiness 1.08 0.90 0.69

Excitement 1.30 0.91 0.64

Surprising 1.29 0.86 0.35

Trust 1.18 0.88 0.71

Accept 1.49 0.93 0.54

Inspiring 1.34 0.93 0.64

Optimism 1.36 0.89 0.70

Pride 1.22 0.92 0.73

Desire 1.37 1.00 0.64

Cont’...

Mao, Zhang, Connaughton, Holland & Spengler An associative learning account 37

There were three criterion variables: attitudinal loyalty, perceived quality and behavioural intentions. Summative scores for attitudinal loyalty and perceived quality were computed by averaging the scores of three attitudinal loyalty items and averaging the scores of two perceived quality items, respectively. As behavioural intention was measured by a single item, its internal reliability was unable to be assessed. Its correlation with attitudinal loyalty was 0.53 and with perceived quality was 0.41.

Construct and Items Cronbach Mean SD Item-to-Total Correlation

Negative Emotions 0.85

Sorrow 1.30 0.90 0.62

Worry 1.54 0.90 0.54

Fear 1.06 0.93 0.59

Shame 1.12 0.95 0.61

Loneliness 0.94 0.88 0.54

Anger 1.28 0.97 0.68

Sad 1.30 0.99 0.66

Event Attitude 0.76

Favourable-Unfavourable 3.63 1.88 0.47

Good-Bad 4.46 1.72 0.67

Positive-Negative 4.70 1.81 0.66

Event-Brand Relatedness 0.51

Using Li-Ning/Nongfu Spring branded products 3.44 0.97 0.34

Event and brand have a similar image 3.02 1.00 0.34

Attitudinal Loyalty 0.73

I consider myself to be loyal to Li-Ning/Nongfu Spring 0.04 1.02 0.58

Li-Ning/Nongfu Spring would be my first choice 0.01 0.92 0.63

I will not buy other brands if Li-Ning/Nongfu Spring is available at the store -0.02 0.92 0.44

Perceived Quality 0.71

The likely quality of Li-Ning/Nongfu Spring is extremely high 0.45 0.95 0.55

The likelihood that Li-Ning/Nongfu Spring would be functional is very high 0.40 0.95 0.55

Behavioural Intention -It is likely that I will buy Li-Ning branded products the

next time? 3.28 1.01

Note: The Cronbach alpha coefficient for positive emotions was 0.92 after deleting “surprising” due to low correlation (r = 0.35)

Canonical correlation analysis

As the number of criterion variables (i.e., three) was of a smaller set in this study when compared with those of the predicting variables, the CCA yielded a total of three functions. For each successive function, the canonical correlation coefficient that measures the bivariate correlation between the derived variates, was 0.57, 0.19, and 0.08, respectively. Collectively, the full model across the functions was statistically significant (Wilks = .65, F (27, 3882) = 22.62, p < .001). Because Wilks represents the variance unexplained by the model, 1 - yields the full model effect size in an R2 metric. For the set of canonical functions, the R2 type effect size was .35, indicating that a total of 35% variance was shared between the predicting and criterion sets of variables. As the dimension reduction analysis allowed testing of the hierarchical arrangement of functions for statistical significance, Functions 1 and 2 were found to be statistically significant (F (27, 3882) = 22.62, p < .001; and F (16, 2660) = 3.53, p < .001, respectively), with 31.93% and 3.52% of variance shared, respectively. Function 3 was not statistically significant. Given the effects for each function, the first two functions were considered noteworthy.

Table 3 presents the standardised canonical function coefficients and structure coefficients for Functions 1 and 2. The squared canonical correlation (Rc

2) and squared structure coefficients (Rs

2) were also provided, along with the communalities (h2) across the two functions for each variable. The standardised canonical function coefficients are the weights assigned to each variable in the process of maximising covariance between the predictor and criterion variates. The canonical structure loading is the correlation coefficient between the observed variable scores and the derived variate. Both standardised canonical function coefficient and canonical structure loading denote the relative importance of a variable in each set of variables. The squared canonical correlation is the simple square of the canonical correlation, which represents the proportion of variance shared by the two variates. The squared structure coefficients are the square of the structure coefficients, indicating the proportion of variance an observed variable linearly shares with the derived variate. Communalities are the proportion of variance in each variable that is explained by the complete canonical solution (Sherry & Henson, 2005; Thompson, 1991). Adopting the criterion of a standardised canonical function coefficient and a structure coefficient equal to or greater than .40 (Stevens, 2009), Function 1 was defined by positive contributions from behavioural intention and attitudinal loyalty as the primary contributors and perceived quality as a secondary contributor that functioned as a synthetic criterion variate. This conclusion was supported by the high squared structure coefficients and h2. For the predictor set of variables in Function 1, emotional experience, image relatedness, event involvement, and event attitude were the primary contributors to the synthetic predictor variate, with function-relatedness and brand experience making secondary contributions, and sport involvement, brand usage, and brand familiarity making no additional contribution to the synthetic predictor variable. In addition to the large canonical function coefficients, this conclusion was supported by the high squared structure coefficients and h2. Noticeably, all of these structure coefficients, with the exception of sport involvement, had positive coefficients, indicating that they were all positively related. The findings were supportive of the theoretically expected relationships between the associative learning and branding effect variables. Considering that they were representations of the evaluative component of the associative learning process, Function 1 was hence termed the “evaluative conditioning” function.

Journal of Customer Behaviour, Volume 12 JCB38

Mao, Zhang, Connaughton, Holland & Spengler An associative learning account 39

TAB

LE 3

Des

crip

tive

stat

istic

s an

d ca

noni

cal s

olut

ion

for

asso

ciat

ive

lear

ning

fact

ors

pred

ictin

g br

andi

ng e

ffec

ts fo

r fu

nctio

ns 1

and

2

Min

Max

Mea

nSD

Skew

ness

Kur

tosi

s

Func

tion

1Fu

ncti

on 2

Vari

able

Coe

fr s

r s2 (%

)C

oef

r sr s

2 (%)

h2 (%

)

Pre

dict

or s

et

Spor

t Inv

olve

men

t1

53.

120.

720.

130.

11-0

.11

-0.0

40.

15-0

.21

-0.1

21.

411.

57

Even

t Inv

olve

men

t1

73.

791.

24-0

.21

0.26

0.25

0.68

46.0

4-0

.52

-0.4

621

.25

67.2

9

Bra

nd U

sage

a1

100

70.4

90.

020.

040.

140.

190.

010.

010.

15

Bra

nd F

amili

arity

15

3.29

0.99

-0.0

8-0

.35

0.01

0.13

1.62

0.09

0.41

16.7

418

.36

Bra

nd E

xper

ienc

e1

53.

310.

710.

290.

440.

270.

3915

.54

0.78

0.78

60.8

976

.43

Emot

iona

l Ex

peri

ence

-33

0.05

1.04

-0.2

10.

470.

310.

7352

.74

-0.2

7-0

.36

13.0

365

.77

Even

t Att

itude

17

4.27

1.49

-0.2

1-0

.28

0.12

0.65

42.8

10.

22-0

.23

5.13

47.9

4

Func

tion-

Rel

ated

ness

15

3.44

0.97

-0.3

7-0

.14

0.12

0.47

22.1

20.

120.

3210

.28

32.4

0

Imag

e-R

elat

edne

ss1

53.

021.

00-0

.13

-0.5

50.

460.

7860

.59

0.02

-0.0

30.

0760

.66

Rc2

31.9

33.

52

Cri

teri

on s

et

Att

itudi

nal L

oyal

ty-2

20.

010.

77-0

.37

0.98

0.58

0.90

81.7

60.

230.

142.

0683

.82

Per

ceiv

ed Q

ualit

y-2

20.

430.

83-0

.20

0.43

0.18

0.64

41.5

30.

890.

6136

.76

78.2

9

Beh

avio

ural

In

tent

ion

15

3.28

1.01

-0.3

20.

060.

430.

8267

.86

-0.9

5-0

.45

20.4

488

.30

Not

e: S

truc

tura

l coe

ffic

ient

s (r

s) gr

eate

r th

an |.

40| a

re u

nder

lined

. Com

mun

ity c

oeff

icie

nts

(h2 )

grea

ter

than

50%

are

und

erlin

ed. C

oef =

sta

ndar

dise

d ca

noni

cal f

unct

ion

coef

ficie

nt (

i.e.,

cano

nica

l w

eigh

t);

r s =

stru

ctur

e co

effic

ient

(i.e

., ca

noni

cal

load

ing)

; r s2

= sq

uare

d st

ruct

ure

coef

ficie

nt;

r c2 =

squa

red

cano

nica

l co

rrel

atio

n; h

2 =

com

mun

ality

coe

ffic

ient

.a B

rand

Usa

ge is

a d

umm

y va

riab

le. T

he s

tand

ardi

sed

cano

nica

l fun

ctio

n co

effic

ient

and

str

uctu

re c

oeff

icie

nt w

ere

not s

ensi

ble

to in

terp

ret.

The

raw

can

onic

al fu

nctio

n co

effic

ient

s w

ere

0.05

and

0.4

1, r

espe

ctiv

ely.

It is

sug

gest

ed th

at b

rand

usa

ge w

as a

con

trib

utin

g fa

ctor

in th

e se

cond

can

onic

al fu

nctio

n.

Regarding Function 2, standardised canonical function coefficients and structure coefficients suggest that the criterion variables of relevance were perceived quality and behavioural intention. The structure coefficients of behavioural intention and perceived quality had different signs, indicating that they were negatively related in Function 2. As for associative learning variables, brand familiarity, brand experience, and event involvement were the primary predictors. These associative learning variables also tended to reveal larger canonical function coefficients. An exception involved brand familiarity, which had very low function coefficients (0.09) but modest structure coefficients (0.41). This was likely due to the high correlation that brand familiarity had with brand experience (i.e., r = 0.45). Furthermore, as brand usage was the only dummy variable in the set of predicting variables, the standardised canonical function coefficient and structure coefficient for this variable were not sensible to interpret; however, judging from the canonical function coefficient (0.41),

Journal of Customer Behaviour, Volume 12 JCB40

FIGURE 2 Canonical functions with primary contributing factors

Canonical Function 1: Evaluative Conditioning Function

Canonical Function 2: Predictive Learning Function

EventInvolvement

EventAttitude

FunctionRelatedness

ImageRelatedness

EmotionalExperience

EventInvolvement

BrandFamiliarity

BrandExperience

BrandUsage

AttitudinalLoyalty

PerceivedQuality

BehaviouralIntention

PerceivedQuality

BehaviouralIntention

SecondPredictorVariate

SecondCriterionVariate

FirstPredictorVariate

FirstCriterionVariate

0.68

0.65

0.73

0.47

0.78

0.57

0.90

0.64

0.82

-0.46

0.41

0.780.19

0.61

-0.45

Mao, Zhang, Connaughton, Holland & Spengler An associative learning account 41

brand usage could be considered a major contributor to the function. Additionally, the structure coefficient of event involvement had a negative sign, suggesting that it was negatively correlated with brand familiarity and brand experience in Function 2. As Function 2 explains the deviance of Function 1, it is reflective of a second process that is independent from the first way of associative learning. These results were generally supportive of the theoretically expected relationships between predictive learning and branding effects, and thus Function 2 was termed the “predictive learning” function. Figure 2 presents these two canonical functions with primary contributing factors.

DISCUSSION

The purpose of sponsorship-linked marketing is to build and communicate associations (Cornwell, 1995). It is the unique association power that distinguishes sponsorship from traditional advertising activities, and creates value for consumers and sponsors. This study stressed the limitations of traditional theoretical accounts in explaining branding effects of sponsorship and proposed that associative learning is a fundamental branding mechanism of sponsorship marketing. Although associative learning has been a primary focus in human behaviour research, it has been lacking in previous sponsorship research; thus, this study was designed to fill the void. The associative learning variables included in this study were actually not novel as they had been repeatedly identified as relevant and important in previous investigations related to sponsorship effectiveness (Gwinner & Swanson, 2003; Hansen et al., 2005; Martensen et al., 2007; Pham, 1992; Roy & Cornwell, 2004; Speed & Thompson, 2000). Nevertheless, associative learning theory provided this study with a consistent and integrated conceptual framework through which the inclusion of the constructs could be justified, the impact of each construct could be inferred, and predictions of branding effects could be made. Furthermore, based on theoretical reasoning (Van Osselaer & Janiszewski, 2001), this study made a conceptual argument that there are two qualitatively distinct ways of learning sponsor associations: evaluative conditioning and predictive learning. Hence, a key contribution of this study is the introduction of an integrated model of branding effects to the sponsorship literature.

This study examined the multivariate relationships between associative learning and branding effects variables through a canonical correlation analysis. Our findings suggest that there are indeed two significant functions, which we termed the “evaluative conditioning function” and “predictive learning function”. In the evaluative conditioning function, image relatedness, emotional experience, event involvement, event attitude, and function relatedness were of relatively high loadings ( ranged from 0.47 to 0.78) on the predictor variate. Brand usage, brand familiarity, and brand experience played a relatively non-significant role in this function. The first predictor variate therefore can be interpreted as the valence of the unconditioned stimulus. On the criterion variate side, attitudinal loyalty, perceived quality, and behavioural intention had relatively high loadings ( ranged from 0.64 to 0.90) with the same sign. Thus, attitudinal loyalty, perceived quality, and behavioural intention were positively correlated with each other. The first criterion variate can be interpreted as the symmetrical branding effects. Overall, the first function suggests that branding effects, particularly attitudinal loyalty and positive behavioural intention, might be achieved through evaluative conditioning in which consumer’s affective response to the sponsored event and the heuristic of relatedness between the

brand and the event may have an impact on consumer’s evaluation of the sponsored brand. This interpretation is consistent with the principles of evaluative conditioning in that the affect towards the unconditioned stimulus is transferable, and the process is less focused and requires less cognitive resources (Van Osselaer, 2008).

In the predictive learning function, brand usage had a relatively high raw canonical function coefficient (0.41), brand experience and brand familiarity had relatively high positive loadings ( was 0.78 and 0.41 respectively), and event involvement had a relatively high negative loading ( = -0.46) on the predictor variate. On the criterion variate side, perceived quality had a positive loading ( = 0.61), and behavioural intention had a negative loading ( = -0.45). Thus, perceived quality was positively correlated with brand usage, brand familiarity, and brand experience; behavioural intention was positively correlated with event involvement; and perceived quality and behaviour intention were negatively correlated in the predictive learning function. It is necessary to note that the second canonical correlation function explains the deviance from the first canonical correlation function. These seemingly paradoxical results suggest that the learning of sponsorship association could have different impacts on perceived quality and behaviour intentions after partialling out of the effects of evaluative conditioning. While the second criterion variate captured this asymmetrical impact of learning on quality belief and behavioural intention, the second predictor variate captured the two contingency conditions of when predictive learning might occur: (a) consumers have low event involvement but favourable brand experience; or when (b) consumers have high event involvement but unfavourable brand experience. For example, when an event is characterised by relatively low consumer involvement, the evaluative conditioning function suggests that brand loyalty, perceived quality, and behavioural intention would be simultaneously diluted. The predictive learning function, on the other hand, suggests a positive impact on perceived quality under the circumstances that consumers were familiar with the brand, and had a relatively favourable brand experience. Alternatively, when an event is characterised by relatively high consumer involvement, the evaluative conditioning function suggests that brand loyalty, perceived quality, and behavioural intention would be simultaneously enhanced. The predictive learning function, on the other hand, suggests that although sponsorship may enhance behavioural intention, the quality perception may still suffer a negative impact if the consumer had an unfavourable brand experience. This interpretation is consistent with the principle that predictive learning is adaptive and more focused in nature, and only takes place when consumers are motivated (Van Osselaer, 2008; Van Osselaer & Alba, 2000; Van Osselaer & Janiszewski, 2001). In addition, consistent with Cornwell et al. (2000), this study did not find sport involvement to be of a direct impact on attitudinal loyalty, perceived quality, and behavioural intention as the loadings of sport involvement on either function were rather low. This finding suggests that in a cross-sectional study, sport involvement might not be a good proxy variable for extrinsic involvement as some previous studies found (Cornwell et al., 2000; Pham, 1992).

Managerial implications

For sponsors, the essence of commercial sponsorship is to acquire the rights of being associated with the sponsored organisation, which can later be leveraged for branding purposes. The current study contends that the commercial value of sponsorship lies with the strength of the association and consumer’s learning of this association. Thus, managing sponsorship is to manage consumer’s learning processes, and there

Journal of Customer Behaviour, Volume 12JCB42

Mao, Zhang, Connaughton, Holland & Spengler An associative learning account 43

are two learning processes: evaluative conditioning and predictive learning, through which marketers can capitalise. The first route is more fundamental and pervasive. It suggests that pairing a brand with an emotionally-charged event can lead to a transfer of valence from the event to the brand. This process, however, is conditional on at least four elements: event attitude, event emotional experience, event-brand relatedness, and event involvement. Through the findings of this study, we have seen how these four elements could have a collective impact on consumer’s attitudinal loyalty, brand loyalty, and behavioural intentions. To facilitate the evaluative conditioning process, sponsorship managers need to work closely with event organisers to enhance event involvement and create a positive event experience. Ideally, a sponsor’s activation and marketing activities should be fully integrated into an event organiser’s operations and event-related campaigns. For example, companies may take part in special activities that event organisers create to nurture fan involvement, or may even help initiate those kinds of activities for the event. To leverage the power of the emotional experiences of fans, a sponsor’s in-stadium/arena advertising and sponsored interactive entertaining activities may focus on emotional experiences that are consistent with the nature of the event and in the meantime, have an objective to build up fan pride, optimism, and excitement. Furthermore, the value of association can be enhanced by an articulated, publicised relatedness between the event and the brand. Sometimes, this relatedness is self-evident to consumers, such as that between Adidas products and the FIFA World Cup. Other times, this relatedness is more subtle, such as that between Hewlett-Packard (HP) and the Boston Marathon (Masterman, 2007). In these cases, extra efforts and creativity should be directed to strategically incorporate the relatedness message in a sponsor’s marketing communications. In HP’s case, by educating the public that HP provided the technology to track every registered participant at the event, and enhanced the experience and quality of the event, a greater positive association can be garnered (Masterman, 2007). Furthermore, when compared with function-relatedness, image-relatedness plays a more important role in the evaluative conditioning process. Therefore, marketers are advised to focus their communications on symbolic brand associations to help create image relatedness, and thus garner better branding effects. Marketing communications should focus on such messages as partnership, collective strategies, high quality, elegance and magnitude of outreach.

The research findings in this study also suggest that the mere fact of being associated with an event may be used as a salient cue for predicting consumption benefits. Although the predictive learning process is not as pervasive as the evaluative conditioning process, it occurs when the contingency conditions are met. Research findings of this study have shown how event involvement, exciting brand usage, experience and familiarity could have different impacts on a consumer’s quality belief and behavioural intentions to purchase a sponsor’s products. It has become a widely adopted practice for potential sponsors to evaluate information about attendance, fan passion and psychographics before they enter into a sponsorship agreement (IEG, 2011); however, this is not sufficient. It is critical for sponsorship managers to monitor the brand experience of the targeted segment of consumers as well as the dynamic interactions between brand experience and event experience. For instance, when sponsorship-linked marketing was used to penetrate a new market or market segment where consumption experience of the products is lacking, sponsors may focus their activation strategy on creating brand experience through on-site free sampling or even building a brand experience centre. By integrating the festivity of an event into one’s consumption experience, consumers may learn this association

cue to predict their consumption benefits. This consumption experience centred strategy can be also effective for companies who use sponsorship-linked marketing to reposition their products or recover from previous failures. Because predictive learning is adaptive in nature, it deepens the level of processing the association information when consumers find the on-site product experience as being at odds with their previous beliefs.

Sponsorship offers unique opportunities for implementing brand strategy. This study offers some support to the popular wisdom that sponsoring a sport event with which the target population is highly involved and towards which the target population holds a positive attitude will be more likely to enhance consumer-based brand equity in the target market. However, sponsoring a top-notch event also requires much heavy investment. We have seen that it costs millions of dollars to secure a TOP deal, not to mention the costs to activate the sponsorship association. For small scale companies, sponsoring hallmark events is often not feasible. However, from the predictive learning perspective, companies may still benefit from an elevated perception of quality even by sponsoring low profile events. The key is to create a positive on-site brand experience for consumers to deepen their processing. Furthermore, this study also suggests that function-relatedness plays a more significant role in the predictive learning process. Therefore, when sponsoring a lower-profile event, marketers would benefit more by creating unique function-relatedness with that event, and by emphasising functional brand associations to tie in with the event and highlight the promotion of superior performance.

Furthermore, the existence of two ways of learning sponsorship association has significant implications for long-term sponsorship relationships. A long-term sponsorship relationship (e.g., Coca Cola and the Olympic Games) is extremely valuable because it can become a sustained resource with competitive edge for the company when compared with those companies with same or similar product categories (Amis, Pant & Slack, 1997; Amis & Slack, 1999). From the predictive learning perspective, however, a long-term sponsorship relationship does not necessarily lead to stronger branding effects because co-occurrence of the brand name and event name does not guarantee the strengthening of associations. The updating of associations only takes place when the existing beliefs no longer accurately predict consumption benefits. Fortunately, for long-term sponsorship managers, they can still rely on the evaluative conditioning process and work with the event organisers to achieve branding objectives. Evaluative conditioning offers an avenue for marketers to achieve branding goals through creating and delivering value-added customer experiences. For event organisers, efforts should be made to nurture consumer involvement and create a positive emotional experience, and reinforce their positive attitude towards the event, in an effort to enhance the value of the event to sponsors. It is undeniable that the building of fan affective involvement is an on-going, painstaking process that requires enormous investment and innovation. Programs and actions that aim to increase fan involvement with the event should be initiated and regularly implemented. This can be done by increasing the performance and service quality of the event, increasing fan involvement with social events and activities, and increasing volunteer opportunities.

Journal of Customer Behaviour, Volume 12JCB44

Mao, Zhang, Connaughton, Holland & Spengler An associative learning account 45

REFERENCES

Aaker, D.A. (1991). Managing brand equity: Capitalising on the value of a brand name. NY: Free Press.

Amis, J., Pant, N., & Slack, T. (1997). Achieving a sustainable competitive advantage. A resource based view of sport sponsorship. Journal of Sport Management, 11(1), 80-96.

Amis, J., & Sla ck, T. (1999). Sport sponsorships as distinctive competence. European Journal of Marketing, 33(3/4), 250-272.

Baeyens, F., De Houwer, J., Vansteenwegen, D., & Eelen, P. (1998). Evaluative conditioning is a form of associative learning: On the artifactual nature of Field and Davey’s (1997) artifactual account of evaluative learning. Learning and Motivation, 29(4), 461-474.

Becker-Olsen, K. (2003). Questioning the name game: An event study analysis of stadium naming rights sponsorship announcements. International Journal of Sports Marketing & Sponsorship, 5(3), 181-192.

Bloch, P., Sherrell, D., & Ridgway, N. (1989). Product knowledge scale. In G. Brunner & P. Hensel (Eds.), Marketing scales handbook: A compilation of multi-item measures. Chicago, IL: American Marketing Association.

Christensen, S.R. (2006). Measuring consumer reactions to sponsoring partnerships based upon emotional and attitudinal responses. International Journal of Market Research, 48(1), 61-80.

Cornwell, T.B. (1995). Sponsorship-linked marketing development. Sport Marketing Quarterly, 4(4), 13-24.

Cornwell, T.G., Relyea, G.E., Irwin, R.L., & Maignan, I. (2000). Understanding long-term effects of sports sponsorship: Role of experience, involvement, enthusiasm and clutter. International Journal of Sports Marketing & Sponsorship, 2(2), 127-143.

Cornwell, T.B., Roy, D.P., & Steinard, E.A. (2001). Exploring managers’ perceptions of the impact of sponsorship on brand equity. Journal of Advertising, 30(2), 41-51.

Cornwell, T.B., Weeks, C.S., & Roy, D.P. (2005). Sponsorship-linked marketing: Opening the black box. Journal of Advertising, 34(2), 21-42.

Crompton, J.L. (2004). Conceptualisation and alternate operationalisations of the measurement of sponsorship effectiveness in sport. Leisure Studies, 23(3), 267-281.

Donahay, B., & Rosenberger, P.J. (2007). Using brand personality to measure the effectiveness of image transfer in Formula One racing. Marketing Bulletin, 18, 1-15.

Geylani, T., Inman, J.J., & Ter Hofstede, F. (2008). Image reinforcement or impairment: The effects of co-branding on attribute uncertainty. Marketing Science, 27(4), 730-744.

Gorn, G.J. (1982). The effects of music in advertising on choice behavior: A classical conditioning approach. Journal of Marketing, 46(1), 94-101.

Greenwald, A.G., & Leavitt, C. (1984). Audience involvement in advertising: Four levels. Journal of Consumer Research, 11(1), 581-592.

Grohs, R., Wagner, U., & Vsetecka, S. (2004). Assessing the effectiveness of sport sponsorships - an empirical examination. Schmalenbach Business Review, 56(2), 119-138.

Gwinner, K.P., & Eaton, J. (1999). Building brand image through event sponsorship: The role of image transfer. Journal of Advertising, 28(4), 47-57.

Gwinner, K., & Swanson, S.R. (2003). A model of fan identification: antecedents and sponsorship outcomes. Journal of Services Marketing, 17(3), 275-294.

Hair, J.F., Anderson, R.E., Black, W.C., & Tatham, R.L. (2009). Multivariate data analysis. Upper Saddle River, NJ: Prentice-Hall.

Hansen, F., Halling, J., & Christensen, L.B. (2006). Choosing among alternative parties to be sponsored for supporting brand strategies, based upon emotional responses. Journal of Consumer Behaviour, 5(6), 504-517.

Hansen, F., Martensen, A., & Christensen, S.R. (2005). Modelling emotional and attitudinal responses as drivers of sponsorship value. International Journal of Sports Marketing & Sponsorship, 7(1), 75-80.

Henseler, J., Wilson, B., Götz, O., & Hautvast, C. (2007). Investigating the moderating role of fit on sports sponsorship and brand equity. International Journal of Sports Marketing & Sponsorship, 8(4), 321-329.

IEG (2008). IEG sponsorship report 2007. Retrieved November 18, 2010, from http:// www.sponsorship.com/documents/SR_Promo_Issue_01-07.pdf

IEG (2011). 11th annual IEG/Performance Research sponsorship decision-makers survey. Retrieved May 18, 2011, from http://www.sponsorship.com

IOC (2009). International Olympic Committee marketing report Beijing 2008. Retrieved November 18, 2010, from http://view.digipage.net/?userpath=00000001/00000004/00040592/&page=34

IOC (2012). International Olympic Committee marketing report London 2012. Retrieved December 20, 2012, from http://www.olympic.org/Documents/IOC_Marketing/London_2012/LR_IOC_MarketingReport_medium_res1.pdf

Janiszewski, C., & Van Osselaer, S.M.J. (2000). A connectionist model of brand-quality associations. Journal of Marketing Research, 37(3), 331-350.

Kaynak, E., Salman, G.G., & Tatoglu, E. (2008). An integrative framework linking brand associations and brand loyalty in professional sports. Journal of Brand Management, 15, 336-357.

Keller, K.L. (1993). Conceptualizing, measuring, and managing customer-based brand equity. Journal of Marketing, 57(1), 1-22.

Kyllonen, P.C., Tirre, W.C., & Christal, R.E. (1991). Knowledge and processing speed as determinants of associative learning. Journal of Experimental Psychology - General, 120(1), 57-79.

Lacey, R., Close, A.G., & Finney, R.Z. (2010). The pivotal roles of product knowledge and corporate social responsibility in event sponsorship effectiveness. Journal of Business Research, 63(11), 1222-1228.

Lattin, J.M., Carroll, J.D., & Green, P.E. (2003). Analyzing multivariate data. Pacific Grove, CA: Thomson Brooks/Cole.

Lukas, B., & Ferrell, O.C. (2000). The effect of market orientation on product innovation. Journal of the Academy of Marketing Science, 28(2), 239-247.

Martensen, A., Gronholdt, L., Bendtsen, L., & Jensen, M.J. (2007). Application of a model for the effectiveness of event marketing. Journal of Advertising Research, 47(3), 283-301.

Masterman, G. (2007). Sponsorship : for a return on investment. London: Butterworth-Heinemann.

McDaniel, S.R. (1999). An Investigation of Match-Up Effects in Sport Sponsorship Advertising: The Implications of Consumer Advertising Schemas. Psychology & Marketing, 16(2), 163-184.

Muehling, D., & Laczniak, R. (1988). Advertising’s immediate and delayed influence on brand attitudes: Considerations across message-involvement levels. Journal of Advertising, 17(4), 23-34.

Orlick, T.D. (1974). An interview schedule designed to assess family sports environment. International Journal of Sport Psychology, 5(1), 13-27.

Pham, M.T. (1991). The evaluation of sponsorship effectiveness: A model and some methodological considerations. Gestion 2000, 4, 47-65.

Pham, M.T. (1992). Effects of involvement, arousal, and pleasure on the recognition of sponsoring stimuli. Advances in Consumer Research, 19(1), 85.

Pham, M.T., & Johar, G.V. (2001). Market prominence biases in sponsor identification: Processes and consequentiality. Psychology & Marketing, 18(2), 123-143.

Pisharodi, R.M., & Langley, C.J. (1991). Interset association between measures of customer service and market response. International journal of physical distribution & logistics management, 21(2), 32-44.

Pracejus, J.W. (1998). An inference-based model of building brand equity through sponsorship (Unpublished doctoral dissertation). University of Florida, Gainesville.

Rao, A., & Ruekert, R. (1994). Brand alliances as signals of product quality. Sloan Management Review, 36, 87-87.

Journal of Customer Behaviour, Volume 12JCB46

47Mao, Zhang, Connaughton, Holland & Spengler An associative learning account

Rescorla, R., & Wagner, A. (1972). Variations in the effectiveness of reinforcement and nonreinforcement. Classical conditioning: Current research and theory. NY: Appleton-Century-Crofts.

Roy, D.P., & Cornwell, T.B. (2003). Brand equity’s influence on responses to event sponsorships. Journal of Product & Brand Management, 12(6), 377-393.

Roy, D.P., & Cornwell, T.B. (2004). The effects of consumer knowledge on responses to event sponsorships. Psychology & Marketing, 21(3), 185-207.

Sherry, A., & Henson, R.K. (2005). Conducting and interpreting canonical correlation analysis in personality research: A user-friendly primer. Journal of Personality Assessment, 84(1), 37-48.

Simonin, B.L., & Ruth, J.A. (1998). Is a company known by the company it keeps? Assessing the spillover effects of brand alliances on consumer brand attitudes. Journal of Marketing Research, 35(1), 30-42.

Sirgy, M.J., Lee, D.J., Johar, J.S., & Tidwell, J. (2008). Effect of self-congruity with sponsorship on brand loyalty. Journal of Business Research, 61(10), 1091-1097.

Speed, R., & Thompson, P. (2000). Determinants of sports sponsorship response. Journal of the Academy of Marketing Science, 28(2), 226-238.

Staats, A.W., & Staats, C.K. (1959). Effect of number of trials on the language conditioning of meaning. Journal of General Psychology, 61(2), 211-223.

Stevens, J. (2009). Applied multivariate statistics for the social sciences (5th ed.). New York: Routledge.

Stuart, E.W., Shimp, T.A., & Engle, R.W. (1987). Classical conditioning of consumer attitudes: Four experiments in an advertising context. Journal of Consumer Research, 14(3), 334-349.

Sweldens, S., Van Osselaer, S.M.J., & Janiszewski, C. (2010). Evaluative conditioning procedures and the resilience of conditioned brand attitudes. Journal of Consumer Research, 37(3), 473-489.

Thompson, B. (1991). A Primer on the Logic and Use of Canonical Correlation-Analysis. Measurement and Evaluation in Counseling and Development, 24(2), 80-95.

Till, B.D., Stanley, S.M., & Priluck, R. (2008). Classical conditioning and celebrity endorsers: An examination of belongingness and resistance to extinction. Psychology & Marketing, 25(2), 179-196.

Tourangeau, R., Rasinski, K.A., Bradburn, N., & D’Andrade, R. (1989). Carryover effects in attitude surveys. The Public Opinion Quarterly, 53(4), 495-524.

Tripodi, J.A., Hirons, M., Bednall, D., & Sutherland, M. (2003). Cognitive evaluation: Prompts used to measure sponsorship awareness. International Journal of Market Research, 45(4), 435-455.

Van Osselaer, S.M.J. (2008). Associative learning and consumer decisions. In C.P. Haugtvedt, P. Herr & F.R. Kardes (Eds.), Handbook of Consumer Psychology (pp. 699-729). Mahwah, NJ: Erlbaum.

Van Osselaer, S.M.J., & Alba, J.W. (2000). Consumer learning and brand equity. Journal of Consumer Research, 27(1), 1-16.

Van Osselaer, S.M.J., & Janiszewski, C. (2001). Two ways of learning brand associations. Journal of Consumer Research, 28(2), 202-223.

Wakefield, K.L., Becker-Olsen, K., & Cornwell, T.B. (2007). I SPY A SPONSOR: The Effects of Sponsorship Level, Prominence, Relatedness, and Cueing on Recall Accuracy. Journal of Advertising, 36(4), 61-74.

Yoo, B., & Donthu, N. (2001). Developing and validating a multidimensional consumer-based brand equity scale. Journal of Business Research, 52, 1-14.

Zaichkowsky, J.L. (1986). Conceptualizing involvement. Journal of Advertising, 15(2), 4-34.Zaichkowsky, J.L. (1994). The Personal Involvement Inventory: Reduction, revision, and

application to advertising. Journal of Advertising, 23(4), 59-70.Zeynep Aksin, O., & Masini, A. (2008). Effective strategies for internal outsourcing

and offshoring of business services: An empirical investigation. Journal of Operations Management, 26(2), 239-256.

Journal of Customer Behaviour, Volume 12JCB48

APPENDIX A

Sample questionnaire

Part 1 - Associative learning variables

Event Involvement: To me the China Basketball Association Games are:

1 Important ____:____:____:____:____:____:____ Unimportant2 Boring ____:____:____:____:____:____:____ Interesting3 Relevant ____:____:____:____:____:____:____ Irrelevant4 Exciting ____:____:____:____:____:____:____ Unexciting5 Means nothing ____:____:____:____:____:____:____ Means a lot to me6 Appealing ____:____:____:____:____:____:____ Unappealing7 Fascinating ____:____:____:____:____:____:____ Mundane8 Worthless ____:____:____:____:____:____:____ Valuable9 Involving ____:____:____:____:____:____:____ Uninvolving10 Not needed ____:____:____:____:____:____:____ Needed

(Source: Adapted from Zaichkowsky, 1994.)

Emotional Experience: To what extent do you have the below feelings toward the China Basketball Association Games when you watch their games (0 = no feeling and 3 = very strong feeling).

1. Hope2. Sorrow3. Worry4. Fear

5. Excitement6. Surprising7. Trust8. Accept

9. Shame10. Loneliness11. Anger12. Sad

13. Inspiring14. Optimism15. Pride16. Desire

17. Joy18. Love19. Enjoyment20. Happiness

(Source: Adapted from Hansen, Halling & Christensen, 2006; Martensen et al., 2007.)

Event Attitude: Do you think the China Basketball Association Games are:

1 Favourable ____:____:____:____:____:____:____ Unfavourable2 Good ____:____:____:____:____:____:____ Bad3 Positive ____:____:____:____:____:____:____ Negative

(Sources: Adapted from McDaniel, 1999; Muehling & Laczniak, 1988)

Consumer Brand Knowledge: The following questions concern your experience with the following five* brands:

1. Have you ever used/owned any Li-Ning brand product? (1) Yes (2) No

2. To what extent do you agree with the following statements. (1) I am familiar with Li-Ning branded products. (a) Strongly disagree (b) Disagree (c) Neutral (d) Agree (e) Strongly agree

49Mao, Zhang, Connaughton, Holland & Spengler An associative learning account

(2) I have favourable experience with the Li-Ning branded products (a) Strongly disagree (b) Disagree (c) Neutral (d) Agree (e) Strongly agree

*The questions for four filler brands (Nike, Lenovo, Adidas, and Sony) are the same, thus omitted in this appendix.

(Sources: Adapted from Bloch et al., 1989; Lacey et al., 2010.)

Brand-Event Relatedness: The following questions concern your evaluation of the relatedness between the brand and the event - that is, whether you see a link between Li-Ning and the China Basketball Association Games if Li-Ning sponsors the event:

It is likely that participants including officials, athletes and volunteers of the China Basketball Association Games use Li-Ning branded products during the games. (a) Strongly disagree (b) Disagree (c) Neutral (d) Agree (e) Strongly agree

The China Basketball Association Games and Li-Ning have a similar image. (a) Strongly disagree (b) Disagree (c) Neutral (d) Agree (e) Strongly agree

(Source: Adapted from Gwinner & Eaton, 1999.)

Sport Involvement

How do you like sports? (a) Strongly dislike (b) Dislike (c) Neutral (d) Like (e) Strongly like

How often do you actually take part in sports? (a) No, never (b) Seldom, once a month or less (c) Sometimes, 2-4 times a month (d) Often, 2-3 times a week (e) Regularly, nearly every day

Do you watch sports on television? How often? (a) No (b) Less than once a month, not too often (c) Once or twice a month, sometimes (d) Once or twice a week (e) Three or four times a week, very often

(Source: adapted from Orlick, 1974.)

Part 2 - Branding effects variables

Nongfu Spring/Li-Ning become official sponsors of CBA

DIRECTIONS: Li-Ning recently announced that the company became official sponsors of China Basketball Association Games. The Li-Ning Corporation owns the brand “Li-Ning” and specialises in the production of sports apparel. In return, Li-Ning is authorised to use the proprietary logo of the events on their product packaging. After you read this announcement and know the sponsorship deal, to what extent your evaluation of the brand Li-Ning would change?

Journal of Customer Behaviour, Volume12JCB50

FactorNegative

ImpactSlightly

Negative Impact

No Impact

Slightly Positive Impact

Positive Impact

Attitudinal Loyalty1. I consider myself to be loyal to Li-Ning

branded products. -2 -1 0 1 22. Li-Ning branded products would be my

first choice. -2 -1 0 1 23. I will not buy other brands if Li-Ning

branded products are available at the store. -2 -1 0 1 2

Perceived Quality4. The likely quality of Li-Ning branded

products is extremely high. -2 -1 0 1 25. The likelihood that Li-Ning branded

products would be functional is very high. -2 -1 0 1 2

(Source: Adapted from Yoo & Donthu, 2001.)

Behavioural Intention6. Due to knowing this sponsorship deal, it is likely that I will buy Li-Ning branded products the next time.(a) Strongly disagree (b) Disagree (c) Neutral (d) Agree (e) Strongly agree

ABOUT THE AUTHORS AND CORRESPONDENCE

Luke Lunhua Mao is a doctoral candidate and Alumni Graduate Fellow in the Department of Tourism, Recreation and Sport Management at the University of Florida. His primary scholarly interest is consumer behaviour in sports and recreation related domains. Thus far, his research has covered several prominent sport contexts, including sponsorship, sports gambling, athletic donation, and sports service.

Corresponding author: Luke Lunhua Mao, Doctoral Student, Department of Tourism, Recreation and Sport Management, University of Florida, P.O. Box 118208, Gainesville, FL 32611-8208, USA

E [email protected]

James J. Zhang, DPE, is a professor of sport management in the Department of Kinesiology at the University of Georgia. His primary research interests include sport marketing, sport consumer behaviours, and sport organisational behaviours.

Dr James J. Zhang, Department of Kinesiology, University of Georgia, Athens, GA 30602, USA

Daniel P. Connaughton, EdD, is a professor in the Department of Tourism, Recreation and Sport Management at the University of Florida. His primary research interests include risk management and legal issues in sport.

51Mao, Zhang, Connaughton, Holland & Spengler An associative learning account

Dr Daniel P. Connaughton, Department of Tourism, Recreation, and Sport Management, University of Florida, P.O. Box 118208, Gainesville, FL 32611-8208, USA

Stephen Holland, PhD, is an associate professor in the Department of Tourism, Recreation and Sport Management at the University of Florida. His primary research interests include ecotourism, recreation management, economic impact studies and visitor behaviour.

Dr Stephen Holland, Department of Tourism, Recreation and Sport Management, University of Florida, P.O. Box 118208, Gainesville, FL 32611-8208, USA

John O. Spengler, JD, PhD, is an associate professor in the Department of Tourism, Recreation and Sport Management at the University of Florida. His primary research interests include health promotion and obesity prevention through sport and recreation, and sport law and policy.

Dr John O. Spengler, Department of Tourism, Recreation and Sport Management, University of Florida, P.O. Box 118208, Gainesville, FL 32611-8208, USA

© of Journal of Customer Behaviour is the property of Westburn Publishers Ltd or itslicensors and its content may not be copied or e-mailed to multiple sites or posted to a listservwithout first obtaining the copyright holder's express written permission. However, users mayprint, download, or e-mail articles for individual, non-commercial use only. This article hasbeen reproduced by EBSCO under license from Westburn Publishers Ltd.