Master’s Thesis
Tailoring on the Basis of Social Influence Strategies (TBSIS):
The case of online advertisements of football club related products
Name: Yasin Tunçbilek Student number: 10617256
Thesis supervisor: Dr. M.J. van der Goot Master’s programme: Communication Science Track: Persuasive Communication
Abstract
This study concentrates on the effects of tailoring on the basis of social influence strategies (TBSIS) in online advertising. The aim of the research was to examine whether online advertisements that are tailored on the basis of social influence strategies (compared to not-tailored online advertisements) positively affect perceived social value of the product, attitude towards the advertisement, attitude towards the product and purchase intention, and to what extent identification with a sports team moderates these effects of TBSIS. An online
experiment (N = 370) was conducted, whereby participants were exposed to an advertisement with a message that either contained a consensus, a scarcity or an authority strategy
implementation. Based on their Susceptibility to Persuasion Scale (STPS) scores, participants were either allocated to the tailored or the not-tailored condition. The three-way ANOVA’s revealed that TBSIS did not result in a greater positive effect on perceived social value of the product, attitude towards the advertisement, attitude towards the product and purchase
intention. Moreover, sport fan team identification did not moderate the effects of TBSIS. As a result, all eight hypotheses were rejected. However, an alternative measure for determining tailoring – TBSIS (alternative) – was proposed, which resulted in a significant main effect on perceived social value of the product; online advertisements that are tailored on the basis of social influence strategies have a greater positive effect on perceived social value of the product (H1).
Keywords: tailoring; social influence strategies; tailoring on the basis of social influence strategies; TBSIS; sport fan team identification; online advertising
Table of contents
Abstract 2
1. Introduction 5
2. Theoretical background 7
2.1. Tailoring 7
2.2. Social influence strategies 10
2.2.1. Consensus 11
2.2.2. Scarcity 12
2.2.3. Authority 14
2.3. Tailoring on the basis of social influence strategies (TBSIS) 15
2.4. Sport fan team identification 17
3. Methods 19 3.1. Design 19 3.2. Sample 20 3.3. Pilot study 20 3.3.1. PSV fan products 21 3.3.2. Texts 22 3.4. Stimuli 23 3.5. Procedure 23 3.6. Measures 25 3.6.1. Dependent variables 25 3.6.2. Moderator 26
3.6.3. Tailoring on the basis of social influence strategies 27 3.6.3.1. The Susceptibility to Persuasion Scale 27
3.6.4. Control variables 29 3.6.5. Manipulation check 30 3.6.6. Demographics 30 4. Results 31 4.1. Randomization checks 31 4.2. Manipulation check 31
4.3. Effects on perceived social value of the product 32 4.4. Effects on attitude towards the advertisement 33
4.5. Effects on attitude towards the product 34
4.6. Effects on purchase intention 35
4.7. Hypothesis testing with alternative measure of tailoring 36
4.7.1. Results with consensus condition 37
4.7.2. Results without consensus condition 38
5. Conclusion & Discussion 38
References 42
Appendices 51
A. Questionnaire Pilot study 51
B. Stimuli 56
1. Introduction
Advertising practitioners have difficulties in persuading consumers in online domains;
advertising clutter, perceived goal impediment and prior negative experiences result in people avoiding advertisements on the Internet (Cho, 2004). Tailoring – ‘fitting’ a message to an individual – could support advertising practitioners in gaining the attention of consumers, and persuading them effectively. Technological developments have provided new opportunities in persuasive communication; computers and Web technologies (e.g., cookies) are able to collect information easily and use this information to tailor a message for an individual (Dijkstra, 2008; Kreuter, Strecher & Glassman, 1999; Maslowska, Smit & Van den Putte, 2013; Noar, Benac & Harris, 2007; Rimer & Kreuter, 2006; Skinner, Campbell, Rimer, Curry &
Prochaska, 1999; Suggs & McIntyre, 2009). The effectiveness of tailoring has been proven by many studies in health communication (Hawkins, Kreuter, Resnicow, Fishbein & Dijkstra, 2008; Kreuter et al., 1999; Rimer & Kreuter, 2006; Noar et al., 2007; Skinner et al., 1999; Suggs & McIntyre, 2009), but only a few studies on tailored (online) advertisements are available (Hirsh, Kang & Bodenhausen, 2012; Maslowska et al., 2013).
Online advertising – a component of e-commerce – makes often use of social
influence strategies, which are persuasive attempts to change an individual’s beliefs, attitudes or behavior (Cialdini, 2001b; Guadagno, 2013; Guadagno & Cialdini, 2005). From a
consumer perspective, these strategies could aid consumers in their online purchase decisions. Consumers experience purchase uncertainty in e-commerce settings due to a lack of
information (Dimoka, Hong & Pavlou, 2012; Utz, Kerkhof & Van den Bos, 2012), and therefore often base their decisions on available cues (i.e. expressions of social influence). The persuasive effectiveness of social influence strategies has been demonstrated at an
average level (i.e. over groups of people) in multiple contexts (Cialdini, 2001a; Cialdini,
(2012) found individual differences in the effects of these social influence strategies, which opened new perspectives in tailoring. Subsequently, Kaptein and colleagues discovered promising results for future research in their studies that focused on tailoring on the basis of social influence strategies (hereafter TBSIS), mainly in the domain of health communication (Kaptein & van Halteren, 2012; Kaptein, de Ruyter, Markopoulos & Aarts, 2012; Mayer, Thiesse & Fleisch, 2014; Sakai, van Peteghem, van de Sande, Banach & Kaptein, 2011). However, the effects of TBSIS in e-commerce are hardly examined.
E-commerce has become one of the major profit-generating avenues for professional sports organizations (Zhang & Won, 2010). This observation is not surprising; sports fans already hold positive attitudes toward their favorite sports team and are willing to spend money on fan products. For example, fans who highly identify with a sports team are more likely to purchase team-licensed merchandise (Fisher & Wakefield, 1998; Kwon, Trail & James, 2007; Wann & Branscombe, 1993). However, fans with low sport fan team
identification require stronger persuasion attempts. Moreover, official kits and match tickets are popular fan products and more likely to be purchased compared to less familiar fan products. Thus, TBSIS could provide a means for online advertising to enhance persuasion in these situations.
The aim of this study is to examine whether tailored online advertisements have a greater persuasive effectiveness on sports fans than not-tailored online advertisements. This exploration is of interest for advertising practitioners and sports marketers to gain insights in tailoring online advertisements and to develop a possible means to predict persuasion.
Additionally, this study would contribute to the literature of the relatively new field of TBSIS in e-commerce. Pre-purchase attitudes and intention of football fans regarding a fan product will be assessed. The research question is as follows: Do online advertisements of football club related products that are tailored on the basis of social influence strategies have a greater
positive effect on perceived social value of the product, attitude towards the advertisement, attitude towards the product and purchase intention, and to what extent does the degree of identification with a sports team moderate these effects?
2. Theoretical background 2.1. Tailoring
Tailor originates from the Latin word talea, which means ‘to cut’ (Kreuter & Skinner, 2000).
In the context of communication, tailoring signifies a message that ‘fits’ an individual (Linn, 2013; Schmid, Rivers, Latimer & Salovey, 2008; Skinner, 2013). A generic message on the contrary is a single communication intended for all people (Hawkins et al., 2008; Kreuter et al., 1999; Kreuter & Skinner, 2000; Noar et al., 2007; Rimer & Kreuter, 2006; Schmid et al., 2008). Tailored communication is generated by using information about a given person to determine what specific content (s)he will receive (Hawkins et al., 2008). Such individual characteristics can be beliefs, traits, abilities, demographics, lifestyle, preferences, values, interests, needs and cognitive style (Kreuter et al., 1999; Maslowska et al., 2013).
Tailoring is used interchangeably, often incorrectly, with other concepts like
personalization or targeting (Kreuter et al., 1999). Personalized messages are often standard
messages with a person’s name incorporated in a message (Kreuter et al., 1999; Kreuter & Skinner, 2000; Noar et al., 2007; Schmid et al., 2008; Suggs & McIntyre, 2009). Some scholars claim that personalized communication is the most basic form of tailoring (Dijkstra, 2008; Hawkins et al., 2008; Schmid et al., 2008). Targeted messages are usually based on a set of demographic characteristics and are directed to a specific subgroup of the general population (Kreuter et al., 1999; Kreuter & Skinner, 2000; Noar et al., 2007; Schmid et al., 2008; Suggs & McIntyre, 2009). The current study is more in-depth with regard to tailoring.
According to Dijkstra (2008), three tailoring-ingredients – elements that differentiate between tailored and not-tailored persuasion – can be distinguished in computer-tailored persuasion: personalization (see previous paragraph), feedback and adaptation. The latter term is also called customization or matching (Maslowska et al., 2013). Feedback refers to providing individual information that relates to important personal goals, whereas adaptation signifies an adjustment of the content of a persuasive message to individual characteristics (i.e. content matching). The present study concentrates on adaptation, which is considered as the most common ingredient in computer tailoring (Dijkstra, 2008; Hawkins et al., 2008). An adapted message gives the impression that the text is intended for a general audience. The tailored aspects of the text might not be evident for a receiver, but this individual may sense a ‘connection’ with the content elements (Dijkstra, 2008).
The core of the persuasive effects of adaptation lies in a match of the content
information with a specific psychological factor, which is expected to enhance psychological processes that are responsible for stronger persuasion (Dijkstra, 2008). The Elaboration Likelihood Model (hereafter ELM) helps to understand the effects of tailoring. The ELM postulates that a receiver processes a persuasive message either centrally or peripherally (Petty & Cacioppo, 1986a; Petty & Cacioppo, 1986b). Tailoring, such as mainly applied in health communication, is intended to increase an individual’s motivation and ability to process a message. People are more likely to assess the content (e.g., the arguments) thoroughly, if they perceive the information as personally relevant (Hawkins et al., 2008; Kreuter et al., 1999; Maslowska et al., 2013; Rimer & Kreuter, 2006; Schmid et al., 2008). A receiver elaborates highly upon the message and follows a central route instead of a
peripheral route, which results in attitudes that are more persistent over time and more likely
to change permanently (Dijkstra, 2008; Hawkins et al., 2008; Kreuter et al., 1999; Maslowska et al., 2013; Petty & Cacioppo, 1986a; Petty & Cacioppo, 1986b).
Studies in health communication have demonstrated several positive outcomes of tailored messages compared to not-tailored messages. Tailoring increased the likelihood of a message to be (i) noticed, read and recalled, (ii) memorized, (iii) discussed with others, and perceived by receivers as (iv) interesting, (v) personally relevant, and (vi) written exclusively for them (Hawkins et al., 2008; Kreuter et al., 1999; Skinner et al., 1999). Moreover, tailoring has proven to encourage a positive attitude towards a message and stimulate desirable
changes in a recipient’s behavior (Kreuter et al., 1999; Rimer & Kreuter, 2006). An array of health-related behaviors are effectively promoted and changed, like smoking cessation, reduced dietary fat consumption, increased fruit and vegetable consumption, opting for a mammography, and physical activity (Kreuter et al., 1999; Noar et al., 2007; Skinner et al., 1999; Suggs & McIntyre, 2009).
Although tailoring has been mostly studied in the field of health communication, research on tailoring within the context of advertising is rising. Research by Hirsh et al. (2012) indicated that a message in a product advertisement tailored to an individual’s personality profile (i.e. the ‘Big Five’ personality dimensions) enhances the persuasive effectiveness of a message. The persuasive effectiveness of an advertisement was composed of an individual’s (i) perceived persuasiveness, (ii) perceived effectiveness, (iii) purchase intention, (iv) overall likeability of the advertisement, (v) interest in the product, and (vi) interest in learning more about the product. Maslowska et al. (2013) examined the
applicability of tailored product advertising within the Netherlands and Poland. Tailoring was only effective for Polish consumers; they perceived the tailored messages as more relevant, became more involved and were less skeptical towards the messages, which led them to have more positive message attitudes, brand attitudes and purchase intentions. The authors argue that tailored advertising may only be effective in less marketing oriented countries than countries with a profound marketing knowledge. Other research has shown that Americans
have a negative attitude towards advertisements that are tailored to their interests (Turow, King, Hoofnagle, Bleakley & Hennessy, 2011).
2.2. Social influence strategies
Social influence is a change in an individual’s beliefs, attitudes or behavior due to real or imagined external pressure(s) (Cialdini, 2001b; Guadagno, 2013; Guadagno & Cialdini, 2005). These forms of influence can be employed as strategies to persuade people. Social influence strategies are also referred to as persuasion principles, influence principles, persuasion strategies or (sales) influence tactics throughout the literature (Kaptein & Parvinen, 2014; Kaptein, Parvinen & Pöyry, 2013). A vast and varied array of influence attempts exists and scholars differ in their taxonomies (Kaptein, 2011; Kaptein, 2012; Mayer et al., 2014; Unal, Temizel & Eren, 2014).
Cialdini (2001a; 2001b; 2001c) categorized all influence attempts in six general principles: (i) consensus (also known as social validation or social proof), (ii) scarcity, (iii) authority, (iv) commitment/consistency, (v) liking/friendship, and (vi) reciprocity. Cialdini’s principles are simple and parsimonious (Kaptein, Lacroix & Saini, 2010; Mayer et al., 2014), and a large body of research on social influence focuses on his taxonomy (Kaptein, 2011; Kaptein, Markopoulos, de Ruyter & Aarts, 2015; Unal et al., 2014). The persuasive effectiveness of these social influence strategies has been proven repeatedly at an average level in multiple contexts (Cialdini, 2001a; Cialdini, 2001b; Cialdini, 2001c; Cialdini & Sagarin, 2005; Kaptein, 2012; Kaptein & van Halteren, 2013; Kaptein et al., 2015; Sakai et al., 2011).
The present research concentrates on the social influence strategies ‘consensus’, ‘scarcity’ and ‘authority’. According to Kaptein and colleagues, these strategies are the most pervasive; their implementation is common in online sales and marketing, whereby a positive
average effect is expected on consumers’ attitudes towards an associated product and their likelihood of product purchases (Kaptein & Duplinsky, 2013; Kaptein & Eckles, 2012).
2.2.1. Consensus
Consensus refers to an individual’s tendency to copy the beliefs, the attitudes or the behavior of a group of people of who (s)he believes that acting in a similar manner is socially
acceptable (Cialdini, 2001a; Cialdini, 2001b; Cialdini, 2001c; Cialdini & Sagarin, 2005; Goldstein, Cialdini & Griskevicius, 2008; Unal et al., 2014). Individuals are more
comfortable to form an opinion or to take an action when they are aware that numerous other individuals have already completed this given action (Kaptein & Dusplinsky, 2013). This effect is even larger when an individual is in an unfamiliar or an uncertain situation; (s)he does not know how to act appropriately within this setting and will therefore seek social proof (Amblee & Bui, 2011). Individuals encounter many similar situations on the Internet (e.g., when ascertaining the quality of a product or a service) (Dimoka et al., 2012; Guadagno, 2013), which makes online domains appealing for applying consensus strategy
implementations.
Common implementations of a consensus strategy in e-commerce settings are popularity claims, which are firm published statements about a product or a service that display its popularity. Concrete examples of popularity claims are “95% of the consumers purchased this product”, “Already 10.000 consumers bought this product”, “This product is a bestseller” and “This product is by the consumers rated as the best”.
Freling and Dacin (2010) revealed that advertisements with popularity claims increase a consumer’s purchase intention. Moreover, this effect depends on the interplay between a high level of consensus in a claim, the group that is presented in a claim to whom a receiver feels connected to, and a high processing motivation of a receiver. Another study
confirmed the effect of a consensus strategy: a popularity claim boosted product quality perception and purchase intention, probably as a result of a lack of available information and the bandwagon-effect of the claim (Jeong & Kwon, 2012). Coulter and Roggeveen (2012) showed that a popularity claim at an online group buying website (e.g., Groupon) enhances the positive effect on perceived product value and anticipated regret, which mediate their effect on purchase intention. Chang (2012) demonstrated that consumers, but only females, generate higher purchase intentions for advertisements with consensus cues as opposed to advertisements without. However, Dean (1999) indicated that a popularity cue does not elicit a positive effect on pre-purchase attitudes (perceived product quality, perceived product uniqueness, manufacturer esteem and perceived corporate citizenship).
2.2.2. Scarcity
Scarcity refers to an individual’s tendency to value a product, a service or an opportunity more when (s)he perceives these as exclusive, rare, limited or one-of-a-kind (Cialdini, 2001a; Cialdini, 2001b; Cialdini, 2001c; Cialdini & Sagarin, 2005; Kaptein & Duplinsky, 2013). A product or a service that is assumed to be scarce, is expected to elicit positive consumer attitudes and to increase the purchase likelihood (Eisend, 2008). Moreover, individuals are inclined to act fast in this given situation, because they strongly believe that this opportunity will not arise again. A prominent explanation for the effects of scarcity is that individuals acquire feelings of personal distinctiveness or uniqueness when possessing scarce products (Lynn, 1991). The relative lack of need of contextual awareness and low elaboration makes scarcity strategy implementations easily applicable within the context of online advertising (Kaptein & Duplinsky, 2013).
Scarcity strategy implementations, regardless of whether these implementations are present in an online or an offline setting, are either based on a time limit or a quantity limit
(Gierl & Heuttl, 2010; Gierl, Plantsch & Schweidler, 2008). Concrete examples of messages with a time limit are “Only today you receive a special discount rate on this product” and “This product is only available this week”. Quantity limit examples are “This product is a limited edition” and “There are only 5 items left of this product”. The present study concentrates on both time limit and quantity limit implementations in commerce.
Bae and Lee (2005) demonstrated that a message with a scarcity strategy
implementation, either a time limit or a quantity limit, in an online shopping environment increases a consumer’s purchase intention. This effect is even greater when both levels of a consumer’s product involvement and product knowledge are low. Another research concluded that messages with a quantity limit in advertisements are more effective in influencing
consumers’ purchase intentions than messages with a time limit; a greater perceived consumer competition, due to the assumed limited product amount, is found to mediate this effect (Aggarwal, Jun & Huh, 2011). Eisend (2008) showed that an advertisement with a scarcity strategy implementation enhances perceived product value, which again impacts purchase intention. Moreover, the impact of perceived product value on purchase intention is mediated by whether the advertisement positively changed an individual’s attitude towards the product, and whether an individual believes that the advertisement positively changed the attitude of other individuals. According to Coulter and Roggeveen (2012), a scarcity claim on a group buying website increases perceived product value and anticipated regret, which mediate their effect on purchase intention. Furthermore, messages with a supply-related scarcity appeal in advertisements are less likely to activate persuasion knowledge than messages with a demand-related scarcity appeal. This effect is moderated by message specificity; stating an appeal in specific terms compared to vague terms decreases the persuasiveness of messages with a supply-related scarcity appeal in advertisements (Aguirre-Rodriguez, 2013). However, Jeong and Kwon (2012) found no significant effect of a limited availability claim on purchase
intention; the authors ascribed this finding to low message credibility and a lack of psychological reactance.
2.2.3. Authority
Authority refers to an individual’s tendency to act in line with an endorsement or a request that is made by a legitimate authority, and to perceive the information that is provided by this authority as credible. Individuals want to make the correct choice and base their actions on the advice of a domain relevant authority of who they expect to possess information that is not available for consumers. As a result, individuals identify authority figures as experts and believe that their statements are ‘true’ (Cialdini, 2001a; Cialdini, 2001b; Cialdini, 2001c; Cialdini & Sagarin, 2005; Guadagno, 2013). An explanation for the effectiveness of authority is the premise that every social community requires some levels of responsibility and
obedience to authority in order to exist (Cialdini, 2001b). Authority strategy implementations are simple and easily applicable (Kaptein & Duplinksky, 2013), and therefore attractive to change and influence consumer behavior in e-commerce domains.
Authority strategy implementations exist in the form of highly valued consumer and official authority endorsements such as reviews or statements on a product or a service. Concrete examples of authority strategy implementations are “We would recommend this toothpaste to anyone. – American Dental Association” and “I would recommend this book to everyone. – Stephen King, famous American author”.
Most of the studies that discuss the ability of authority figures or experts to influence attitude or behavior change are grounded in social psychology, such as the well-known Milgram experiments (Milgram, 1974). Sundar, Xu and Oeldorf-Hirsch (2009) revealed that an authority cue in an online shopping environment positively influences purchase decisions. Moreover, an expert source in commercial settings leads to greater positive attitudes towards
the endorser and the advertisement than a non-expert source (Braunsberger, 1996). Cialdini emphasized that an authority strategy implementation is successful when used as a decision heuristic on the Internet (Guadagno & Cialdini, 2005).
2.3. Tailoring on the basis of social influence strategies (TBSIS)
TBSIS is based on the expected effects of different social influence strategies separately for a specific individual (Kaptein, 2011; Kaptein, 2012; Kaptein et al., 2015). Kaptein and Eckles (2012) discovered individual differences in the effects of social influence strategies. TBSIS, as proposed by Kaptein (2012), implies using a social influence strategy of which is expected to be most effective for an individual.
There are two distinct methods to estimate which social influence strategy will be the most effective for each individual: implicit tailoring and explicit tailoring (Kaptein et al., 2015). Implicit tailoring is established with the Susceptibility to Persuasion Scale (hereafter STPS; the STPS will be discussed more extensively in the methods section). The STPS is a questionnaire for measuring an individual’s degree of susceptibility to each social influence strategy in order to determine tailoring (Kaptein et al., 2012; Kaptein et al., 2015). Explicit
tailoring is guided by persuasive technologies; an adaptive system records an individual’s
actual responses to social influence strategy implementations over a period of time in order to adequately tailor future interactions. A combination of both methods can also be applied for tailoring (Kaptein et al., 2015). For the experiment in the present research, the implicit tailoring method will be adopted due to its feasibility. Nonetheless, the effects of tailoring with both methods are discussed in the next paragraphs.
Only a few studies are available on TBSIS as suggested by Kaptein, mainly in the domain of health communication. Kaptein et al. (2012) examined snacking behavior with a two-week text-messaging intervention. Participants were asked to self-report their snack
consumption during this period of time. The study revealed that messages tailored to an individual based on the a priori STPS scores (consensus, scarcity, authority and
commitment/consistency) result in a greater decrease in snack consumption than messages with a random social influence strategy implementation or messages that are not-tailored (i.e. contra-tailored). Mayer et al. (2014) confirmed this effect of tailoring on snacking behavior. However, there was no significant difference in snack consumption between a tailored message and a random message. Sakai et al. (2011) studied stair and elevator usage amongst office workers in a real life setting for a period of five weeks. Results demonstrated that messages tailored to an individual (consensus, authority and commitment/consistency) increase stair climbing.
Another study (Kaptein & van Halteren, 2012), more or less in the field of health communication, examined user engagement in a health and lifestyle service for a period of six months. Users are coached via a web service, whereby their activity data is only analyzed when uploaded to the web service. To encourage docking – the uploading of the activity data to the web service – docking reminders are sent via e-mail. The outcomes displayed that tailoring is partly successful: e-mails tailored to an individual (consensus, scarcity and authority) result in a higher docking rate than e-mails without a social influence strategy implementation or with the best pre-tested implementation (authority). The docking rate was also higher in the tailored condition than in the random condition, but the difference was insignificant. Moreover, the average click through rate in the tailored condition was higher than in the no strategy condition. Lastly, a significant interaction effect was found between the timing of the e-mail and the condition, indicating that tailored e-mails become more effective over time.
Only one research was conducted in the field of e-commerce; Kaptein (2011) assessed the click through and conversion rates on an online store for children’s clothing
(www.kinder-kleertjes.com) with an adaptive persuasive system in comparison to the same online store without an adaptive persuasive system. Results indicated that product presentations tailored to an individual (consensus and authority) lead to an increase in click through and conversion rates, whereas in the not-tailored condition the rates were lower due to the presentation of an ‘incorrect’ social influence strategy.
Although there is little empirical evidence on the effect of TBSIS, and more specifically in the domain of e-commerce, the expectation is that tailoring will positively influence pre-purchase attitudes and intention; particularly perceived social value of the product, attitude towards the advertisement, attitude towards the product and purchase intention. This expectation is grounded on the previously presented evidence of tailoring and social influence strategies in general, and the promising first results of studies that focused on TBSIS. The following hypotheses, also displayed in the conceptual model (Figure 1), are therefore constructed:
Online advertisements that are tailored on the basis of social influence strategies have a greater positive effect on perceived social value of the product (H1), attitude towards the advertisement (H2), attitude towards the product (H3) and purchase intention (H4) than not-tailored online advertisements.
2.4. Sport fan team identification
Sport fan team identification is defined as the degree to which an individual fan feels psychologically linked to a sports team (Greenwood, Kanters & Casper, 2006; Madrigal, 2000; Wann & Branscombe, 1995), or is concerned about it (Madrigal, 2000; Wann &
Branscombe, 1995). The concept of team identification originates from social identity theory, which points out the communality of a group of fans in relation to their favorite sport team
(Kwon et al., 2007). Individual fans are more likely to identify with a sport team that
represents the attributes they have assigned to their own self-concepts (Fink, Parker, Brett & Higgins, 2009).
There is sufficient empirical evidence available of sport fan team identification having positive effects on many sport fan behaviors (Greenwood et al., 2006). More precisely, sport fan team identification is a strong predictor of sport fan consumption behavior (Fink, Trail & Anderson, 2002). Highly identified fans, compared to fans low in identification, are more likely to (i) attend matches (Fisher & Wakefield, 1998; Wann & Branscombe, 1993; Wann, Roberts & Tindall, 1999), (ii) follow matches via media (Fisher & Wakefield, 1998; Kwon et al., 2007), (iii) spend time and money when they watch their team play (Wann &
Branscombe, 1993; Wann et al., 1999), and (iv) purchase team-licensed merchandise (Fisher & Wakefield, 1998; Kwon et al., 2007; Wann & Branscombe, 1993). Özer and Argan (2006) demonstrated that identification with the team is the most important factor in the process of purchasing team-licensed merchandise. Lastly, sport fan team identification also elicits a positive attitude towards a sponsor’s product (Madrigal, 2000). In line with this literature, fans high in identification are expected to have a greater perceived social value of the product, attitude towards the advertisement, attitude towards the product and purchase intention than fans low in identification.
However, the effect of TBSIS on the dependent variables should be stronger for fans with low sport fan team identification compared to fans with high sport fan team
identification. This assumption is based on the fact that highly identified fans already possess greater positive attitudes than fans low in identification. They are therefore less susceptible to TBSIS, because there is less opportunity for a positive change in their attitudes. The
The effect of tailoring on perceived social value of the product (H5), attitude towards the advertisement (H6), attitude towards the product (H7) and purchase intention (H8) is stronger for participants with low sport fan team identification compared to participants with high sport fan team identification.
Figure 1. Conceptual model
3. Methods 3.1. Design
The study featured a 2 (tailoring on the basis of social influence strategies (TBSIS): tailored vs. not-tailored) X 3 (type of social influence strategy: consensus vs. scarcity vs. authority) X 2 (sport fan team identification: low vs. high) between-subjects factorial design with
perceived social value of the product, attitude towards the advertisement, attitude towards the product and purchase intention as dependent variables.
H1 H4 H2 H3 TBSIS: tailored vs. not-tailored Perceived social value of the product Attitude towards the advertisement Attitude towards the product Purchase intention Sport fan team
identification: low vs. high
3.2. Sample
The sample of the main study consisted of fans of the Dutch professional football club PSV Eindhoven. Respondents were required to understand Dutch and had to be at least 18 years old in order to partake in the online experiment, which was completed by 372 participants. A PSV FANstore voucher with a value of €10,- was raffled among them. Two participants were excluded from the initial sample. One participant indicated in the end of the survey that he supported the Dutch professional football club Ajax Amsterdam and had responded to the questions as if they were targeted at an Ajax fan. Another participant answered consequently in a straight line on all the items of all the scales and filled out a fake age (i.e. 99). There were no other significant cases detected on the measured variables.
The final sample contained 370 participants, and consisted of 322 males (87 %) and 48 females (13 %). The age of the participants ranged from 18 to 71 years (M = 32.31, SD = 11.18). The education level of the respondents was mainly higher professional education (N = 172, 46.50 %) and scientific education (N = 112, 30.30 %), followed by intermediate
vocational education (N = 64, 17.30 %), secondary education (N = 20, 5.40 %) and primary education (N = 2, .50 %).
3.3. Pilot study
An online pilot study (see Appendix A) in Dutch among 36 PSV fans was conducted in Qualtrics (www.qualtrics.com) to gain input for designing the stimuli for the main study. The pilot study provided an answer on (i) which PSV fan product (PSV backpack or PSV
Museum) to implement in the stimuli, and (ii) whether the suggested texts about the presented fan products reflected the social influence strategies (consensus, scarcity and scarcity) as intended and were suitable to incorporate in the stimuli.
The sample of the pilot study consisted of 28 males (77.80 %) and 8 females (22.20 %). The age of the participants ranged from 18 to 66 years, with a mean age of 35.25 years (SD = 13.45). The education level of the respondents was mainly intermediate vocational education (N = 15, 41.70 %) and higher professional education (N = 14, 38.90 %), followed by scientific education (N = 4, 11.10 %) and secondary education (N = 3, 8.3 %).
3.3.1. PSV fan products
The PSV backpack and the PSV Museum were selected as the fan products for the pilot study because of their (i) difference in product type (search vs. experience), (ii) expected relative unfamiliarity among the PSV fans compared to other fan products within their product type (e.g., search: a PSV kit or scarf, experience: a PSV match or stadium tour), (iii) low
involvement character, and (iv) compatibility with the three selected strategies (e.g., a scarcity strategy implementation for free services like a subscription for the PSV newsletter or the official app is not realistic). Participants were in a random order exposed to both an image of the PSV backpack and the PSV Museum, which were obtained from the official PSV websites (www.psvfanstore.nl and www.psv.nl).
Participants were asked two questions after exposure to the image of the PSV
backpack: (1) “Do you already have the PSV backpack in your possession?” (answer options; yes/no), and (2) “How likely is it that you will purchase the PSV backpack (again) in the nearby future?” (answer options on a 7-point bipolar scale; 1 = very unlikely, 7 = very likely). Similar questions were asked for the PSV Museum, except then focused on the visiting
behavior.
The pilot study revealed that the PSV Museum was the best option to implement in the stimuli. None of the participants possessed the PSV backpack, and a participant purchasing the PSV backpack was unlikely (M = 1.86, SD = 1.29). On the contrary, 15 participants
(41.70 %) already visited the PSV Museum, whereas 21 participants (58.30 %) did not. The likelihood of a participant visiting the PSV Museum (again) was moderately high (M = 5.22,
SD = 1.51). A paired samples t-test (t (35) = -9.86, p = < .001) showed that the likelihood of
participants visiting the PSV Museum (again) was significantly higher than the likelihood of participants purchasing the PSV backpack.
3.3.2. Texts
Participants were requested to indicate their opinion on statements about the PSV Museum that examined to what degree the created texts (in Dutch) represented a particular social influence strategy1: (i) texts: “Popular” and “The PSV Museum has been already visited by many PSV fans!”; statement: “I think the PSV Museum is visited often” (consensus), (ii)
texts: “Special offer” and “Only this month the first 500 PSV fans receive a discount rate
when visiting the PSV Museum!”; statement: “I think the tickets with a discount for the PSV Museum are limited available this month” (scarcity), (iii) texts: “Recommended” and “The PSV Museum is a must visit for all PSV fans! – Toon Gerbrands, General Manager at PSV”;
statement: “I think Toon Gerbrands is an expert regarding the PSV Museum” (authority 1),
and (iv) texts: “Recommended” and “The PSV Museum is a must visit for all PSV fans! – The Dutch Museum Association”; statement: “I think the Dutch Museum Association is an expert regarding the PSV Museum” (authority 2). The statements were assessed on a 7-point Likert scale (1 = strongly disagree, 7 = strongly agree).
The texts represented the social influence strategies moderately strong, but sufficiently to incorporate in the stimuli. Participants perceived the texts as follows: (i) consensus (M = 5.11, SD = 1.98), (ii) scarcity (M = 4.61, SD = 1.64), (iii) authority 1 (M = 4.44, SD = 1.66), and (iv) authority 2 (M = 4.31, SD = 1.80). A paired samples t-test indicated that there was no
significant difference between how participants perceived Toon Gerbrands (authority 1) and The Dutch Museum Association (authority 2) as an expert with regard to the PSV Museum, t (35) = .48, p = .632. However, Toon Gerbrands was perceived slightly more as an expert than The Dutch Museum Association, and was therefore chosen for the stimuli. The amount of 500 in the scarcity strategy implementation was reduced to 100 to strengthen the magnitude of this social influence strategy.
3.4. Stimuli
Three stimuli (see Appendix B) were designed for the main study. The advertisements were designed as if they could have been shown on the splash page of PSV’s official website (www.psv.nl). The image of the PSV Museum used in the stimuli was obtained from the webpage of the PSV Museum on the PSV website. The texts in each advertisement were in Dutch and reflected one of the three social influence strategies. The consensus strategy advertisement contained the texts “Popular” and “The PSV Museum has been already visited
by many PSV fans!”. The texts for the scarcity strategy advertisement were “Special offer”
and “Only this month the first 100 PSV fans receive a discount when visiting the PSV
Museum!”. An authority strategy implementation in the advertisement was embodied by the
texts “Recommended” and “The PSV Museum is a must visit for all PSV fans! – Toon
Gerbrands, General Director of PSV”. All other characteristics of the advertisements were
held equal across the designs to minimize the impact of extraneous factors.
3.5. Procedure
The experiment was conducted online via Qualtrics (see Appendix C). An invitation to participate in the experiment was sent out to PSV fans via Whatsapp, a PSV forum website (psv.netwerk.to), and social media platforms (Facebook, LinkedIn and Twitter). The
invitation was placed in Facebook groups and a LinkedIn group solely for PSV fans. A private Facebook event was created for the experiment, whereby PSV fans were able to invite Facebook friends who also supported PSV. In addition, employees of PSV, footballers of PSV, journalists and famous Dutch PSV fans were via Twitter asked to share the experiment with their followers. A link to the Qualtrics experiment was included in the invitation, and where possible, a short description of the study was provided. Participants were able to complete the survey at their own pace and in their own environment.
Before starting the survey, participants were requested for their informed consent to legitimize the use of data by the Communication Science department of the University of Amsterdam. Respondents were reminded to contact the researcher or the Amsterdam School of Communication Research (ASCoR) by e-mail in case of a complaint, question or remark about the study. After agreement, participants were randomly assigned to either an
advertisement with a consensus, scarcity or authority strategy implementation. They were asked to assess the presented advertisement carefully as if the advertisement was shown when visiting the official PSV website. Subsequently, all respondents filled out the dependent measures in the following order: perceived social value of the product, attitude towards the advertisement, attitude towards the product and purchase intention. In addition, the moderator measure sport fan team identification was examined. Next, TBSIS was assessed with several items derived from the STPC. Furthermore: control variables questions were administered; participants were asked what their thoughts were on the purpose of the study; and a
manipulation check was conducted to confirm the manipulation. Finally, participants answered three demographic questions. In the end, respondents were not only thanked for their time and effort, but were also given the opportunity to leave comments and their e-mail address to receive the results of the experiment and compete for the voucher.
3.6. Measures
3.6.1. Dependent variables
Perceived social value of the product – The measure for perceived social value of the
product was adopted from the validated multidimensional PERVAL (i.e. perceived value of a product or a service) scale of Sweeney and Soutar (2001). Perceived social value of the product is one of the four dimensions on the PERVAL scale. The other dimensions are ‘quality/performance (functional value)’, ‘price/value for money (functional value)’ and ‘emotional value’, and were not assessed in this study. Perceived social value of the product was measured with four items: (1) “A visit to the PSV Museum would help me to feel accepted”, (2) “A visit to the PSV Museum would improve the way I am perceived”, (3) “A visit to the PSV Museum would make a bad impression on other people” (reverse coded), and (4) “A visit to the PSV Museum would give the visitor social approval”. The items were administered on a 7-point Likert scale (1 = strongly disagree, 7 = strongly agree). After recoding the third item, all items loaded on the same factor (EV = 2.64, R2 = .66). The scores on the individual items were averaged to compute a scale score (α = .65, M = 3.27, SD = .93).
Attitude toward the advertisement (Aad) – Aad was measured with the validated scale
of Mitchell and Olson (1981). Four items were assessed on a 7-point semantic differential scale (bad/good, dislike/like, not annoying/annoying (reverse coded) and
uninteresting/interesting). All items, after recoding the third item, loaded on the same factor (EV = 2.45, R2 = .61). The scores from the separate items were averaged to calculate an overall Aad score (α = .77, M = 4.39, SD = 1.15).
Attitude toward the product (Ap) – Ap was examined with the same validated scale of
Mitchell and Olson (1981); the items were now targeted at the product. Again, four items were measured on a 7-point semantic differential scale (bad/good, dislike/like, not
item, all items loaded on the same factor (EV = 2.72, R2 = .68). The average of the four items were merged into an overall Ab score (α = .83, M = 4.78, SD = 1.21).
Purchase intention – Two altered items from the willingness to buy indicators of
Dodds, Monroe and Grewal (1991) were used to measure purchase intention: (1) “I am considering ordering tickets for the PSV Museum”, and (2) “I would definitely order tickets for the PSV Museum”. The items were administered on a 7-point Likert scale (1 = strongly disagree, 7 = strongly agree). The scores on the individual items were averaged to compute a scale score (α = .85, M = 3.20, SD = 1.49).
3.6.2. Moderator
Sport fan team identification – Sport fan team identification was measured with the validated
Sport Spectator Identification Scale (SSIS) of Wann and Branscombe (1993). The original 7 items of the SSIS were slightly adapted for this study and assessed in the following order: (1) “How important to you is it that PSV wins?” (1 = not important, 7 = very important), (2) “How strongly do you see yourself as a fan of PSV?” (1 = not at all a fan, 7 = very much a fan), (3) “During the season, how closely do you follow PSV via any of the following: a) attending matches or trainings, or on television, b) on the radio, c) television news, a newspaper or magazine, or d) the internet?” (1 = almost every day, 7 = never) (reverse coded), (4) “How strongly do your friends see you as a fan of PSV?” (1 = not at all a fan, 7 = very much a fan), (5) “How important is being a fan of PSV to you?” (1 = not important, 7 = very important), (6) “How much do you dislike PSV’s greatest rivals?” (1 = dislike very much, 7 = do not dislike) (reverse coded), and (7) “How often do you display PSV’s name or insignia outside matches or trainings, like at your work and/or in private life?” (1 = never, 7 = always). The questions are normally administered on an 8-point Likert scale, but were in this study measured on a 7-point scale to maintain consistency with the rest of scales. All items
loaded, after recoding the third and sixth item, on the same factor (EV = 3.74, R2 = .53). Averaging the individual items resulted in a scale score (α = .82, M = 5.56, SD = .91).
A median split on the sport fan team identification scores was performed to create two groups (low vs. high) in order to assess the hypotheses concerning an interaction effect of sport fan team identification. The median of sport fan team identification was 5.71.
Participants with a sport fan team identification score lower than the median were assigned to the low group (N = 179), whereas the remaining participants were allocated to the high sport fan team identification group (N = 191). As a conclusion, participants in the former group can be seen as the ‘average’ PSV fans, and participants in the latter group as the more ‘passionate’ PSV fans.
3.6.3. Tailoring on the basis of social influence strategies 3.6.3.1. The Susceptibility to Persuasion Scale
The validated, 26-items, multidimensional STPS of Kaptein et al. (2012) examines to what extent an individual is susceptible to a given social influence strategy. Only the items of the subscales ‘consensus’ (4 items), ‘scarcity’ (5 items) and ‘authority’ (4 items) of the STPS were assessed. An average susceptibility score on each subscale was calculated.
Consensus – Consensus was measured with four items: (1) “If someone from my
social network notifies me about a good book, I tend to read it”, (2) “When I am in a new situation I look at others to see what I should do”, (3) “I often rely on other people to know what I should do”, and (4) “It is important to me to fit in”. The items were administered on a 7-point Likert scale (1 = completely disagree, 7 = completely agree). A factor analysis was not performed; the validity of the subscale has been proven (Kaptein et al., 2012). Averaging the scores on the four individual items resulted in a scale score (α = .40, M = 4.09, SD = .93).
However, to increase the internal consistency of the scale, the first item was excluded (α = .60, M = 3.96, SD = 1.13).
Scarcity – Scarcity was assessed with five items: (1) “I believe rare products (scarce)
are more valuable than mass products”, (2) “When my favorite shop is about to close, I would visit it since it is my last chance”, (3) “I would feel good if I was the last person to be able to buy something”, (4) “When my favorite shampoo is almost out of stock I buy two bottles.”, and (5) “Products that are hard to get represent a special value”. The items were measured on a 7-point Likert scale (1 = completely disagree, 7 = completely agree). There was no factor analysis conducted due to the confirmed validity of the subscale (Kaptein et al., 2012). The scores from the separate items were averaged to compute a scale score (α = .60, M = 4.23, SD = 1.00).
Authority – Authority was measured with four items: (1) “I am very inclined to listen
to authority figures”, (2) “I always obey directions from my superiors”, (3) “I am more inclined to listen to an authority figure than a peer”, and (4) “I am more likely to do
something if told, than when asked”. The items were measured on a 7-point Likert scale (1 = completely disagree, 7 = completely agree). Again, a factor analysis was not required because of the established validity of the subscale (Kaptein et al., 2012). The scores on the individual items were averaged to calculate a scale score (α = .65, M = 3.64, SD = 1.02).
3.6.3.2. Tailoring
An independent variable ‘TBSIS’ with two conditions (tailored vs. not-tailored) was
computed in order to test the hypotheses. Allocating participants to either the tailored or the not-tailored condition was determined on the basis of which condition (consensus vs. scarcity vs. authority) a participant was exposed to, in conjunction with whether a participant’s
subscales (e.g., a participant in the consensus condition WITH a highest average susceptibility score on the consensus subscale, compared to the scores of the other two subscales, was allocated to the tailored condition). When average susceptibility scores on two (or on all three) subscales were highest and thus equal, a participant was also assigned to the tailored condition. All other situations resulted in participants being placed in the not-tailored condition. Of the 370 valid cases, 145 participants (39.20 %) were allocated to the tailored condition, whereas 225 participants (60.80 %) were assigned to the not-tailored condition.
3.6.4. Control variables
Product involvement – The validated involvement scale of Zaichkowsky (1985) was used to
assess involvement with the product category ‘museum’. The adjusted scale consists of ten items that were measured on a 7-point semantic differential scale: (1) important/unimportant (reverse coded), (2) boring/interesting, (3) relevant/irrelevant (reverse coded), (4)
exciting/unexciting (reverse coded), (5) means nothing/means a lot, (6) fascinating/mundane (reverse coded), (7) appealing/unappealing (reverse coded), (8) worthless/valuable, (9) not needed/needed, and (10) involving/uninvolving (reverse coded). The individual items were averaged to compute an overall product involvement score (α= .92, M = 4.29, SD = 1.16) (1 = very low involvement, 7 = very high involvement).
Product familiarity – Product familiarity with the PSV Museum was measured with
the question: “Did you, before you started participating in this study, know of the existence of the PSV Museum?” (answer questions; yes/no). A majority of the participants (N = 337, 91.10 %) was prior to the experiment familiar with the PSV Museum, whereas 33 participants (8.90 %) were not.
Product experience – Product experience with the PSV Museum was examined with
yes/no). 88 Participants (23.80 %) already visited the PSV Museum in the past, whereas 282 participants (76.20 %) did not.
Website involvement – Website involvement with the official PSV website was
assessed by asking how often participants visited the official PSV website (“How often do you visit the official website of PSV?”). Answer options included: 1 = never, 2 = less than once a month, 3 = one to three times a month, 4 = less than once a week, 5 = one to three times a week, 6 = four to six times a week, 7 = and on a daily basis. The average website involvement score was 4.24 (SD = 1.69) (1 = very low involvement, 7 = very high involvement).
3.6.5. Manipulation check
At the end of the experiment a manipulation check was conducted to check whether the participants perceived the manipulation as intended. The participants were asked to indicate their opinion on three statements related to the PSV Museum in the presented advertisement: (i) “Based on the advertisement I have just seen, I think the PSV museum is popular among the PSV fans”, (ii) “Based on the advertisement I have just seen, I think there is a special offer when visiting the PSV Museum”, and (iii) “Based on the advertisement I have just seen, I think the PSV Museum is recommended by a person or an authority”. The items were measured on a 7-point Likert scale (1 = completely disagree, 7 = completely agree).
3.6.6. Demographics
Lastly, age, gender and education level (answer options; no education, primary education, secondary education, intermediate vocational education, higher professional education and scientific education) were administered. Education level was divided in a group with low education (N = 86, 23.20 %) and a group with high education (N = 284, 76.8 %).
4. Results 4.1. Randomization checks
There was no significant difference between the two conditions of TBSIS on product involvement (t (368) = -.49, p = .627), product familiarity (X2 (1, N = 370) = .16, p = .690), product experience (X2 (1, N = 370) = 1.26, p = .262), website experience (t (368) = 1.14, p = .255), age (t (368) = -.41, p = .683), gender (X2 (1, N = 370) = .33, p = .566) and education level (X2 (1, N = 370) = .11, p = .744). Therefore, these variables were not taken into account as control variables in the analyses.
4.2. Manipulation check
Three one-way ANOVA’s were conducted to assess whether the manipulation of the three social influence strategies succeeded. The first ANOVA (F (2, 367), = 22.53, p = < .001) showed that participants in the scarcity condition (M = 4.98, SD = 1.68) perceived the scarcity strategy implementation in the advertisement significantly higher on the presence of a special offer, and thus as intended, than participants who were confronted with a consensus strategy implementation (M = 3.79, SD = 1.55) or an authority strategy implementation (M = 3.84, SD = 1.59) in the advertisement. A Bonferroni test revealed that participants in the scarcity condition scored significantly higher on scarcity strategy perception in comparison to participants of both the consensus condition and the authority condition (both ps = < .001).
The second ANOVA (F (2, 367) = 19.18, p = < .001) demonstrated a significant difference on authority strategy perception between participants in the authority condition (M = 5.16, SD = 1.44), participants in the consensus condition (M = 4.10, SD = 1.58) or the scarcity condition (M = 4.11, SD = 1.55). A Bonferroni test confirmed that participants who were exposed to the advertisement with an authority strategy implementation perceived the given implementation significantly greater on the presence of a recommendation by an
organization or a person in the advertisement than participants in both the consensus condition and the scarcity condition (both ps = < .001.).
The third ANOVA (F (2, 367) = .02, p = .983) displayed that participants in the consensus condition (M = 3.97, SD = 1.41) perceived the consensus strategy implementation in the advertisement not significantly higher on the PSV Museum as a popular fan product as presented in the advertisement, and thus not as intended, than participants in the scarcity condition (M = 3.95, SD = 1.53) or participants in the authority condition (M = 3.94, SD = 1.46). In conclusion, the manipulation in the scarcity condition and the authority condition was successful, but the consensus strategy implementation was not.2
4.3. Effects on perceived social value of the product
The first hypothesis (H1) stated that participants who were exposed to online advertisements that are tailored on the basis of social influence strategies would perceive a greater social value of the product than participants who were exposed to not-tailored online
advertisements. According to the fifth hypothesis (H5), this effect would be stronger for participants with low sport fan team identification compared to participants with high sport fan team identification.
A three-way ANOVA3 with TBSIS, type of social influence strategy and sport fan team identification as the three factors displayed an insignificant main effect of TBSIS, F = (1, 358) = 1.16, p = .282, η2 = .003. Furthermore, there was no significant main effect of type
2 The hypotheses were also tested without the participants in the consensus condition. The same results were found regarding the hypotheses.
Therefore, only the results with all three conditions of type of social influence strategy are reported.
3 Two important assumptions of a three-way ANOVA to provide valid results are (i) a normal distribution of the dependent variable, and (ii)
a homogeneity of variances across each combination of the groups of the three factors. Due to the large amount of groups (3 X 2 X 2 = 12), ‘perceived social value of the product’ was checked on a normal distribution on the complete sample size (N = 370). A Shapiro-Wilk’s test (p < .001), and an examination of the histogram, the Q-Q plot and the box plot, showed that perceived social value of the product was not normally distributed, with a skewness of .853 (SE = .127) and a kurtosis of .066 (SE = .253). However, these values are acceptable (between 1 and -1) for conducting a three-way ANOVA. The homogeneity of variances for each combination of the groups of the three factors was assessed. The F-value for Levene’s test was 2.17, with a significant p-value of .016. The assumption of homogeneity of variances was
of social influence strategy, F = (2, 358) = 1.54, p = .216, η2 = .009, and sport fan team identification, F (1, 358) = 3.21, p = .074, η2 = .009.
The analysis demonstrated no significant interaction effect between TBSIS and sport fan team identification, F (1, 358) = .18, p = .68, η2 < .001. In addition, there was no
significant interaction effect between TBSIS and type of social influence strategy, F (2, 358) = 1.18, p = .310, η2 = .007, nor an interaction effect between type of social influence strategy and sport fan team identification, F (2, 358) = .24, p = .791, η2 = .001. Lastly, no significant interaction effect occurred with all factors combined, F (2, 358) = .54, p = .582, η2 = .003.
The insignificant results of the main effect of TBSIS and the interaction effect between TBSIS and sport fan team identification lead to a rejection of H1 and H5.
4.4. Effects on attitude towards the advertisement
The second hypothesis (H2) indicated that participants who were exposed to online
advertisements that are tailored on the basis of social influence strategies would have a greater positive attitude towards the advertisement than participants who were exposed to not-tailored online advertisements. Following the sixth hypothesis (H6), this effect on attitude towards the advertisement would be stronger for participants with low sport fan team identification
compared to participants with high sport fan team identification.
A three-way ANOVA4 with TBSIS, type of social influence strategy and sport fan team identification revealed an insignificant main effect of TBSIS, F = (1, 358) = .59, p = .444, η2 = .002. Additionally, no significant main effect of type of social influence strategy was found, F = (2, 358) = .33, p = .728, η2 = .002, but for sport fan team identification it was,
F = (1, 358) = 6.38, p = .012, η2 = .017. An independent samples t-test (t (368) = -2.68, p =
4 A Shapiro-Wilk’s test (p < .001) indicated that ‘attitude towards the advertisement’ was not normally distributed. Observing the histogram,
the Q-Q plot and the box plot resulted in a different conclusion: attitude towards the advertisement was normally distributed, with a skewness of -.362 (SE = .127) and a kurtosis of -.009 (SE = .253). A Levene’s test revealed a F-value of .57, with an insignificant p-value of
.008) presented a significant difference in attitude towards the advertisement between participants with high sport fan team identification (M = 4.55, SD = 1.13) and participants with low sport fan team identification (M = 4.23, SD = 1.16).
The analysis exposed an insignificant interaction effect between TBSIS and sport fan team identification, F (1, 358) = .25, p = .618, η2 = .001. Also, there was no significant interaction effect between TBSIS and type of social influence strategy, F (2, 358) = .83, p = .437, η2 = .005, nor an interaction effect between type of social influence strategy and sport fan team identification, F (2, 358) = .57, p = .5608 η2 = .003. Finally, a combination of all factors did not elicit a significant interaction effect, F (2, 358) = .86, p = .422, η2 = .005.
The main effect of TBSIS and the interaction effect between TBSIS and sport fan team identification were insignificant; H2 and H6 are therefore not maintained.
4.5. Effects on attitude towards the product
The third hypothesis (H3) stated that participants who were exposed to online advertisements that are tailored on the basis of social influence strategies would have a greater positive attitude towards the product than participants who were exposed to not-tailored
advertisements. According to the seventh hypothesis (H7), this effect would be stronger for participants with low sport fan team identification compared to participants with high sport fan team identification.
A three-way ANOVA5 with TBSIS, type of social influence strategy and sport fan team identification displayed an insignificant main effect of TBSIS, F = (1, 358) = 1.32, p = .252, η2 = .004. In addition, there was no significant main effect of type of social influence
5 A Shapiro-Wilk’s test (p < .001) showed that ‘attitude towards the product’ was not normally distributed. On the contrary, a visual
inspection of the histogram, the Q-Q plot and the box plot proved a normal distribution of attitude towards the product, with a skewness of -.333 (SE = .127) and a kurtosis of .009 (SE = .253). The F-value for Levene’s test was 1.11, with an insignificant p-value of .349; the
strategy, F = (2, 358) = .71, p = .491, η2 = .004, and sport fan team identification, F = (1, 358) = 4.41, p = .036, η2 = .012.
The analysis presented no significant interaction effect between TBSIS and sport fan team identification, F (1, 358) = .03, p = .864, η2 < .001. Furthermore, no significant
interaction effect between TBSIS and type of social influence strategy occurred, F (2, 358) = .42, p = .655, η2 = .002, nor an interaction effect between type of social influence strategy and sport fan team identification, F (2, 358) = .58, p = .560, η2 = .003. Lastly, the interaction effect with all factors together was insignificant, F (2, 358) = .03, p = .967, η2 < .001.
Both the main effect of TBSIS and the interaction effect between TBSIS and sport fan team identification were not significant; H3 and H7 are thus rejected.
4.6. Effects on purchase intention
The fourth hypothesis (H4) indicated that participants who were exposed to online advertisements that are tailored on the basis of social influence strategies have a greater purchase intention than participants who were exposed to not-tailored advertisements. Following the eighth hypothesis (H8), this effect would be stronger for participants with low sport fan team identification compared to participants with high sport fan team identification.
A three-way ANOVA6 with TBSIS, type of social influence strategy and sport fan team identification revealed no significant main effect of TBSIS, F = (1, 358) = .17, p = .680, η2 < .001. Moreover, the main effect of type of social influence strategy was not significant, F = (2, 358) = 1.96, p = .144, η2 = .011, but of sport fan team identification was, F = (1, 358) = 9.43, p = .002, η2 = .026. An independent samples t-test (t (368) = -3.53, p < .001) exposed a significant difference in purchase intention between participants with high sport fan team
6 A Shapiro-Wilk’s test (p < .001) showed that ‘purchase intention’ was not normally distributed; the histogram and the box plot confirmed
this result. The Q-Q plot, in conjunction with a skewness of .297 (SE = .127) and a kurtosis of -.762 (SE = .253), ultimately determined that purchase intention was normally distributed. A Levene’s test displayed a F-value of 1.13, with an insignificant p-value of .337. The
identification (M = 3.46, SD = 1.50) and participants with low sport fan team identification (M = 2.92, SD = 1.44).
The analysis demonstrated an insignificant interaction effect between TBSIS and sport fan team identification, F (1, 358) = .67, p = .412, η2 = .002. Additionally, there was no significant interaction effect between TBSIS and type of social influence strategy, F (2, 358) = 1.56, p = .213, η2 = .009, nor an interaction effect between type of social influence strategy and sport fan team identification, F (2, 358) = 1.40, p = .249, η2 = .008. Conversely, the interaction effect between all factors combined was significant, F (2, 358) = 3.43, p = .034, η2 = .019.
As a conclusion, H4 and H8 are not supported due to the insignificant analysis results of the main effect of TBSIS and the interaction effect between TBSIS and sport fan team identification.
4.7. Hypothesis testing with alternative measure of tailoring
The hypotheses were also tested7 with an alternative measure for determining tailoring; a new independent variable ‘TBSIS (alternative)’ with two conditions (tailored vs. not-tailored) was created. Assigning participants to either the tailored or the not-tailored condition was
determined on the basis of which condition (consensus vs. scarcity vs. authority) a participant was exposed to, in conjunction with whether a participant’s average susceptibility score for this given condition was higher than the midpoint (i.e. 4) of the subscale (e.g., a participant in the consensus condition WITH an average susceptibility score of 4.25 on the consensus subscale was placed in the tailored condition). Allocation to the tailored condition was even the case when this average susceptibility score was not the highest compared to the scores of the other two subscales. On the contrary, participants with scores of 4 or lower on this
particular subscale were assigned to the not-tailored condition. This alternative measure of tailoring resulted in 171 participants (46.20 %) in the tailored condition, and 199 participants (53.80 %) in the not-tailored condition. Without the participants in the consensus condition, 116 participants (46 %) were placed in the tailored condition, and 136 participants (54 %) in the not-tailored condition.
4.7.1. Results with consensus condition
A three-way ANOVA8 with TBSIS (alternative), type of social influence strategy and sport fan team identification displayed a significant main effect of TBSIS (alternative) on perceived social value of the product, F = (1, 356) = 5.16, p = .024, η2 = .014. An independent samples t-test (t (335.29) = -3.25, p = .001) demonstrated a significant difference in perceived social value of the product between participants in the tailored condition (M = 3.44, SD = 1.00) and participants in the not-tailored condition (M = 3.12, SD = .85). As a result, H1 is accepted.
4.7.2. Results without consensus condition
A three-way ANOVA9 with TBSIS (alternative), type of social influence strategy and sport fan team identification revealed a significant main effect of TBSIS (alternative) on perceived social value of the product, F = (1, 243) = 4.40, p = .037, η2 = .018. An independent samples t-test (t (226.60) = -2.98, p = .003) exposed a significant difference in perceived social value of the product between participants in the tailored condition (M = 3.43, SD = .97) and participants in the not-tailored condition (M = 3.09, SD = .82). Therefore, H1 is maintained.
8 Perceived social value of the product was normally distributed. Age and website experience were implemented as covariates in the analysis.
However, a Levene’s test revealed a F-value of 2.59, with a significant p-value of .001. The assumption of homogeneity of variances was therefore not met.
