[Document subtitle]
EXAMINING THE INFLUENCE OF ENDORSERS ON
CONSUMER ATTITUDE AND THE MODERATING
EFFECT OF THE DIFFERENT LEVELS OF
INVOLVEMENT WITHIN THE BEAUTY INDUSTRY
Nevena Gospodinova 11385936 Final Thesis
MSc Business Administration: Marketing 2017-2018 Supervisor: Toon Meulemans
Word count: 17124 University of Amsterdam, Business School, Plantage Muidergracht 12, 1018 TV, Amsterdam, The Netherlands 26th January 2018
Statement of Originality
This document is written by Nevena Gospodinova who declares to take full responsibility for the contents of this document.
I declare that the text and the work presented in this document are original and that no sources other than those mentioned in the text and its references have been used in creating it.
The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.
Contents Abstract ... 4 1. Introduction ... 5 2. Literature review ... 8 • 2.1 Attitude formation ... 8 • 2.2 Type of endorsers ... 10
• 2.3 Elaboration Likelihood Model (ELM) ... 17
3. Methodology ... 20
• 3.1 Design and Participants ... 21
• 3.2 Independent variable ... 21 • 3.2.1 Level of involvement ... 22 • 3.2.2 Type of endorser ... 23 • 3.3 Dependent variable ... 24 • 3.4 Control variables ... 25 • 3.5 Procedure ... 26 4. Results ... 28
• 4.1 Pre-test for Type of endorser ... 28
• 4.2 Pre-test for Level of Involvement... 29
• 4.3 Main survey ... 29
• 4.3.1 Validity and Reliability check ... 30
• 4.3.2 Correlation analysis ... 30
• 4.3.3 Factorial ANOVA ... 33
• 4.3.4 Hypothesis check ... 36
• 4.3.5 Additional Attitude analysis ... 39
• 4.3.6 Regression analysis ... 40
5. Conclusion and General Discussion ... 41
• 5.1 Conclusion ... 41
• 5.2 Implications ... 44
• 5.4 Future research ... 46
Reference ... 47
Appendix 1: Pre-test endorser ... 53
Appendix 2: Pre-test level of involvement ... 54
Appendix 3: Makeup Experiment ... 55
Appendix 4: Makeup Tutorials ... 62
Abstract
Marketers recognize different effects on consumer’s attitudes in regards to celebrity endorsers, as compared to non-famous ones. Extensive research on the Elaboration Likelihood Model (ELM) and the effect it has on consumer perceptions has been conducted throughout the years. The literature and the research on both topics shows the benefits of using the right type of endorser and right type of ELM cues as individuals respond differently to the same advertisement based on their created attitude. This research analyzes how the different types of endorsers (famous, non-famous) influence consumer attitudes. Additionally, this study measures the effects that different levels of involvement has on the aforementioned relationship. The study is based on empirical research, which used a sample of 281 participants from three big universities in Bulgaria and the Netherlands. The data is analyzed by using factorial ANOVA. Further research was also done by conducting Regression processes. The findings indicated no moderating effect of involvement. However, the results consolidated the growing trend of using a non-celebrity endorser as promotion of a product. The finding are beneficial for both managers and academicians such that, the former generates clearer idea of what social media and type of endorser they should use. Furthermore, the latter is provided more food for thought on the perception of the celebrity endorser in the beauty sphere.
Key words: Endorser, Elaboration Likelihood Model (ELM), Consumers’ attitude, Social Media Influencer (SMI), Social media, Tutorial, Makeup
1. Introduction
Nowadays, social media is one of the most engaging sources of information. One can easily find information related to the culinary sphere, traveling, do-it-yourself (DIY) tutorials, cosmetics, and much more. The information comes in the form of pictures or videos, making it easy to access and interpret. Since customers are increasingly willing to include social media as a research tool in regards to purchase decisions (Powers, Advincula, Austin, Graiko & Snyder, 2012), brands should evolve from simply being present on social media channels to generating intense communication with their consumers. In the past, the most used channels were YouTube, Facebook, Twitter. However, over the last couple of years, Instagram has become increasingly popular within society, making it the fastest growing social network site globally (Sheldon & Bryant, 2016). With this new player among social media, a new type of infomercials was developed. Accordingly, instamercial is a new term still not thoroughly examined within the academic context. Essentially, instamercial is a short infomercial or tutorial on Instagram. The content can vary, however, the most popular of them involve makeup videos, in which cosmetics fans demonstrate how, what, and when to use different products and brands (Fischer, 2014; Hu, Manikonda, & Kambhampati, 2014). Since this incorporates product placement, many brands use the already established Instagram profiles of makeup fans as a channel of communication with their customers. Although instamercials are extremely popular, there is no detailed research on whether, and to what extent, this new communication format is improving customer equity. Whether utilizing a non-famous professional endorser is a more beneficial tactic than the well-known strategy of using a celebrity endorser to promote a certain product, is still an ongoing discussion between marketing academicians (Sternthal, Dholakia, & Leavitt, 1978; Gaied & Rached, 2010; Ngnoubamdjum & Zahn, 2016 ).
The decision regarding what type of endorser to be used for a brand is always time-consuming, usually with no correct answer. In the 20th century, when this trend began, the endorsement of a celebrity was something rare and unique (Erdogan, 1999), making the endorsed brand stand out from its competitors. Consequently, many brands believed that using a star as an influencer will result in more positive attitudes towards their products (Kaikati 1987; Ohanian, 1991; Tripp, Jenson, and Carlson, 1994; Goldsmith, Lafferty, and Newell, 2000; Erdogan, Baker, and Tagg, 2001). However, in today’s world, consumers are bombarded with brand messages every day (Comiati & Plaias, 2010). The high exposure makes them more used to seeing famous faces, thus, it is harder to win their attention (Erdogan, 1999). This market communication maneuver results in a fundamental shift from preferences of the comsumers for the famous and attractive endorser (Sternthal et al., 1978), to the not-so-famous, credible endorser (Ngnoubamdjum & Zahn, 2016). This hard to attract attention leads to another problem. The ever-decreasing attention span of consumers‘ results from the dynamic modern environment in which we live in. Due to the latter, consumers prefer short, easy to process, interactive messages (Smith, 2012). The attitude formation and attention given depends on the consumers‘ level of involvement. Petty and Cacioppo (1983) are the fathers of the Elaboration Likelihood Model (ELM), introducing acedimia to the two routes of message processing: peripheral and central. They also discuss the outcomes from the use of two different types of cues, such that under low involvement the customers will follow peripheral cues when processing a brand message and focus on the secondary components, such as attractiveness of endorser or the jingle of the commercial. Oppositely, under high involvement, the consumer will be willing to cognitively process the brand message, using central cues, such as the strength of the arguments or the quality of the product. Zhang and Zinkhan (2006) also studied the phenomenon of the involvement conditions, introducing a
third component, which holds various implications on its own (i.e. moderate level of involvement).
Although Petty and Cacioppo (1983) studied attitude formation under the two different involvement conditions, combined with the two different types of endorser (famous and non-famous), an interesting gap appears in regards to whether the results can be translated within the beauty industry. In that industry, some of the usually peripheral cues, such as attractiveness of the endorser, can also serve as central ones, due to the nature and reasons to use of the product. Thus, the research question for this study that will serve as a foundation for this paper is:
With the use of Instamercials, and the ever-decreasing attention span of a millennial, how does having different levels of involvement moderate the relationship between the two different types of the endorsers and the total attitude (towards the endorser, ad, and brand) when dealing with makeup tutorials?
In order to study this question in depth, this paper is organized in the following way. First, an intensive literature review is conducted and presented, incorporating numerous articles of the type of endorser (famous/ non-famous) and the different pros and cons for each type and Elaboration Likelihood Model (Low/Moderate/High involvement). Moreover, the attitude formation process is mentioned, as well the attention span of the Millenials. Secondly, based on the findings of the articles, a gap question and conceptual model is created that will be the backbone of this paper. Thirdly, the methodology section is presented, including the two pretests conducted for the independent variables (i.e. type of endorser (famous/non-famous) and level of involvement (low/moderate/high)), as well as the procedure and the design of the survey. Following, the results are discussed, based on the factorial ANOVA conducted to either support or reject the hypotheses formulated for this study. Additionally, and based on the findings and directionality of the results, further research was conducted, observing the
difference between the different types of attitude (towards endorser, ad, and brand), from which the total attitude was formed. Although not included in the conceptual model, age and gender proved to produce interesting results, which were further analyzed using the Regression Process analysis, to determine a possible moderation effect of those two variables.
Finally, a discussion section concludes the paper, as limitations, implications (both managerial and academic), is presented, as well as suggestions for further research. The implication of the findings for the managers will be the generation of clearer idea of what social media and type of endorser they should use. While for academicians – providing more food for thought on the perception of the celebrity endorser in the beauty sphere.
2. Literature review
In this section, previous literature covering the topics of type of endorser, the advantages and disadvantages of using a famous endorser, as well as a new type of endorser, the social media influencer (SMI), will be discussed in depth. Following attitude formation, the decreasing attention span of consumers and the Elaboration Likelihood Model (ELM) will be studied. The topics will be observed in the context of online media, more precisely social media in the beauty industry. First, clear and understandable definitions of attitude and celebrity endorser will be presented and analyzed. In addition, different articles supporting and disproving the use of famous endorsers will be discussed. Finally, articles covering the electronic Elaboration Likelihood model (eELM) will be observed.
2.1 Attitude formation
One of the most famous communication model is the DAGMAR (Defining Advertising Goals for Measuring Advertising Results) introduced by Colley (1961). Its function is to clearly set and measure the communication objectives of a company. Its greatest benefit is that, instead of sales goals that are hard to correlate to the communication expenditures, it uses other
measures for effectiveness, some of which are brand awareness and brand attitude (De Pelsmacker, Geuens & Van den Bergh, 2007). This study will focus on one of the elements (attitude).
Attitude is a measure that is quite popular among the advertising and marketing studies. Spears and Singh (2004) discuss the two main reasons for that. First, they believe attitude is useful in prediction of consumer behavior. Secondly, the availability of theoretical frameworks for the study of attitude is vast and ever expanding. In their paper, they filtered the most useful and precise attitude measures, as their two main requirements for the tools were to be an individual internal evaluation of the brand and to be generalizable to a wide range of products. Thus, they came up with 31 items that can measure the attitude of consumers, some of which (e.g. useful/useless; likely/unlikely) were used in this survey (see
Appendix3: Makeup Experiment).
According to De Pelsmacker, Geuens & Van den Bergh (2007), attitude is “the person’s overall evaluation of an object, a product, a person, an organization, an ad”. It is believed that the more positive the attitude, the higher the chance the consumer will ultimately purchase the product or experience. The total attitude can be formed by combining different subtypes, such as attitude towards the endorser, attitude towards the ad, attitude towards the brand (Zhang & Zinkhan, 2006), which were also used in this research.
With all the messages with which the consumers are bombarded and the ever decreasing attention span (Comiati & Plaias, 2010; Donnison, 2007), the use of the right communication model is the key to success. Anderson and Rainie (2012) and Donnison (2007) study the Millennials and their attention creation process. They characterize this generation with thirst for instant gratification, settling for quick choices, lack of patience, short attention span, and higher propensity of boredom. Thus, exposed to a commercial, generating a positive attitude
from the customers has become an arduous task for the companies, as well as the communication tasks between their consumers.
One of the elements where communication originates is the source of information. This is the person who conveys the information. This source can vary from a consumer, a celebrity endorser, an expert, the CEO, or an independent source (McCracken, 1989). In most cases, the endorser is presented in a field in which he or she is an expert. The match between the field and the endorser’s interest is proved to have a positive influence on the perception of the advertisements and respectively the consumers’ attitude (Kamins, 1990).
Another element, studied by Smith (2012), are graphics of the material presented to the audience. Among the six website features observed, graphics scored highest for successfully grabbing the Millennials attention. With the ever-decreasing attention span, the audience is looking for short, personalized, and professional content that will save them time and effort. Therefore, clear, precise, and easy to process interactive messages will be the best for communication with this dynamic generation (Anderson & Rainie, 2012).
2.2 Type of endorsers
Due to the ever increasing use of social media in the 21st Century, targeting consumers through a particular online channel becomes necessary. Kim and Ko (2012) examined the effect of the social media marketing (SMM) activities of luxury fashion brands on customer equity and purchase intentions. They found that the brands’ SMM activities have a more positive effect on brand equity than the old-fashioned, traditional marketing approach. Additionally, Myers and Robonson (2010) observed the effects in a well-established industry (beauty products). They tried to discover the effect social media has on the beauty world. Kim and Ko (2012) findings is in line with the one of Myers and Robonson (2010) who state that the social media reach and impact has great success in the beauty industry. However, that success would not be possible without the right endorser for the products at hand.
Celebrity endorsers are regularly used to promote an endless list of products including cars, luxurious products, makeup, and more (Mowen & Brown, 1981). A number of factors have been mentioned as important in using endorsers, such as the endorser's personal qualities (i.e. reputation, believability, likeability), since many consumers decide to use them as personal influencers and reasons to purchase a product (Phang & Furuoka, 2007; Amos, Holmes & Strutton, 2008; Ohanian, 1991; Hudha & Hidayat, 2009). Consequently, many brands have bought into the premise that a famous endorser positively impacts consumer attitudes towards an advertisement and subsequently the brand, consumers’ purchase intention, as well as other measures of effectiveness (Kaikati 1987; Ohanian 1991; Tripp et al. 1994; Goldsmith et al. 2000; Erdogan et al. 2001). Therefore, celebrity endorsers play an important role in the marketing communications strategy of many big companies. Scholars believe that the above-mentioned endorser’s qualities will result in desirable outcomes for the companies. Having mentioned that, the first hypothesis is developed as:
H1: Famous endorsers lead to more positive total attitude (towards endorser, ad, and brand) than non-famous endorsers.
Depending on product type, an endorser can come from several different sources. From a celebrity, to a supplier, to the customer, and in some cases, the CEO of a company himself can be considered as an endorser. However, using a famous celebrity endorser was perceived as the better option for many years. According to Erdogan (1999), using a celebrity for endorsement dates back to the nineteenth century, giving an example of the association between Queen Victoria and Cadbury's Cocoa. However, he also discusses the limited supply of “stars” during that period, making it harder to use that strategy. It was not until the 1970s’ when the number of available famous endorsers increased substantially (Schimmelpfennig & Hollensen, 2016).
After the use of celebrities became more popular, academicians started being interested in the pros and cons of using this strategy within the marketing context. According to Erdogan (1999), although a star can polish the image of the brand and create positive brand attitude, the overuse of one and the same “faces” can decrease the beneficial effect. In addition, in some cases the endorser can overshadow the product, essentially hindering the accumulation of the desired results. In relation, the author decides to compare this strategy with one where a created endorser (i.e. non-famous professional endorser) is used. He states that in some cases the non-famous spokesperson generates more attention towards the brands and higher involvement, while in the first strategy the customer would be more focused on the famous endorser, rather than the product (Erdogan, 1999). Thus, at times, celebrity qualities may not be the best choice. Several authors observe the choice between celebrity and non-famous endorser, trying to find out which of them has a stronger effect on consumers’ attitude and in which situation is more beneficial to use each respected type.
It is said that more than 25% of the advertising in the USA use celebrity endorsers (Erdogan, 1999; Erdogan et al., 2001), and around 10% of the budget is allocated for paying those celebrities (Agrawal, Kamakura, 1995). These actions suggest that the USA companies tend to believe that this strategy will have a positive impact on the consumers’ attitude towards the advertisement and the associated brand (Shimp, 2000). Friedman and Friedman (1979) define a celebrity as “an individual who is known by the public for his or her achievements in areas other than that of the product class endorsed”. There are four reasons why using a famous endorser is useful: 1. stars attract attention, 2. make the advertising message more powerful, 3. boost the brand image, and 4. make it easier to enter international markets (Karasiewicz & Kowalczuk, 2014).
Phang and Furuoka (2007) also study and prove the positive effect of the tendency of using a celebrity endorser. They explain the attention the stars continuously gather and eventually
influence the brand preferences within society, resulting in increased purchase intentions. However, they acknowledge that the extensive use of celebrities as an endorser is bounded by spending significant amounts of money and time to accumulate such a resource. Therefore, they suggest that taking a decision to invest in famous source should be well thought and measured in order to be financially worth the investment (Agrawal & Kamakura, 1995).
According to Sternthal et al. (1978), the most important source credibility features of the famous endorser are trustworthiness, attractiveness, and credibility. Their main finding led to the conclusion that the credible source is more persuasive than the low credible one. In line with these results, Friedman and Friedman (1979) found that those three characteristics could enhance the purchase intentions, brand attitudes, and the attitude towards the advertisement. However, Sternthal et al. (1978) also found out that in situations when the audience is already predisposed to a message, a less-credible source could also generate more positive attitudes.
Understanding the effectiveness of the endorser is of high importance during the selection of the most efficient marketing communication strategy. Erdogan et al. (2001) observe five characteristics that are important when choosing a famous endorser, 1. perceived trustworthiness, 2. perceived expertise, 3. perceived physical attractiveness, 4. perceived familiarity and 5. perceived likability. He finds that the first two characteristics are the crucial for managers when selecting a celebrity endorser for a technical/ attractiveness-unrelated product as opposed to a nontechnical/ attractiveness-related product. The opposite was found to be true in regards to the final three characteristics.
Although the attractiveness is a main reason for the success of the ads with celebrity endorsers, it would not always play on the celebrity behalf (Erdogan, 1999). The physical attractiveness of the endorser will influence the attitude of the consumer only if the product characteristics match-up with the image of the endorser (Till & Busler, 1998). Advertising a product via a celebrity whose image is highly congruent with the brand leads to greater
advertiser and celebrity believability, compared with a situation in which there is low congruence (Kamins & Gupta, 1994). Thus, for beauty products, the use of attractive promoters will significantly enhance the measure of the spokesperson credibility and attitude towards the product, relative to the use of an unattractive one (Kamins, 1990).
Based on “what is beautiful is good” (Kamins, 1990), advertisers often decide to combine the celebrity status with physical attractiveness, in order to achieve more positive attitudes towards the ad and the brand itself (Friedman & Friedman 1979; Singer, 1983). Till and Busler (1998) examine the effectiveness of physical attractiveness and expertise as relevant matchup factors, as well as the role of endorsers-product fit. They find that the use of an attractive spokesperson has a positive effect on the brand attitude. However, when compared to expertise, attractiveness comes second, as expertise may be a more appropriate factor for matching products. In this particular case, the authors did not observe which products are used to enhance the attractiveness of the endorser, as in the beauty products industry. However, they assume that in the beauty industry it is more likely for customers to conclude that using the product can result in enhanced attractiveness.
Besides the endorser’s characteristics, the type of the product is also important when choosing an endorser. Early research findings state that the effectiveness of the endorser varies by product type (Friedman & Friedman, 1979). The match up hypothesis (Till & Busler, 1998) studies the attractiveness of the endorser and the match with the products. Essentially, the attractive of the endorser is more effective when promoting products used to enhance one’s attractiveness, such is the case with makeup industry (Kamins, 1990). However, the paper results were that the expertise dimension might be more useful than the attractiveness in matching endorsers with brands. The speculations are based on the fact that for products that are meant to enhance one’s beauty, the effect may be not visible with endorsers that are already attractive (Bower and Landreth, 2001). In addition, Wen, Tan, and Chang (2009)
examined the endorser effectiveness between two different types of products: hedonic and utilitarian. The findings proved that the hedonic products are better influenced by strong-tie endorsers than weak-tie endorsers, while for utilitarian products, high-expertise endorser were more effective than low-expertise ones.
While the above-mentioned authors perceive the celebrity endorser as a successful element of communication strategy, concluding that the attractiveness and credibility are strong reasons, McCracken (1989) believe the true reasons go beyond those characteristics. The celebrity endorser plays a huge role in the transfer process from attitude towards the influencer to the attitude towards the product. According to McCracken (1989), the celebrity endorsement has a special power that makes it the better choice. The author believes that the stars bring extra meaning and depth to the message. In addition, the famous endorsers are said to add precision, as well as personality and lifestyle to the ad, which makes them a powerful source within media.
The use of a celebrity endorser comes with a price. As with every managerial decision, this advertising strategy poses risks on their own. Karasiewicz and Kowalczuk (2014) found five different risks managers face when applying celebrity endorsement: 1. mismatch between celebrity and product, 2. brands being overshadowed by the endorser, 3. overuse of a particular celebrity for too many brands, 4. negative reputation of an endorser may transfer to the reputation of the brand, and 5. decrease interest in a brand once the endorser loses publicity.
While most of the articles on that topic compare famous and non-famous endorsers without specifying what type is the non-famous endorser, Freberg, McGaughey, and Freberg (2010) introduce a brand new type of endorser, the Social Media Influencer. According to Hearn and Schoenhoff, the SMI works to generate a form of “celebrity” capital by cultivating as much attention as possible and crafting an authentic “personal brand” via social networks, which
can then be used by companies and advertisers for consumer outreach. The success of SMIs are determined based on the return on influence as marketers seek them out to capitalize on their wide social networks and benefit from the intimate, more ‘trustworthy’ relationships SMIs have created (Khamis, Ang & Welling, 2017). In addition, SMIs rely heavily on the number of daily hits their profile creates, number of times a post is shared, or the increase in the number of followers (Freberg et al., 2010).
From the early 2000s, self-brand was practiced, not just by the publicly famous people, but also by the ‘ordinary people’ (Khamis, Ang & Welling, 2017). This was, and is, possible due to the contemporary convergent technologies, in particular the social media platforms. While the creation of self-branding started with the creation of blogs and websites, which still required some basic programing knowledge, nowadays, the most popular platforms are Facebook, Instagram, and YouTube. In addition, the popularity and prominence are no longer dependent on the traditional gatekeepers, such as the producers and PRs (Khamis, Ang & Welling, 2017). With over a 150 million-user base, Instagram is one of the most used social platforms, making it easy to achieve Instafame, the condition of having great number of followers (Marwick, 2015). While the most followed people are still the real stars, such as Beyoncè (110 million followers) and Drew Barrymore (8.1 million followers), more and more ordinary users have become Instagram famous just by the help of streaming eye-catching storylines, instamercials, and selfies (Marwick, 2015).
Freberg et al. (2010) examines the self-branding of the Social Media Influencers. With online media more accessible and its consumer-centric atmosphere, more and more individuals seek out the specific resources that spark their interests. Enabling ordinary people to assert strong identities, many high public profiles were formed with millions of followers. The key features those profile makers have is the ability to easily attract the public and entertain them in order to maintain the built relationship with the audience for a low cost. Thus, with easy access to
social media and the equal, even better performance, the non-famous endorsers are increasingly gaining preponderance over the celebrity ones (Freberg et al., 2010).
In addition, Ngnoubamdjum and Zahn (2016) conducted research in the cosmetic field among Millennials, the findings of which showed that the generation preferred the non-famous endorser over the celebrity one. Their study found that the customers are less focused on the level of famousness of the influencer and more on the product attributes. Therefore, the consumer preferred the non-famous professionals in regards to ads with cosmetics. Supporting this statement, Benedic and Granjon (2017) discuss the marketing campaign of L’Orèal Paris, in which the brand purposefully used influencers with moderate levels of famousness for more authentic partnership. The interest in collaboration with smaller endorsers is perceived as a growing trend in the beauty industry (Chen, 2016; Benedic & Granjon, 2017). The new type of endorser brings the question, are celebrity endorsers still a successful communication strategy, and are they cost-efficient?
2.3 Elaboration Likelihood Model (ELM)
Whether a consumer will pay more attention to the strength of argument, the source of information, the background, or the details of an ad depends on their level of involvelment. In order for the company to chose the best combination of elements for its communication plan, the target consumers preferences should be clear and well defined.
The pioneers of involvement of the consumer are Petty and Cacioppo (1983). Consequently, they found that consumers can process information from an ad using two different routes: central and peripheral. The former is a result from attention to the details of the information that is central to what the person feels or wants. In this case, the consumer is willing to spend more time to actually get informed and spend more time on the quality of the information he or she receives (Petty & Cacioppo, 1983). In other words, the person is going to cognitively process the information. Contrasting the central route, the peripheral route to attitude change
is associated to any positive or negative cues, or ones that are based on the level of attractiveness of the endorser. In this case, the consumer makes his or hers decision without the need for engaging in any extensive, issue-relevant thinking. This type of consumer attention can be grasped without presenting them any relevant product information (Petty & Cacioppo, 1983).
In order to prove their hypotheses, Petty, Cacioppo, and Schumann (1983) ran a research with undergraduates, each of whom was exposed to one of the six conditions, formed by combining the factors: high vs. low involvement, strong vs. weak arguments, and famous vs. non-famous endorser. Their conclusion was that under low involvement, people are more influenced by the type of the endorser and do not pay attention to the central cues (i.e. the arguments). While under high involvement, the respondents did not care who is the endorser, as far as his or her arguments were strong.
Authors that are also interested in the Elaboration Likelihood Model are Zhang and Zinkhan (2006), who discuss the relationship between the audience involvement and the responses to humorous ads. This paper is one of the first to introduce the third level of involvement (i.e. moderate). According to the authors, while a consumer may start processing a message using the peripheral route, once he or she is intrigued, the processing may turn into a central one. In other words, for the consumers that are moderately involved, the ads with content based on the emotional appeal, such as humor (De Pelsmacker, Geuens & Van den Bergh, 2007), the consumer may turn highly involved based on this content.
With the rise of the SMI (Freberg et al., 2010), online advertising has become a hygene factor for success. Therfore, the online component that may influence the persuasion, is a topic of a great interest for the academicians in the marketing field. One such is Hershberger (2003), who developed an electronic ELM in order to find out whether there will be any change in the audience attitude. His research in the online context did not find any moderating effect of
level of involvement in the formation of the attitudes. However, the findings of Cho (1999) differ greatly with that of Hershberger (2003). The former did a similar experiment as the latter, trying to prove that the Elaboration Likelihood model is applicable in the online context, as well. His fidnings supported his hypothesis such that, under low involvement, the internet surfers are more willing to open banners that are bigger and more dynamic, while under high involvement, the size and the dynamism do not matter for the the audience, as they are willing to cognitively and in-depth process the given imformation if it is relevant (Cho, 1999).
Other similar factors that can influence consumers’ attitude formation and the choice of elaboration route are the motivation, ability, and opportunity with which the prospect serves (De Pelsmacker, Geuens & Van den Bergh, 2007). Motivation is related to the willingness of one to spend time, pay attention to the details, and process carefully the given information. Additionally, ability is the capacity of the consumer to process the information in the desired way. Finally, opportunity is the extent to which the situation enables the person to have access to the message. When the three factors are all high, the information will be processed using the central route. However, when they are low, the information will most probably be perceived using the peripheral route. Based on the prior research, especially on the Petty and Cacioppo (1983) theory of ELM, and the observation for the two different type of endorser famous and non-famous SMI, three hypotheses can be generated:
H2a: The effect of the type of endorser on the total attitude (towards endorser, ad, and brand) is moderated by level of involvement, such that under low involvement, the effect of the famous endorser is stronger than the one under medium or high involvement.
H2b: The effect of the type of endorser on the total attitude (towards endorser, ad, and brand) is moderated by level of involvement, such that under moderate involvement,
the effect of the famous endorser is less strong than the one under low involvement, but stronger than the one under high involvement.
H2c: The effect of type of the endorser on the total attitude (towards endorser, ad, and brand) is moderated by level of involvement, such that under high involvement, the effect of the famous endorser is weaker than the one under medium or low involvement.
The different processing routes lead to different attitude formation, where the increase in the positive attitude towards the endorser, ad, or brand depends on how well the cues in an ad match the level of involvement of the customer. The greater the match, the more positive the attitude, and vice versa. Based on the prior literature review observations and the generated hypothesis, a conceptual model was developed. The model is formulated in the following way:
3. Methodology
In the following section, the methodology of the current study will be discussed. Several topics will be covered. The pre-test design and analysis, as well as the implication and results found during those initial pre-testing. A paragraph will be dedicated to the design and the participants of the main survey. Additionally, the independent, dependent, and control variables will be observed, focusing on the description of how each was operationalized throughout the research. Finally, the procedure of the main study, including how it was formulated and presented to the participants, will conclude this section.
3.1 Design and Participants
281 people with various background and country of origin participated in this study (see
Appendix 3: Makeup Experiment). They were randomly assigned to cell of a 3 (involvement: low, moderate and high) x 2 (endorser: famous and non-famous) factorial design. Thus, this randomized, between-subject design contained 6 cells with at least 40 participants in each experimental condition. A Dior product was chosen as an experimental product in order to match the two different type of endorsers. This decision was taken after considering a variety of different products through tutorial analysis. The independent variables in the survey is the type of endorser, with a moderating effect from level of involvement. The dependent variable being measured is the Attention towards the tutorial, brand, endorser, resulting in a total attitude measure.
The respondents of the study were 281 students and faculty members from the Universiteit van Amsterdam in the Netherlands, University of National and World Economy in Sofia, Bulgaria and Plovdiv University, Bulgaria. The ages ranged from 18 to 75, mainly focusing on the students from 20 to 26 years old, who are more involved with social media (Lenhart, Purcell, Smith & Zickuhr, 2010). In order to acquire access to the database of student emails of these universities, I communicated with professors (one from each university) to gain approval. All three of the academicians were from the school of Economics and Business at their respected universities. They were asked to distribute the survey among their students during their lectures. In return, they were thanked and sent a letter of gratitude.
3.2 Independent variable
In the following section, the independent variables of this study are presented and discussed. This section will give and overview of how each variable was operationalized for the main study. Two different independent variables are used, type of endorser and a moderating effect from level of involvement.
3.2.1 Level of involvement
Following the model of Petty, Cacioppo, and Schuman (1983) and Zhang and Zinkhan (2006), this study used a direct approach to operationalize involvement. The participants were exposed to three different involvement conditions. The three levels of involvement were manipulated by varying the descriptions given to the subjects before their exposure to the tutorials. A pretest for the involvement manipulation was conducted. The description manipulation is discussed below.
3.2.1.1 Description manipulation
The descriptions for the tutorial were pretested among 25 participants by randomly exposing them to one of the three different options. After watching the tutorial, they were asked to state their level of involvement prior to the tutorial itself. Analyzing the data, the results showed that the description designed to put the participants in low-involvement worked as intended, such as all respondents said that they were low-committed. While the other two descriptions’ results turned out to be different from what was intended. The moderate-involvement text was not clear enough. Although some of result turned out to be as intended, some of the participants said it was not clear what the focus of the condition was. The high-involvement description worked more as a moderate involvement, since most of the subjects said they are “averagely involved”.
Based on the result from the pretest, the descriptions were modeled and changed based on the parts that were not clear or did not function as intended. In the low-involvement condition, the participants were given the following instruction:
In the next page you will observe a makeup tutorial. This video will participate in a competition for best infomercial. Please skim through the video, as you normally would do in your regular time on social media. (Please turn your sound on!)
Here, I neither mentioned the product, nor did I give detailed information about the tutorial. In the moderate-involvement condition, subjects were exposed to the same text as in the low-involvement condition. However, they were also told that in exchange for participating in the study, each participant would have the chance of winning a 10% discount for Dior makeup. In the high-involvement condition, in addition to the description given in the low-involvement condition, the subjects were told that:
The endorser will use mascara Dior Show. This video will participate in a competition for best makeup infomercial, organized by Dior. Please focus on the level of professionalism with which the endorser does her makeup. Each participant that can answer all of the video-related questions after the tutorial will win a Dior product. (Please turn your sound on!)
Thus, the low-involved subject were led to believe that there is no benefit of focusing on the tutorial, whereas the high-involved subjects were led to believe that a prize is certain, making them pay more attention while watching the tutorial. As for the moderate involved-subjects, they were induced that they have a chance of winning, however, no certainty is given.
3.2.2 Type of endorser
To identify and guarantee an appropriate and famous endorser, a second pre-test was conducted among 30 respondents. They were shown eight different endorsers with different levels of celebrity status. Four of them were well known celebrities (the actress Brew Barrymore and the models Kendall Jenner, Adriana Lima and Bella Hadid). The other four endorsers were non-famous makeup artist professionals and vloggers. The participants were first asked to answer whether they are familiar with the person and then answer which is the most famous and influential among them. According to the responses, the most famous was Drew Barrymore, as N= 29 of participants could recall her name. The second after her were
Adriana Lima and Kendall Jenner, which N= 18 of the respondents recognized. The other endorsers were not familiar to most of the people.
Based on the results (see Appendix 4: Makeup Tutorials), Drew Barrymore was used for the famous endorser, while for the non-famous endorser, a new social media influencer to the makeup sphere was chosen. In that way, this study achieves high amplitude of celebrity status between the two endorsers.
In the main survey, the two endorsers tutorial were kept relatively the same, in terms of brand used (Dior Show mascara), length (approximately 30 seconds), sex (both female), background (plain), style of makeup (simple everyday technique), and tone of voice (informal and entertaining). Trying to keep the videos as similar as possible, changing only the celebrity status, I made sure different features of the infomercials would not affect the results, providing higher validity to the results. The respondents were randomly exposed to one of the two tutorials (the one of Drew and the one of the non-famous professional).
3.3 Dependent variable
The dependent variable of this study is the attitude of the respondents toward the overall tutorial. As discussed in the literature review section, Spears and Singh (2004) found the ultimate 31 items for a precise measurement of the consumer attitude. Three measures that will match the context of the research was chosen and used: attitude towards the ad (Aad),
attitude towards the brand (Abr) and attitude towards the endorser (Aend). Aad was measured
with 7-point Likert scale (extremely useful/ extremely useless). Abr was measured with a
similar item scale (like a great deal/ dislike a great deal). Aend was measured with two 7-point
Likert scales with the following options: extremely professional/extremely unprofessional and in term of strength of arguments provided by the endorser: extremely strong/ extremely weak.
3.4 Control variables
Several other variables that are believed to influence the results are discussed as control variables below.
3.4.1 Age
According to Lenhart et al (2010), the most active users of social media are the young adults between 18 and 29 years old. In addition, the same target group is said to be the same for the beauty lovers, who get inspired namely by the social media (Kumar, 2005). 63% of the current survey respondents were of age between 20 and 26 years old (i.e. those most susceptible to social media influence). Therefore, controlling for age will ensure legitimacy of the data.
3.4.2 Gender
According to Kumar (2005) and Comiati et al. (2010) the main target group of beauty industry is the young women. With such a high percentage of respondents of this study being women (80.8%), controlling for gender will further increase the credibility and legitimacy of the results.
3.4.3 Level of Prior Interest in the makeup sphere
The level of prior interest is important to be included as a control variable in order to take into consideration the participants that are already fans of the beauty industry, and determine whether these participants will be influenced by the descriptions or not. For this group of the respondents, the involvement manipulations may not work, due to the fact that they were
already highly involved prior the experiment. This variable is measured with a 7-point Likert scale consisting of bipolar anchors: far above average/ far below average.
3.4.4 Level of Prior interest in Dior makeup
The level of prior interest in the Dior brand was also measured, in order to control for people who are already highly involved with the brand and therefore the manipulation will not work for them. This variable was also measured with a 7-point Likert scale ranging from far above average to far below average.
3.4.5 Manipulation check
Several manipulations were included as manipulation checks. Level of familiarity of the endorser question, as well as two attention-check questions were included. In the first question, the participants were asked whether they are familiar with the endorser, while the second type of questions were related to the type of the endorser and the type of the opportunity given prior the tutorial (Involvement level). Measuring those variables, will further ensure that the participants are truly focused on the study, cementing the results as valid.
3.5 Procedure
The participants were asked to use their laptop, cellphone, or tablet during the class, and to open the link shared by the professor five minutes before the end of the class. Throughout the experiment, the respondents were asked to work independently and remain focused on the questions. At the beginning of the experiment, a short description was provided, including information about the research, the time required, and the creator behind the questions. In the case in which the participant consented to the terms and conditions, he or she was randomly allocated to one of the six conditions (i.e. low involvement with famous endorser, low involvement with non-famous endorser, moderate involvement with famous endorser,
moderate involvement with non-famous endorser, high involvement with famous endorser, high involvement with non-famous endorser). Each respondent was asked to read the description carefully and follow the required steps. After each of the descriptions, the participants were asked to turn the sound on for the following video. While watching the tutorial, depending on which of the three description they were shown, the respondents had to pay attention to different features. After completing the reading and watching the video, manipulation checks were included to determine the level of celebrity of the endorser and the level of prior experience with the brand and with the makeup sphere in general. These questions were aimed to determine the correct level of involvement and level of familiarity of the endorser. Based in those levels, the participants were then asked to state what is their attitude toward the tutorial, the brand, the endorser’s level of professionalism, the strength of the endorser’s arguments and purchase intention.
After the section with the tutorial-related questions, several demographic questions were asked (i.e. age and gender). In addition, an attention check was included to make sure that the level of involvement with the survey was satisfactory and the participants were paying attention to the study. The questions asked were related with the type of endorser that the respondent was shown and the type of opportunity he or she was given prior the video. Finally, a section was included to explain the nature of the mock questions and the inability to fulfill the promised prize in the description. In order to remain within the ethical standards, the participants were compensated with an offer to participate in a raffle for $20 Amazon gift card, as each person that wanted to participate, was asked to leave his or her e-mail address. This section was included after the survey itself, in order not to influence the level of involvement, and, therefore, the results of the experiment. After the data collection phase, all the e-mails of the people that wished to participate in the lottery were put in a software for a random selection. Aiming at transparency and fair winning, the selection of the lucky e-mail
was recorded and published on social media. Finally, the participants were thanked and debriefed.
4. Results
The experiment included 281 participants, including males (19,2%) and females (80,8%). In this section of the research, several different tests were conducted as part of the overall analysis of the data collected. The average age of the participants was 28 years old. The majority of the participants had some prior interest in the makeup sphere, as 49.1% of the respondents were highly involved in the sphere beforehand. This ensures high credibility of the data, since most of the respondents had basic knowledge of the studied sphere. In this section, the analysis of the two pre-tests will first be discussed, including the frequency for the type of endorser and one-way ANOVA for the level of involvement. For the main survey, frequencies, validity and reliability tests, correlation, and a factorial ANOVA, were conducted, observed, and analyzed.
4.1 Pre-test for Type of endorser
In the current survey (see Appendix 1: Pre-test endorser), the participants were presented one of the two types of endorsers: famous and non-famous. However, in order to make sure that the amplitude of the level of famousness will be substantial between the different conditions, a pre-test was conducted. Each of the 30 respondents was presented with four famous and four non-famous endorsers. In order to ensure that the famous endorser will be familiar to the majority of the participants, a frequency analysis was done. After comparing the data of all the eight endorsers, the one with the highest frequencies was chosen. The star Drew Barrymore was recognized by 96.2 % (25 out of 26) of the respondents, followed by Adriana Lima and Kendal Jenner with 69.3%, Bella Hadid 46.1%, Nikkie Turorials 23.1%, Anna Lovesu and Militza Yovanka with 7.7% and Lola Liner that was not recognized by anyone from the participants. Since the aim of the non-famous endorser was to be unfamiliar to the
majority, another makeup artist was picked due to the large amount of followers of all of the above-mentioned influencers. Therefore, another new beauty influencer with fewer followers was used for the main research – Roxette Arisa with 115 thousand fans, which still makes my decision credible.
4.2 Pre-test for Level of Involvement
In order to ensure the operationalization effect of the three descriptions for low, moderate, and high involvement, from which the respondents will read before the tutorial, a one-way ANOVA was conducted. The results (see Appendix 2: Pre-test level of involvement) prove that there is a statistically significant effect based on the different descriptions on the level of involvement, F= 3,84, p<.05. Observing the descriptive results, a difference was noticed between the means of the three different conditions: low, moderate and high involvement conditions. Thus, the means for the low involvement is M= 3.55, moderate resulted in M= 4.57 and high equated to M= 5.88. While the difference between the first two conditions is substantial, the one between the second and the third is not so distinctive. That is why in the original research the description for the high involvement was changed, in order to bring the respondents of that scenario in the desired level of involvement.
4.3 Main survey
Prior to the analysis, several steps were taken to clean and systemize the data. Missing data was not noticed, however, some of the values of the answers were inversed. For instance, low involvement being measured as 7 on the 7-point Likert scale. Therefore, the values of those variables were transformed by recoding them in order for the value and the numerical indicator to match, such that:
1-3 = Low (i.e. negative or unlikely)
5-7 = High (i.e. positive or likely)
In total, 281 participants completed the study. In terms of the level of normality of the distribution of the data, the answers of the questions are reliable such that the skewness and kurtosis close to zero. A check for outliers was also included as only two outliers were found. However, after comparing the analysis of the data with and without those two extreme values, no substantial difference was noticed. Due to the normal distribution of each of the variables, a factorial ANOVA was chosen for the main analysis of the data.
4.3.1 Validity and Reliability check
Factor analysis was conducted to ensure that the data collection techniques and procedures yielded consistent findings. The results show that the attitude towards the tutorial (Total) scale has high reliability, with Cronbach’s Alpha = .845. (a>.70 being the norm). The corrected item-total correlations indicate that all the items have a good correlation with total score of the scale (all above .30). Also, none of the items would substantially affect reliability if they were deleted (all bellow below .10).
4.3.2 Correlation analysis
A correlation analysis was conducted in order to observe the intensity and the meaning of the relationship between the twelve variables of this study. The used variables were Level of Involvement, Type of Endorser, Attitude Towards the Brand, Attitude Towards the Endorser’s Professionalism, Attitude Towards the Endorser’s Arguments, Attitude Towards Tutorial, Purchase Intentions, Level of Interest in the Makeup Sphere, Level of Interest in Dior Makeup, Age, Gender, and Total Attitude. The means, standard deviation and correlation data can found in Table 1, as the variables are presented in the same order as enumerated above. In the table, there are some correlations that are highlighted in a gray color. The ones with the dark gray highlight have significant correlation at the 0.01 level (2-tailed), while the one with
the light gray highlight have significant correlation at the 0.05 level (2-tailed). The significance of the correlation with the dark gray is stronger than the one with the light one.
The first noticeable significant correlation is with of the Attitude Towards the Endorser’s Professionalism. Based on the analysis, that variable is negatively and significantly correlated with the Type of Endorser (r= -0,185, p < .01), meaning that when the participants were exposed to the famous endorser, their opinion on the level of professionalism was more negative, while when exposed to the non-famous endorser, their beliefs were more positive in terms of the influencer’s competence. Additionally, the more positive the Attitude Towards the Level of Expertise of the promoter, the higher the Attitude Towards the Brand (r=0,523, p < .01). The same positive correlation can be observed between the strength of the arguments provided and the Attitude Towards the Professionalism (r=, 579, p < .01) and the Brand (r=0,549, p < .01). Logically, the overall attitude towards the tutorial is also correlated positively and substantially to the other attitudes: Attitude Towards Brand (r=0,579, p<.01), Attitude Towards Endorser’s Professionalism (r=0,532, p<.01), Attitude Towards Endorser’s Argument (r=0,690, p<.01). However, although not so significant, the relation between the tutorial attitude and the type of the endorser is negative (r= -, 129, p < .05). Purchase Intentions are in line with the attitudes due to the positive and significant correlation. Out of the four control variables, only Age does not have a significant correlation with any of the
other variables. The only moderate relation is between Age and Attitude Towards the Tutorial (r=0.119, p<.05). However, a strongly significant relationship is noticed between Gender and Attitude Towards Brand (r=0,161, p<.05), Level of Interest in the Makeup Sphere (r=0,58, p<.05) and Level of Interest in Dior Makeup (r=0,327, p<.05). The level of prior interest in the beauty industry and in the Dior makeup is substantially related to the attitude of the participants and their purchase intention. That means that the higher the prior interest, the higher their attitude, especially for the one particularly interested in the brand (Pind=0,288,
p<.01; Pbr=0,368, p<.01). Observing the Gender data, as expected, the female participants’
Attitude Towards the Brand was more positive (r=0,161, p<.01), as well as their Prior Interest Towards the Beauty Sphere (r=0,58, p<.01) and the Interest in Dior Makeup (r=0,327, p<.01).
Examining the table, there is not a substantial correlation between the Total Attitude of the respondents and the Level of Involvement (r=0,097), which disproves Hypothesis 2a, 2b and 2c. However, more tests (i.e. Factorial ANOVA) were ran and will be discussed in the following section. Nevertheless, a moderate significant correlation was noticed between Total Attitude and Type of the Endorser (r= -,12, p < .05). Based on the negative value of the relationship, the more famous the endorser, the lower the attitude. This finding may partially prove Hypothesis 2b and 2c. However, to consolidate this statement, a Factorial ANOVA will be conducted. The correlation between the total attitude and the other types of attitudes, as well as the purchase intention is substantial and positive, which proves that they are in-line and interconnected. However, it is also important to recognize that correlation does not always lead to causation in certain situations. The full results are shown in Table 1 above. In the following section, the main hypotheses and effects are fully discussed in great detail.
4.3.3 Factorial ANOVA
To examine the research question, a factorial ANOVA was performed. This type of analysis is used to analyze the main effect of each independent variable. In this analysis, the Total Attitude will be compared to both the Type of Endorser and Level of Involvement. The first independent variable has two groups: famous and non–famous. The second independent variable has three separate conditions: low, moderate, and high involvement. The ANOVA will use the F-test, which will allow an overall comparison on whether group means differ. In other words, whether the effect of one of the independent variables on the dependent one is the same across all levels of the second independent variable, or if there is any interaction between the independent variables. Considering an alpha of 0.05, if the calculated results for the hypothesis F-value is larger than the critical F-value a, the hypothesis will be disproven.
Levene’s test of quality of error variables was used to test whether k samples have equal variances (Table 2). In other words, to test the level of homogeneity of variances. In this case the significance level is higher than 0,05 (r=,543,
p<0,05), which means that the variance between each condition is equivalent to each other, not resulting in significant differences in variance. Therefore, the results of this study are valid in terms of initial tests, allowing progression to the other various tests that were conducted.
In Table 3, a between-subjects effect test is presented, which studies the coefficients that quantify the association between the dependent and independent variables. According the rule of thumb, when the values of the level of significance and the partial Eta Squared are below .06 the effect size is low, between 0.06 and 0.14 the effect is moderate and above 0.14 the effect is high. Having said that, under Partial Eta Squared, the effect size is moderate for the
Table 2: Levene's Test of Equality of Error Variancesa
F df1 df2 Sig.
Prior Interest in Dior Makeup (r=.076; p < .01). With the results close to that significance level, the effect is low for the Type of Endorser (r= .024, p< .05), the effect is moderate for Level of Involvement (r=0,121, .06<p<.14), and high for the variables Prior Interest in Makeup Sphere (r=.712, p>.14), Age (r=.185, p>.14), Gender (r=.361, p>.14), even though the significance level for the last three mentioned variables are well above the significance level of p = .05. The last two variables (Age and Gender) are interesting to be further observed and studied.
Table 3 Source Type III Sum of Squares Df Mean Square F Sig. Partial Eta Squared Noncent. Parameter Observed Powerb Corrected Model 60,527a 9 6,725 5,790 ,000 ,161 52,110 1,000 Intercept 150,105 1 150,105 129,23 0 ,000 ,323 129,230 1,000 PrIntMU ,159 1 ,159 ,137 ,712 ,001 ,137 ,066 PrIntDior 25,813 1 25,813 22,223 ,000 ,076 22,223 ,997 Age 2,050 1 2,050 1,765 ,185 ,006 1,765 ,263 Gender ,971 1 ,971 ,836 ,361 ,003 ,836 ,149 LvlInv 4,952 2 2,476 2,132 ,121 ,015 4,263 ,435 TypEndorser 5,979 1 5,979 5,147 ,024 ,019 5,147 ,618 LvlInv * TypEndorser 2,602 2 1,301 1,120 ,328 ,008 2,240 ,246 Error 314,777 271 1,162 Total 6805,938 281 Corrected Total 375,304 280
a. R Squared = ,161 (Adjusted R Squared = ,133) b. Computed using alpha = ,05
In order to examine whether there was significant difference between the different types of involvement, pairwise comparison was conducted. In Table 4, such substantial difference is noticed only between low and moderate involvement with a significant different of 0,043 (p<0,05).
Table 4
(I) LvlInv (J) LvlInv
Mean
Difference (I-J) Std. Error Sig.b
95% Confidence Interval for Differenceb
Lower Bound Upper Bound
Low Moderate -,315* ,155 ,043 -,620 -,009 High -,206 ,161 ,201 -,522 ,110 Moderate Low ,315* ,155 ,043 ,009 ,620 High ,109 ,163 ,506 -,213 ,430 High Low ,206 ,161 ,201 -,110 ,522 Moderate -,109 ,163 ,506 -,430 ,213
*. The mean difference is significant at the ,05 level.
In Table 5, the same pairwise comparison was analyzed for the second independent variable, which studies two types: famous and non-famous. Here, the difference is significant between the two types of influencers with resulting values of 0,024 (p<0,05).
Table 5 (I) TypEndorser (J) TypEndorser Mean Difference (I-J) Std. Error Sig.b 95% Confidence Interval for Differenceb
Lower Bound Upper Bound
Non-famous Famous ,295* ,130 ,024 ,039 ,551
Famous Non-famous -,295* ,130 ,024 -,551 -,039
*. The mean difference is significant at the ,05 level.
In order to prove or disprove the four hypothesis, the level of significance, the means, as well as whether the direction of the output were all observed and taken into account in the overall analysis of the data.
4.3.4 Hypothesis check Hypothesis 1
Hypothesis 1 suggested that the famous endorsers lead to a more positive attitude than non-famous endorsers. Although the pairwise comparison showed that there is significant difference between the two conditions (r=,024, p<0.05), the mean of the famous endorser (M= 4,642) and the mean of the non-famous endorser (M=4,937) disproves the hypothesis. Additionally, the results are opposite from the expectations, with attitude towards the famous endorser lower for the low-involved participants and higher for the highly involved individuals.
Hypothesis 2a
The second hypothesis was formed to prove that the effect of the type of endorser on the attitude is moderated by level of involvement, such that under low involvement, the effect of
the famous endorser is stronger than the one under medium or high involvement. Based on factorial ANOVA, and the calculated mean for the low involvement (M=4,616), the first hypothesis is not supported. While the expectations were that under low-involvement the means will be the highest and gradually decreasing by the increase of the involvement, the case is not that. What was noticed from the pairwise comparison is that the only significant difference is between the low and moderate involvement (r=,043, p<,05). However, in order to prove the 2b hypothesis, the results from the ANOVA analysis were separately studied for this case.
Hypothesis 2b
This hypothesis predicted that the effect of the type of endorser on the attitude is moderated by level of involvement, such that under moderate involvement, the effect of the famous endorser is less strong than the one under low involvement, but stronger than the one under high involvement. The mean for the moderate involvement (M=4,930) are, indeed, higher than the mean for the higher-involvement (M=4,822). However, it is not, as expected, lower than the mean for the low-involvement condition. Additionally, they are substantially different, the moderate-involvement being the one with the higher value. That being said, this hypothesis is still not supported, based on the reverse results that were not predicted in the initial analysis.
Hypothesis 2c
It was predicted that the effect of type of the endorser on the attitude is moderated by level of involvement, such that under high involvement, the effect of the famous endorser is weaker than the one under medium or low involvement, resulting in lower overall attitudes. However, when the mean of this condition (M=4,822) is compared with the other two, it is noticeable
Plot 2
Graph 2
that the expectations are not supported. The mean is neither lower than the low-involvement condition, nor it is significantly lower than the moderate-involvement case.
Based on the analysis, two plots were created. The first one (Graph 1) shows that the attitude towards the non-famous
endorser is more positive in all the three level of involvement. What could also be seen is that the expected linear function for the famous endorser is not negative but downward V-shaped.
In the second plot (Graph 2), the difference between the endorsers is more visible. The graph shows that under
non-famous influencer the attitude is not significantly different between the different levels of involvement. In addition, in this condition all the attitudes are positive. Under the famous endorser condition, the difference becomes significant. While in the moderate involvement
Plot 1
condition, the positive attitude is not changed significantly with the famous influencer, high and especially in the low involvement condition, the attitude does significantly change to more negative reported levels.
4.3.5 Additional Attitude analysis
In order to find whether the results differ for the different type of attitudes, three separate factorial ANOVAs with dependent variables of attitude towards endorser, attitude towards brand, and attitude towards advertisement, were conducted.
In the analysis where the dependent variable is Attitude Towards the Endorser, no significant difference was found between the three different levels of involvement and their means (Mlow=3,253; Mmod=3,053; Mhigh=3,138). However, there was a significant difference
between the two types of endorsers (r=.002, p<.01) and their means (Mnon=2,904,
Mfam=3,392). In addition, based on these means, Hypothesis 1 can supported for Attitude
Towards the Endorser measure. The value for the non-famous endorser are lower than the one of the famous one. However, Hypothesis 2a, 2b, 2c could not be supported in this situation.
In the test with the dependent variable as Attitude Towards the Advertisement, no significant difference was found for the Type for the Endorser (r=.073, p<0,05). Therefore, Hypothesis 1 is not supported under this measure. While still being insignificant, the means and the directionality prove to be slightly more promising such that the mean for the non-famous endorser scored lower (M=3,252) than the one for the famous endorser (M=3,528). In terms of level of involvement, the only significant difference is the one between the low and moderate conditions (r=.041, p<.05), where the mean of the former (M=3,589) is larger than the one of the latter (M=3,213). Thus, hypothesis 2a and 2b are partially supported. However, hypothesis 2c could not be supported, because there was no significant difference, as well as the mean of the high condition (M=3,368) was not lower than the two other conditions.
