Online advertising effectiveness : luxury fashion brands and their Instagram accounts

(1)

Online Advertising Effectiveness:

Luxury Fashion Brands and their Instagram Accounts

Sarah Noor van de Kraats 10368507

June 23, 2016

MSc Business Administration:

Entrepreneurship and Management in the Creative Industries

University of Amsterdam Amsterdam Business School Supervised by Ms. I. Rozentale

#LouisVuitton

#Burberry

(2)

Statement of Originality

This document is written by Student Sarah Noor van de Kraats who declares to take full responsibility for the contents of this document.

I declare that the text and the work presented in this document is 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.

(3)

This research contributes to theories of social media posting effectiveness. The main objective was to get more insights into the influences of creating a favorable brand attitude and increasing a consumer’s intent to purchase. The subjects of this research are the Instagram-accounts of three different luxury fashion brands, namely Louis Vuitton, Burberry and Ralph Lauren. In specific, it was researched whether and how brand awareness and the credibility, entertainment and informational content of an Instagram-accounts’ posts influence a person’s brand attitude towards a luxury fashion brand. Also, the influence of this brand attitude on purchase intention was examined. The data was obtained with the aid of an online survey. Path analysis with multiple regression analysis was used to determine these effects. For all the brands taken together, positive significant effects were found for the influences of brand awareness, credibility, entertainment and information on brand attitude. However, for the specific brands credibility and informational content were not always found to be of significance. Brand attitude was found to have a significant positive association with purchase intention for all the brands individually and together. This relationship between brand attitude and purchase intention was found to be the strongest of all significant relationships. This research contributes to the developing knowledge about online advertising effectiveness and indicates how luxury fashion brands can best make use of their Instagram-accounts to increase a consumer’s intent to purchase.

(6)

1. Introduction

The Internet has changed the way in which brands communicate with consumers. Unlike a decade ago, consumers now spend much more time behind their computers. This has led to more people watching series on the Internet and reading news articles online (Heine & Berghaus, 2014). As an effect of this, traditional advertising through television and magazines has become less effective because consumers are watching less television and do not get their information from traditional magazines or newspapers anymore (Schivinski & Dabrowski, 2014).

This change in consumer behavior especially has large consequences for companies and organizations operating in industries that rely heavily on advertising for their existence. One of those industries is the luxury fashion industry. Luxury fashion brands resemble prestige and exclusivity and strongly depend on a premium brand image for their profitability (Theng So, Grant Parsons and Yap, 2013). To create this premium brand image, it is beneficial that unique advertisements are created and marketed (Theng So et al., 2013).

Therefore, as consumers are moving online, luxury fashion brands have to shift their focus to online advertising as well. Besides advertisements on websites, many luxury brands have established their own social media accounts in order to engage consumers more and persuade consumers to buy their products (Halzack, 2016). Facebook and Twitter are the two social networks with the most users (Newcom Research & Consultancy, 2015). Research about advertising effectiveness for luxury fashion brands on Facebook and Twitter is extensive and shows that these social media can be beneficial to create brand engagement, which in turn increases the firm’s financial performance (Lin et al., 2015; in Yang, Lin, Carlson, & Ross, 2016; Taecharungroj, 2016).

Another social media platform that is used for online advertising by luxury brands is Instagram (Instagram, 2016; Halzack, 2016). Instagram is currently the most widely used social medium by luxury fashion brands, as an Instagram posts generates 11.5 times as many interactions as a Facebook post does and even more than 55 times as many interactions as a Twitter post. Especially for luxury fashion brand it is found that posts on Instagram generate much more interactions than posts on Facebook and Twitter (Halzack, 2016). Despite the popularity of the platform, extensive research on how Instagram influences brand attitude, brand engagement and purchase behavior cannot be found yet. Thus, because of the increasing importance of Instagram as an advertising channel, this thesis intends to look at Instagram-advertising effectiveness in particular.

(7)

The most frequently used way of measuring advertising effectiveness is through measuring the effects of brand awareness, brand attitude and purchase intention (Uribe, 2015). However, as this study focuses on modern technologies, Instagram posts’ credibility, information and entertainment values are included as well. This is because social media can differ widely among these aspects and they are found to influence a person’s attitude towards a post (De Vries, Gensler & Leeflang, 2012). It is investigated to what extent these factors, together with brand awareness, influence brand attitude, and how brand attitude influences purchase intention. Thus, this paper examines the ways in which luxury fashion brands can make use of Instagram in order to increase consumer’s intention to purchase a luxury fashion brand’s product.

As Instagram is relatively new and used by many people and brands these days (Instagram, 2016), this thesis contributes on theories concerning the effectiveness of online advertising and social media strategies. This research gives new perspectives on the influences of brand attitude and purchase intention in a digital setting, specifically for luxury fashion brands. Besides this, this thesis is beneficial for luxury fashion companies as the results provide companies with valuable information about how their social media posting-strategy is received and perceived by current and potential customers. Furthermore, luxury fashion brands can gain insight into what strategies will work best for them if they want to achieve certain goals with their Instagram postings.

This thesis proceeds by reviewing the relevant literature regarding luxury fashion brands, social media and Instagram usage of luxury fashion brands. Afterwards, brand awareness, brand attitude, credibility, entertainment content, informational content and purchase intent are explained and the hypotheses are formed. Following this, the research method and data collection are explained and the hypotheses tested. Then, the results are discussed and the implications and limitations of this research are mentioned. Finally, areas for further research are outlined.

2. Theoretical Framework

As this research aims to determine how Instagram posts by luxury fashion brands influence a consumer’s brand attitude and purchase intention, it is beneficial to first have a good definition of what advertising is, how it is perceived in modern times and what luxury fashion brands exactly are and why they rely heavily on advertising. Therefore, the following paragraph will give a detailed explanation of this. Afterwards, social media and how companies can profit from social media usage are explained.

(8)

2.1 Advertising and luxury fashion brands

Advertising is a marketing tool and is part of the marketing communications mix, which aims to manage profitable customer relationships (Kotler & Keller, 2014). Marketing is the process by which companies create value for their customers in order to build strong relationships with them so that they can capture value from them in return (Ebbers & Pruppers, 2012). Advertising’s traditional definition states that advertising is any paid form of non-personal communication by an identified sponsor designed to spread information about ideas, goods or services (Uribe, 2015). Firms conduct advertising via media. These media are widespread and can be direct mail, television, print media and the Internet (Ebbers & Pruppers, 2012). Advertising conveys a message to its audience that is encoded in language, symbols and other attributes that may have distinctive meanings (Ebbers & Pruppers, 2012).

Traditional advertising is of great importance for luxury fashion brands. This is because these brands depict prestige and exclusivity, as mentioned before (Theng So et al., 2013). According to Vigneron and Johnson (2004 in Li et al., 2012) luxury is a term used to describe the top category of prestigious brands in the academic literature. Luxury goods are associated with wealth, exclusivity and power and offer high levels of symbolic and emotional value (O’Cass & McEwenm, 2004 in Li et al., 2012). Luxury products in general refer to products that provide pleasure or comfort but are not necessary for the survival of a human being (Fuchs, Prandelli, Schreier & Dahl, 2013). Therefore, luxury fashion brands can be defined as brands that offer premium-priced fashion items with the highest level of quality. Examples of luxury fashion brands are Chanel, Louis Vuitton, Gucci and Prada. In contrast, mainstream fashion brands offer products for lower prices, but the quality of the products is lower as well (Fuchs et al., 2013). Examples of these brands include Zara and Topshop. With the aid of advertising, luxury fashion brands are able to create a premium brand image that is unique and allows them to ask premium prices as well (Theng So et al., 2013).

However, during the last couple of years consumers have shifted away from traditional advertising and have focused more on getting brand information at their convenience, which means that there has been a shift towards online advertising and brand content (Schivinski & Dabrowski, 2014). As an effect of this, the traditional definition of advertising as a paid form of communication does not hold entirely true anymore; with the occurrence of the Internet, many valuable advertisements are unpaid or indirectly paid (Tuten, 2008). Besides this, the Internet has changed the way advertisements are send out and

(9)

received (Tuten, 2008; Ebbers & Pruppers, 2012). Traditionally, advertising was done on a mass scale in order to reach a mass audience. Now, with the development of for instance email, messages that are more personal can be crafted and sent to individual customers directly. Another difference is that advertising has moved from one-way communication to two-way communication (Bakshi, Shamma & Gilbert, 2014; Tuten, 2008). Consumers respond to the online advertisements by engaging in conversations on Facebook-pages and sharing ads with their friends, among others (Berger & Milkman 2012).

Therefore, many luxury fashion brands have moved online during the past decade as well. A tool that especially allows for powerful online advertising and can generate a lot of buzz online is social media (Chen & Berger, 2013). Although it is a possibility on most social media platforms to pay the platform to show advertisements to selected people, in this thesis the postings of the companies are referred to as the advertisements.

2.2 Social media

Social media are “the two-way communication platforms that allow users to interact with each other online to share information and opinions” (Kim & Ko, 2010, p. 166). Overall, global social media usage has increased rapidly over the last decade. Compared to 2005, when only 10 per cent of all Internet users made use of social networking sites, 76 per cent of all Internet users made use of social media in 2015 in the United States (Perrin, 2015). In the Netherlands, a similar increase is found (Turpijn, Kneefel, van der Veer, 2015). Although young adults have been the largest group of social media users since Facebook and Twitter went online, there has been an increase of over 33 per cent of social media usage of people older than 65 years and over in the past decade (Perrin, 2015). Furthermore, it is found that social media users are most often people with high education levels and higher household incomes (Marketing Unity, 2016). This is an important fact for this research, as this paper looks at the Instagram usage of luxury fashion brands, which are usually expensive and whose target audience consists of people with higher income and higher education (Marketing Unity, 2016).

This increase in social media usage seems unlimited and social media has established itself as a mass phenomenon (Bruhn et al., 2012). As social media are thus also replacing more traditional media, the question has arisen to which extent marketers can best make use of social media in order to reach their advertisement objectives. This is because there is still little knowledge about how some social media strategies influence consumers and consumer behavior (Schivinski & Dabrowski, 2014).

(10)

2.2.1 Types of social media

There are many different forms of social media including wikis, micro blogs, pictures, podcast, weblogs and video (Richter & Koch in Kim & Ko, 2012). These social platforms vary from a technological perspective and in how people use them (Weinberg & Pehlivan, 2011). For instance, Facebook allows its users to post very long entries, while Twitter has a limited number of characters per message sent out. This indicates that the channels can differ in richness and vividness (De Vries et al., 2012). Besides this, other social media focus exclusively on careers (e.g. LinkedIn) while others focus on (short) films (e.g. YouTube).

Three of the most popular social media applications used today are Facebook, Twitter and Instagram (Newcom Research & Consultancy, 2015). Out of these three social media platforms Facebook was established first and still remains most popular with more than 1.5 billion monthly users worldwide (Zehoria Digital Marketing, 2015). It has been found that companies perceive Facebook presence as crucial for the survival of their organization. This is because the odds that your competitors are on Facebook are high and you have to make sure that you use Facebook correctly in order to stand out from the competition (State of Inbound Marketing, 2012 in Zehoria Digital Marketing, 2015). Facebook allows its users to posts pictures, text, videos and links to other websites on their page and on those of others as well. Therefore, Facebook posts can differ in richness and the posts can be visible on a person’s page for as long as you want (Facebook, 2016).

Twitter is the second-most used social media platform these days, with a number of 500 million tweets sent out daily and 320 million active users monthly (Twitter, 2016). Twitter is a micro-blogging application that allows its users to post messages with a maximum of 140 characters. This means that the posts are relatively short compared to Facebook.

The third-most used social media platform is Instagram. The application was launched on October 6, 2010. As of January 2016, Instagram has more than 400 million actives users monthly and more than 80 million photos are shared everyday (Instagram, 2016). Although this number of photos shared per day is still less than the daily number of tweets sent out, Instagram has more active users monthly than Twitter. The users of Instagram are diverse in age and gender, although there are slightly more females than males (Instagram, 2016).

Instagram posts receive most attention within the first few hours after they appear (Carah & Shaul, 2016). Users who follow a specific account may see their posts in their feed, which includes all posts by the pages they follow ranked on time and popularity. They may scroll through this feed, or they may go to the page they are interested in and look at all the

(11)

posts of this account without being interfered by posts from other accounts. However, most people will only see the posts in their feed and do not take the time to visit a specific account page (Carah & Shaul, 2016). Therefore, the glance at a post is usually momentary. This is a characteristic of society nowadays where people are exposed to expanding flows of images and people’s attention spans decrease (Wissinger, 2007). Therefore, to seek the attention from other Instagram users, it is of importance that the posts are entertaining and informative. To accomplish this, Instagram users frequently use filters, videos, and hashtags and make sure to post at times when attention is most high (Carah & Shaul).

After reviewing these three most popular social media channels, it is apparent that they all have their own strengths and can serve different purposes; both for individuals and advertising purposes. The next paragraph looks at how companies can best make use of social media as an advertising tool.

2.2.2 Social media for companies

Companies have noticed the potential benefits social media usage can have on their profitability and thus budgets directed towards social media are continuously growing (Tsimonis & Dimitriadis, 2014). Moreover, because of the low costs related to social media advertising, it can be very profitable for companies (Tsimonis & Dimitriadis, 2014). Social media platforms such as Facebook, Twitter and Instagram are accessible to everyone at no costs and since many firms are online and have access to free social media platforms, firms can reach a mass audience with one message (Schivinski & Dabrowski, 2014). In this way, a brand’s name and reputation can spread with ease and limited costs and people become familiar with the brand (O’Flynn, 2012 in Tsimonis & Dimitriadis, 2014).

Especially for Facebook a mass audience can be reached easily because of the amount of users. Companies and academics are aware of this and therefore research about companies’ use of Facebook is extensive. It is found that Facebook posts that are controversial to some degree have a high tendency to be talked about (Chen & Berger, 2013). Furthermore, if companies want their consumers to share their posts with a large audience on social media, they should target people that want to look good. This is because Barasch and Berger (2014) found that consumers have a tendency to broadcast posts they feel make them look good in other people’s eyes. This sharing by the audience is also found to be beneficial as more people can see them in this way. This links to research carried out by Chen and Berger (2013) about E-word-of-mouth. They found that when people respond to posts and share it with others they are able to generate a buzz for a product or company. This again

(12)

influences consumer’s perceptions and has a positive effect on their brand attitude. This positive brand attitude has a very strong influence on purchase intention (Schivinski & Dabrowski 2014).

Social media posts can have a negative influence on consumer perceptions when the posts are not up-to-date. Kim and Ko (2012) found that Facebook posts have to be up-to-date in order for consumers to see them as relevant and credible. Besides this, consumers can perceive the Facebook posts by companies as uninteresting and do not feel any attachment or closeness to the brand (Papasolomo & Melanthiou, 2012). If this is the case, the posts are not found to enhance brand attitude and purchase intention. This not only holds true for posts on Facebook, but for other social media as well, such as for example Twitter.

For Twitter, research is extensive as well. As Twitter is a micro-blogging site that enables consumers to send messages directly to companies that are also on Twitter, consumers often send them their compliments and complaints. It is found that if companies respond to these complaints, customer relationships improve and consumers will have a more favorable attitude towards the brand (Ma, Sun & Kekre, 2015). Besides this, Schivinski and Dabrowski (2014) found that for Twitter, user-generated social media communications have a greater effect on consumer’s perception of brands then firm-created communication as well. Thus, for companies on Twitter it is necessary to have a positive and informative strategy and to carefully manage the content consumers create.

Altogether, previous research on Facebook and Twitter is extensive and determines effective strategies for these media in order to increase brand attitude and purchase intention. This does not hold true for Instagram. Therefore, this research investigates the extent to which the company’s advertising strategy influences a consumer’s brand attitude and purchase intention, specifically for the luxury fashion industry. The next paragraph outlines how this will be measured.

2.3 Advertising effectiveness

Throughout the years, the most frequently used way of researching the effectiveness of advertisement is through a trilogy of communication effects (Uribe, 2015). This trilogy consists of cognitive, affective, and conative/behavioral dimensions (Hutchinson & Alba, 1992; Li, Daugherty & Biocca, 2002; Uribe, 2015). For each of these dimensions, one specific indicator is used. For the cognitive dimension this is brand awareness, while the affective dimension refers to brand attitude and the behavioral impact of social media posts is measured with a consumer’s purchase intention. In multiple researches, these three concepts

(13)

are found to influence each other in a sequential order (Bruhn et al., 2012; Li et al., 2002). Thus, in order for brand attitude to develop a person first needs to be aware about that brand. Then, when a positive attitude towards the brand is developed the intention to purchase can increase.

Li, Daugherty and Biocca (2002) used the framework to prove that 3-D advertising is, just as traditional advertising, able to influence the product knowledge, brand attitude and purchase intention of the consumers (2002). Bruhn, Schoenmueller and Schafer (2012) found that for social media advertising, these three concepts also influence each other. Since this paper investigates the effectiveness of social media advertising as well, it seems appropriate to use this framework as it has already been proven useful.

To further account for other effects of the Instagram posts that can influence purchase intention this research adds three more aspects to the framework. These are the luxury fashion brands Instagram’s credibility, entertainment and informational content. The

decision to include these aspects comes from earlier research on social media advertising. This is because social media posts allow for a different way of advertising than traditional advertising. For example, social media posts can differ in informational content, vividness and interactivity (De Vries et al., 2012). All these aspect influence the overall attractiveness of a brand’s Instagram account.

Thus, the aforementioned aspects are all taken into account when looking at how Instagram-accounts’ posts influence a consumer’s purchase intention. These concepts and the hypotheses formed as a result of them are discussed next.

2.3.1 Brand awareness

Brand awareness concerns the ability of a consumer to recognize or recall the brand with sufficient detail to make a purchase (Kotler & Keller, 2012). This cognitive measure is used to determine the ability of an Instagram post to attract attention and create knowledge about a specific brand and the offerings of that brand, namely the products and services (Li, Daugherty & Biocca, 2002). According to multiple studies, brand awareness consists of two components: brand recognition and brand recall (Percy & Rossiter, 1992; Uribe, 2015; Keller, 1993; Lu et al., 2014).

The ability of consumers to identify a brand when the brand name is given as a clue is referred to as brand recognition (Keller, 1993). It concerns a memory of when the brand is heard of before (Lu et al., 2014). Thus, in order for a consumer to have brand recognition the consumer needs to correctly define a brand as having been seen or heard of before (Keller,

(14)

1993).

In contrast to brand recognition, brand recall concerns the recall of brands without being in the direct environment of that brand (Kotler & Keller, 2012). Brand recall occurs when, for instance, a consumer has a specific product need first. Then, a consumer relies on his or her memory to come up with possible solutions for brands to purchase from. Thus, a consumer must recall from their memory in order to be able to make a decision about buying (Percy & Rossiter, 1992).

2.3.2 Brand attitude

A brand attitude is a summative evaluation of an object and is beneficial in aiding consumers to evaluate a brand’s perceived ability to meet a need (Kotler & Keller, 2012). According to existing literature, advertisements can create a positive brand attitude (Uribe, 2015). In order for a person to have an attitude towards a brand, a person first needs to be aware about a brand. Thus, brand awareness influences brand attitude. Multiple researches have found that this is a positive relationship (MacKenzie & Lutz, 1989; Uribe, 2015). Laroche et al. support this claim (1996). In their research concerning brand familiarity and purchase intention they found that a consumer’s brand attitude is affected by their brand familiarity.

These findings are also supported by Li, Chang and Chang (2014). They found that if consumers have high brand awareness relating to a product recommend online by bloggers, this has a positive association with a consumer’s attitude towards this brand. Since blogs and Instagram posts are both similar in respect to online content that is provided to generate sales and recognition, it is expected that high brand awareness also has a positive association with a high brand attitude concerning luxury fashion brands on Instagram. Therefore, the following hypothesis is formed:

H1: Brand awareness will positively influence the brand attitude of luxury fashion brands on Instagram.

2.3.3 Credibility

Ling, Piew and Chai (2010) found that in order for consumer’s to have a positive attitude towards advertisements, the sources of the advertisements have to be credible, trustworthy and believable. Credibility thus refers to the extent to which a consumer perceives the claims made in the post, either about the brand or product, as truthful, trustworthy, reliable and believable (MacKenzie & Lutz, 1989).

(15)

According to Laroche et al., (1996) and Mackenzie and Lutz (1989), credibility affects traditional brand attitude for brands. As mentioned, in this thesis a post is seen as an advertisement. Thus, as it is believed that credibility is found to influence brand attitude in previous advertising studies (Kim & Ko, 2010; MacKenzie & Lutz, 1989) the following hypothesis is formed:

H2: Credibility will positively influence the brand attitude of luxury fashion brands on Instagram.

2.3.4 Entertainment

Furthermore, the content of the social media posts can have multiple cues. These can either be visual, auditory or verbal (Instagram, 2016). Given the fact that consumers want to remain entertained with advertisements and that new types of advertisements are found to positively influence brand attitude (e.g. Li et al., 2012; Taylor, Lewin, and Strutton 2011; Raney et al. 2003, in De Vries et al., 2012), it is assumed that Instagram entertainment also influences brand attitude. Entertaining advertisements are ads that are perceived to be fun, exciting and flashy and are known as a hedonic measure. It is found that if advertisements are more entertaining, people create a more positive attitude towards the ads since they enjoy viewing them. As mentioned, this again influences people’s perceptions of the products and the brand as well in a positive way (Olney, Holbook and Batra, 1991; Fennis, Das & Fransen, 2012).

As brands can make use of different types of post, including photos and short videos (Carah & Shaul, 2016), they can influence the entertainment levels of their overall Instagram page. They can do this with the inclusion of posts that are diverse concerning the format of the posts and have varying levels of vividness. As this is found to positively influence an attitude towards a post (Coyle & Thorson, 2001; De Vries et al., 2012), it is assumed that this will also positively affect the attitude towards the brand. Therefore, the following hypothesis is formed:

H3: Instagram accounts with posts that are perceived as entertaining have a positive association with brand attitude.

2.3.5 Informational content

Finally, information given out by organizations also affects the attitude towards a brand. This information can, for instance, include links to websites, prices, information about the events in the posts and names of the celebrities or models used in the images or videos.

(16)

Ducoffe (1992) found a positive relation between informativeness and attitude towards advertisements (in Wang & Sun, 2010). Bruhn et al. (2012) found similar results. Their research concluded that if people perceive advertisements as more informative, their attitude towards the advertisement and the brand would increase. Therefore, the following hypothesis is formed:

H4: Instagram accounts with posts that are perceived as informative have a positive association with brand attitude.

2.3.6 Purchase Intention

According to Kim and Ko, “purchase intention is defined as the consumer’s possibility of purchasing in the future” (2012, p. 167). Multiple studies found that brand attitude is related to purchase intention (Kim & Ko, 2012; Wang & Sun, 2010). For example, Wang and Sun (2010) found in their study about consumer’s beliefs, attitudes and behavioral responses towards online advertising that a more positive attitude towards an online advertisement predicted the frequency of both clicking on the links to a for online shopping and online shopping itself.

Furthermore, Lu, Chang & Chang (2014) found in their research about blogger’s posts that a positive brand attitude towards the advertisers indeed positively influences the purchase intention of consumers. These findings are supported by research by Laroche, Kim and Zhou (1996). They found that in order for a consumer’s purchase intention to increase, a consumer has to have confidence in the brand. Finally, Schivinski & Dabrowski (2014) found a positive effect of brand attitude on purchase intention for the clothing industry in their research. Since the luxury fashion industry mainly promotes clothing, these results indicate that the same social media advertising effects can be found for the luxury fashion industry. Thus, considering these previous studies, the following hypothesis is formed:

H5: A positive brand attitude towards a luxury fashion brand posting on Instagram will lead to a positive purchase intention.

To sum up, it is assumed that brand awareness, credibility, entertainment and information all have a positive influence on brand attitude and brand attitude has a positive association with purchase intention. However, since brand awareness, credibility, entertainment and information are expected to have a direct positive relationship with brand attitude and brand attitude is expected to have a direct positive relationship with purchase intention, it will be

(17)

tested whether the first 4 independent variables also have a direct positive relationship with purchase intention. To do so, path analysis is used.

Figure 2.1 illustrates the conceptual framework of this thesis. The purple lines indicate the hypotheses that are tested, while the green lines represent the direct effects that are measured as well because of the path analyses.

Figure 2.1 Conceptual Framework

3. Method

This research focuses specifically on (potential) consumers’ responses towards the Instagram posts/advertisements by luxury fashion brands. The aim is to see if there are certain strategies that increase consumer’s brand attitude and purchase intention. This section focuses on the empirical setting of the research.

3.1 Research design

As this thesis wants to establish whether there are positive relationships between the variables of interest, explanatory research will be carried out. In particular, a survey design is chosen as this fits the purpose of this research, which is to determine the relationships between the multiple variables (Saunders, Lewis and Thornhill, 2007). An advantage of data collection with a survey is that it allows the researcher to suggest possible reasons for particular

(18)

relationships among variables and the results are generally quite generalizable (Saunders et al., 2007). Furthermore, the survey strategy allows for the collection of data from a large number of people in a cost-effective manner (Sanders & Lewis, 2012). As this thesis aims to gather data from a sample of 125 people, using a survey is less expensive and time-consuming than talking to every individual in person. Lastly, another advantage is that a survey strategy requires less skill and sensitivity from the researcher to undertake than other methods of data collection, such as in-depth interviews (Saunders et al., 2007).

However, there are also limitations to the use of surveys. Firstly, it is necessary to ensure that the research sample is representative and that the response rate is sufficient for the survey to be usable. This process can be very time-consuming. Besides, since respondents need to fill out the survey, the researcher’s progress is partly dependent on when the respondents complete the survey (Saunders et al., 2007).

Secondly, the data that can be collected with the aid of a survey is limited (Saunders et al., 2007). This is because only a limited amount of questions can be asked and a limited amount of images can be shown without the respondents losing their interest and goodwill. Also, when a respondent gives a detailed answer to an open question there is no chance for the researchers to get deeper into this topic and ask what the respondents actually mean. Thirdly, the researcher always has to be very careful when constructing the survey in order for the data to be useful. According to Fan and Yan (2010) it is of importance to ensure that the questions of the survey are ordered and displayed in a comprehensible manner. When this is not the case the survey data will be of little use.

To make sure that the survey is made and distributed in a way that allows for the collection of useful data, the aforementioned issues are taken into account when designing the survey. For instance, to ensure that there is a representative sample with a quick response rate, potential respondents are asked in person to complete the survey as soon as possible. Also, since many of the respondents, who are acquaintances of the researcher and fit the target group, have a large group of friends and acquaintances that also fit they target group, they will be asked to do the same with their friends. In this way, people are aware of the need for quick completion of the survey. Furthermore, to deal with the limited amount of data that can be collected, the researcher made sure all the necessary questions were included and that the survey does not take longer than 10 minutes, which is mentioned beforehand to the respondents.

(19)

3.2 Data sources and data collection 3.2.1 Questionnaire

The survey data is gathered with the aid of an online questionnaire. This questionnaire is constructed with the survey-design tool Qualtrics (qualtrics.com) and is distributed online. This is a cost-effective and efficient way of data collection (Saunders & Lewis, 2012).

A web survey has some negative aspects and important facts that have to be taken into account. First of all, web surveys have on average approximately 10 per cent lower response rates than a telephone or mail survey (Fan & Yan, 2010). Furthermore, since pictures are shown it is decided to keep this number at a minimum in order for the respondents to not lose interest. Therefore, the two most recent images targeting men and the two most recent images targeting women for each individual brand are shown. Also, the survey is structured in a comprehensible manner. For all the brands the same questions are asked in the same order.

Moreover, since this research focuses on the usage of Instagram, a mobile application, it is crucial that the survey can be displayed easily on a mobile phone. In this way, the target audience is more inclined to complete the survey instantly (Saunders et al., 2007). Finally, anonymity is guaranteed to the respondents as this builds trust, which makes the respondents inclined to answer more truthfully (Saunders et al., 2007).

3.2.2 Research setting

The Instagram accounts of real luxury fashion brands are used for this study. The brands selected are Louis Vuitton, Burberry and Ralph Lauren. These brands are chosen because they are the three highest ranked fashion brands in the top 100 of the international luxury brand ranking (Interbrand, 2015).

The brands all have a different country of origin. Louis Vuitton hails from France, Burberry is British and Ralph Lauren is American. All these brands have a verified Instagram-account. As of April 21, 2016 Louis Vuitton has the most followers of these brands on Instagram with 10.5 million followers. Burberry has 6.5 million followers, while Ralph Lauren has the lowest amount of followers with 3.3 million (Instagram, 2016).

The number of followers on Instagram indicates that they are well-known already. Therefore, it is likely that all respondents will indicate high levels of brand awareness. The downside of using a well-known brand that is generally seen as a positive, successful brand is that consumers will also indicate positive brand attitudes. These positive brand attitudes towards well-known brands are likely to be found because of the reputation of the brand

(20)

(Uribe, 2015). This can lead to the outcome of this study not reporting a significant effect between the posts credibility, entertainment levels and information levels.

However, since all types of people are asked to participate in the survey, there may be many respondents who are not knowledgeable about the fashion industry. In this case, there may be respondents who do not know the brands mentioned in the questionnaire. Thus, to decrease this risk of only having respondents indicating high levels of brand awareness, people from all different kinds of backgrounds are asked to participate.

3.2.3 Sampling strategy

To target the respondents, convenience sampling is used. This is a non-random sampling technique (Saunders & Lewis, 2012). Specifically, self-selection sampling is used for this thesis. This is a type of volunteer sampling, which allows individuals to determine if they want to take part in the research (Saunders et al, 2007). The link to the questionnaire is distributed through the social network of the researcher with no extra incentives than aiding the researcher.

An advantage of self-selection is that people who have strong feelings about the research questions usually answer (Saunders et al., 2007). Furthermore, non-probability sampling offers a practical way to collect data in a limited amount of time (Saunders & Lewis, 2012). To ensure that a good representation of the population is captured in this research, a diverse group of respondents was asked to participate. This is explained in the next paragraph.

3.2.4 Sample description

Many luxury fashion brands offer products to both men and women. The target audience is aged between 18 and 65-year old, as this category covers both men and women with the income to spend on luxury fashion brands and people that have access to Instagram (Patterson, 2015). All respondents who complete the survey are included in the sample. Furthermore, although this research focuses on Instagram accounts, non-users of Instagram are also asked to complete the survey. This is done to be able to make a distinction between non-users and users with the results.

The survey was available from April 21, 2016 until May 17, 2016. A total of 253 respondents opened this survey by clicking on the link. However, not all respondents finished the questionnaire. Out of the 253 people opening the survey only 175 respondents completed the survey. This means that 78 people started the survey but did not proceed to finish it.

(21)

Therefore, the dropout rate of this survey is 30.8%. However, due to a flaw in the survey left unnoticed by the researcher, data of 50 respondents was unusable. This means that the total sample size for this thesis is 125 respondents.

3.3 Measurement design of variables

The next paragraphs outline the measures that are used in this thesis and explain how they are used.

3.3.1 Measurement of variables

The following variables are used in this thesis. All variables are measured with a 5-point Likert-scale ranging from strongly disagree to strongly agree. Appendix A gives an overview of the survey sent out.

Brand Awareness: To measure brand awareness both brand recognition and brand recall is measured. To do so, the variables used by Lu et al (2014) are used.

Credibility: To measure credibility concerning each brand’s Instagram posts, respondents are asked to rate three statements concerning credibility taken from Greer (2003). To have a comprehensive measure of credibility, both credibility towards the brand and towards the posts are included in this measure (Ling et al., 2010).

Entertainment and Information: The entertainment value of the Instagram posts from the luxury fashion brands is measured using three questions adapted from the study by Mathwick, Malhotra and Rigdon (2001). This measurement is also adapted to measure information. Instead of focusing on entertainment, there is a focus on information in this part. Brand Attitude: Brand attitude is measured using four statements. These statements are taken from Mathwick and Rigdon (2004). To make these statements more fitting regarding the content of this research, slight changes in the wording are made.

Purchase Intention: Purchase intention is measured with four statements adopted form Bian and Forsythe (2012) who used a 7-point Likert-scale. However, to make the measurement more fitting regarding the rest of the data, a 5-point Likert scale is used.

After generating the survey data, all the items of each variable, such as the 3 items of credibility for Louis Vuitton, are added to create one variable concerning credibility for Louis Vuitton. This is done for all the Likert-scale items in order to generate comprehensive constructs. Thus, 6 (brand awareness, credibility, entertainment, information, brand attitude and purchase intention) x 3 (Louis Vuitton, Burberry and Ralph Lauren) =18 variables are

(22)

constructed. Moreover, as this thesis wants to focus on the overall influences of brand awareness, credibility, entertainment, information, brand attitude on purchase intention and is not only interested in the differences between the brands, 6 constructs are created that represent the total scores for the six items. These are labeled BAwTOT, CRTOT, ETTOT, INTOT, BAtTOT and PITOT.

3.3.2 Control variables

The control variables included concern the characteristics of the (potential) consumers. One of these control variables is age. As mentioned previously, this research focuses on both men and women and in order to look at differences between males and females, respondents have to indicate their gender. Another control variable used in this survey is age. This variable is included because it was previously assumed that especially middle-aged women buy luxury brand products but there has been a shift towards younger people in the recent decade (Unity of Marketing, 2016).

Furthermore, the respondents are also asked to indicate whether they are Instagram users and whether they purchased products from luxury fashion brands. This is done to control for the effect people already are familiar with both Instagram and luxury fashion brands. The last control variable this thesis uses is income since it is expected that people with higher incomes are more likely to purchase luxury fashion products.

In the survey, more control variables are asked, as can be seen in Appendix A. However, these variables are not used for the analyses. They are only used to generate a better view of the whole sample, as is explained in the next chapter.

3.3.3 Method of analysis

To test all the hypotheses, multiple regression path analysis is used. Path analysis is a causal modeling approach to explore the correlations within a defined model. Path analysis can be used to estimate the magnitude and strength of effects within a hypothesized causal system (Lleras, 2005). The models of path analysis cannot prove causation, but they do reflect theories about causation and they can inform the researcher as to which hypothesized model fits the data best. In this thesis, multiple regressions are used to determine the relationships. The standardized regression coefficients (beta) that are found with the regressions show the direct effect of an independent variable on a dependent variable in the path model. It is examined whether these relations are strong and significant and thus if there is support for the hypotheses.

(23)

For the hypothesized model in this thesis, two layers of linear regressions are necessary. The criterion in the first regression is brand attitude while the predictors are brand awareness, credibility, entertainment and information. For the second regression, the criterion is purchase intention, while the predictors are brand awareness, credibility, entertainment, information and brand attitude. The model is researched using SPSS. Besides this, the control variables age, gender, having an Instagram account, having purchased a luxury fashion item before and income will be included in the regressions as well. The results are displayed with the aid of path diagrams. The standardized regression coefficient (beta) shows the direct effect of an independent variable on a dependent variable in the path model.

Furthermore, for all the analyses a 95% significance level is used. Thus, to be of significance the p-value has to be 0.05 or smaller than 0.05 (p < 0.05). For all the analyses, the corresponding p-values can be found in the appendix when they are not given in the text.

4. Results

This section presents the findings of the analyses performed in this thesis. Firstly, the descriptive statistics are explained, followed by the correlations between the variables. Finally, the regression analyses are explained and performed.

4.1 Descriptive statistics

As mentioned before, the total sample size for this thesis is 125 respondents. In the next sections, the characteristics of these respondents are discussed.

4.1.1 Gender, age and nationality

Out of these 125 respondents, 82 are female (65.6%) while the remaining 43 respondents are male (34.4%). The largest part of the sample is aged between 18-25 years old (59.2%). The variable age is measured in 5 categories. The dataset used in this study does not include any participants from the first category, therefore three dummies are included to control for age. These are dummies for the categories 2 until 4. Table 4.1 gives an overview of the different age categories and lists the dummy variables.

(24)

Table 4.1. Age

Age Frequency Percent Dummy

<18 0 0.0% - 18-25 74 59.2% Control Group 26-34 18 14.4% AgeGroup2 35-54 22 17.6% AgeGroup3 >54 11 8.8% AgeGroup4 Total 125 100% -

The sample consists of people from 22 different nationalities. The largest part of this sample, 96 respondents, is of Dutch origin (76.8%). Appendix B gives an overview of all the nationalities of the respondents.

4.1.2 Education, job title and income

The highest level of education obtained by the respondents of the questionnaire was a Master’s Degree (31.2%), while the most frequent level of education obtained is a Bachelor’s Degree (33.6%). Table 4.2 gives an overview of the education obtained.

Table 4.2. Education

Level of education obtained Frequency Percent

Master’s Degree 39 31.2% Bachelor’s Degree 42 33.6% HBO 21 16.8% MBO 6 4.8% VWO 13 10.4% Other 4 3.2% Total 125 100%

Out of these 125 people, more than half is a student (53.6%). Table 4.3 gives an overview of the job titles and functions of the respondents.

(25)

Table 4.3. Job

Current job title Frequency Percent

Student 67 53.6%

Intern 5 4.0%

Employee 24 19.2%

Executive employee 8 6.4%

Self-employed 11 8.8%

None of the above 10 8.0

Total 125 100%

The biggest part of the respondents earns less than €10.000 a year (52.8%). For income, there are also multiple categories. Since the respondents are divided over all the different categories and income is used as a control variable, 5 dummy variables are used for the categories 1 until 5. Table 4.4 gives an overview of the distribution of income earned by the respondents and names the dummy variables.

Table 4.4. Income

Average yearly income Frequency Percent Dummy

<€10.000 66 52.8% Income1 €10.000-€20.000 14 11.2% Income2 €20.000-€30.000 7 5.6% Income3 €30.000-€40.000 14 11.2% Income4 €40.000-€50.000 8 6.4% Income5 >€50.00 16 12.8% ControlVariable Total 125 100% - 4.1.3 Instagram users

Out of the 125 respondents, 97 have an Instagram-account (77.6%). This item is included as a control variable, since it is believe this can have an influence on the results. Graph 4.1 gives a representation the percentage of respondents with an Instagram account.

(26)

Graph 4.1. Instagram users

Of these Instagram-users, the biggest part indicated that they do not follow a luxury fashion brand on Instagram (64.9%), see table 4.5.

4.1.4 Luxury fashion items purchase history

Finally, respondents were asked about their purchase history concerning luxury fashion brands. 86.4 per cent indicated that they had bought a luxury fashion product at least once in the past. This was also used as a control variable, since it is expected that people who have already bought luxury fashion items are more inclined to do so in the future. The percentages and numbers are outlined in table 4.6.

Table 4.5. Luxury fashion brands

Follows luxury fashion brands Frequency Percent

Yes 34 35.1%

No 63 64.9%

Total 97 100%

Table 4.6. Bought luxury fashion product Has ever bought a luxury

fashion product

Frequency Percent

Yes 108 86.4%

No 17 13.6%

(27)

4.2 Reliability analysis

Before any analysis could be carried out on the individual drivers of brand attitude and purchase intention, the scales comprising these drivers were tested on their reliability. Reliability is the extent to which data collection techniques produce consistent findings. The most common measure of scale reliability is Cronbach’s alpha (α). A Cronbach alpha higher than .7 is acceptable (Fields, 2005). In the table 4.8 reliability coefficients are giving for the 6 scales of each of the three luxury fashion brands.

Table 4. 8. Reliability coefficients

α Louis Vuitton Burberry Ralph Lauren Overall

Brand Awareness .739 .774 .759 .791 Credibility .869 .903 .945 .913 Entertainment .832 .938 .927 .888 Information .745 .806 .851 .873 Brand Attitude .847 .885 .900 .900 Purchase Intention .952 .948 .939 .904

The brand awareness, credibility, entertainment, information, brand attitude and purchase intention scales for each of the three brands were found to have high reliability (α > 0.7). Also, all the corrected item-total correlations indicate that all the items have a good correlation with the total score of the scale (all above 0.30). Therefore, no items were removed for any of the constructs.

4.3 Correlations

Besides the reliability tests the correlations between all the variables in this thesis are shown before testing the hypotheses as well. The correlation matrix in Appendix C shows how all the different variables are correlated to each other. It also shows the means and standard deviations of the variables.

Regarding the first hypothesis of this thesis, it can be concluded that for Louis Vuitton, Burberry, Ralph Lauren and all the brands combined, there is a significant positive relationship between brand awareness and brand attitude. The same positive significant relationship can be observed for hypothesis 2, 3 and 4 for all the brands individually and together.

(28)

brand attitude and purchase intention are also all found to be significant for each brand individually and for all the brands together. All these coefficients show positive relations.

Most of the correlations with the control variables are not significant for brand attitude and purchase intention. Only the correlations between gender and brand attitude and purchase intention for Burberry are found to be significant (0.28 and 0.25). Furthermore, for all the brands together the correlation between gender and brand attitude was also found to be significant (0.18). None of these correlations are exceptionally high. When testing the hypotheses, the variables will be controlled for by other variables in the model held constant. This is done next.

4.4 Hypothesis testing 4.4.1 Louis Vuitton

Firstly, the hypotheses for Louis Vuitton were tested without the control variables. For the first step, the predictors were brand awareness, credibility, entertainment and information while the criterion was brand attitude. The full model with the path coefficients is found in figure 4.1.

Figure 4.1 Full model Louis Vuitton without control variables

These path coefficients measure the relative strength and sign of the effect from a predictor variable to criterion variable. When more than one variable is present, the coefficients represent partial regression coefficients that measure the effect of one variable on the other, controlling for the other variables (Lleras, 2005).

(29)

As the results show, the relationships of brand awareness (.234), credibility (.138), entertainment (.288) and information (.194) are all positive. However, the relationship between credibility and brand attitude is the only one not significant (p = 0.101). Furthermore, the beta coefficients measure how strongly each predictor variable influences the criterion variable, relative to the predictor variables. This standardized beta is measured in units of standard deviation. This means that for example for brand awareness, a beta value of .234 indicates that a change of one standard deviation in brand awareness will result in a change of .234 standard deviation in brand attitude. Thus, a higher beta value indicates a greater impact of the predictor variable on the criterion variable. The strongest influence on brand attitude is found to be of entertainment.

The effect found between brand attitude and purchase intention was also found to be positive and significant (p = 0.00). This relationship was very strong with a beta coefficient of 0.710. This means that for brand attitude, a beta value of .710 indicates that a change of one standard deviation in brand attitude will result in a change of .710 standard deviation in purchase intention. Thus, a higher beta value indicates a greater impact of the predictor variable on the criterion variable. This is the standard way of interpreting this data. This is also done for the next analyses.

This strong effect was not found for brand awareness (-.075), credibility (-.011), entertainment (.119) and information (-.018) and their relationships with purchase intention. These beta coefficients were mostly negative and not significant (p < 0.05) and they were also found to be very small. As mentioned, the corresponding p-values can be found in the appendix.

Thus, from this path analysis, it can be concluded that in this model for Louis Vuitton, brand awareness, entertainment and information all influence brand attitude and that brand attitude influences purchase intention. The relationship between brand attitude and purchase intention is also by far the strongest relationship found. Because of this relationship between brand attitude and purchase intention, brand awareness, entertainment and information also have an indirect influence on purchase intention through brand attitude, but there is no direct relationship. For credibility, no significant relationship is found with brand attitude or purchase intention, neither direct nor indirect. In Appendix D, a table containing the SPSS output of the hypotheses testing for Louis Vuitton can be found.

For Louis Vuitton, the full model was also analyzed using control variables. This model with the path coefficient is found in figure 4.2.

(30)

Figure 4.2 Full model Louis Vuitton with control variables

The path coefficients did not differ significantly from the coefficients generated by the analysis without control variables for brand awareness (.247), credibility (.131), entertainment (.280) and information (.196). The results are also the same for the relationships between brand awareness, entertainment and information; significant but small and moderately strong (p < 0.05). Furthermore, for credibility no significant relationship was found either. Again, the only significant relationship with purchase intention was brand attitude (.678). Here the coefficient is slightly lower, but still strong and significant (p = 0.00). Furthermore, neither of the control variables was found to be significant. This means that when controlling for the variables mentioned, no differences are found between the relationships. This can also be concluded by the small differences between the models analyzed without and with control variables.

For both the models with and without control variables, the reduced model was also tested. The graphs and the tables containing the results can be found in Appendix D. This was done to test whether the reduced model fits the data better compared to the full model. Overall, it was found that the reduced models fit the data as well as the full model does. That is, a causal model deleting the direct influence of the brand awareness, credibility, entertainment and information on purchase intention did not fit the data more poorly than did the model including these paths.

(31)

4.4.2 Burberry

First the model for Burberry was tested first without the control variables. The path diagram can be found in figure 4.3. The SPSS results concerning the diagram are found in Appendix E.

Figure 4.3 Full model Burberry without control variables

The relationships between brand awareness (.356), credibility (.216), entertainment (.301) and information (.203) and brand attitude are positive for Burberry, too. Although these relationships all show small beta coefficients, the correlation coefficients show moderate strength and all the relationships were found to be significant (p < 0.05). For Burberry, brand attitude is most strongly influenced by brand awareness out of the four variables, followed closely by entertainment.

The relationships between brand awareness (.092), credibility (-.029), entertainment (.101) and information (-.011) and purchase intention were either positive or negative, very small, while not significant (p > 0.05). Again, only a positive, significant relationship was found between brand attitude (.750) and purchase intention (p = 0.00). Thus, a change of one standard deviation in brand attitude will result in a change of .750 standard deviation in purchase intention. This relationship is very strong as well with a high beta and correlation coefficient. From this path analysis and model it can be concluded that for Burberry, brand awareness, credibility, entertainment and information all influence brand attitude and that brand attitude influences purchase intention. This also means that brand awareness, credibility, entertainment and information have an indirect influence on purchase intention

(32)

through brand attitude, but there is no direct relationship.

For Burberry, the full model was also analyzed using control variables. This model with the path coefficient is found in figure 4.4.

Figure 4.4 Full model Burberry with control variables

The path coefficients did not differ significantly from the coefficients generated by the analysis without control variables for Burberry. The results changed because of the control variables and were found to be brand awareness (.309), credibility (.206), entertainment (.270) and information (.246) and brand attitude. However, the relationships remain similar: significant but small (p < 0.05). Brand awareness still showed the strongest influence on brand attitude. Also, the only significant relationship with purchase intention was again with brand attitude (.715), meaning a change of one standard deviation in brand attitude will result in a change of .715 standard deviation in purchase intention, as explained before. Here the coefficient is still strong and significant (p = 0.00).

Furthermore, one of the control variables turned out to be significant (p = 0.005). This was Income Group 3. The coefficient was -.246, which indicates a negative and moderately strong relationship with brand attitude. Thus, a person with an income of between €20.000 and €30.000 per year score -.246 standard deviations lower on brands attitude than people from any other income category.

(33)

tested. The graphs and the tables containing the results can be found in Appendix E. This was done to test whether the reduced model fits the data better compared to the full model. Overall, it was found that the reduced models fit the data as well as does the full model.

4.4.3 Ralph Lauren

For Ralph Lauren the same steps are undertaken as mentioned before for Louis Vuitton and Burberry. The model can be found in figure 4.5 and the SPSS output can be found in Appendix F.

Figure 4.5 Full model Ralph Lauren without control variables

Again, the relationships between brand awareness (.339), credibility (.160), entertainment (.269) and information (.164) with brand attitude were tested first. Similar to the results from the previous two brands, the coefficients were moderately strong, with brand awareness most strongly influences by brand awareness. However, of these 4 relationships, only two were found to be significant at the 5 per cent significance level. These were brand awareness (p = 0.000) and entertainment (p = 0.003). Furthermore, none of these 4 variables had a significant direct relationship with purchase intention. The beta coefficients were very low and also varied between positive and negative. They were; brand awareness (.014), credibility (-.043), entertainment (-.032) and information (.016). The only significant relationship found with purchase intention was again brand attitude (.851), with p = 0.00. Thus, a change of one standard deviation in brand attitude will result in a change of .851 standard deviation in

(34)

purchase intention. This relationship was found to be very strong with a high correlation coefficient.

Thus, for Ralph Lauren, brand awareness and entertainment influence brand attitude and brand attitude influences purchase intention. Brand awareness and entertainment also have an indirect influence on purchase intention through brand attitude, but there is no direct relationship. For credibility and information, there is no significant direct or indirect relationship with either brand attitude or brand awareness.

Similar results were found when taking the control variables into account. Figure 4.6 displays the model concerning the path analysis with control variables.

Figure 4.6 Full model Ralph Lauren with control variables

These first four variables and their relationships with brand attitude were all positive, namely brand awareness (.359), credibility (.136), entertainment (.276) and information (.160). However, again not all were significant. Only brand awareness and entertainment being significant at the 5 per cent level. Again, the relationship between brand awareness and brand attitude was found to be stronger than the relationship between entertainment and brand awareness. Furthermore, some control variables showed positive relationships, while others showed negative relationships. None of these control variables were significant, with all of them p > 0.05. Also, only one relationship with purchase intention was statistically significant with p < 0.05. This was brand attitude (.864). Thus, a change of one standard deviation in brand attitude will result in a change of .864 standard deviation in purchase intention. All the others were not statistically significant, with p > 0.05. The exact p-values

(35)

can be found in the appendix.

Again, the reduced models were also tested. The output of this can also be found in Appendix F. With the aid of these data it is concluded that the reduced model fits the data as well as does the full model and the control variables were not found to be of significance for Ralph Lauren.

4.4.4 Luxury fashion brands

The hypotheses for all brands combined were tested without and with the control variables as well. The full model with the path coefficients without control variables is found in figure 4.7.

Figure 4.7 Full model all brands combined without control variables

The path coefficients were found for the relationships between brand awareness (.252), credibility (.179), entertainment (.312) and information (.238) and brand attitude. All these relationships were found to be statistically significant at the 5 per cent level. Again, these were all small relationships, but they were found to be moderate to strong with the aid of the correlation coefficients. The influence of entertainment was found to be the strongest one of the four on brand attitude, with the influence of brand awareness coming next.

The relationship found between brand attitude (.805) and purchase intention was also found to be positive and significant (p = 0.00). This relationship was very strong as well. Thus, a change of one standard deviation in brand attitude will result in a change of .805

Online advertising effectiveness : luxury fashion brands and their Instagram accounts