This post, again? : A study on the effect of post repetition on multiple social media platforms on attitudes and engagement : the moderating role of product type and product involvement

(1)

A study on the effect of post repetition on multiple social media platforms on attitudes and engagement: The moderating role of product type and product involvement

Author: Vivian Henny Engbertien (V.H.E.) ter Haar Student number: 11035536

Date: 20th of June

Version: Final

Qualification: MSc in Business Administration – Digital Business track Institution: University of Amsterdam / Amsterdam Business School Supervisor: F. Javier Sese

(2)

2

Statement of originality

This document is written by Vivian ter Haar 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)

3

Acknowledgments

First of all, I would like to thank my thesis supervisor Javier Sese. His approach and feedback were very structured and helpful. Javier allowed this paper to be my own work but steered me in the right direction whenever he thought I needed it.

I would also like to thank all the people who took their time to participate in my online experiment; without their fast responses and large numbers, my thesis would not be what it is today.

(4)

4

4.2 Stimulus material ... 36 4.3 Pre-test ... 37 4.4 Sample ... 38 4.5 Procedure ... 39 4.6 Measures ... 40 4.6.1 Dependent variables ... 40 4.6.2 Moderating variables ... 41 4.6.3 Control variables ... 42 5. Results ... 43 5.1 Randomization check ... 43 5.2 Manipulation check ... 43 5.3 Correlation analysis ... 45 5.4 Preliminary ANOVA’s ... 46 5.5 Main analysis ... 49 5.6 Overview results ... 52

(5)

5

6. Discussion ... 53

6.1 Theoretical implications ... 54

6.2 Managerial implications ... 56

6.3 Limitations ... 58

6.4 Recommendations for future research ... 59

7. Conclusion ... 60

References ... 62

Appendices ... 71

Appendix I – Pre-test stimuli ... 71

Appendix II – Pre-test questionnaire ... 73

Appendix III – Used material ... 77

Appendix IV – Questionnaire online experiment ... 79

(6)

6

Abstract

Firms are increasingly relying on social media as a marketing communication channel, with an estimated social media marketing spending globally near 36 billion dollars in 2017. However, a common practice of social media marketeers is to post roughly identical content on every owned social media platform. But is this the right way to leverage the platforms? The objective of this study is to examine the effect of identical versus similar post repetition on multiple social media platforms (Facebook, Twitter, and Instagram) on brand attitude and engagement. The author further investigates the moderating effects of product type (hedonic versus utilitarian) and high versus low product involvement. An online experiment among 195 participants divided into four conditions has been conducted. Via preliminary ANOVA’s and more robust regression analyses, the results indicate that there is a significant effect of post repetition on post attitude; if participants saw a similar post on multiple social media platforms, their attitude was significantly higher than the attitude of participants who saw identical posts on multiple platforms. There was no significant effect found in the proposed moderating relationship of product type and product involvement and the three-way

interaction between the variables. The study contributes and lent support to the existing (traditional) literature on message repetition and offers critical managerial insights regarding how to leverage social media for better customer outcomes.

(7)

7

1. Introduction

The average person will spend approximately a total of five years and four months over a lifetime on social media, with 40 minutes on YouTube, 35 minutes on Facebook and 15 minutes on Instagram (Business Insider, 2016; Social Media Today, 2017). Social Media Today (2017) states that “currently, total time spent on social media beats time spent eating and drinking, socializing and grooming”. No wonder why social media are still a top priority for managers in 2018. In several non-academic articles (Entrepreneur, 2017; Forbes, 2018) as well as academic articles (Barger & Labrecque, 2013; Peters, Chen, Kaplan, Ognibeni & Pauwels, 2013; De Vries, Gensler & Leeflang, 2017) authors make this very clear. Social media are an important part of managers’ marketing strategy to communicate and engage with all sorts of customers. The estimated social media marketing ad spending globally is projected to reach nearly 36 billion dollar in 2017 (eMarketer, 2015). The U.S. – the biggest spender – will spend 17,34 billion dollar on social media marketing in 2019 wherein 2014 it was “only” 7,52 billion (Statista, 2018). For managers, however, it is difficult to create engaging content that attracts customers to the brand (Frankwatching, 2017). A content marketing strategy is crucial for the success of social platforms. But the phenomenon is hyped and a lot of marketing managers do not know what the goal is. Top priorities nowadays for business to consumer content creators are creating more engaging content, better understanding of which content is effective – and what is not, finding more or better ways to repurpose content and optimization of content (Frankwatching, 2017). 44% of managers wants to find better methods to reuse content. A current practice of brands is, namely, to post roughly the same, identical content on each social channel they own (Barger & Labrecque, 2013). This means leveraging the message of one platform on another social media platform, identical (exactly or highly the same words, phrasing and photos/videos are used) or similar (the same words (or synonyms) and phrasing are used, but in a different order). This phenomenon is seen by

(8)

8 several companies on social media; big, small, Dutch or international. For example, phone giant Samsung. On Facebook they posted a message with “There is a new star in town. And it’s going to change the way you experience everything. Introducing the #GalaxyS9 & #GalaxyS9Plus.” On Instagram the brand posted a highly identical message, with the same photo. A big Dutch bank, ING, also posted identical messages on three different social media platforms: Twitter, Facebook, and Instagram. The same is true for big Dutch retailers like Albert Heijn, Dirk van den Broek and Jumbo. All these brands post identical or similar messages on two or three different social media platforms.

There is extensive literature about (traditional) message repetition (Goldberg, 1954; Wilson & Miller, 1968; Johnson & Watkins, 1971; McCullough & Ostrom, 1974; Cacioppo & Petty, 1980; Lim, Ri, Egan & Biocca, 2015), it is commonly known that there is an effect. Scholars have looked at, among others, traditional message repetition, online message repetition and identical versus similar message repetition. When looking at the identical versus similar literature, there is a disagreement on the outcomes in traditional media. Some scholars found similar message repetition to be more effective (Sears & Freedman, 1965; McCullough & Ostrom, 1974), others found effects for neither identical nor similar message repetition (Mitchell & Olson, 1981; Burnkrant & Unnava, 1987). Exposure to messages in traditional advertising is, however, entirely different from exposure to messages on social media since social media are a lot faster and more part of our lives than traditional advertising ever was or will be. Therefore, it is essential to take identical and similar messages into account as well on these new, online media. When it comes to firm-generated content (FGC), scholars have looked to different sorts of FGC and the effects on a variety of dependent variables. De Vries et al. (2012), for example, looked at the vividness and interactivity of FGC and Ashley and Tuten (2015) examined different message strategies. And so there are a lot more (Gensler, Völckner, Liu-Thompkins & Wiertz, 2013; Kumar, Bezawada, Rishika,

(9)

9 Janakiraman & Kannan, 2016; Schivinksi & Dabrowski, 2016). However, none of these scholars have looked at the repetition of FGC.

There has been no research on what the effects are of posting identical versus similar messages on multiple social media channels. Do customers (or followers) like being exposed to identical messages on multiple platforms? Or does it annoy them? Reinforces similar repetition the message or will people stop following the brand because of it? Will people see the brand as intrusive or as “the brand which is only repeating himself”? Since companies are investing a lot of money in social media, this is an urgent matter. In this study, a contribution to this stream of literature is made by examining the effect of posting identical versus similar messages on multiple social media platforms. The author examines – by conducting an experiment among 195 participants – the effect on brand attitude and the willingness to engage with the post, with attitude towards the post as mediator and with possible moderation of product involvement and product type. Why these (in)dependent variables have been chosen will be explained in the next paragraphs. The following research question follows logically from the preceding:

RQ What is the effect of brands posting identical versus similar messages on multiple social media platforms on brand attitude and willingness to engage with the post, and do product involvement and product type moderate these effects?

This research question is both academically as well managerially relevant. Since there has been no research on leveraging identical versus similar posts on social media, the lack of knowledge will be reduced. For managers, the results of this study give them insights into how they should arrange their social media (content) strategies. Should they continue posting identical messages on multiple platforms? Or should they stop it immediately and leverage

(10)

10 only similar messages? Or vice versa? Since a vast amount of money is involved in this

matter, it is essential to know the answers to these questions.

This paper is structured as follows. In the next paragraphs, existing literature on the different concepts present in the research question will be discussed. Secondly, a conceptual framework is given that explains the effects and connections between the concepts. This conceptual framework is followed by the method section describing the research design. In the third part, results will be presented. The paper ends with a discussion and conclusion.

2. Literature review

In this section, I give an overview of the existing literature on three relevant topics: firm-generated content, customer engagement, and message repetition.

2.1 Firm-generated content

Firm-generated content (FGC) is a well-known topic in recent literature. Several scholars have addressed this topic in recent years, and they have given a variety of names to firm-generated content, like ‘firm-firm-generated content’ (Kumar, Bezawada, Rishika, Janakiraman & Kannan, 2016), ‘firm-created social media communication’ (Schivinksi & Dabrowski, 2016), ‘firm-generated brand stories’ (Gensler, Völckner, Liu-Thompkins & Wiertz, 2013) ‘firm-to-consumer social messages’ (De Vries, Gensler & Leeflang, 2017), ‘(creative) branded social content’ (Ashley & Tuten, 2015) and ‘marketer generated content’ (Goh, Heng & Lin, 2013). In these studies, different definitions – but with quite the same meaning – are given. For example, the definition of FGC made by Schivinski and Dabrowksi (2016, pp. 191): “firm-created social media communication is understood as a form of advertising fully controlled by the company and guided by a marketing strategy agenda.” Goh et al. (2013) define ‘marketer generated content’ as content that is made by marketers on behalf of their firm to engage consumers actively. FGC is content the brand made itself. You can see FGC on social

(11)

11 media as so-called owned media (Edelman & Salsberg, 2010). Owned media are media the company uses or creates by itself to, for example, advertise on. Think about corporate

websites and social media pages (Edelman & Salsberg, 2010). In this study, FGC is defined as ‘a post, advertisement or another type of communication the brand made and posted itself on one or more social media platforms.’ Social media platforms are, for example, Facebook, Instagram, Twitter, and Pinterest.

It is crucial for a firm to understand – for social customer relationship management (CRM) to be effective – how customers respond to FGC (Kumar et al., 2016). The explicit messages from the study of Kumar et al. (2016) are that social media marketing matters. To communicate and build and maintain a relationship with customers, managers should embrace it. Their results state that “investing in developing a social media community with a dedicated fan base can significantly strengthen customer–firm relationships and can lead to a definitive impact on the firm’s revenues and profits.” (Kumar et al., 2016, pp. 15). The same is stated by De Vries et al. (2012, pp. 1): “social media outlets constitute excellent vehicles for

fostering relationships with customers.” De Vries et al. (2012) examined the effectiveness of, among others, firm-to-consumer (F2C) social messages for brand building and customer acquisition. They looked at brand post popularity. Vividness and interactivity of brand posts appeared to be important for customers to comment and like firm-generated content (De Vries et al., 2012). These authors appealed to other scholars and encouraged them to gain more knowledge about the underlying processes that explain the effectiveness of FGC.

Just like the articles mentioned above, Ashley and Tuten (2015) looked at FGC as well. They call it ‘branded social media messages’ and studied different message strategies. Which one is effective? They looked at all kind of message strategies like interactivity, exclusivity, and forms of appeal and what their effect is on, among others, engagement. They also looked at integrated content, this means whether a brand leverages the content of existing

(12)

12 campaigns or makes unique content for social media (Ashley & Tuten, 2015). They find in their study that 10 of the 28 brands leveraged their existing media campaigns on social media platforms. This means that a person who follows the brand on multiple platforms will be exposed with a high likelihood to the same message multiple times. Unfortunately, they did not study the effect of this.

De Vries et al. (2012) already addressed it in their study: studies on FGC provide dispersed insights into the relative effectiveness of FGC on behavioral outcomes. It is a new field of research where important gains can still be made. I have analyzed several studies on FGC and the effect of this content on different dependent variables. However, like in the literature on message repetition, scholars have never looked at FGC repetition on multiple social media platforms.

In the next paragraph, I will elaborate on the behavioral outcomes ´customer

engagement´ and ´brand attitude´ that can be influenced by FGC. In the end, companies invest in FGC to create a better bond with customers, to make them feel right about the brand and to engage with them.

2.2 Customer engagement and brand attitude

Customer engagement has been discussed extensively in marketing in the last decade by both academicians and practitioners. It is a hot topic and nowadays the primary focus of many firms. Different studies have highlighted the benefits of engaging customers in the

marketplace. In consumer electronics industry, for example, fully-engaged shoppers make 44% more visits per year than actively disengaged shoppers do. The engaged shopper spends $373 per shopping trip, while disengaged consumers spend “only” $289. The same

phenomenon is seen in the restaurant, hospitality, insurance, and retail banking sector (Pansari & Kumar, 2017).

(13)

13 Customer management has evolved over the years (Pansari & Kumar, 2017).

Marketing was until the 1990s focused on customer transactions where it was about share-of-wallet, recency, frequency, and monetary value. In the late 1990s and early 2000s, customer management was about establishing a positive relationship between customers and brand, where trust and commitment were important. But, as Pansari and Kumar (2017, pp. 294) state: “it is not enough to simply satisfy the customer to make him/her loyal and profitable.

Profitable loyalty and satisfaction need to be evolved to a higher level.” Therefore, customer management evolved further from establishing positive relationships to engaging customers in all possible ways. This led to the rise of the term “customer engagement” (CE) (Pansari & Kumar, 2017). Practitioners define it, from the perspective of the organization, as “activities facilitating repeated interactions that strengthen the emotional, psychological or physical investment a customer has in a brand” (Sedley, 2010, pp. 7). Academics define CE in several ways as well: “the mechanics of a customer’s value addition to the firm, either through direct or/and indirect contribution” (Pansari & Kumar, 2017, pp. 295; Kumar, Aksoy, Donkers, Venkatesan, Wiesel & Tillmanns, 2010). Vivek, Beatty and Morgan (2012) have discussed CE from the perspective of customer attitudes toward the brand; they define CE as “the intensity of an individual’s participation in and connection with an organization’s offerings or organizational activities, which either the customer or the organization initiates” (p. 127). Van Doorn, Lemon, Mittal, Nass, Pick, Pirner and Verhoef (2010) define CE as a

“customer’s behavioral manifestation towards a brand or firm, beyond purchases, resulting from motivational drivers” (p. 253).

Pansari and Kumar developed a framework that focusses on how customer engagement can be gained. They linked the contributions of CE on firm performance outcomes. The tangible benefits of customer engagement can be seen in the form of firm performance, such as higher profits, revenue or market share. Besides that, a tangible indirect

(14)

14 benefit of CE is customer discussions about the brand on social media that can affect a wide group of protentional customers (Hogan, Lemon & Libai, 2003; Pansari & Kumar, 2017). This feedback has an indirect impact on the firm performance. An intangible benefit of CE is that a highly engaged customer is willing to provide the firm with more information about themselves since the engaged customer has heightened trust. The firm can use this

information to “better understand its customers and engage with them accordingly” (Pansari & Kumar, 2017, pp. 306). Vivek et al. (2012) made a theoretical model of CE as well. The authors explored the nature and scope of CE which they say is a vital component of

relationship marketing. They offered a model where value, trust, affective commitment, word of mouth, loyalty, and brand community involvement are potential consequences of CE.

Studies mentioned above show that it is essential to know about CE and to work on it actively since CE affects firm performance. As Pansari and Kumar (2017, pp. 298) say: “it is important to know how customers can be engaged for maximizing firm performance.”

One way to engage customers is, at this moment a very hot topic as well, via social media. Scholars often study customer engagement as a dependent variable when examining the effect of FGC on social media. Customer engagement is, as described, not a new concept in de field, but can be interpreted differently. Schultz and Peltier (2013) show in their study that the most appropriate definition of engagement is subject to debate. In the social media area, engagement is often seen as a consumer taking some action beyond viewing or reading (Paine, 2011), like liking a post, commenting on a post or sharing a post with friends. It can also be an indicator of the overall level of consumer interest in a brand’s post (Barger & Labrecque, 2012). In this study, engagement is seen as how likely a consumer is to like, share or comment on a brand’s post. After seeing (a) specific post(s), is the consumer more willing to engage or less willing? Although scholars looked at how various social media content habits affect customer engagement (Van Doorn et al., 2010; Ashley & Tuten, 2015; Barger,

(15)

15 Peltier & Schultz, 2016), they never looked at how message repetition on social media affects customer engagement.

The same applies to brand attitude. Brand attitude is a well-known and highly used concept when looking at different fields of research. In social media marketing and message repetition fields as well; however, scholars never combined these two. Fishbein and Ajzen (1975, pp. 222) define a person’s attitude as “a function of his salient beliefs at a given point in time." Olson and Mitchell (1981) describe brand attitude as ‘consumer’s overall evaluation of a brand.’ Schivinski and Dabrowski (2016) describe the term as ‘consumer perceptions of brands’. In their study, Schivinski and Dabrowski (2016) looked at FGC and the influence on brand attitude and hypothesized that FGC would have a positive influence on brand attitude; their expectations proved to be true. McCullough and Ostrom (1974) looked at traditional communications and the effect of these messages on attitudes. The overall conclusion of their study was that when similar (rather than identical) messages were employed, repetition did have an immediate positive effect on attitudes (McCullough & Ostrom, 1974). Both studies mentioned above looked at FGC or message repetition and the effect on brand attitude, and they both found a significant effect. However, scholars failed – again – to look at the combination of these two: FGC and repetition and the effect on brand attitude.

To encourage the research on this topic, Barger and Labrecque (2013) state the

following in their study as an important implication for future research: “How should a brand deal with users who follow the brand on multiple social channels? A current practice of many brands is to post roughly the same content on each social channel.” (Barger & Labrecque, 2013, pp. 21). They ask the question if repetition of identical messages reinforces the message or simply annoys the consumer who has been exposed to the same messages across multiple platforms. In the next paragraph, I will elaborate further on the topic message repetition.

(16)

16 2.3 Message repetition

Message repetition is a common phenomenon in both psychology and – traditional and online – advertising literature. “Messages are more effective when repeated” seems like common knowledge since extensive literature has studied the effects of message repetition. However, when diving deeper into the message repetition literature, contradictory results show up. In this paragraph, I give an overview of the existing literature on message repetition, structured in three different topics: repetition in a traditional environment, repetition in an online environment and identical versus similar message repetition.

2.3.1 Repetition in traditional environment

One of the oldest references goes back to 1890, where Thomas Smith published his book ‘Successful advertising’. In this book, he stated that it took twenty times for a person to hear a message before buying. In contradiction to what Smith wrote, Krugman (1965) stated that three times is adequate. In 1969, Grass and Wallace examined the effects of repeated exposures to TV commercials on participant’s interests in and attention to the commercials. Results show that as the number of exposure increased, participant’s interests and attention first increased to a satiation point and then declined to a stable level. The two-factor theory of Berlyne (1970) is the leading theory for the effects of repeated exposures to messages. His analysis states that message repetition is nonmonotonically related, like an inverted U, to communication effectiveness: if you increase the number of exposures to a message from low to moderate it is expected to enhance message effectiveness, whereas further exposures are expected to cause communication effectiveness to decline (Berlyne, 1970; Cacioppo & Petty, 1979; Anand & Sternthal, 1990). Cacioppo and Petty (1979) elaborated further on the two-factor theory. Their results showed that as message repetition increased, participant’s agreement with the message first increased and then decreased. Because of this, in the first stage, a favorable attitude toward the message appeared. In the second stage, a less favorable

(17)

17 attitude appeared. The same results were found in the study of Batra and Ray (1986). These scholars found that participants raised favorable attitudes toward the advertised brands rapidly during the first two commercial exposures, but their attitude dropped from the fourth

exposure. In psychology, the relationship between ad exposure and ad effectiveness is often explained by the mere-exposure effect (Zajonc, 1968; Bornstein & D’Agostino, 1992). This means that in terms of communication, repeating a message leads to familiarity, and people show preference for things they are familiar with. Message repetition, therefore, might be effective, since someone developed a preference for it and is, therefore, more likely to take action.

In contradiction to studies and findings mentioned above, in the years ’50-’70, Goldberg (1954), Wilson and Miller (1968) and Johnson and Watkins (1971) found that repeated presentation of a persuasive message does not produce more immediate attitude change than does a single presentation. The same applies to the outcomes of the study of Belch (1982). He found in his lab experiment examining the impact of TV commercial repetition on judgments that “neither attitudes nor intentions were affected by the level of advertising exposure.” (pp. 63). The same results apply to the study of Mitchell and Olson (1977). They found that there was no effect on belief strength, attitude, and purchase intention when two types of print ads were repeated.

2.3.2 Repetition in online environment

Chatterjee (2005) studied the effects of repeated ad exposures on the internet. The author investigated the effect of banner ad repetition level and same and varying banner ad executions on behavioral response (click rate) and memory-based outcomes (recall and recognition). The participants were asked to search for information on a website and memorize the information. In the meantime, ads were repeated on different pages of the website either four or fifteen times. Results show that there was no significant effect of banner

(18)

18 ad executions or repetition level on behavioral response and recall. Recognition was found to be higher under high repetition and with the same ads (Chatterjee, 2005). In line with the study of Chatterjee (2005), Manchanda, Dubé, Goh, and Chintagunta (2006) looked at

whether banner advertising affects purchasing patterns on the internet. They expected that the more the participant was exposed to an ad, the higher the probability that the participant would purchase. As they state: “prior research has shown that viewing a series of

advertisements leads to higher recall and more positive attitudes” (Pechmann & Stewart, 1988, pp. 103). This is why Manchanda et al. (2006) expected that “the probability of purchase is higher for consumers who are exposed to advertising on many different websites and pages” (pp. 103). In contradiction to Chatterjee, Manchanda et al. found that the number of exposures has a positive effect on repeat purchase probabilities. The same was found by Fang, Singh, and Ahluwalia (2007). These authors investigated within the online environment as well. They examined whether the mere-exposure effect is able to account for the persuasion effects of banner advertising. Results showed that “there was a significant main effect of exposure frequency in the positive evaluation condition” (pp. 99). A significant linear trend in positive evaluations was discovered as exposure frequency increased (Fang et al., 2007). A more recent study that aligns with studies mentioned above is from Lim, Ri, Egan, and Biocca (2015). These scholars examined the effect of digital video advertising through mobile TV, internet, and television on message, ad and brand credibility. Their study showed that participants who were exposed to repetitive ads on paired media of internet, television, and mobile tv have greater perceived advertising effectiveness than counterparts exposed to repetitive ads from only one medium. The multiple-media repetition also generated more positive attitude toward the brand than the single-medium repetition (Lim et al., 2015). The most recent study that examines the impact of advertising frequency on consumer attitudinal response is of Hussain, Ferdous, and Mort (2018). These scholars examined whether

(19)

19 advertising type and appeal moderate the relationship between web banner advertising

frequency and consumer attitudes. Based on a lab experiment involving 400 participants, they found that advertisement type, static or dynamic (pop-up), acted as a moderator between frequency and brand attitude. They recommended that if the objective is to create brand attitude through static banner advertisements, advertisers should focus on advertising frequency.

2.3.3 Repetition of identical versus similar messages

Often the distinction between similar and identical messaging has been made within the literature of message repetition. An identical post is a post that is exactly or highly the same, with identical words and phrasing and identical visual elements. Similar posts are posts that are about the same topic but are not identical. They differ in word and phrase order and/or use synonyms (McCullough & Ostrom, 1974). As mentioned before, Goldberg (1954), Wilson and Miller (1968), and Johnson and Watkins (1971) did not find an effect of repeated messages on attitude change. However, as McCullough and Ostrom (1974) state, “an

important similarity among these failures to obtain a repetition effect is that they all repeated identical communications, and consequently, the subjects did not receive new information or arguments as a function of repetition” (pp. 395). Sears and Freedman (1965) for example used similar rather than identical messages in their experiment. They added slightly different information in their communication. They found that a positive repetition effect can be obtained when similar (non-identical) messages are used (Sears & Freedman, 1965).

McCullough and Ostrom (1974) build further on this topic, and they showed respondents five similar ads that used a similar basic appeal but differed in phrasing and the order of the message arguments. Results show that when similar rather than identical messages were used, message repetition did have an instant positive effect on post-exposure attitudes (McCullough & Ostrom, 1974). The same results were found in the studies of Gorn and Goldberg (1980),

(20)

20 Grass and Wallace (1969), and Schumann, Petty and Clemons (1990). In the latter study, the scholars looked at cosmetic and substantive variation conditions; they found that cosmetic as well as substantive variation in repeated ads have a great impact on overall attitudes. Studies mentioned above support the use of similar ads (or variated ads) rather than identical ads to avoid tedium or decrease in effectiveness. However, there are some studies that report less positive results. Burnkrant and Unnava (1987) used the “encoding variability hypothesis” to explain why variation in an ad would be more effective than expose participants to a single ad. They measured brand recall and attitude. Recall was better under varied conditions, however, there was no significant difference with respect to attitude. The study of Mitchell and Olson (1981) failed as well to find any significant influence with regard to variation of repeated print ads.

As becomes clear from the literature review of message repetition above, there has been done extensive research on message repetition; some scholars looked at message repetition in traditional media and the effect on attitude, purchase intention or evaluation. Others made the distinction between similar versus identical messages more prominent. For again others, message repetition was studied in the online environment examining online (banner) advertisements. This may be seen as there is nothing left in this research field, but that is certainly not the case. None of these scholars mentioned above have looked at social media and the influence of message repetition on these platforms. This is a big gap in the literature of message repetition.

2.4 Summary

In sum, I can conclude that there is extensive literature about repetition of messages and firm-generated content on all sorts of dependent variables. Scholars have looked to types of FGC (posts) on brand attitude and customer engagement but never looked at repetition of identical versus similar posts on multiple social media platforms. Besides that, despite the extensive

(21)

21 literature on message repetition, no research has been conducted considering message

repetition on social media. In this study, I extend and contribute to this stream of literature by examining the effect of posting identical versus similar messages on multiple social media platforms. In the next part, a conceptual framework is given that will explain the underlying theories and where hypotheses follow logically.

3. Theoretical development

In the first paragraph of this section, the conceptual model of this study will be presented and explained. Secondly, the hypotheses of the study will be introduced. 3.1 Conceptual framework

Figure 1: Conceptual model

The proposed framework, as shown in figure 1, is organized in the following manner. In this framework, the goal is to understand the connection between post repetition on multiple social media platforms on several outcome variables. I propose that post repetition on multiple platforms has an impact on the attitude towards the post. This ‘post attitude’ is a mediating mechanism that is connected to brand attitude and engagement. I also propose that product involvement and product type have a moderating role in the main effect. Lastly, I expect a three-way interaction effect between post repetition, product type and product involvement.

(22)

22 Post repetition can refer to many things: showing people ten times the same

commercial on television, exposing people to three similar ads in a magazine or showing people the same banner ad repeatedly on different websites. In this study, post repetition refers to exposing people to a repeated post on multiple social media platforms. A post is a message placed on social media by a brand (firm-generated content), for example, a tweet, a timeline-story on Facebook or a post on an Instagram-feed. Wang and Nelson (2006) looked at similar versus identical messages on multiple platforms as well. They called the distinction VMSMs: varied multiple-source messages and IMSMs, identical multiple-source messages. They defined (pp. 110) VMSMs as “different or varied brand or product messages originated from multiple sources” and IMSMs as “similar or identical brand or product messages presented by different sources”. In this study, post repetition can occur in two types as well: in an identically and similarly manner. An identical post is a post that is exactly or highly the same, with identical words and phrasing and identical visual elements. Similar posts are posts which point out the same topic or product but use different phrasing and word order and make possible use of synonyms (McCullough and Ostrom, 1974). In the research design section, I will elaborate on this.

This research is also considering the moderating role of product type. Product type refers to the distinction in utilitarian versus hedonic products. Consumers see utilitarian products more as useful rather than fun. Utilitarian products are more instrumental, goal-oriented, and accomplish functional or practical tasks (Batra and Ahtola, 1990; Mano & Oliver, 1993). Examples are laundry detergent (for practical cleaning), toilet paper,

toothpaste, and bread since people consume these products for practical or primary purposes. Hedonic products are products for which consumption is associated with fun, entertainment, pleasure, and/or affective experiences (Bart et al., 2014; Schulze et al., 2014), like movie tickets, designer clothes, high-end gadgets/wearables or smartphones and cosmetics. Voss et

(23)

23 al. (2003) say: products that are perceived as necessary and helpful reflect mainly utilitarian value whereas products that trigger enjoyment and excitement are considered by consumers mostly as hedonic. In the traditional advertisement field as well as in the “new” internet and social media field, the distinction is often made and investigated and is seen as an important construct in understanding marketing effectiveness (Dhar & Wertenbroch, 2000; Okada, 2005; Bart et al., 2014; Schulze et al., 2014). As Petty and Cacioppo (1986) presented in their Elaboration Likelihood Model (ELM), people pay less attention to utilitarian products and more attention to hedonic (fun and entertainment-oriented) products. Furthermore, the literature suggests that reactions are different for utilitarian and hedonic items (O’Curry and Strahilevitz, 2001; Schulze et al., 2014). The latter scholars, for example, studied whether the same approach for utilitarian and less utilitarian (hedonic) products is useful. They looked at the app FarmVille (fun and entertaining, a hedonic product) and utilitarian products. The scholars looked at viral marketing reactions within their studies. Does this differ for the type of product? The findings of the study of Schulze et al. (2014) show that “the same sharing mechanism that made FarmVille so successful is the worst possible mechanism for promoting primarily utilitarian products.” This implies that marketing managers should know about and apply different strategies for different product types. However, scholars have never taken into consideration the distinction between the product types in combination with social media and message repetition. This is why I choose this variable to consider in the conceptual

framework. When people are being exposed to messages several times on different social media platforms, are the effects different for hedonic versus utilitarian goods?

The second moderating variable considering in this research is product involvement: do people mind the identical posts on multiple platforms less when they are more involved with the product? It is expected that different amounts of involvement (high versus low) have an impact on how someone perceives the post repetition on multiple social media platforms.

(24)

24 The construct has been defined differently by several scholars. Cushing and Douglas-Tate (1985) define it as ‘how the product fits into that person’s life’. Greenwald and Leavitt (1984) as well as Petty and Cacioppo (1984) state that high involvement products are related to personal relevance or importance. In the highly cited study of Zaichakowsky (1985, pp. 342), the author defines product involvement as “a person’s perceived relevance of the object based on inherent needs, values and interests”. As can be concluded logically from the foregoing definitions, product involvement is subject to one’s personal experiences and life. Although the aforementioned is true, there has been found a pattern in what low and high involvement products are. Low involvement products can be seen as products which are bought more often, as a habit or as a routinized response behavior and require limited

consideration (Kumar et al., 1992). They are repeatedly purchased, for example, toilet paper, salt, and soap. High involvement products are products that are infrequently purchased, for example, cars and homes (Pansari & Kumar, 2010). Since people invest more time and resources in understanding details about high involvement products, he or she is more likely to notice that things went right or wrong (Anderson, 1994).

Product involvement is an important moderating variable to consider in the model since in previous research product involvement proved to be an important moderator in both message repetition literature as well in customer engagement literature (Pansari & Kumar, 2010). Besides that, according to Kushwaha and Shankar (2013) and Bian and Moutinho (2009), the classification of high versus low involvement is a fundamental base and a very important framework to understand consumer decision-making behavior. As Bian and Moutinho (pp. 396) state: “product involvement is a central framework, vital to

understanding consumer decision-making behavior as well as associated communications”. Delgado-Ballester and Muneara-Aleman (2001) show the relationship between customer engagement and customer involvement. “Involvement indicates the customer’s level of

(25)

25 motivation to seek information that may be used to manage and moderate any potential risk inherent in the decision-making process” (Pansari & Kumar, 2010, pp. 296). This would occur before the customer decides whether to like, share or comment a post (engage with the post) when looking at this study.

As mentioned, ad attitude is a mediating variable that connects brand attitude and engagement with the post (MacKenzie, Lutz & Belch, 1986). Since in this study

firm-generated content is examined, ‘ad’ attitude is called ‘post’ attitude. Post attitude is defined as an individual's internal evaluation of an object (Mitchell & Olson, 1981), such as a social media post. As described in the literature review section, attitude towards the ad has been an important outcome variable in the literature of message repetition (Pehlivan, Sarican & Berthon, 2011; De Vries et al. 2017). Although attitude can be seen as old-fashioned and not original these days, it is still important in this matter, since message repetition on multiple social media platforms have never been examined on any outcome variables. This study wants to examine if a person finds a post annoying, attractive or whatsoever; therefore, post attitude is the best suitable construct.

Brand attitude refers to the consumer’s overall evaluation of a brand (Olson &

Mitchell, 1981). As described in the introduction, social media are important for managers; a vast amount of money is invested in social media strategies. It is, therefore, crucial to know what effects social media habits have on brand attitude; it can either make or break the brand. Several studies have looked at, for example, the impact of firm-generated content on social media on different customer behavior outcomes (Pehlivan et al., 2011; Schivinski & Dabrowski, 2016). However, brand attitude has never been measured in combination with message repetition on multiple social media platforms.

(26)

26 To conclude, an explanation of the construct ‘engagement’ that can be found on the right-hand side of the framework. Engagement was a hot topic the past decade and still is, as well in traditional literature and online literature. Engagement is in the literature being recognized as an important performance measure of social media (Van Doorn et al., 2010; Ashley & Tuten, 2015; Barger et al., 2016) and these days it is still one of the main concerns of marketing managers. Therefore, in this study, engagement is one of the outcome variables since it is of high importance and no study ever conducted research on message repetition on multiple platforms and how customers would perceive that. The construct is conceptualized as ‘willingness to engage with the post’. How likely is a consumer to like, share or comment on a brand’s post? After seeing (a) specific post(s), is the consumer more willing to engage or less willing?

(27)

27 Table 1: Overview of variables used in the conceptual framework

Variables used in the study Definition Suggested scale

Post repetition A social media post repeated multiple times on different social media platforms (Facebook, Twitter, and Instagram)

Identical posts. Posts that are exactly or highly the same, with

identical words and phrasing and identical visual elements.

Similar posts. Posts that are about the same topic, but are not

identical. They differ in words and phrasing and vary in (order of) arguments.

-

Post attitude An individual's internal evaluation of an object, such as a social media post (Mitchell & Olson, 1981).

7-point Likert scale: ‘not attractive/attractive’, ‘irritating/not-irritating’, ‘not interesting/interesting’, ‘bad/good’, ‘pleasant/unpleasant’, ‘likable/unlikable’. (Mackenzie et al., (1986); Chattopadhyay & Nedungadi, 1992).

Brand attitude The consumer’s overall evaluation of a brand (Mitchell & Olson, 1981).

7-point Likert scale: ‘Do you like brand X? ‘Do you think brand X is bad?’ ‘Do you think brand X is good?’ ‘Do you think brand X is annoying?’ ‘Do you think brand X is boring?’

Customer engagement Engagement of the customer with the social media posts. How likely is a consumer to like, share or comment on a brand’s post.

7-point Likert scale: ‘How likely are you to like this post?’, ‘How likely are you to comment on this post?’, ‘How likely are you to share this post?’ (Barger & Labrecque, 2013).

Product type Utilitarian products. Useful, rather than fun. Instrumental,

goal-oriented and accomplishes functional or practical tasks (Batra & Ahtola, 1990; Mano & Oliver, 1993). Hedonic products. Products for which consumption is associated with fun, entertainment, and/or pleasure. Trigger enjoyment and excitement (Bart et al., 2014; Schulze et al., 2014).

-

Product involvement A person’s perceived relevance of the object based on inherent needs, values and interests (Zaichakowsky, 1985).

7-point Likert scale: ‘unimportant/important’, ‘irrelevant/relevant’, ‘means nothing/means a lot’, ‘boring/interesting’, ‘unexciting/exciting’, ‘unappealing/appealing’, ‘not needed/needed’ and ‘uninvolving/involving’ (Zaichkowsky, 1985).

(28)

28 3.2 Hypotheses

In this section, I will elaborate on (the directions of) the relationships between the different constructs proposed in the conceptual model.

3.2.1 Effect of post repetition on multiple platforms on post attitude

In the literature, I elaborated on the two-factor theory of Berlyne (1970). This theory was and still is the leading approach when it comes to advertising message repetition (Belch, 1980). Berlyne proposed the inverted U-curve: if you increase the number of exposures to a message from low to moderate it is expected to enhance message effectiveness, whereas further

exposures are expected to cause communication effectiveness to decline (Berlyne, 1970; Cacioppo & Petty, 1979; Anand & Sternthal, 1990). Although various scholars found that message repetition does not have an (positive) effect at all, according to the literature review (Mitchell and Olson, 1977; Belch, 1982; Chatterjee, 2005) the most scholars found the opposite (Cacioppo and Petty, 1979; Batra and Ray, 1986; Manchanda et al., 2006). When looking at Lim et al. (2015), these scholars even found that multiple-media repetition, where this study is about as well, generated more positive attitude toward the brand than a single-medium repetition did. However, another distinction must be taken into consideration as well: identical versus similar messaging. When looking at similar versus identical ads, Sawyer (1981, pp. 257) states “it is well established that repetition of similar (non-identical) ads is more effective than repetition of identical ads in terms of both recall and persuasion” (Schumann et al., 1990). When looking at similar versus identical messaging in the literature as identified in the literature review, this saying of Sawyer seems to be accurate. More scholars found that identical ads are less effective (Sears & Freedman, 1965; McCullough & Ostrom, 1974; Gorn & Goldberg, 1980) than the opposite (Burnkrant & Unnava, 1987; Mitchell & Olson, 1981). For example, Wang and Nelson (2006) found that VMSMs, varied multiple-source messages, had a more significant impact on message effectivity than IMSMs,

(29)

29 identical multiple-source messages had. They discovered that VSMSs increased consumers’ purchase intention as well. When combining aforementioned theories and outcomes, I propose the following:

H1 Post repetition on multiple social media platforms will have a positive effect on post attitude, but only if the posts are similar rather than identical.

3.2.2 Impact of moderating variables

As I explained in section 3.1, I expect a moderating role of both product type and product involvement. These are two possible interpretations of a difference between two product categories. In some research (Voss, Spangenberg & Grohmann, 2003), scholars combine the two categories and see them as interacting variables. First, I will look at the individual effects of the moderators. In the last part of this section, I will elaborate on a possible three-way interaction effect.

Impact of product type

Product type is important to examine since sharing mechanism characteristics that are being used for hedonic products might be ineffective for utilitarian products’ marketing (Schulze et al., 2014). Schulze et al. (2014, pp. 2) state in their article, based on schema theory of

Aronson, Wilson and Akert (2012) and Bartlett (1932) that in a fun- and entertainment-oriented environment (in the current study Facebook, Instagram, and Twitter) for primarily utilitarian products consumers will pay little attention to (viral) marketing messages. They state that, in line with social psychology theories, consumers do not visit Facebook – or other social media platforms – to learn about utilitarian products. Facebook or other social media are used for fun and entertainment, rather than for functional or useful tasks. When

considering the elaboration likelihood model (ELM) of Petty and Cacioppo (1986) the

(30)

30 on Facebook (and so expected on other social media) do not correspond with the customers’ expectations or schema (Bartlett, 1932; Schulze et al., 2014) and, as logically can be

concluded, therefore be perceived as unexpected and less wanted or suited than hedonic product messages. In another study, Mano and Oliver (1993) proposed that products which are evaluated higher as hedonic will lead to higher product-elicited arousal. Utilitarian

evaluation, as they propose, will be uncorrelated with arousal. Their results show that hedonic evaluation was closer to the consumers’ affective experience for both pleasantness and

positive affect scales. Another study that is in line with the studies mentioned above, and sees hedonic products as perceived more positive by consumers, is the study of Chitturi,

Raghunathan, and Mahajan (2008). They looked at the relationship between hedonic and utilitarian product benefits and customer delight and satisfaction. They found that products that meet or exceed customers’ hedonic wants enhance, among others, customer delight. Customer delight has, in turn, a positive effect on customer loyalty (Chitturi et al., 2008) and therefore on brand attitude.

As mentioned before, hedonic goods are about sensory experience, pleasure, fantasy, and fun. Utilitarian goods are goal-oriented, functional, and practical (Hirschman &

Holbrook, 1982). When looking at identical versus similar messages, it was stated that a similar message varies; this message differs in words and arguments and brings more

“experience”. Identical messages do not vary and are static, they do not add new experiences. When looking at the literature, it can be stated that for hedonic products often deviant

messages are better for increasing attitudes. For example, Rossiter, Percy, and Donovan (1991) and Youn (1998) suggested that emotional advertising is important for hedonic products; for utilitarian goods, a non-emotional approach seems to be better. Spotts, Weinberger, and Parsons (1997) showed that humorous appeals for hedonic products are effective. For utilitarian products, a more rational and message-oriented appeal is used,

(31)

31 without any fuss (Geuens et al., 2010). Therefore, it is expected that for hedonic goods,

similar messages that stand out and differ from each other will increase post attitude. Besides this, another reason for the expected interaction effect between similar post repetition and hedonic products is based on Dhar and Wertenbroch (2000). These scholars showed that loss aversion is greater for hedonic than for utilitarian goods. To avoid this uncertainty of loss, people prefer to obtain more information about a hedonic product (Vishwanath, 2003). Similar messages can supply this need better than identical messages can.

When looking at studies mentioned above, I expect that posting similar messages will be more effective for hedonic products than utilitarian products. Therefore, I propose the following:

H2 Similar post repetition on multiple social media platforms will have a positive effect on post attitude, but this effect is stronger when the product is hedonic rather than utilitarian.

Impact of product involvement

As elaborated on in paragraph 3.1, product involvement is an important construct in examining consumer behavior. The ELM is a well-known and the leading theory when it comes to product involvement. This model states that product involvement has a moderating effect on information processing (Petty & Cacioppo, 1984). The model of Petty and Cacioppo suggests that, depending on the customer’s product involvement, the same information can be processed in different ways, via the central route and peripheral route. The central route is “followed” when consumers have high motivation, high ability and high opportunity to process information, also called high-elaboration likelihood. People process information via

(32)

32 the peripheral route when the effort and involvement are low (Petty & Cacioppo, 1986;

Chung & Zhao, 2003).

The concept product involvement can be distinguished in high involvement and low involvement. Since high involvement products or services are of higher importance to the consumer, they require more information (Zaichkowsky, 1987). The more someone is involved with a product, the more he or she is motivated to process the product information. Or, as other scholars state, when consumers are higher involved with a product, they are more motivated to dedicate cognitive effort to evaluating the product (Bian & Moutinho, 2015). Besides that, involvement leads to heightened arousal and increased cognitive elaborations as well (Mano & Oliver, 1993). In high involvement situations, consumers look for more

personal, experimental and symbolic gain than they do in low involvement situations (Solomon, Surprenant, Czepiel & Gutman, 1985). Lower involvement products require minimal thought since low involvement products are more likely to become a buying habit. Chung and Zhao say the following: when people are exposed to a low involvement product, they are less able to make a great deal of effort to process information (Chung and Zhao, 2003).

Consumers who are high involved with a product are more likely to research the product extensively. As Suh and Yi (2006) name in their article: high involved people have gathered extensive knowledge about the product, so they are also more likely to provide the company feedback. In this study, feedback can be seen as a like or a comment a consumer gives a brand’s social media post. Besides that, since high involvement products are lifestyle products with higher status, consumers are more likely to share their experiences with their network (Suh and Youjae, 2006; Pansari & Kumar, 2017). In this study, ‘sharing experiences’ can be seen as a ‘share’ on social media.

(33)

33 It can be stated that when product involvement is high, consumers are more focused and into the product or message then when product involvement is low. In high involvement situations, consumers want more information, which they process deliberately. When looking at identical posts on multiple platforms, where customers do not get any new information, this would be a bad experience for high involved people. Similar messages, however, bring for example new arguments, that high involved people would prefer. When looking and combining aforementioned literature, I propose the following effect:

H3 Similar post repetition on multiple social media platforms will have a positive effect on post attitude, but this effect is stronger when product involvement is high rather than low.

When people are low involved they do not have the motivation or ability to process the information presented and heuristic thinking is likely. They process information via the peripheral route and will be sensitive to heuristic cues (Petty & Cacioppo, 1986). Familiarity is considered a heuristic cue (Tversky & Kahneman, 1974; Wyer & Srull, 1984). When consumers are exposed to the same message multiple times, this repetition increases

familiarity and familiarity may increase perceived validity (Wyer & Srull, 1984). Therefore, the hypothesis works the other way around as well (low involved and identical rather than similar).

As noted before, the most used categorizations regarding products are product involvement and the distinction between hedonic and utilitarian products. Therefore, some scholars (Mano & Oliver, 1993; Spangenberg, Voss & Crowley, 1997; Chung & Zhao, 2003; Geuens, De Pelsmacker & Faseur, 2010; Bart et al., 2014) state that there is an interaction effect between hedonic and utilitarian characterizations and high and low involvement. In this study, a three-way interaction effect is expected between post repetition, product type, and

(34)

34 product involvement. Based on previous arguments discussed at hypotheses 2 and 3, the following is expected:

H4 Similar post repetition on multiple social media platforms will have a more positive effect on post attitude for people who are higher involved and when the product is hedonic.

3.2.3 Effect of post attitude on brand attitude

Research has shown that attitude toward the ad – in this study attitude towards the post – is a causal mediating variable in the process through which advertising influences brand attitudes (Mitchell & Olson, 1981; MacKenzie, Lutz & Belch, 1986). In the early 80’s, Mitchell and Olson (1981) found that attitude toward the ad, defined as a construct that represents consumers’ feelings of (un)favorability toward the ad itself, is a mediating influence on purchase intention and brand attitude. In many other studies the role of ad attitude as mediator has been found continuously (Belch & Belch, 1983; Park & Mittal, 1985; Zhang, 1996). The next hypothesis follows logically from aforementioned since post attitude is an outcome variable in the previous hypotheses.

H5 Positive post attitude will positively affect brand attitude. 3.2.4 Effect of post attitude on customer engagement

Since engaging with a post means people will give the post a like, a comment or a share, it is only logical to conclude these people have a favorable attitude towards the post – otherwise, the reader will not engage with it. This also applies the other way around: when people have a positive attitude towards the post, they are more likely to engage with it. Therefore, the undermentioned hypothesis will be expected.

(35)

35 3.2.5 Overview of hypotheses

A structured overview of the hypotheses mentioned in the previous paragraph. Table 2: Overview of the hypotheses

H1 Post repetition on multiple social media platforms will have a positive effect on post

attitude, but only if the posts are similar rather than identical.

H2 Similar post repetition on multiple social media platforms will have a positive effect

on post attitude, but this effect is stronger when the product is hedonic rather than utilitarian.

H3 Similar post repetition on multiple social media platforms will have a positive effect

on post attitude, but this effect is stronger when product involvement is high rather than low.

H4 Similar post repetition on multiple social media platforms will have a more positive

effect on post attitude for people who are higher involved and when the product is hedonic.

H5 Positive post attitude will positively effect brand attitude.

H6 Positive post attitude will positively effect customer engagement.

4. Methodology

4.1 Research method and experimental design

The best-suited research design for the research question is an online experiment using a survey. An experiment is the best method since two different situations have to be compared to each other. Different groups of subjects have to receive different treatments to examine the effect on their responses. In an (online) experiment it is possible to manipulate the

independent variable(s) and to measure the various dependent variables (Boeije, ‘t Hart & Hox, 2009). Besides that, time and costs have to be taken into consideration since it is only limited. An online experiment offers a solution to these “problems” since this method enables to collect data in a short period without making (many) costs. The experiment is designed in Qualtrics, a research tool the University of Amsterdam made available for students. An online

(36)

36 experiment with a 2 (type of repetition: identical versus similar) x 2 (product type: hedonic versus utilitarian) subjects design has been conducted for this study. It is a between-subjects experiment since performances or scores of different participants are compared; a between-subjects experiment enables the researcher to examine the differences between the multiple conditions without any possible bias of exposure to other types of social media posts (Charness, Gneezy & Kuhn, 2012). There are four experimental conditions where the

independent variables – repetition and product type – are manipulated. See table 3 for an overview.

Table 3: Experimental design

2x2 Hedonic Utilitarian

Identical Group 1: identical posts with

hedonic product

Group 2: identical posts with utilitarian product

Similar Group 3: similar posts with hedonic product

Group 4: similar posts with utilitarian product

4.2 Stimulus material

The experimental material is posts of different brands promoting utilitarian and hedonic products on multiple social media platforms (Instagram, Facebook and Twitter). Existing brands and real posts are used for this experiment since this is better for the external validity than making fake posts of fake brands. The stimuli material for group 1 is identical posts about a utilitarian product. For group 2, the material is identical posts about a hedonic product. Group 3 was exposed to similar posts about a utilitarian product and group 4 was exposed to similar posts with a hedonic product. See table 3 for an overview. In the period before the experiment the material was acquired by searching the mentioned platforms for these four types of posts. This – and making screenshots – was done by phone, since most of the people use social media on their phone (Google Consumer Barometer, 2018). Posts were

(37)

37 photoshopped in order to create the right stimuli, they had all the same structure and

approximately the same amount of words. The layouts in which the posts in the different conditions were presented did not differ from each other in size, order and design. To come as close to the real situation, real Facebook, Twitter, and Instagram timelines were recreated. For the sake of internal validity as well, condition 1 and 3 saw the same brand and product, the posts only differed in phrasing: in condition 1, the phrasing is identical and in condition 3, the phrasing is similar. The same applies to condition 2 and 4.

4.3 Pre-test

A pre-test was conducted to investigate whether the manipulated material is usable for the experiment. The pre-test consisted of two conditions (condition 1: N=15, condition 2: N=15), both conditions got three post-examples of either hedonic or utilitarian product posts with either identical or similar content. Participants had to assess these post-examples on an identical versus similar scale and hedonic versus utilitarian scale. See Appendix I for an overview of the pre-tested stimuli material and which condition contained which examples. For every condition, three different examples of posts were tested in order to not rely on only one range of posts, with one product/content type. On a 7 point-Likert scale from ‘strongly agree’ to ‘strongly disagree’, participants were asked if they perceived the posts as identical or similar. They were also asked about their thoughts of the products, based on the HED/UT scale of Voss et al. (2003). The analysis of the identical and similar questions was based on descriptive statistics, since only one item was measured. See the fourth column of table 4. For the scales, first a reliability analysis was conducted based on Cronbach’s Alpha (α). For a reliably scale the threshold is α > .70. There were no counter-indicative items in the dataset. The corrected item-total correlations indicate that all the items have a good correlation with the total score of the scale (> .30) for most of the scales. For most of the scales, none of the items would substantially affect reliability if they were deleted. For UT/example 2, however,

(38)

38 for one item the corrected item-total correlations were < .30. α substantially improved when removing the ‘I think the product is practical’ item. The same applied to UT/example 3, removing the ‘I think the product is functional’ item increased α by .10. A factor analysis is helpful to further evaluate the scales, however, since the pre-test shows highly reliable scales, this factor analysis was not conducted for the pre-test. Table 4 presents all the outcomes.

Table 4: Reliability and means of HED/UT and ID/SIM post examples

Example Scale

reliability HED/UT

Mean scales HED/UT (1-7)

Mean scale ID/SIM (1-7) Ut/identical example 1 α = .94 M=5.25 (SD=1.29) M=6.27 (SD=1.75) Ut/similar example 1 α M=6.20 (SD=1.08) Ut/identical example 2 α = .84 M=5.73 (SD=.91) M=6.07 (SD=1.71) Ut/similar example 2 α M=6.07 (SD=1.34) Ut/identical example 3 α = .70 M=5.4 (SD=.68) M=5.93 (SD=1.39) Ut/similar example 3 α M=6.40 (SD=1.39) Hed/identical example 1 α = .91 M=5.2 (SD=.91) M=5.87 (SD=1.25) Hed/similar example 1 α M=5.27 (SD=1.79) Hed/identical example 2 α = .87 M=4.81 (SD=.96) M=5.87 (SD=1.25) Hed/similar example 2 α M=5.67 (SD=1.45) Hed/identical example 3 α = .84 M=4.99 (SD=.90) M=5.67 (SD=1.80) Hed/similar example 3 α M=5.60 (SD=1.40)

Compute means was used to form scales of the reliable scale items, and again descriptive statistics were run to find the means. As becomes clear from the table, example 2 of

UT/IDSIM (Swiffer) and example 1 of HED/IDSIM (Nike shoes) scored the highest on the means. Therefore, this material was used in the online experiment. For a complete overview and description of the final stimulus material used in the experiment, I refer to Appendix III. For the pre-test questionnaire, see Appendix II.

4.4 Sample

The sample is a so-called ‘convenience sample’; a type of non-probability sample where respondents are selected because of their convenient accessibility and proximity (Emerson, 2015). Via a personal approach on Facebook, friends and fellow students of the author were

(39)

39 asked to participate in the experiment. The only term for participation was that people had to be familiar with two of the three social media platforms (Facebook, Twitter and/or

Instagram). People who do not know about at least two of these platforms were excluded from the sample for the sake of external validity. People who do not know enough of the platforms cannot imagine the “real life” situation. In total there were 205 participants. However, some participants did not complete the questionnaire. They were excluded from the sample. Participants who only were familiar with one platform (or with no platform at all) were excluded from the sample as well. The total remained 195: condition 1 (N = 49), condition 2 (N = 50), condition 3 (N = 46) and condition 4: (N = 50). 76,4% of the participants were female (N=149), and 23,6% were male (N = 46), with an average age of 25,84 (SD = 7,77). 51,3% has a Bachelor’s degree in college. The largest part, 43,6%, spends 2-4 hours on social media. 48,7% of the participants are so-called ‘bystanders’ (see paragraph 4.6.3 for an

explanation). An overview of the frequencies can be found in Appendix V. 4.5 Procedure

The online experiment was designed in Qualtrics and distributed on Facebook with a link. When people clicked on the Qualtrics-link, the first thing they saw was an “informed

consent”. When people agreed with this consent, they became participants in the experiment. Participants were asked to imagine a situation of them scrolling through several social media platforms where they encounter posts of particular brands. Since real brands and post will be used, there is a possibility that people have a certain pre-existing attitude towards the brand that can influence internal validity. Therefore the question ‘what is your attitude towards brand X’ was asked before the stimulus material. Participants were exposed to the stimulus material (identical/similar post of hedonic/utilitarian product) and were asked to examine the posts thoroughly and to imagine that they are “scrolling” through social media like they normally would do. Afterward, participants were asked questions about their post attitude,