It’s not what we share, it’s how we share it:
Predicting engagement on Instagram via Social Media Influencers Leslie Ann Wheeler
11736739
Master’s Thesis
Graduate School of Communication Entertainment Communication Science
Jessica Taylor Piotrowski, Ph.D. February 1, 2019
Abstract
Millennials are influencing the expansion of mass-self communication by posting and engaging with User-Generated Content (UGC) on social media. UGC platforms such as Instagram are seeing tremendous growth at an accelerated rate. Consequently, having an online persona is not only an option anymore, but a necessity for success in personal and professional lives. Knowledge of what inspires online engagement is vital in these unprecedented times. And yet, research in this space has only just begun. To improve our understanding of successful strategies in the digital space, this study focuses on predicting engagement by analyzing self-presentation strategies of Social Media Influencers (SMIs). Source Credibility Theory (STC) guided the study by looking at three main source characteristics: Authenticity, Dynamism, and Attractiveness. A quantitative content analysis was conducted with 554 SMI posts on Instagram. Results from a multiple regression model indicate that the three factors are collectively predictive of engagement. Although individually, they are not significant. Additionally, four controlling factors negatively predicted engagement. Finally, study limitations are recognized, and directions for future research are discussed based on these findings.
Keywords: mass communication, User-Generated Content (UGC), engagement, self-presentation, Social Media Influencer (SMI), Source Credibility Theory (SCT), Instagram
It’s not what we share, it’s how we share it
“I am about to do what old people have done throughout history: call those younger than me lazy, entitled, selfish and shallow. But I have studies! I have statistics! I have quotes from respected academics!” (Stein, 2013)
Here's the cold, hard data: 1,000 selfies are uploaded on Instagram every 10 seconds (Cohen, 2016). The above statement from TIME Magazine (Stein, 2013) continues to spark controversy across the world, but few deny that Millennials grew up as digital natives in unprecedented times with the internet and society’s consequential changes (Prensky, 2001). Consequentially, what was once discussed as mass communication has now expanded into mass self-communication (Valkenburg, Peter, & Walther, 2016). Mass communication describes the number of people one may communicate to via various media channels, whereas mass self-communication adds the self-generated, self-directed, and self-focused aspect of the modern, reciprocal media (Valkenburg et al., 2016). Self-expression outlets are anywhere and everywhere for those looking, and it turns out that many Millennials aren’t shy. Self-presentation on social media is even encouraged now that having an online presence is becoming fundamental to relationship building (Marwick, 2010) and professional success (Khamis, Ang & Welling, 2017). And yet, the study of self-presentation remains woefully under-developed—despite its increasing importance in personal and professional improvement. Lack of research in this area emphasizes the crucial need for rigorous, empirical scholarship focusing on self-presentation strategies in the digital space.
User-Generated Content (UGC) inherently offers the opportunity to analyze and understand self-presentation strategies. UGC sites are interactive media spaces for consumers to create individualized, personal media about themselves or others and send it to whomever they
wish (Solis & Kutcher, 2011). UGC is not only shifting how we create, consume, and share information, but it is forever changing the way brands and professional content publishers think about the people who define their markets (Solis & Kutcher, 2011). What many people consider(ed) amateur Millennials are social media influencers (SMIs) who make a living off of self-branding and creating content to publish to UGC platforms such as Instagram. Considering the estimated $2 billion industry value and the projected 80% increase by 2020 (Contestabile, 2018), daily articles are produced worldwide speculating how SMIs have accumulated “loyal fans” (Marwick, 2010). Indeed, the freedom to produce content on the internet is now seen as not only an option, but a necessity for the success of brands and people (Gandini, 2016) and is argued in the commercial space to be an “inspirational and seemingly replicable” practice (Khamis et al., 2017, p. 194). And yet, despite this, the academic literature on SMIs’ tactics in detail is nearly non-existent. To contribute to this knowledge based on both self-presentation in general and SMIs in particular, this study takes the important next step by asking “What strategies are associated with successful SMIs?” Such information will do more than provide valuable information to industry leaders interested in, relying on, and bolstering their use of SMIs; it will advance our theoretical understanding of self-presentation.
Literature Review Social Media Influencer (SMI)
The online world today is increasingly driven by User Generated Content (UGC) more so than by professional advertising publishers (Marwick, 2010). Senft (2008) coined the term “Micro-celebrity” for users who have attained a certain level of fame by creating content on social media. This term has been used by researchers in the past (Khamis et al., 2017; Djafarova & Trofimenko, 2018) as well as numerous other terms, especially in reference to platforms such
as “YouTubers,” “Instagrammers,” and “bloggers” (Gumbrecht, Nardi, Schiano, & Swartz, 2004; Abidin, 2013; Kip & Uzunoglu, 2014). However, the most widely-recognized term is Social Media Influencer (SMI) (Freberg, Graham, McGaughey, & Freberg, 2011; Khamis et al., 2017). A SMI is someone who influences attitudes and behavior via social media by creating an online persona for the purpose of delivering consistent media to a loyal audience, regardless of the size of the audience (Marwick, 2010; Freberg et al., 2011; Djafarova & Trofimenko, 2018).
A user can become “instafamous” by producing UGC and acquire millions of dollars and followers (Djafarova & Trofimenko, 2018). These users have been recognizable in the early stages by the size of their followership, as it can extend to the millions, but as people explore this industry, professionals are noticing the powerful influence they have over the attitudes and behaviors of their audience (Freberg et al., 2011). They are now recognized as the most effective and trustworthy third-party endorsements (Freberg et al., 2011). However, the concept of online, self-made opinion leaders is so modern that researchers and professionals struggle to distinguish it.
SMIs’ particular differentiation from television or sports celebrities is that rather than the traditional spectator/spectacle dichotomy, they have personal interaction and online exchanges with their audience daily (Marwick, 2010). Traditional celebrities give the illusion of interaction and access, while the SMI practice involves sharing personal information with others as one would with a close friend or relative to increase direct feelings of friendship or closeness (Marwick, 2010). SMIs appear to be more authentic than traditional celebrities (Djafarova & Rushworth, 2017). SMI implies that there is an online audience to strategically uphold through ongoing UGC, communication, and interaction (Khamis et al., 2017). Instagram
Although many UGC sites exist, a few of the most commonly used platforms for SMIs include Instagram, YouTube, and Facebook (Marwick, 2010). The level of entry to these social media platforms are comparatively low and the potential to reach and influence a large audience is high and fair-game for any user (Solis & Kutcher, 2011; Khamis et al., 2017). For this reason, people are constructing online (me)mes to make their (self)ies stand out, which has resulted in unprecedented levels of social interaction and online participation (Khan, 2016). The fastest-growing UGC platform amongst them is Instagram, however research in this space is limited (Djafarova & Rushworth, 2017; Djafarova & Trofimenko, 2018).
Users interact differently on each UGC site in respect to their unique affordances (Khan, 2016). Instagram is being utilized for entertainment purposes and social interaction (Khan, 2016). The platform encompasses a culture of broadcasting the self via self-presentation, which makes it especially intriguing from a research perspective. From a practical standpoint, Instagram is the most important social media platform for working with SMIs, as it is the platform of preference for 87.1% of SMIs (“Influencer Marketing Statistics” n.d.). An estimated 27,000 SMIs are on Instagram with a combined 37 billion followers (“Influencer Marketing Statistics” n.d.). Instagram is a popular resource for sharing and interacting with photos and short videos, and SMIs aren’t the only ones choosing the platform. It attracts one billion users, of which 500 logs in daily and contributes to the ninety-five million posts uploaded each day (Clarke, 2018). Central to Instagram’s success is that it doesn’t simply invite users, it engages users; It lures them in, fascinates them, and captures their attention (O’Brien & Toms, 2008). Engagement
In the past, researchers deemed personal interaction as face-to-face contact and suggested that proximity was crucial to a leader’s influence and success (Kip & Uzunoglu, 2014). Today,
social media have inspired “new thinking about how to create lasting, flexible, and evolving relationships with consumers” (Khamis et al., 2017, p. 195). Online and offline lives are blending due to the manifold relationships that have formed and/or were maintained completely within the cyber space in just over a decade. Entire communities now connect and thrive on the internet with platforms to facilitate their communication and leaders to guide their organization. Users trust organizations that generate high levels of interactivity on social media (Yang & Lim 2009). Thus, media organizations’ primary self-branding objective is increasingly focused on creating and delivering interesting content to engage their audience online (Marwick, 2010). SMI success being largely operationalized as engagement.
To engage is defined by Dictionary.com (n.d.) as to “participate or become involved in.” In the online space, engagement is a quality of user experiences characterized by features such as feedback, interactivity (O’Brien & Toms, 2008), and social interaction (Khan, 2016). In a study on YouTube engagement, Khan (2016) defined engagement as “comprising behavioral aspects or click-based interactions (participation), as well as simple content viewing and reading (consumption)” (p. 237). Today, researchers and industry leaders argue that engagement is far more valuable in predicting influence of a SMI than number of followers is (Davis, 2016; Main, 2017; Djafarova & Trofimenko, 2018; Mathew, 2018). Engagement is understood to be the leading contributor to positive attitudes and even audience reach on social media (Kietzmann, Hermkens, McCarthy, & Silvestre, 2011).
Despite the clear importance that industry and SMIs place on engagement, it remains inherently unclear as to what predicts engagement with SMIs (Djafarova & Rushworth, 2017). Rather, there are numerous speculations and conjecture that punctuate the commercial landscape. For instance, some believe it can be improved in eight steps (Hughes, 2018), nine steps
(Canning, 2019), 14 steps (Barker, 2018), or 21 steps (Phillips, 2018). Others have concluded that it is engagement itself that produces engagement (Hughes, 2018), but fail to mention how the first round of engagement is inspired. Conferences such as VidCon, Social Media Week, and Content Marketing Summit are held worldwide attracting large crowds willing to pay thousands of dollars (“2019 Digital Marketing Conferences” 2019) to find the formula to reach this “holy grail” of digital marketing. Unfortunately, no two popular press conjectures are consistent in predicting engagement. Furthermore, no researcher has yet attempted to crack the code of engagement strategies of the SMI in scientific literature. As the social media influencer industry grows, it becomes increasingly important for researchers to investigate these nuances in interaction and explore SMI engagement on particularly prominent platform for SMI users— Instagram.
Theoretical Approaches Towards Predicting SMI Engagement
Crucial to understanding SMI success online is studying the source strategies of the message. When focusing on Instagram, the message is the post consisting of a photo and caption, while the source is the SMI who creates the content. The SMI chooses how he/she presents his/herself in each post and the audience chooses which presentation to engage with. Although predicting this engagement can be approached in various manners, persuasion literature has been applied throughout history to study the source of a message and visual techniques (suitable to those offered on Instagram) affecting attitude or behavior change (Pornpitakpan, 2004). Thus, theory in persuasion research serves as a guide through this research to interesting and probable predictors of engagement on Instagram.
Visual representation, such as in photographs, has become one of the most influential forms of self-expression, more so than text alone (Djafarova & Trofimenko, 2018).
Self-expression is central to SMIs’ practice on Instagram, as they construct a social identity via written and visual content creation. In theory, this concept is referred to as self-presentation (Goffman, 1959). Self-presentation is the way in which an individual presents him or herself in a social setting according to how he or she wants to be perceived by others (Goffman, 1959).
Aristotle’s concept of ethos, pathos, and logos is amongst the first theories in persuasion communication that placed the source itself squarely at the center (Demirdögen, 2010; Higgins & Walker, 2012). Ethos refers to the persona the message source wished to portray; pathos refers to the emotion appeal of a message; and logos refers to the argument with facts or reason (Higgins & Walker, 2012). From a translation, Aristotle proclaimed that when it comes to engaging an audience, “his character may almost be called the most effective means of persuasion he possesses” (Demirdögen, 2010, p. 191). This was in regards to the Ethos of a message. Source Credibility Theory (SCT) (Hovland & Weiss 1951; Ohanian, 1990) uses the concept of Ethos to argue that the influence of communication strongly depends on the personal appeal of the speaker (Giffin, 1967; see all Pornpitakpan, 2004).
SCT affirms that the source of the message is the key factor to what is communicated (Giffin, 1967). A communicator uses a physical image and nonverbal messages to create his/her persona (Demirdögen, 2010). Although SCT suggests three main characteristics; expertise, trustworthiness, and attractiveness of the communicator (Ohanian, 1990), various factors have been widely accepted and utilized in SCT studies.
For example, Yang & Lim (2009) conducted research into SCT and studied blogger credibility as seen through the audience’s eyes. No significant effect was found. Researchers in that study used a traditional SCT measure that included six factors: trustworthiness, expertness, experience, professionality, reliability, and intelligence (Yang & Lim, 2009). Similarly, Kang
(2010) conducted a survey investigating effects of blog credibility, which yielded indicators including: authentic, insightful, informative, consistent, and fair. These expanded on Lenhart & Fox (2006) qualitative study that theorized blog-specific attributes of SMI credibility including: passion, authenticity, transparency, influence, and insight. In this same vein, Djafarova & Rushworth (2017) expanded on SCT via qualitative theorizing. They concluded that “the attractiveness and quality composition of images are of significant importance to participants and a key influencer on whether or not users decide to follow new profiles” p. 17. However, it is important to note that study focused on followership rather than engagement, which we now know to be more central to SMI success.
A study into engagement on Instagram stimulated by the sender’s self-presentation should analyze all visual aspects of a post that the source has control over—including the photo and the caption. Three traditional, but adaptable characteristics of SCT are considered in this study. Highly relevant factors in the context of visual content in social media include a communicator’s trustworthiness, dynamism, and attractiveness (Giffin, 1967). Trustworthiness online has expanded into the notion of authenticity (Khamis et al., 2017), used to determine whether the SMI appears to have honest intentions. The SMI should portray to the user that the SMI has no hidden agenda or manipulation intentions for creating content (Khamis et al., 2017). Authenticity is supported by various researchers as being key to a SMI’s loyal audience (Lenhart & Fox, 2006; Kang, 2010; Djafarova & Trofimenko, 2018). In as much, it is expected that:
H1: Posts where the SMI presents herself as being authentic will produce higher levels of engagement.
In addition, dynamism is one of the firsts source credibility factors utilized to measure persuasion impact on an audience (Giffin, 1967). It gives the impression of being confident and
may provoke favorable impressions from an audience. It has been argued as one of the most crucial engagement skills for presenters (Palis, 2017). As such:
H2: Posts where the SMI presents his/herself as having high levels of dynamism will produce higher levels of engagement.
And lastly, attractiveness is deemed as one of the most influential characteristic of a source (Giffin, 1967; Ohanian, 1990; Pornpitakpan, 2004; Djafarova & Trofimenko, 2018). It has been considered by various researches in the SCT to be one of the most dominant features in persuasion, especially in reference to visual media (Giffin, 1967; Ohanian, 1990; Pornpitakpan, 2004; Djafarova & Trofimenko, 2018). It refers to the source’s physical appearance (Djafarova & Trofimenko, 2018). Physical characteristics of the body (waist, legs, bust, midsection, etc.) play a significant role in determining attractiveness (Garza, Heredia, & Cieslicka, 2016). As does self-presentation of the body in terms of facial expressions and body language (Joo, Li, Steen, & Zhu, 2014). However, attractiveness is a complex concept; difficult to measure, but important to study. As such, it is expected that:
H3: Higher levels of attractiveness of the SMI in the post will produce higher levels of engagement.
Method Design
To answer the question of what visual strategies affect a SMI’s engagement, a quantitative content analysis was conducted with SMI posts on Instagram.
Sampling
Quota sampling was used to ensure sufficient diversity and representation of posts that are considered high/low engagement along with different types of SMIs (micro/macro).
Engagement was operationalized as click-based interaction in the form of likes and comments. The engagement rate is the percent of followers who like and/or comment on the photo. High engagement is considered above 3% and low engagement is considered below 3%, due to industry standards (Mee, n.d.). In regards to followership, Micro SMIs are considered SMIs with low followership, while Macro SMIs are considered SMIs with high followership (Hatton, 2018). According to industry standards, an account with up to 10,000 followers is considered to be a Micro influencer (Hatton, 2018). However, “Influencer Tiers” (n.d.) argues that a Micro SMI has between 10,000 and 50,000 followers. This study considered both suggestions and operationalized a Micro influencer as a SMI with less than 30,000 followers. Macro influencers were considered having 30,000 followers or more. Together, this resulted in 60 SMIs that were selected for inclusion (30 macros; 30 micros, of which 15 in each group were considered high engagement and 15 were considered low engagement accounts. Due to the overwhelming number of females who meet the sample qualifications and the predominantly female population on Instagram (Djafarova & Rushworth, 2017), the sample of SMIs is 100% female.
Research has shown that when selecting a SMI to follow, users generally look through the profile posts to decide whether the content is engaging and worth following (Djafarova & Trofimenko, 2018). The 12 most recent photos are shown on the profile page, while the remaining posts are hidden until the user actively scrolls to view more. Thus, the first 12 photos on each profile were coded resulting in 720. Once video posts and “not applicable” or “cannot be determined” cases were dropped, there were 554 valid cases (N=554). The unit of analysis was each shared post. The photo was analyzed, as well as the caption.
Procedure
was chosen. From the selected profile, the researcher found another suitable SMI under one of the “recommended” accounts. This process continued until the entire sample was collected.
Various qualifications were met for the SMI to be selected. First, it is noted that SMIs specialize in different fields (food, fashion, beauty, motherhood, etc.) (Djafarova & Trofimenko, 2018) and that focusing on one niche will better ensure reliability of the study (Lenhart & Fox, 2006). Thus, all samples were drawn from those who created content centered around “lifestyle”, as this is the category found to be most liked by users (Djafarova & Trofimenko, 2018). Secondly, this study considered only achieved SMIs to ensure their online self-presentation is the main cause for the loyalty of their audience, rather than offline fame (ascribed SMIs) (Marwick, 2010). As such, traditional celebrities (sports, actors, etc.) were not analyzed. Finally, the language of the Instagram profile had to be English.
The codebook was designed in Qualtrics. Although one coder coded the data, three coders conducted a pilot study to assess codebook reliability. Four SMIs with 12 posts were selected. Thus, 96 units of analysis were coded to test the inter-coder reliability for 48 variables. Each variable in the codebook scored > 0.7 as per the rule of thumb for Krippendorff’s alpha, with the exception of one. Colors in photo was intended to measure the colors in the photo on a nominal scale with 10 dichotomous variables to see if such visuals influenced engagement. After coder training, it was concluded that to determine which colors existed and which did not was inconsistent due to the thousands of differences in color shades and this variable was not of utmost importance to the research question. This variable was removed from the codebook. See appendix for the complete codebook.
Main variables for coding are from the RQ: (1) engagement and (2) visual strategies. The independent variable is the visual strategies, while engagement serves as the dependent variable in this study. Codes consisted of 43 variables measured at the categorical level and five measured at the continuous level (N=48).
A visual strategy is defined as any method or sequence of visible tactics implemented to obtain a specific outcome (Dictionary.com, n.d.). It serves as the self-brand image framework within which a SMI decides “what to share and how to share it” (Brand It Beautiful, n.d.). For this study, visual elements are also considered in this category, as the SMIs physical features will be analyzed, as well. Categorical, nominal and ordinal scale codes record the presence and/or type of visual strategy (N= 43). These strategies are segmented into three main subcategories: authenticity, dynamism, and attractiveness. These codes will be used to measure H1, H2, and H3.
Measurement
Social media measurement is in its infancy, (Murdough, 2009). Therefore codes are adjusted to the social media context as efficiently as possible. Descriptive meta-data is used to measure variables. A continuous scale measured the SMI’s engagement rate for the post (N=1). The engagement percentage is measured at the ratio level by the tool WeFind. For examples of measures, see Appendix for codebook.
Authenticity refers to the trusted intentions that lead to the “consumer's confidence in the source for providing information in an objective and honest manner” (Ohanian, 1990, p. 47). To measure authenticity, dichotomous, categorical variables were measured at the nominal level to see if a promotional strategy was present or not in the post. These tactics included (a) Branded entertainment, (b) Public Service Announcement (PSA), (c) Product placement, (d) “Sponsored”
advertisement, and (e) Giveaway. Branded entertainment refers to any written acknowledgement of an event sponsored by a particular company or brand. PSAs in the social media context involve the mention of a product. This can be accompanied by a sales announcement, opinion, and/or call to action. Product placement refers to when a specific product is placed within the frame of the photo or tagged in a promotional manner. Furthermore, a photo is “Sponsored” when the word “Sponsored” appears at the top of the photo or in the caption. This is may also appear in the caption as “#sponsored” or “#ad”. Finally, a giveaway is where a SMI offers a prize to the audience if they follow certain accounts. These are indicators of the SMIs intention to persuade the audience, so were expected to be essentially non-existent for high engagement. The more persuasion tactics present, the lower the level of authenticity.
Attractiveness is categorized into two main variables: (1) body language, and (2) skin display. Thus, heavy emphasis is put on this code measured by categorical scales at the nominal or ordinal level. Body language measured popular body language techniques including arm and hand positions in relation to the body (Joo et al., 2014). Women may dress revealing because they believe it is attractive (Moor, 2010), so skin display measures where and approximately to what extent the SMI chooses to reveal bare skin in the photo. This was measured on a dichotomous scale of showing/not showing, using lined drawings as vivid examples (Garza et al., 2016).
Dynamism refers to the perceived energy and arousal levels of the communicator (Giffin, 1967). Energy was measured at the ordinal level from low to high. This ranges from when (1) SMI appears to be sleeping or laying down; (2) SMI is standing; (3) SMI appears to be physically moving; (4) SMI is engaging in slow-paced, but physical activity; and (5) SMI
appears to be engaging in a fast-paced physical activity. Arousal was measured at the ordinal level from low to high according to the SMI’s facial expression.
The visual image in terms of production quality (Luo & Tang, 2008; Djafarova & Rushworth, 2017) was analyzed and used as controlling factors for the analysis. Photo quality was measured based on the contrast of the camera lens in the photo (Luo & Tang, 2008). Scene context (Joo et al., 2014) & caption type (Gumbrecht et al., 2004) were measured as categorical variables to control for an interaction effect during analysis. As well as whether or not the caption directly asked for engagement (Palis, 2017).
Results
A multiple regression analysis was conducted to test whether post engagement could be predicted by the SMI’s authenticity, dynamism, and attractiveness strategies as independent variables—controlling for the post’s caption type, caption engagement, scene context, and
production quality. The model explaining post engagement was significant, R = .26, F(14, 553) = 2.34, p < .001, with 4.4% (adj. R2= .044) of the variance in post engagement explained.
Nonetheless, authenticity (b* = -.08, t = -1.66, p = .098, 95% CI [-.96,.08]), dynamism (b* = .07, t = 1.55, p = .123, 95% CI [-.08, .71]), and attractiveness in terms of body language (b* = -.05, t = -1.03, p = .303, 95% CI [-.53, .17]) and skin display (b* = .08, t = 1.89, p = .060, 95% CI [-.01, .29]) were not significant predictors of post engagement. Hence, H1, H2, and H3 are rejected, and their null hypotheses are retained. It is important to note that three caption types were significant; commentary or opinion (b* = -.12, t = -2.10, p = .037, 95% CI [-2.73, -.09]), ideas (b* = -.09, t = -2.01, p = .045, 95% CI [-4.75, -.05]), and relatable (b* = -.10, t = -2.20, p = .028, 95% CI [-6.03, -.35]). The Scene Context: Alone was a significant predictor as well (b* = -.09, t = -2.11, p = .035, 95% CI [-2.65, -.10]). Table 1 summarizes the model.
Table 1
OLS predicting post engagement on Instagram – unstandardized coefficient Post Engagement
Authenticity -.44 (.27)
Attractiveness: Body Language -.18 (.18)
Attractiveness: Skin Display 1.42 (.08)
Dynamism .31 (.20)
Caption Type: Life Documentation -1.11 (.63)
Caption Type: Commentary or Opinion -1.41 (.67)*
Caption Type: Expressing Emotion .85 (.81)
Caption Type: Ideas -2.40 (1.20)*
Caption Type: Relatable -3.19 (1.45)*
Caption Type: Giveaway -.57 (.80)
Scene Context: Alone -1.37 (.65)
Scene Context: Setting .23 (.47)*
Production Quality .13 (.53)
Engaging Caption .64 (.51)
Standard errors in parentheses *p < .05. ** p < .01. ***p < .001
N = 554
Discussion Conclusion
Millennials are shaping digital communication via UGC platforms (Solis & Kutcher, 2011). This generation, and generations to follow, will be increasingly prominent on social media and the consequences are already taking effect (Prensky, 2001). Today, there are 7.7 billion people in the world, of which more than half already uses the internet (Smith, 2019). On average, these 4.2 billion internet users accumulate more than seven social media accounts each. Consequently, self-presentation on social media is becoming necessary for personal and professional development. And yet, though it is vital to build an online persona while this technological evolution magnifies at full-speed (Gandini, 2016), research in this space has only just begun. To advance theoretical knowledge on self-presentation, this study analyzed SMIs’ visual tactics in detail to find the strategies that could harm or improve engagement on Instagram.
Content analysis was conducted to analyze visual self-presentation strategies by three SCT characteristics; authenticity, dynamism, and attractiveness. A model was created to determine the relationship between these main factors and engagement generated on a posted Instagram photo. Controlling factors were also added to the model. Altogether, the model was significantly predictive of engagement. On the other hand, neither authenticity, dynamism, nor
attractiveness individually predicted engagement. Conversely, controlling factors did prove to be significant independent predictors of engagement on Instagram posts. Captions that provided opinions, offered ideas, or implied relatable statements each predicted lower percentages of engagement. Furthermore, when a SMI involved another person in the photo, engagement percentage increased. The model as a whole was significant in predicting post engagement, but not substantially. Although this is standard for social science research, these key findings should be interpreted with caution.
Key findings are consistent with SCT in that there are several contributing factors to a source’s credibility (Giffin, 1967; Ohanian, 1990). However, authenticity and attractiveness findings were not consistent with research claims that suggest they will alone have a significant influence on source credibility (Lenhart & Fox, 2006; Kang, 2010; Djafarova & Trofimenko, 2018) No significant results were found for attractiveness in terms of body language (Joo et al., 2014) nor skin display (Garza et al., 2016). And as any study, it accepts its limitations in terms of procedure, operationalization, and offline strategies that future researchers should consider. Limitations
Firstly, sampling was conducted during the winter, whereas codes such as skin display, energy level, and setting may have been influenced. Future researchers may wish to sample posts representative of each season. Although 720 posts were sampled and analyzed on Instagram, more than 100 cases were dropped due to missing values. Videos were not considered for coding and answer categories in the codebook such as “cannot be determined” or “not applicable” supported considerable uncertainty. Future research should conduct a larger pilot coding, check percentages for values as such, and adjust the codebook accordingly. Researchers should also construct measurements for coding short video posts on social media, so all posts may be
considered for analysis.
Social media is challenging to measure and at times, seemingly impossible (Murdough, 2009). Especially for abstract concepts such as authenticity, dynamism, and attractiveness that can be studied in different contexts and operationalized in various manners. These concepts may yield different results when considering other verticals (outside of “lifestyle”) or sample gender. Dynamism, for instance, may have a stronger influence on males in the fitness vertical. Codes may also translate differently on different platforms. This study measured authenticity by persuasion tactics presented by the SMI, which may be overlooked by the audience due to the normalization of persuasion messages on Instagram (Djafarova & Rushworth, 2017). The audience could potentially consider the promotional tactics as honest recommendations. Future research may consider the various options of operationalizing key concepts. Engagement is measured in likes and comments, however, direct messaging is also one of Instagram’s interactive features—although it cannot be seen to the viewer. Furthermore, researchers should consider non-visible strategies in the future.
Social media is inherently visual, however, there are behind-the-scene strategies that play a role in engagement online (Murdough, 2009). Messaging is an interactive feature of Instagram that is hidden to the eye, so online tools such as WeFind do not take all engaging aspects into account. Engagimanually by the SMI, or with software that automatically engages with other posts via liking, commenting, or messaging in order to receive engagement in return. Many offline factors cannot be measured via content analysis. Thus, qualitative interviews or quantitative experimentation is recommended for future research to study both visual and behind-the-scenes strategies on social media. Despite its limitations, this study yields significant findings that improve our understanding of self-presentation strategies of SMIs.
About more than 500 people became a new user of social media just in the average time it takes to read the first word of this manuscript to the last (Smith, 2019). Knowledge of what inspires online engagement is vital in such rapidly-changing times. With the importance of a strong social media presence increasing by the minute, successful self-branding online “is not only possible or desirable,” Khamis et al. (2017) argues, “it is imperative and inevitable” p. 192. Understanding engagement strategies is key to the aspiring professional looking for a competitive advantage in the overly crowded marketplace (Khamis et al., 2017). Professionals and academics alike can benefit from these findings and the advancement this study made into the field of self-presentation and SMI research. With this study, an important stepping stone has been put into place for the future of communication science.
Sources
Abidin, C. (2013). Cyber-BFFs*: Assessing women’s ‘perceived interconnectedness’ in Singapore’s commercial lifestyle blog industry *Best Friends Forever. Global Media Journal, 7,1
Ayeh, J.K. (2015). Travellers’ Acceptance of Consumer-Generated Media: An Integrated Model of Technology Acceptance and Source Credibility Theories. Computers in Human Behaviour, 48, 173-180. DOI:10.1016/j.chb.2014.12.049
Barker, S. (2018, August 7). 14 of the Best Strategies to Increase Instagram Engagement. Retrieved from https://shanebarker.com/blog/increase-instagram-engagement/
Brand It Beautiful. (n.d.). WHY YOU NEED A VISUAL STRATEGY FOR YOUR BRAND IMAGERY. Retrieved on January 13, 2019 from
http://www.studiobicyclette.com/blog/why-you-need-a-visual-strategy
Bright, L., Daugherty, T., & Eastin, M.S. (2008). Exploring Consumer Motivations for Creating User- Generated Content. Journal of Interactive Advertising, 8, 2, 16‐25. DOI:
10.1080/15252019.2008.10722139
Canning, N. (2019, January 27). 9 Ways to Increase Instagram Engagement in 2019. Retrieved from https://later.com/blog/how-to-increase-instagram-engagement/
Clarke, T. (2018, October 5). 24+ Instagram Statistics That Matter to Marketers in 2019. Retrieved from https://blog.hootsuite.com/instagram-statistics/
Cohen, D. (2016, January 6). Selfies, Narcissism, and Social Media. Retrieved from https://www.adweek.com/digital/rawhide-selfies-infographic/
Contestabile, G. (2018, January 15). Influencer Marketing in 2018: Becoming an Efficient Marketplace. Retrieved from https://www.adweek.com/digital/giordano-contestabile-activate-by-bloglovin-guest-post-influencer-marketing-in-2018/
Davis, J.D. (2016, May 24). Why engagement is more important than follower count. Retrieved from https://unionmetrics.com/blog/2016/05/engagement-vs-follower-count/
Demirdögen, Ü.D. (2010). The Roots of Research in (political) Persuasion: Ethos, Pathos, Logos and the Yale Studies of Persuasive Communications. International Journal of Social Inquiry, 3, 1, 189-201.
Digital Marketing Conferences – 400+ Events Yearly (+Our Discounts). (2019, January 29). Retrieved from https://www.marketingterms.com/conferences/
Djafarova, E. and Rushworth, C. (2017). Exploring the credibility of online celebrities' Instagram profiles in influencing the purchase decisions of young female users. Computers in Human Behavior, 68, 1-7.
Djafarova, E. & Trofimenko, O. (2018). ‘Instafamous’ – credibility and self-presentation of micro-celebrities on social media. Information, Communication & Society, 1-15. DOI: 10.1080/1369118X.2018.1438491
Engage [Def. 2]. (n.d.). In Dictionary.com, Retrieved December 1, 2018 from
https://www.google.com/search?q=define+engage&oq=define+engage&aqs=chrome..69i 57j69i60j0l4.3302j1j7&sourceid=chrome&ie=UTF-8
Freberg, K., Graham, K., McGaughey, K., & Freberg, L. A. (2011). Who are the social media influencers? A study of public perceptions of personality. Public Relations Review, 37, 90-92. DOI:10.1016/j.pubrev.2010.11.001
Gandini, A. (2016). Digital work: Self-branding and social capital in the freelance knowledge economy. Marketing Theory, 16, 1, 123-141. DOI: 10.1177/1470593115607942 Garza, R., Heredia, R.R., & Cieslicka, A.B. (2016). Male and Female Perception of Physical
Attractiveness: An Eye Movement Study. Evolutionary Psychology, 1-16. DOI: 10.1177/1474704916631614
Giffin, K. (1967). THE CONTRIBUTION OF STUDIES OF SOURCE CREDIBILITY TO A THEORY OF INTERPERSONAL TRUST IN THE COMMUNICATION PROCESS. Psychological Bulletin, 68, 2, 104-120.
Goffman, E. (1959). The presentation of self in everyday life. New York, NY: Anchor Books. Greer, J.D. (2009). Evaluating the Credibility of Online Information: A Test of Source and
Advertising Influence. Mass Communication and Society, 6, 1, 11-28. https://doi.org/10.1207/S15327825MCS0601_3
Gumbrecht, M., Nardi, B.A., Schiano, D.J., & Swartz, L. (2004). Why We Blog. Communications of the ACM, 47, 12, 41-46. DOI: 10.1145/1035134.1035163 Hatton, G. (2018, February 13). Micro Influencers vs Macro Influencers. Retrieved from
https://www.socialmediatoday.com/news/micro-influencers-vs-macro-influencers/516896/
Higgins, C., & Walker, R. (2012). Ethos, logos, pathos: Strategies of persuasion in social/environmental reports. Accounting Forum, 36, 194-208.
Hovland, C., & Weiss, W. (1951). The influence of source credibility on communication effectiveness. Public Opinion Quarterly, 15, 635-650. DOI:10.1086/266350
Hughes, C. (2018, August 29). Increase Your Instagram Engagement in 8 Easy Steps. Retrieved from https://blog.iconosquare.com/increase-your-instagram-engagement/
Influencer Marketing Statistics. (n.d.) Retrieved on January 25, 2019 from
https://influencermarketinghub.com/influencer-marketing-statistics-quotes-facts/#toc-4 Influencer Tiers for the Influencer Marketing Industry. (n.d.). Retrieved on January 13, 2019
from http://mediakix.com/influencer-tiers/#gs.bL5ygWZN
Joo, J., Li, W., Steen, F.F., & Zhu, S.C. (2014). Visual Persuasion: Inferring Communicative Intents of Images, IEEE Computer Society Conference on Computer Vision and Pattern Recognition, Colombus, Ohio, September 2014, ResearchGate.
Kang, M. (2010). Measuring Social Media Credibility: A Study on a Measure of Blog Credibility. Institute for Public Relations.
Khamis, S., Ang, L. & Welling, R. (2017). Self-branding, ‘micro-celebrity’ and the rise of Social Media Influencers, Celebrity Studies, 8, 2, 191-208. DOI:
10.1080/19392397.2016.1218292
Khan, M.L. (2016). Social media engagement: What motivates user participation and consumption on YouTube? Computers in Human Behavior, 66, 236-247.
Kietzmann, J.H., Hermkens, K., McCarthy, I.P., & Silvestre, B.S. (2011). Business Horizons, 54, 2411-251. DOI:10.1016/j.bushor.2011.01.005
Lenhart, A., & Fox, S. (2006). Bloggers. A portrait of the internet’s new storytellers. Pew Internet & American Life Project. Retrieved January 12, 2019, from
https://www.pewtrusts.org/en/research-and-analysis/reports/2006/07/19/bloggers-a-portrait-of-the-internets-new-storytellers
Main, S. (2017, March 30). Micro-Influencers Are More Effective With Marketing Campaigns Than Highly Popular Accounts. Retrieved from https://www.adweek.com/digital/micro-influencers-are-more-effective-with-marketing-campaigns-than-highly-popular-accounts/ Marwick, Alice. (2010). STATUS UPDATE: CELEBRITY, PUBLICITY AND SELF-BRANDING
IN WEB 2.0. (Unpublished Doctoral Dissertation). New York University, New York, New York.
Mathew, J. (2018, July 30). Understanding Influencer Marketing And Why It Is So Effective. Retrieved from https://www.forbes.com/sites/theyec/2018/07/30/understanding-influencer-marketing-and-why-it-is-so-effective/#3fe4c2f271a9
Mee, G. (n.d.). What is a Good Engagaement Rate on Instagram? Retrieved on January 16, 2019 from https://blog.scrunch.com/what-is-a-good-engagement-rate-on-instagram
Moor, A. (2010). She Dresses to Attract, He Perceives Seduction: A Gender Gap in Attribution of Intent to Women’ s Revealing Style of Dress and its Relation to Blaming the Victims of Sexual Violence. Journal of International Women’s Studies, 11, 4, 115-127
Murdough, C. (2009). Social Media Measurement. Journal of Interactive Advertising, 10, 1, 94-99. DOI: 10.1080/15252019.2009.10722165
O’Brien, H.L., & Toms, E.G. (2008). What is User Engagement? A conceptual Framework for Defining User Engagement with Technology. Journal of the American Society for Information Science and Technology, 59(6), 938-955.
Ohanian, R. (1990). Construction and Validation of a Scale to Measure Celebrity Endorsers' Perceived Expertise, Trustworthiness, and Attractiveness. Journal of Advertising, 19, 3, 39-52.
Palis, A.G. (2017). A fundamental (often neglected) lecture skill: presenting with credibility. Annals of Eye Science, 2, 38, 1-5. DOI: 10.21037/aes.2017.05.06
Phillips, A. (2018, December 4). 21 Tips to Massively Increase Instagram Engagement in 2019. Retrieved from https://www.falcon.io/insights-hub/topics/social-media-strategy/21-tips-increase-instagram-engagement/
Pornpitakpan, C. (2004). The persuasiveness of source credibility: A critical review of five decades' evidence. Journal of Applied Social Psychology, 34, 2, 243-281.
Prensky, M. (2001). Digital natives, digital immigrants. On the horizon, 9, 5, 1-6
Senft, T.M. (2008). Camgirls: celebrity and community in the age of social networks. New York: Peter Lang. Retrieved from Google Books
Shan, Y. (2016). How credible are online product reviews? The effects of self-generated and system-generated cues on source credibility evaluation. Computers in Human Behavior 55, 633-641. https://doi.org/10.1016/j.chb.2015.10.013
Smith, K. (2019, January 2). 122 Amazing Social Media Statistics and Facts. Retrieved from https://www.brandwatch.com/blog/amazing-social-media-statistics-and-facts/#section-2 Solis, B., & Kutcher, A. (2011). Engage!: The Complete Guide for Brands and Businesses to
Build, Cultivate, and Measure Success in the New Web, Revised and Updated [Kindle version]. Retrieved from Amazon.com
Stein, J. (2013). The Me Me Me Generation. Time Magazine. Retrieved December 16, 2018 from http://time.com/247/millennials-the-me-me-me-generation/
Strategy [Def. 4]. (n.d.). In Dictionary.com, Retrieved January 13, 2019 from https://www.dictionary.com/browse/strategy?s=ts
European Conference on Computer Vision, Marseille, France, October 12-18, 2018. ResearchGate.
Uzunoglu, E., & Kip, S.M. (2014). Brand communication through digital influencers:
Leveraging blogger engagement. International Journal of Information Management, 34, 592–602. http://dx.doi.org/10.1016/j.ijinfomgt.2014.04.007 0268-4012/
Valkenburg, P.M., Peter, J., & Walther, J.B. (2016). Media effects: Theory and Research. Annual Review of Psychology, 67, 315-338. doi: 10.1146/annurev-psych-122414-033608
Visual [Def. 4]. (n.d.). In Dictionary.com, Retrieved January 13, 2019 from https://www.dictionary.com/browse/visual?s=ts
Yang, S.U. & Lim, J.S. (2009). The Effects of Blog-Mediated Public Relations (BMPR) on Relational Trust. Journal of Public Relations Research, 21, 3, 341-359. DOI: 10.1080/10627260802640773
Appendix Codebook
https://uvacommscience.eu.qualtrics.com/jfe/form/SV_dmB5q0jnDhWPUIl
Q1 Coder ID: (Indicate the number of the individual who is coding this sheet, according to the coder ID list)
Q2 Date: (dd/mm/yy) (Indicate the date of coding this photo) SMI account information
Q3 SMI ID: (Indicate the number of the SMI, according to the SMI ID list) Q4 Profile link: (Paste the Instagram URL of the selected SMI’s profile) Q5 Number of account followers: (As indicated at the top of the profile)
Q6 Profile engagement: (Percentage of total profile engagement using WeFind tool)
Post information
Q7 Post ID: (Fill in the post ID, as indicated on the name of the saved post photo) Post engagement: (percentage of post engagement using WeFind tool)
Q8 Type of post: (Photo, video, or album?) (An album is a set of photos) 1= Photo
2= Video 3= Album
(If the post is a video, skip coding and proceed to following post) (If there are a set of photos in
one post—referred to as an album—, code only the first photo)
Source
Q9 SMI present in selected photo? (If SMI is present, proceed to following codes. If SMI is NOT present, proceed directly to the Content section)
0= No 1= Yes
Authenticity (Perceived intention)
(Promotion strategies consist of when the SMI acknowledges branded entertainment, describes a
brand or product, announces a sponsored post, or places a branded product in the post. In a verbal or written manner. Find definitions of each promotion tactic below)
Q10 Branded entertainment: (Any acknowledgement of an event sponsored by a particular
brand) 0= No 1= Yes
Q11 PSA: (Public Service Announcement) (When the caption mentions a product with the
brand—with or without discount code—it is considered a PSA) 0= No
1= Yes
Q12 Product placement: (A branded product or prominent brand name placed in the frame of the
photo. Product or brand must be tagged in the photo to be considered. If a brand is only in the photo as a tag, this is a 1)
0= No 1= Yes
Q13 “Sponsored” ad: (The mention of “Sponsored” content above the post will also be
considered a public service announcement. Hashtags with the word #ad also count in this category)
0= No 1= Yes
Q14 Giveaway: (The mention of a large giveaway of money or prizes if the user follows certain steps to win)
0= No 1= Yes
Dynamism
(Energy and arousal levels)
Q15 Energy level:
2= Person is standing
3= Person is physically moving (Example: walking. Legs are apart in a forward walking motion, usually outdoor on a sidewalk or in the middle of a street)
4= Person is engaging in slow-paced physical activity (Examples: yoga &/or handstands) 5= Person is engaging in fast-paced physical activity (Examples: jumping, dancing, running &/or surfing)
5= Not applicable (Example: headshot where it is not apparent whether person is sitting or standing)
Q16 Arousal level:
1= Person has little to no facial expression (Examples: blank expression &/or pursing lips in a kissing or sassy manner)
2= Person is showing mild emotion in her facial expression (Examples: Smiling without showing teeth, or smile with few teeth showing)
3= Person is expressing strong emotions in her facial expression (Examples: Mouth open appearing to be laughing or screaming &/or tongue out)
5= Not applicable (Cannot see the person’s face at all, or person is facing to the side and cannot clearly see expression) (If 4, skip facial display and proceed to Gestures)
Attractiveness: (Body language) Q17 Touching head or hair
0= No 1= Yes
5= Cannot be determined
Q18 Are arms open or closed? (Arm(s) are not covering body or arms are covering body) 0= Closed (Both arms are inward over body)
2= Both (One arm is over body, the other is away)
9= Cannot be determined (Photo where arms are not visible. Example: headshot)
Q19 Hand (Hand is placed on hip or leg) 0= No
1= Yes
5= Cannot be determined (Photo where arms are not visible. Example: headshot
0= No 1= Yes
Q21 Are arms raised outward?
0= No 1= Yes
5= Cannot be determined
Q22 Holding something in hand(s): 0= Nothing
1= Yes
9= Cannot be determined
Attractiveness: Skin display
Skin showing: (Whether a certain part of skin is showing in photo. When select body parts are not visible, mark 0) (No need to mark when neck, hands, and feet are showing)
Back-
Q23 Upper back: (Area from neck to under bust)
0= No 1= Yes
Q24 Middle back: (Area from under bust to waist)
0= No 1= Yes
Q25 Lower back: (Area from waist to hips)
0= No 1= Yes Bust-
Q26 Upper bust: (Area from neck to shoulder to waist)
1= Yes
Q27 Middle bust: (Area from shoulder to waist to bust) 0= No
1= Yes
Q28 Lower bust: (Area from bust to
under bust) 0= No 1= Yes Midsection-
Q29 Upper midsection: (Area from under bust to waist)
0= No 1= Yes
Q30 Middle midsection: (Area from waist to hip height)
0= No 1= Yes
Q31 Lower midsection: (Area from hip height to hips)
0= No 1= Yes Legs-
Q32 Upper leg: (Area from hips to thigh) 0= No
1= Yes
Q32 Middle leg: (Area from leg length to calf) 0= No
1= Yes
Q34 Lower leg: (Area from calf down) (No need to mark when skin is showing on the foot)
0= No 1= Yes Arms-
Q35 Upper arm: (Area on arm from shoulder to bicep)
0= No 1= Yes
Q36 Middle arm: (Area from bicep to elbow) 0= No
1= Yes
Q37 Lower arm: (Area from just below elbow to wrist) (No need to mark when skin is showing on hands)
0= No 1= Yes
Caption: Q38 Type:
1= Life documentation (Sharing or asking for updates or news from one’s life) (Examples: “I am doing X today.” “Guess where my next vacation will be.” “What’s going on in your life?” 2= Commentary or opinion (Sharing or asking for opinions or recommendations) (Example: “I love this product.” “I prefer to…” “What do you think?”)
3= Expressing emotion (Sharing or asking specifically about emotions) (Example: “I am so
excited!” “X makes me happy”)
4= Ideas (Sharing or asking for new ideas or offering tips) (Example: “To look like X, I…”) 5= Relatable (Sharing or asking if one relates to a certain situation) (Examples: “Am I the only one who…” “Does anybody else…” “I think we all…”)
6= Giveaway (Offering a large prize for followers to do XYZ) 7= Other
Q39 Directly asking for engagement? 0= No (Caption does not ask a question)
1= Yes (Caption directly asks for input from the audience)
Production quality (visual quality of photo) Q40 How is the photo quality?
2= Clear (Photo has no blurred aspect)
3= Contrasted (Aspects of the photo are very clear, while other parts of the photo are blurred. The focus is clear or depth is emphasized)
9= Not applicable (Example: photo of a quote or only containing text) Scene context Q41 Setting: 1= Indoor 2= Outdoor 9= Not applicable
Q42 Is the SMI alone in the photo? (Is there another person or animal as part of the photo focus?)
0= No 1= Yes
