1. INTRODUCTION
This study endeavored to understand the celebrity culture in contemporary Bangladesh, centering on social media. Studies on online celebrity culture have been growing worldwide, though their focus is primarily on the Global North (Lau, 2021). Besides this, global academic scholarship has overlooked some regions and countries like Bangladesh, where celebrity culture is thriving. As with regional bias, celebrity studies also seem limited to a few specific disciplines, such as business studies and psychology (Abidin, 2018; Aljasir, 2019; Bennett, 2014; Chung & Cho, 2017; Hou, 2019; Jacobsen, 2020; Kim & Kim, 2020; Ledbetter & Redd, 2016). These two issues have created a knowledge vacuum that demands further scholarly attention and exploration. Since social media usage has dramatically increased worldwide in recent years, transforming the global celebrity culture, research on social media celebrities has become highly relevant and imperative. Previous research shows that social media celebrities can influence followers more than traditional celebrities can (Ahmadi et al., 2022). For this reason, many celebrities are also interested in using social media to reach an expanded audience (Selkie, 2022). However, social sciences and humanities scholars have paid limited attention to online celebrity culture. As a result, a functional and appropriate theoretical framework remains underdeveloped (Couldry, 2015). Taking these issues into account, this research studies online celebrity culture in Bangladesh.
Why is a study on Bangladesh necessary? In Bangladesh, 49.55 million people (29.7% of the population) use social media, and 4.6 million users (+10.1%) increased its ranks in 2021-22. The community of social media users is expanding rapidly, allowing celebrities to exert immense influence over their followers’ behavior (Selkie, 2022). However, due to a lack of proper academic investigation, insights regarding online celebrities are limited, impeding the understanding of people’s choices and the cultural trends of the country. Not only culture is influenced; celebrities often knowingly or unknowingly endorse products and information that may negatively affect people’s health (Selkie, 2022). The parasocial relationship between celebrities and followers can also inspire followers toward positive actions. During the COVID-19 pandemic, for example, national cricket stars donated a large amount of money to the government fund to combat the crisis, which was well-received by their followers and later motivated more people to do the same (Sakib, 2020).
Based on these rationales and in the absence of relevant research, this study attempted to answer four research questions in the context of Bangladesh: Who are the celebrities, how popular are they, what do they post on social media platforms, and how do their followers react to them? Following a quantitative content analysis method and descriptive statistical analysis, this study found that the most popular celebrities on social media are actors, musicians, and sports stars. Also, celebrities mainly share photos on social media, perhaps to recreate their public image and provide regular updates to their followers. The following section discusses relevant literature to provide a background for this research.
2. BACKGROUND
2.1. Who Are the Celebrities?
In its simplest form, celebrity means being popular to many people, who are known as followers or fans. Theoretically, celebrity-ism is “understood to be a quality of individuals” (Rojek, 2015, p. 1). A celebrity is an individual who receives public attention. Are celebrities and influencers the same? While some scholars use the terms interchangeably, some argue that the two are different and that an influencer is the epitome of a celebrity (Abidin, 2018; Hou, 2019). However, this paper does not maintain a strict and subtle division between the concepts. Celebrities can be of three types: ascribed celebrities, who are popular due to their hereditary status; achieved celebrities, who become popular through their recognized talents; and celetoids (a portmanteau of celebrity and tabloids), who are short-lived celebrities, ascending to fame often through scandal, crime, or exceptional performances (Driessens, 2021; Rojek, 2015). In the age of social media, the prevalence of celetoids seems to have increased.
Who are and can be celebrities nowadays? Some studies have investigated the professions of celebrities who are popular in the public domain and their level of popularity (Schimmelpfennig, 2018). Most studies we found relevant were from business and economics disciplines. For example, Um (2013), following a quantitative content analysis method, investigated the dominant professions of celebrity endorsers in the United States (US) and Korea. They explored the following five professions of celebrities: actors, athletes, musicians, comedians, and others. In both countries, the result shows that actors (59.7% in Korea and 63.8% in the US) were the most popular celebrities for commercials. Also, female celebrities (46.8% in Korea and 61% in the US) were more popular than male celebrities (31.7% in Korea and 33.3% in the US). Um’s study further illustrated that celebrities’ professions were more varied in the US than in Korea, since the US “modeled such professions as CEOs, chefs, supermodels, and TV personalities” (Um, 2013, p. 167). The study did not mention what professions the others category includes.
In a similar study, Schimmelpfennig (2018) explored popular celebrities’ professions in German advertisements: actors, athletes, models, musicians, and broadcast stars. Of them, according to their appearance, actors (23%) were the most popular, followed by athletes (22%) and models (14%) (Schimmelpfennig, 2018, p. 7). Following a quantitative content analysis of female magazine advertisements, St. James (2010) studied celebrities’ gender- and profession-based proportions. The study found as the five top following celebrity professions: musicians (42.4%), actors (40.1%), athletes (12.1%), TV personalities (3.5%), and models (2%). Of them, 94.2% were female, and 8.2% were male celebrities. Note that previous studies preferred to use the term athletes instead of sports stars.
In another study, Gergaud et al. (2012) analyzed the top 100 celebrities published by Forbes. Professions of celebrities in the study were grouped into four broad categories with their frequencies: actors/broadcast stars (n=25), musicians/singers (n=7), sports (n=7), and others (n=10). Of them, 35 were males, 14 were females, 42 were Caucasians, 3 were Hispanics, 2 were Afro-Americans, and 2 were Asians. Like Um (2013), this study also did not explain what the others category includes. Also, gender and ethnicity were conflated, further limiting the study. Beyond the business discipline, celebrities’ professions are also a concern in health studies. Ahmad et al. (2020) investigated deceased celebrities from 1999 to 2017 in the US due to drug overdose. Their result shows the 10 top professions of celebrities: athletes, musicians, actors, entertainers, models, writers, DJs, directors, activists, and wrestling managers.
Overall, these studies cannot provide at least two relevant insights. First, they dealt with celebrities’ statistics collected from traditional media instead of social media. Second, they did not analyze celebrities’ level of popularity. Instead, they relied only on celebrities’ frequencies. Therefore, this study asks the following research questions:
2.2. Celebrities and Social Media
Media plays a pivotal role in creating celebrities because recurrent media representation of individuals shapes and reconstructs their public image (Driessens, 2021). In the past, traditional media was the sole agent of producing celebrities. However, social media now occupies an important space in this discussion (Abidin, 2018). It empowers users to self-represent themselves and their talents and to reach a large number of fans easily. Such self-branding and direct, instantaneous, and regular communication seem effective, which has changed how we perceive celebrity culture in the pre-digital era (Bennett, 2014).
It is important to remember that the pervasiveness of social media is now erasing the distinction between real-life celebrities and social media celebrities. Nowadays, while many celebrities emerge from social media, many others are real-life celebrities who may wish to maximize their fandom through social media platforms. Given the present context, more studies are required to identify the features of contemporary online celebrity culture.
Meanwhile, researchers, mainly from psychology and business studies disciplines, have produced numerous papers focusing on celebrity endorsement, the psychology of fandom, branding, and advertising intertwining social media (Abidin, 2018; Aljasir, 2019; Bennett, 2014; Chung & Cho, 2017; Hou, 2019; Jacobsen, 2020; Kim & Kim, 2020; Ledbetter & Redd, 2016). Also, previous studies dealing with celebrity culture relied on only a few specific theories. For example, online celebrity studies from the psychology discipline incorporated parasocial interaction theories (Bennett, 2014; Chung & Cho, 2017; Kim & Kim, 2020; Kim & Song, 2016; Ledbetter & Redd, 2016). Again, studies from business and marketing disciplines relied on celebrity endorsement theory (Aljasir, 2019; Arefin, 2018; Chung & Cho, 2017; Schimmelpfennig, 2018; Shabnaz et al., 2019; St. James, 2010; Um, 2013). However, pure social theory is still absent in celebrity studies (Couldry, 2015), which compels communication scholars to depend on theories from other fields (Hou, 2019; Kim & Kim, 2020; Ledbetter & Redd, 2016). Moreover, classical theories used in celebrity studies seem less competent and less applicable to celebrity-fan interactions on social media platforms (Aljasir, 2019). These conceptual limitations call for theoretical and methodological experimentations.
Another limitation existing in previous literature is that celebrity research is highly West-centric, leaving other regions and countries out of focus (Shabnaz et al., 2019). For example, two keyword searches (Search 1: “celebrity” OR “celebrities” AND “social media” OR “online” OR “Facebook” OR “Twitter” AND “Bangladesh”; Search 2: “celebrity” OR “celebrities” AND “Bangladesh”) in the Scopus database yielded no result, indicating no significant research on the celebrity culture of Bangladesh.
However, we found some underexposed studies conducted by some native scholars from the business discipline. For example, Shabnaz et al. (2019) surveyed 568 women from Dhaka, the capital of Bangladesh, to investigate how celebrity endorsements influence consumers’ intention to choose personal care products. Like similar previous studies (Schimmelpfennig, 2018; Um, 2013), this Bangladesh-based study focuses on consumer behavior and product marketing. In addition, similar research explored Bangladeshi consumers’ music preferences (Arefin, 2018). The growing online community, impactful social media celebrity culture, the prevalence of celetoids, and the absence of proper scholarly investigation provide rationales for the present study. Therefore, this exploratory study further asks two questions:
2.3. Framing Social Media Reactions
Two dominant theories from psychology are used widely in scholarly literature to frame and understand human reactions: the circumplex model of affect, and basic emotion theory. The Circumplex Model of Affect by Russell (1980) includes two primary variables: valence and arousal. Valence is the inclination of emotion, which could be positive and negative, and arousal is the level of emotional intensity, which could be high or low. Combining these two variables, this model includes 28 different emotional reactions in four categories: high arousal positive valence, low arousal positive valence, high arousal negative valence, and low arousal negative valence (Russell, 1980).
In contrast, the basic emotion theory proposed by Ekman (1992) seems more simplified, integrating fewer broad reactions in the framework: happiness, sadness, disgust, fear, anger, and surprise. Like the circumplex model, this theory also relies on human facial expressions. Interestingly, theorists cannot agree on universal emotions to date, allowing researchers to produce different sets of basic emotions (Kowalska & Wróbel, 2017). In the study of social media reactions, however, the basic emotion theory seems more relevant than others. For example, in February 2016, Facebook, the world’s most populated social media platform, introduced five major reactions (i.e., love, wow, haha, sad, and angry) influenced by the basic emotion theory (Al-Rawi, 2020; Dewey, 2016). Previous studies also used this theory as a theoretical lens to analyze social media users’ reactions to different issues (Eberl et al., 2020; Freeman et al., 2019, 2020; Geboers et al., 2020; Giuntini et al., 2019; Jost et al., 2020; Tran et al., 2018; Varanasi et al., 2018).
3. MATERIALS AND METHODS
We prepared two datasets for two different analyses: The first answered RQ1 and RQ2, and the second answered RQ3 and RQ4. We focused on Facebook because most Bangladeshi social media users use this platform (86.73%) compared to the others, such as Twitter (7.73%) and YouTube (3.58%) (StatCounter, 2020). Moreover, of the total Facebook users, 41.4% are found to be highly addicted to it, making the platform influential and worth exploring (Raisa, 2018).
For dataset one, we collected the popularity data of Bangladeshi social media celebrities from SocialBakers ( http://socialbakers.com), an AI-powered social media marketing platform. It has a rich collection of Facebook data, which is used widely in scholarly literature (Brosnan & Gavin, 2015; Mamur, 2015). From this website, we collected data on the top 200 most popular Bangladeshi Facebook celebrities based on their followers. The data included the names of the celebrities, who they are, their numbers of followers, and the monthly growth of their followers. However, information about some celebrities was missing from the website. In such cases, we collected it from their official Facebook fan page or by searching the Internet.
Although the website attributed nine professions to celebrities, we found it problematic from at least two aspects. First, some celebrities are known in more than one profession (e.g., Tahsan is a musician and actor). In these cases, we tried to determine their most popular and accepted identities to their fans, which often contradicted the attributed professions on the website. Second, some celebrities were incorrectly identified on the website, so we amended them (e.g., Sajeeb Wazed is a politician, not a businessman). Also, we found some information incorrect. For example, the list included Abdur Razzak, a non-Bangladeshi and Pakistani lawyer. Therefore, we eliminated such celebrity profiles (n=5) from our dataset. Our final dataset included 195 celebrity profiles and their details. This dataset was optimized to analyze the professions and genders of the most popular celebrities (RQ1) and their popularity patterns (RQ2). For RQ1, we coded the professions based on previous studies and the website’s attributed professions, along with a few newly attributed professions (Ahmad et al., 2020; Gergaud et al., 2012; Schimmelpfennig, 2018; St. James, 2010; Um, 2013).
For dataset two, we collected celebrities’ Facebook posts and users’ engagement in those posts. To understand what the celebrities post on Facebook (RQ3) and how users react to them (RQ4), we selected the five most-followed celebrity profiles: Shakib Al Hasan (sports star; 14.20 million followers), Mushfiqur Rahim (sports star; 12.01 million followers), Pori Moni (actress; 9.09 million followers), Tahsan (singer; 8.89 million followers), and Hanif Sanket (broadcast star; 8.77 million followers). We set the time range between 21 May 2011 and 20 May 2021 and extracted all posts’ data from their official Facebook pages. In this data extraction process, we used CrowdTangle ( http://crowdtangle.com), a public insights tool owned and operated by Facebook (CrowdTangle, 2021). Researchers now widely use it, and we have authorized access to harvest Facebook data. We collected 9,441 posts (MCelebrity=1,882) and 475.44 million users’ interactions (MCelebrity=95.09): 50,359.26 reactions per post.
CrowdTangle has built-in filters/features for extracting Facebook data. It categorizes the content types as links, statuses, photos, videos, live videos complete, live videos scheduled, native videos, and YouTube videos. To avoid discrepancy and confusion, we merged the five video-related categories (i.e., videos, live videos complete, live videos scheduled, native videos, and YouTube videos) and named them as videos. We measured the interactions with comments, shares, like, love, wow, sad, haha, angry, and care attributes. Facebook’s five main reaction buttons (i.e., love, haha, wow, sad, and angry) allow users to express their emotions (Dewey, 2016). In the light of previous literature, for the present study, we reformulated the meaning and valence of these reaction buttons as follows:
-
Love (valence: positive; meaning: positivity, affection, empathy, support, kindness, liking, happiness, amused, proud, confident, satisfied, contented, and comforted).
-
Sadness (valence: negative; meaning: empathy, grief, tragedy, support, misery, guilt, and depression).
-
Haha (valence: positive; meaning: happy, laughter, fun, mockery, and rejection).
-
Wow (valence: positive; meaning: muted disdain, surprise, shock, skepticism, interest, confused, and aroused).
-
Anger (valence: negative; meaning: denial, aggressiveness, strong and overt disdain, dislike, disturbance, restlessness, and affliction) (Dewey, 2016).
We used descriptive statistical analysis to yield frequency counts (n), percentages (%), mean (M), and standard deviation (SD) of the relevant variables. It is a basic statistical analysis to extract descriptive insights from a dataset (Denis, 2018). Also, we performed inferential statistical analyses (i.e., correlation analysis) where necessary. The datasets were prepared using Microsoft Excel, and the statistical analyses were performed using IBM Statistical Package for the Social Sciences (SPSS) 25 (IBM Co., Armonk, NY, USA).
4. RESULTS AND DISCUSSION
This study aimed to explore four aspects of online celebrity culture: celebrities from those professions and genders that are the most popular, their popularity, what they post on social media, and how their fans react to them.
The result shows that 13 types of celebrities are more prevalent in Bangladesh (Table 1). Of them, actors (33.85%) have the highest frequency, followed by singers (24.10%) and musicians (14.36%). Despite their proximity, singers and musicians are categorized separately: singers are indicated as the persons who only sing using their vocals, and musicians write and compose music and often sing. We combined both categories: music-related celebrities constitute 38.46% of the total celebrities.
Table 1
Descriptions of the most popular celebrities
| Celebrity | n | % | Followers | Growth | |||||||
|---|---|---|---|---|---|---|---|---|---|---|---|
| Followersa) | % | Meana) | Standard deviationa) | Followersa) | % | Meana) | Standard deviationa) | ||||
| Actor | 66 | 33.85 | 118.21 | 39.3 | 1.79 | 1.89 | 9.55 | 48.5 | 0.15 | 0.23 | |
| Singer | 47 | 24.10 | 54.19 | 18.0 | 1.15 | 1.61 | 6.26 | 31.8 | 0.13 | 0.46 | |
| Musician | 28 | 14.36 | 13.16 | 4.4 | 4.70 | 0.54 | 0.18 | 0.9 | 0.00 | 0.02 | |
| Sports star | 17 | 8.72 | 66.76 | 22.2 | 3.93 | 4.27 | 1.65 | 8.4 | 0.10 | 0.10 | |
| Writer | 10 | 5.13 | 14.40 | 4.8 | 1.44 | 2.17 | 0.46 | 2.3 | 0.05 | 0.08 | |
| Model | 6 | 3.08 | 1.81 | 0.6 | 0.30 | 0.24 | 0.02 | 0.1 | 0.00 | 0.00 | |
| Broadcast star | 4 | 2.05 | 10.72 | 3.6 | 2.68 | 4.10 | 0.71 | 3.6 | 0.18 | 0.20 | |
| Entrepreneur | 4 | 2.05 | 0.69 | 0.2 | 0.17 | 0.11 | -0.00 | 0.0 | -0.00 | 0.00 | |
| Islamic scholar | 4 | 2.05 | 8.86 | 2.9 | 2.22 | 1.50 | 0.34 | 1.7 | 0.08 | 0.05 | |
| DJ | 3 | 1.54 | 3.61 | 1.2 | 1.20 | 1.13 | 0.12 | 0.6 | 0.04 | 0.04 | |
| Politician | 3 | 1.54 | 4.52 | 1.5 | 1.51 | 1.34 | 0.03 | 0.1 | 0.00 | 0.01 | |
| YouTuber | 2 | 1.03 | 2.96 | 1.0 | 1.48 | 0.68 | 0.20 | 1.0 | 0.10 | 0.10 | |
| Doctor | 1 | 0.51 | 0.10 | 0.3 | 0.10 | 0 | 0.17 | 0.9 | 0.17 | 0.00 | |
Sports stars (8.72%) also have a higher frequency than some other celebrities, such as writers (5.13%) and models (3.08%). Although sports stars are in the fourth position on the list, their followers (22.2%) are the second highest after actors (39.3%). More interestingly, sports stars have the highest mean value (M=3.93 M), followed by Islamic scholars (M=2.22 M), ranking ninth in the list with 2.05% frequency and 2.9% followers. In terms of average monthly followers’ growth, actors have the highest percentage (48.5%), followed by singers (31.8%) and sports stars (8.4%). However, mean growth indicates that broadcast stars (MCelebrity=0.18 M) have the highest average growth, followed by actors (MCelebrity=0.15 M) and singers (MCelebrity=0.13 M). Studies on Korean (59.7%), American (63.8%), and German (23%) celebrity endorsements of brands suggest that actors are the most popular celebrities (Schimmelpfennig, 2018; Um, 2013), which supports the findings of this study.
Another study (St. James, 2010) shows the seemingly contradictory result that musicians (42.4%) were more popular than actors (40.1%) for advertising purposes. However, unlike other studies and the present one, this research combined singers and musicians into a single category. On the other hand, Gergaud et al. (2012) combined actors and broadcast stars as the most popular celebrity category, making it difficult to determine the share of actors. These results infer that the popularity of actors is not limited to only advertisements and marketing. Instead, they are overall the most popular celebrities. Also, the popularity of actors seems not country-specific.
This study also found that male celebrities (64.10%) have higher frequencies than female celebrities (35.90%), which corresponds to their number of followers as well (Table 2). Previous studies provided limited insights regarding celebrities’ gender. A relevant study suggests that male celebrities are more common than female celebrities in advertising (Um, 2013), which is similar to the gender-based frequency of the present study. However, the average number of followers of female celebrities (Mgender=1.60 M) is slightly higher than for male celebrities (Mgender=1.51 M), which also corresponds to their followers’ growth. It indicates that though female celebrities are fewer in number, they have a higher number of fans on average than male celebrities. An Eta coefficient test between celebrity professions and followers shows a strong association, and between professions and follower growth shows a weak association. Again, associations between gender and followers and gender and follower growth are weak or negligible (according to Quinnipiac University; Akoglu, 2018).
Table 2
Descriptions of genders of the most popular celebrities
| Gender | n | % | Followers | Growth | |||||||
|---|---|---|---|---|---|---|---|---|---|---|---|
| Followersa) | % | Meana) | Standard deviationa) | Followersa) | % | Meana) | Standard deviationa) | ||||
| Male | 125 | 64.10 | 189.17 | 62.9 | 1.51 | 2.34 | 9.21 | 46.8 | 0.07 | 0.24 | |
| Female | 70 | 35.90 | 111.70 | 37.1 | 1.60 | 1.87 | 10.48 | 53.2 | 0.15 | 0.31 | |
Female actors are higher in frequency (n=45) with a higher share of followers (73.3%) compared to male actors (n=21 and 26.7% followers) (Table 3). Their mean value is also higher than for male actors. Since actors are the most popular celebrities, female actors’ higher frequency and number of followers again suggest that female celebrities might have higher popularity than male celebrities. We can observe a somewhat similar result for music-related celebrities. Most of the singers (n=34) and musicians (n=27) are male, although the mean value of female singers’ followers (M=1.47 M) is higher than for the male singers’ followers (M=1.03 M). Interestingly, it seems some celebrity professions are gender specific. For example, celebrities who are entrepreneurs, doctors, Islamic scholars, politicians, and YouTubers are all males, whereas only models are all females. It could also be an outcome of limited data. A chi-square test (χ2=72.474) between celebrities’ genders and professions shows a strong positive association (p<0.05).
Table 3
Gender-based descriptions of the most popular celebrities
| Celebrity | Gender | n | Followers | Growth | |||||||
|---|---|---|---|---|---|---|---|---|---|---|---|
| Followersa) | % | Meana) | Standard deviationa) | Followersa) | % | Meana) | Standard deviationa) | ||||
| Actor | Female | 45 | 86.64 | 73.3 | 1.92 | 2.09 | 7.13 | 74.6 | 0.20 | 0.25 | |
| Male | 21 | 31.57 | 26.7 | 1.50 | 1.34 | 2.42 | 25.4 | 0.11 | 0.19 | ||
| Broadcast star | Female | 2 | 1.68 | 15.6 | 0.84 | 0.78 | 0.37 | 52.7 | 0.19 | 0.25 | |
| Male | 2 | 9.04 | 84.4 | 4.52 | 6.01 | 0.34 | 47.3 | 0.17 | 0.24 | ||
| DJ | Female | 1 | 0.82 | 22.6 | 0.82 | 0.00 | 0.07 | 61.5 | 0.07 | 0.00 | |
| Male | 2 | 2.79 | 77.4 | 1.40 | 1.53 | 0.05 | 38.5 | 0.02 | 0.03 | ||
| Doctor | Male | 1 | 0.10 | 100 | 0.10 | 0.00 | 0.17 | 100 | 0.17 | 0.00 | |
| Entrepreneur | Male | 4 | 0.69 | 100 | 0.17 | 0.11 | -0.00 | 100 | -0.00 | 0.00 | |
| Model | Female | 6 | 1.81 | 100 | 0.30 | 0.26 | 0.02 | 100 | 0.00 | 0.00 | |
| Islamic scholar | Male | 4 | 8.86 | 100 | 2.22 | 1.49 | 0.34 | 100 | 0.08 | 0.05 | |
| Musician | Female | 1 | 0.23 | 1.7 | 0.23 | 0.00 | 0.02 | 8.40 | 0.02 | 0.00 | |
| Male | 27 | 12.93 | 98.3 | 0.48 | 0.55 | 0.17 | 91.6 | 0.00 | 0.01 | ||
| Politician | Male | 3 | 4.52 | 100 | 1.51 | 1.34 | 0.03 | 100 | 0.00 | 0.01 | |
| Singer | Female | 13 | 19.07 | 35.2 | 1.47 | 1.47 | 2.84 | 45.4 | 0.22 | 0.56 | |
| Male | 34 | 35.13 | 64.8 | 1.03 | 1.66 | 3.42 | 54.6 | 0.10 | 0.42 | ||
| Sports star | Female | 1 | 1.07 | 1.6 | 1.07 | 0.00 | 0.03 | 1.80 | 0.03 | 0.00 | |
| Male | 16 | 65.68 | 98.4 | 4.11 | 4.35 | 1.62 | 98.2 | 0.10 | 0.10 | ||
| Writer | Female | 1 | 0.40 | 2.7 | 0.40 | 0.00 | -0.00 | -0.20 | -0.00 | 0.00 | |
| Male | 9 | 14.01 | 97.3 | 1.56 | 2.27 | 0.46 | 100 | 0.05 | 0.09 | ||
| YouTuber | Male | 2 | 2.96 | 100 | 1.48 | 0.68 | 0.19 | 100 | 0.10 | 0.10 | |
Regarding Facebook content, celebrities mostly share photos (56.49%) on Facebook, followed by links (17.43%) and videos (14.23%) (Table 4). They are less likely to post statuses (11.85%). Most photos are related to their everyday activities and achievements, perhaps to build a positive image in front of their fans, which is a common practice among celebrities (Abidin, 2018). Interestingly, the percentage gap between photos and other content further suggests how important and favored sharing photos is to celebrities. Links shared on Facebook contain mainly news items (Baresch et al., 2011), suggesting that celebrities share a significant percentage of news links that might or might not be relevant to themselves. Further studies are required to explore what topical content celebrities share on Facebook and how they are relevant to their public identity.
Table 4
Facebook content posted by celebrities
| Types | n | % |
|---|---|---|
| Photo | 5,333 | 56.49 |
| Link | 1,646 | 17.43 |
| Video | 1,343 | 14.23 |
| Status | 1,119 | 11.85 |
| Total | 9,441 | 100 |
The posts from celebrities generate mainly like reactions (85.46%), followed by love (8.90%) (Table 5). Angry reactions (0.03%) are at the bottom of the list, meaning followers rarely prefer reacting angrily to the celebrities’ posts. One reason for the like reaction button’s highest value is that the five major reaction buttons were introduced only in 2016 (and the care reaction button in 2020), allowing Facebook users to express their diverse emotional reactions (Dewey, 2016). Before 2016, users had to express their emotions through only the like button.
Table 5
Descriptions of the followers’ interactions
| Interaction metrics | Interactionsa) | % | Mean | Standard deviation |
|---|---|---|---|---|
| Likes | 406.30 | 85.46 | 43,035.74 | 71,188.21 |
| Love | 42.32 | 8.90 | 4,482.08 | 18,120.84 |
| Comments | 10.97 | 2.31 | 1,161.80 | 4,642.80 |
| Shares | 6.32 | 1.33 | 669.08 | 4,282.89 |
| Haha | 3.60 | 0.76 | 381.40 | 2,288.23 |
| Sad | 2.50 | 0.53 | 264.57 | 4,468.34 |
| Wow | 2.02 | 0.42 | 213.48 | 821.79 |
| Care | 1.30 | 0.27 | 136.84 | 10,063 |
| Angry | 0.13 | 0.03 | 14.26 | 71.11 |
| Total | 475.44 | 100 |
Regarding expressiveness, the like button is ambiguous and difficult to interpret because no specific meaning can consistently be attributed to this button. For example, the like button can simultaneously express positivity and negativity regarding emotional valence and happiness, sadness, anger, surprise, disgust, and fear regarding specific emotional reactions (Giuntini et al., 2019). For such reasons, previous studies also excluded this button from interpretation (Jost et al., 2020); we did the same in this study.
Therefore, from the results, we can infer that the followers are most likely to express affection and liking and least likely to express disapproval and rage. Also, most of their reactions belong to positive valences, meaning followers are almost always positive about their celebrity icons. A higher percentage of comments suggests that followers express their emotions through reaction buttons and written text.
A contingency table shows that photos from celebrities generate the highest share of interactions (80.79%). Although links have a higher frequency than videos, they generate a lower share of interactions (5.22%) than video (10.55%) (Table 6). This suggests that photos have the highest frequency among all content and attract more followers. A point-biserial correlation also shows that photos and total reactions are positively associated (rpbis=0.290; p< 0.01), unlike others (Table 7). However, all other engagement measures except sharing have the highest percentages for photos; followers are more likely to share celebrities’ videos (59.2%) than photos or other content. This further suggests that although followers interact more with photos, they tend to circulate their icons’ videos more often.
Table 6
Different interactions in different types of content
| Types | Total interactionsa) |
% | Commentsb) | Sharesb) | Likesb) | Loveb) | Wowb) | Hahab) | Sadb) | Angryb) | Careb) |
|---|---|---|---|---|---|---|---|---|---|---|---|
| Photo | 384.13 | 80.79 | 76.0 | 37.0 | 82.1 | 78.4 | 80.6 | 64.2 | 77.1 | 70.1 | 72.8 |
| Video | 50.17 | 10.55 | 13.2 | 59.2 | 8.9 | 17.8 | 13.1 | 16.1 | 8.0 | 19.7 | 24.1 |
| Link | 24.80 | 5.22 | 4.2 | 2.0 | 5.6 | 1.7 | 3.4 | 17.6 | 0.6 | 8.0 | 1.6 |
| Status | 16.35 | 3.44 | 6.7 | 1.8 | 3.5 | 2.1 | 2.9 | 2.1 | 14.3 | 2.2 | 1.5 |
Table 7
Associations between content types and reactions
| Types | Total interactions |
Comments | Shares | Likes | Love | Wow | Haha | Sad | Angry | Care |
|---|---|---|---|---|---|---|---|---|---|---|
| Photo | 0.290** | 0.098** | -0.062** | 0.312** | 0.109** | 0.126** | 0.026* | 0.025* | 0.055** | 0.045** |
| Video | -0.062** | -0.007 | 0.201** | -0.093** | 0.025* | -0.008 | 0.009 | -0.011 | 0.031** | 0.038** |
| Link | -0.191** | -0.087** | -0.064** | -0.189** | -0.102** | -0.096** | 0.001 | -0.026* | -0.050** | -0.057** |
| Status | -0.154** | -0.040** | -0.048** | -0.157** | -0.075** | -0.072** | -0.050** | 0.004 | -0.060** | -0.044** |
The number of shares is also positively associated with only videos (rpbis=0.201; p<0.01). After photos, links (17.6%) posted by celebrities receive more haha reactions, followed by videos (16.1%). It suggests that celebrities share links to something either hilarious or joyful, which is why fans express laughter, or something unacceptable, which is why fans express denial and mockery. Status updates (14.3%) from celebrities receive a higher percentage of sad reactions than videos (8%) and links (0.6%). It could be possible that celebrities’ Facebook statuses often contain tragic elements so that their fans express empathy. However, the association between status and sad reactions seems statistically not significant. Of the significant associations, the correlation coefficient analysis shows that only photos’ association with likes is moderate (rpbis=0.312; p<0.01), while most of the other associations are either weak or negligible (Akoglu, 2018). We presume that such smaller correlation values are the consequences of outliers’ influence, a common phenomenon when a study deals with a large amount of dispersed data (Grubbs, 1969; Penn State University, 2021).
5. CONCLUSION
As stated in the previous section, this exploratory study investigated the most popular Facebook celebrities’ professions and genders, their popularity patterns, shared content types, and followers’ reactions to them. The results highlight that actors, music-related celebrities, and sports stars are the most popular celebrities both in Bangladesh and around the globe. Male celebrities are more in numbers, but female celebrities are more popular than male celebrities in average followers. It seems some professions of celebrities are gender specific. Celebrities mostly share photos on Facebook, perhaps to recreate their public image and provide regular updates to their followers. The followers also engage the most in photos with affection and surprise reactions. Overall, followers’ reactions to celebrities’ Facebook content are highly positive.
Beyond these novel findings, this study is different from others, at least in two ways. First, to the best of our knowledge, no prior celebrity-related studies in Bangladesh have focused on celebrities’ Facebook content, popularity patterns, and followers’ reactions. Previous studies in other contexts, except for a few (Bennett, 2014; Ledbetter & Redd, 2016) were mostly concerned with celebrities’ relations to various psychological and economic issues (Aljasir, 2019; Arefin, 2018; Chung & Cho, 2017; Schimmelpfennig, 2018; Shabnaz et al., 2019; Um, 2013). Second, no research dealt with celebrity culture and celebrity-fan communication in Bangladesh. In fact, Bangladesh, where social media and online celebrity culture are thriving, has been under-researched in celebrity studies.
Apart from these broad scholarly contributions, this study theoretically and methodologically contributes to information studies by analyzing Facebook’s interaction data. Although many may consider such data as less significant metadata, it provides essential insights regarding human communication. In information studies, the priority of such interaction data is increasing, and the present study contributes to this growing body of literature. Furthermore, global scholarship of information studies mainly focuses on social media usage in prominent countries, primarily from the Global North, leaving the Global South behind the curtain. This research is a response to this academic favoritism and knowledge gap.
Despite offering some novel insights and scholarly contributions, this study has a few limitations that call for further academic investigation. First, we presume that a larger sample would yield more reliable and generalizable results. However, this study analyzed a decent amount of data, which is larger enough than many previous studies considered. For example, Gergaud et al. (2012) examined only 100 celebrities for their category-based frequencies, which is smaller than the present study. Second, another limitation of this study is related to the data source. Although SocialBakers promised to provide comprehensive data to their clients, it has some limitations. This AI-based platform could not include some influential celebrities on its list. For example, Sohel Taj, a renowned politician with 1.1 million Facebook followers, was not on the list and should have been in the 80th position. Therefore, future studies should collect and analyze more inclusive data for more comprehensive results. Third, we analyzed interaction data for only a few celebrities, which may impact the generalizability of these results. Hence, we recommend not overgeneralizing the findings of RQ3 and RQ4. Lastly, this study only investigated the context of Bangladesh. The country’s unique communication infrastructure, online climate, and celebrity culture might differ from other countries, limiting the applicability of our findings to other contexts.
REFERENCES
, , , (2011) Friends who choose your news: An analysis of content links on Facebook International Symposium on Online Journalism, 1(2), 1-24 https://isoj.org/wp-content/uploads/2018/01/Baresch2011.pdf.
CrowdTangle (2021) Facebook https://www.crowdtangle.com/
(2016) A quick, no-nonsense guide to using Facebook's new reactions https://www.washingtonpost.com/news/the-intersect/wp/2016/02/24/a-quick-no-nonsense-guide-to-using-facebooks-new-reactions
, , , , (2020) Why buttons matter: Repurposing Facebook's reactions for analysis of the social visual International Journal of Communication, 14, 1564-1585 https://ijoc.org/index.php/ijoc/article/view/11657.
, , (2012) Success of celebrities: Talent, intelligence or beauty? Economics Bulletin, 32(4), 3120-3127 https://ideas.repec.org/a/ebl/ecbull/eb-12-00415.html.
, , (2020) Populism fuels love and anger: The impact of message features on users' reactions on Facebook International Journal of Communication, 14, 2081-2102 https://ijoc.org/index.php/ijoc/article/view/13400.
(2015) Critical pedagogical approaches to visual culture in Turkish preservice education Studies in Art Education, 56(4), 355-368 http://www.jstor.org/stable/45185098.
Pennsylvania State University (2021) Cautions about correlation and regression https://online.stat.psu.edu/stat800/lesson/cautions-about-correlation-and-regression
(2018) Impact of Facebook obsession among university students in Bangladesh International Journal of New Technology and Research, 4(2), 89-94 https://www.ijntr.org/download_data/IJNTR04020063.pdf.
(2020) Bangladesh: Celebrities, youth resolve to fight COVID-19 https://www.aa.com.tr/en/asia-pacific/bangladeshcelebrities-youth-resolve-to-fight-covid-19/1779034
StatCounter (2020) Social media stats Bangladesh https://gs.statcounter.com/social-media-stats/all/bangladesh