Public reactions to celebrity cancer disclosures via social media: Implications for campaign message design and strategy

Abstract

Objective:

The aim of this study was to analyse social media users’ reactions to a celebrity’s cancer announcement in order to inform future cancer-related campaigns.

Design:

A content analysis of Facebook users’ written responses to the actor Hugh Jackman’s 2013 post announcing his skin cancer diagnosis.

Setting:

Facebook’s application programming interface (API) software was used to compile all 14,534 comments posted by Facebook users under Jackman’s Facebook post between 21 November 2013 and 6 January 2014.

Method:

The Facebook API captured number of post likes and gender of the poster, while two trained coders also analysed the posts for emotional reactions, emoticon use and mentions of cancer-related behaviours.

Results:

Hope was the most common text-based emotional expression, and happy emoticons were the most used visual emotional expression in user comments in response to Jackman’s diagnosis. Posts mentioning detection behaviours were more likely to receive likes than those that did not mention them. Additionally, female users were significantly more likely to mention detection behaviours related to skin cancer than were male users.

Conclusion:

Celebrity cancer announcements may serve as de facto cancer awareness campaigns as well as highlight how to effectively craft coordinating strategic campaigns launched after a celebrity cancer disclosure.

Keywords

Audience involvement cancer communication emotion gender public communication campaigns social media

From Betty Ford’s 1974 announcement of her breast cancer surgery and later alcohol dependency (Lerner, 2006) to Patrick Swayze’s and Steve Jobs’ deaths from pancreatic cancer (Noar et al., 2013), media coverage of public personas facing cancer can focus the public’s attention on the disease with an intensity unmatched by typical news or campaign messages. Surveys, public health records, and search engine analytics have all captured the existence of post-celebrity-cancer-announcement effects on the public’s health-related behaviours (Noar et al., 2014). However, social media offer users more interactive platforms to respond to mass-mediated news of a celebrity’s cancer announcement.

For example, on 21 November 2013, actor Hugh Jackman posted a photo of himself on his Facebook wall with a large bandage covering his nose. Jackman, famous across the globe for his portrayal of the Marvel Comics’ superhero Wolverine in the X-Men movies, wrote a message alongside the photo for the 6.5 million Facebook users who had ‘liked’ his page: ‘Deb said to get the mark on my nose checked. Boy, was she right! I had basil [sic] cell carcinoma. Please don’t be foolish like me. Get yourself checked. And USE sunscreen!!!’ Within 6 weeks of this post, nearly 15,000 Facebook users had written comments back to Jackman. The interest in Jackman’s post about cancer mirrors larger trends in how people find and interact with health information: more than 7 in 10 (72%) adult Internet users have searched online for health information, while 1 in 4 (26%) of them have read or watched someone else’s health experience in the past year via digital platforms like social media (Fox, 2014).

Jackman’s cancer disclosure was selected as a case study to provide insight into how social media users respond to this form of direct communication with fans. With nearly 23 million people following Jackman’s Facebook page, his personal platform offers significant potential reach and connectivity with his fan base. In this new media landscape, social media provide opportunities to assess the public’s immediate responses to celebrity cancer disclosures, whereas previous work in this area has largely relied on retrospective self-report measures. User-generated comments posted shortly after notable events occur are important data sources because ‘they reflect spontaneous, unsolicited opinions not affected by study demand characteristics or researcher bias’ (Regan et al., 2014: 311). Online comments have been shown to be ecologically valid measures of audience reactions to health-related events, too (Holton et al., 2014).

Analysing social media users’ written responses to these events is also important because they provide insights as to how social media may be used to educate and motivate people to improve their own health. Public health advocates could examine social media responses to celebrity announcements to gauge the potential of strategic messages to ride the wave of attention cancer receives when linked with a popular celebrity. Moreover, comments made on social media can influence other users’ health-related perceptions (Shi et al., 2014), thereby magnifying the potential impact of public comments under celebrities’ posts about cancer. In short, social media use after a celebrity cancer disclosure can serve as a de facto campaign for promoting cancer prevention and detection behaviours.

Given that previous work has found that the public is very receptive to celebrities as spokespeople for disseminating health-related information (Berry et al., 2010), public health organisations could take advantage of these naturally occurring instances where celebrities insert themselves in issues of public health to launch simultaneous awareness and/or fundraising campaigns. However, understanding the nuances of audience responses to celebrity cancer disclosures via social media, including the emotional expressions social media users display, is vital for any campaign hoping to effectively craft and then target its message.

In order to better understand the phenomenon of increased public attention to cancer after a celebrity’s social media disclosure, this study employed a content analysis of online responses to Jackman’s Facebook post about his skin cancer diagnosis and removal. This method was used to investigate the interrelationships between social media users’ emotional reactions (as evidenced in the language they use to write a response to Jackman’s post), and the skin cancer–related information they include in their comments, as well as the potential for users of different genders to respond differently to celebrity cancer disclosures. The findings present provide valuable information for those wanting to understand how celebrity cancer disclosures influence the public, with the potential to motivate behaviour change that improves public health. They can also help inform theory regarding audience involvement with celebrities via social media posts related to health threats. Moreover, cancer-related organisations and public advocates could benefit from understanding the complex emotional reactions social media users have to celebrity cancer announcements as those will likely shape how they interact with subsequent cancer-related information.

Literature review

Users following celebrities on various social media platforms can cultivate mediated relationships through the two forms of audience involvement discussed below (i.e. parasocial interaction [PSI] and identification). A theoretical foundation for both of these forms of audience involvement aids in the understanding of how celebrity cancer announcements on social media can elicit strong audience reactions and also how they might promote health behaviour change.

PSI

Horton and Wohl (1956) first developed the concept of PSI. They argued that audiences develop relationships with mediated persona. Even though these relationships were one-way and non-reciprocal, they observed that there was an ‘illusion of intimacy’ between spectator and performer. Giles (2003) later defined PSI as happening when ‘we respond to a media figure as if he/she/it were a real person’ (p. 188). Eventually, instances of PSI can foster parasocial relationships with media personalities (Rubin and McHugh, 1987). Kosenko et al. (2016) found that women who had a parasocial relationship with Angelina Jolie, coupled with a cancer diagnosis within their family, were the mostly likely to consider genetic testing for cancer prevention.

Identification

J Cohen (2001) defines identification with mediated characters as ‘a mechanism through which audience members experience reception and interpretation of the text from the inside, as if the events were happening to them’ (p. 245). While PSI is the process of viewing a mediated persona as a friend or companion, identification is the process of the audience member’s identity merging with the media persona’s (Brown, 2015). When identifying, the audience member feels similar to a celebrity (Brown and Fraser, 2004). Identification has been a key mediator of behavioural outcomes related to news of celebrity health issues, from encouraging those who identified with Magic Johnson to reconsider their sexual health risk perceptions and practices (Basil, 1996; Brown and Basil, 1995) to motivating information seeking about pancreatic cancer in the wake of Steve Jobs’ death from the disease (Myrick et al., 2014).

Public responses to celebrities and cancer

Celebrities strategically capitalise on the affordances of social media by offering what appears to be backstage access: revealing personal information, creating a sense of intimacy, acknowledging fans, and using pop culture references to engage followers (Marwick and boyd, 2011). Celebrity disclosures of any illness, including cancer, can provide social media users with role models from whom they can learn about disease prevention, detection, and/or treatment behaviours (Brown, 2015). Moreover, social cognitive theory (Bandura, 2004) would argue that social modelling allows media consumers, including social media users, to vicariously learn from others, which in turn can improve their own self-efficacy if or when they encounter a similar health issue. For example, previous research has found that using local celebrity chefs who have direct experience with diabetes can help provide relatable personal testimony and increased self-efficacy to those at risk for the disease (Peek et al., 2014).

However, when a celebrity discloses a cancer diagnosis via social media, little empirical research has examined how their social media followers respond via this type of platform. Social media allow individuals, who may never have a chance to actually meet a celebrity in person, to now keep daily, even hourly, tabs on their favourite stars and even comment under those posts as if they had an actual relationship (Cohen and Tyler, 2016). In the context of a celebrity making a social media post about an illness, there is the potential for social media users to become motivated to take health-related actions themselves (e.g. Brown and Basil, 1995; Myrick et al., 2014). Yet, most research on public reactions to celebrity health announcements has focused on responses to news spread via traditional news outlets (Brown and Basil, 2010), where commenting on a celebrity’s social media announcement is more difficult and less likely to be seen by others in the commenter’s social network.

Additionally, this lack of knowledge about how social media users respond to celebrity social media disclosures of disease may portend missed opportunities to tailor coordinated strategic messages about cancer to the public during a time – immediately following news coverage of the celebrity disclosure – when the public’s interest is higher than normal. Examining how much of the social media conversation concerns the actual health topic that motivated the post would be an initial step in understanding how effective these de facto cancer awareness campaigns are in generating cancer-related conversations. Therefore, the following research questions are proposed:

RQ1. Of the Facebook comments made in response to a celebrity cancer disclosure, what percentage of those comments mentions a diagnosis of skin cancer (for the user or a close friend/family member)?

RQ2. Of the Facebook comments made in response to a celebrity cancer disclosure, what percentage of those comments mentions disease prevention or detection?

The role of emotions

In addition to assessing disease-related mentions in comments to a celebrity’s social media-based cancer disclosure, it is helpful to see what psychological factors are also associated with these comments as these concepts could help health campaign planners to better understand the current state of their target audiences. Emotions are a particularly important psychological factor to consider when studying how users react to celebrity cancer disclosures, because the disease is inherently emotion evoking (Mukherjee, 2010), and they are often better predictors of health-related behaviours than is mere knowledge about health threats (Myrick, 2015).

Emotions are inherently motivational (Roseman, 2011), that is, their adaptive function is to inspire us to take action based on our situation after we have appraised it. Appraisal theories of emotion posit that those actions motivated by emotions differ based on which specific emotions someone is experiencing (Lazarus, 1991). For example, Lazarus states that the action tendency of fear is to motivate people to avoid threat while that of sadness is to motivate people to reflect and avoid physical activity. In the context of celebrity cancer disclosures, user comments expressing certain emotions may indicate that the user is more or less inclined to take subsequent action to seek cancer information or change their own behaviour.

For instance, Myrick et al. (2013) found that young adults who identified with Steve Jobs were more likely to experience feelings of sadness after the Apple executive’s death from pancreatic cancer. This emotional reaction to news about Jobs’ death predicted interpersonal communication and information seeking related to Jobs and cancer. This study demonstrates that demographic similarity is not the only variable that fosters emotional reactions to public figures – the student sample reported an emotional response to Jobs’ death even though they were not near his age and unlikely to have cancer in the near future. As such, we ask the following research question regarding the role of emotions:

RQ3. What is the relationship between types of specific emotional expressions (i.e. hope, empathy, surprise, sadness, and the use of different emoticons) and mentions of skin cancer in Facebook comments made in response to a celebrity cancer disclosure?

Demographic differences

In another study that instead surveyed non-student adults, individuals who identified with Jobs after hearing about his pancreatic cancer death were more likely to experience feelings of worry, which in turn predicted interpersonal communication and information seeking related to cancer (Myrick et al., 2014). While younger individuals (i.e. college students) felt sadness instead of worry after hearing about Jobs’ cancer death, these two studies together reveal that different emotions may arise in different audiences after hearing about a celebrity cancer event. Therefore, studying multiple types of emotional responses to a celebrity cancer disclosure and assessing if or how they differ based on demographics could help public health advocates to strategically target their cancer-related messages in the wake of a well-publicised celebrity cancer disclosure.

Gender and health have long been intertwined, and women are often socialised to be more concerned about health and men to be less openly concerned about such issues (Annandale and Hunt, 1990). In an interview study, Sloan et al. (2010) found that men tend to deny any sort of direct or open interest in issues related to health, precisely because they are often viewed as feminine. The authors found that, instead, men may justify any sort of health-related action by reframing it as an action-taking orientation, sports activity, or appearance concern. While this may be the case outside of the media context, research has yet to find if this pattern of women and men speaking differently about health carries over into the social media arena, particularly if it is a male celebrity who has used the platform to announce an illness.

While many Facebook users do not include their ages in their posts, it is easy to acquire Facebook users’ gender through content scraping software. Moreover, health-related gender roles were directly implicated in Jackman’s Facebook disclosure; he mentions that his wife, Deb, is responsible for him seeking a diagnosis and receiving early treatment. This situation corresponds with traditional gender roles regarding health: women are the family caretakers and responsible for monitoring family members’ health while men often avoid healthcare (e.g. Annandale and Hunt, 1990). Moreover, previous research has established that women share their emotions with a wider range of people, both men and women, than do men (Eagly and Wood, 1991). We therefore propose the following two hypotheses:

H1. Female social media users will be more likely to emote in their responses to a celebrity cancer disclosure post on Facebook than will male social media users.

H2. Female users will mention skin cancer more often in their Facebook comments in response to a celebrity cancer disclosure than will male users.

Other users’ reactions to user-generated comments

Furthermore, emotional expressions may predict how other social media users respond to comments under the celebrity’s original cancer-related post. Other users’ responses are important because greater social media engagement with user-generated comments about cancer (e.g. likes and sharing of the content with others) could translate into stronger social norms about the importance of seeking information or talking to a physician about cancer. The emotional tone of a comment could influence how much other users engage with it. Previous work shows some support for this notion. In a study of members of a hashtag-based cancer community on the social media site Twitter, Myrick et al. (2016) found that tweets with expressions of negative emotions were less likely to be retweeted or favourited than tweets without a negative emotional expression. Other studies have found that positively toned digital content is more likely than negatively toned content to be shared (Berger and Milkman, 2012; Choi and Toma, 2014). The current study proposes that this relationship will likely hold true in the context of celebrity illness disclosure via social media, and therefore, we pose the following final hypothesis:

H3. Positive emotional expressions and mentions of skin cancer will be positively associated with number of likes on Facebook comments made in response to a celebrity cancer disclosure.

Methods

A content analysis was conducted to address the research questions and hypotheses. Facebook’s application programming interface (API) software was used to compile all 14,534 comments posted by Facebook users under Jackman’s Facebook post between 21 November 2013 and 6 January 2014. Posts written in languages other than English and posts containing foreign characters were removed, leaving 11,917 posts. Every third comment from this population was then selected to create a sample totalling 3,953 comments. The codebook included many emotional responses that could potentially be expressed by commenters after a celebrity cancer disclosure, including emoticons. The lead researcher proctored three training sessions with two coders using comments not included in the sample. Once the coders established adequate reliability, both coded 10% of the sample. Each variable exhibited acceptable intercoder reliability (see Table 1). Because of their low frequency in the intercoder sample (only appearing once or twice, if at all, in 400 posts), the following emotion-related variables were removed: anger, fear, sadness, humour, and anger emoticons.

Table 1.

Descriptions, intercoder reliabilities and frequencies of variables in the study.

Variable	Sample comments/descriptions	Cohen’s Kappa	Percent agreement	Frequency (out of 3,953)
Number of comment likes	Automatically coded by the Facebook API	–	–	0 likes = 3,484 1 likes = 395 2 likes = 53 3 likes = 14 4 or more likes = 7
Gender of commenter	Automatically coded by the Facebook API	–	–	Male = 857 Female = 3,078 Unidentified = 18
Surprise	Oh wow, OMG, Oh my!	0.80	97.7	220
Hope/inspiration	Hopeful, I hope …, hope you get better, you’ve inspired me to …	0.69	93.2	1,670
Love	I love you! love ya	0.77	97.7	234
Empathy	I feel your pain, awww, poor thing! Xoxo, You okay? How are you doing/feeling?	0.79	94.5	499
Happy emoticon	:):-););-):-D	0.89	98.2	409
Love emoticon	<3, :*	0.97	99.7	172
Sad emoticon	:( :-( :’( D:	0.93	99.5	160
Surprise emoticon	:-O, o.0, o_0	0.86	99.7%	15
Skin cancer involvement A	Poster has had some type of skin cancer or mole removal	0.88	99.0%	215
Skin cancer involvement B	Poster mentions that family or friend has had/has skin cancer	0.91	99.5%	108
Prevention	Posts saying people should wear sunscreen, avoid the sun, avoid tanning beds, or do any other type of action aimed at preventing skin cancer	0.61	96.7%	209
Detection	Posts saying people should go to the dermatologist or perform self-checks on their skin	0.66	98.0%	217

API: application programming interface.

Measures

The Facebook API collected data on the gender of each Facebook user in the sample (78.2% female). Eighteen individuals did not report their gender to Facebook, though. The API also collected the number of likes each comment received (M = 0.18, standard deviation [SD] = 0.62). All other human-coded measures were binomial (i.e. coded as ‘0’ for not present and ‘1’ for present; see Table 1).

Plan of analysis

Research questions that address frequency of different types of content will be tested using Z tests for proportions, while research questions and hypotheses that ask about the interrelationships between two variables will be assessed using chi-square analyses. For H2 and H3, which both involve multiple statistical tests, the hypothesis will be supported only if more than half of the tests are significant in the predicted way.

Results

The first two research questions asked about the frequency of skin cancer–related comments in response to Jackman’s cancer disclosure. The results revealed that 5.4% of comments mentioned having skin cancer or a mole removal, 2.7% mentioned a family member or friend with skin cancer, 5.3% advocated for prevention activities, and 5.5% recommended detection activities.

Additional descriptive results revealed that hope was the most common text-based emotional expression in the sample, appearing in 42.3% of all posts (see Table 1). Z tests for proportions revealed that hope occurred in the posts significantly more often than empathy (Z = 31.30, p < .001), love (Z = 41.72, p < .001), or surprise (Z = 42.35, p < .001). Beyond emotion-laden text, nearly one in five posts (n = 294) included an emoticon, with happy emoticons the most common (found in 10.4% of posts). Happy emoticons appeared significantly more often than did love emoticons (Z = 10.29, p < .001), sad emoticons (Z = 10.92, p < .001), and surprise emoticons (Z = 20.17, p < .001). There was no significant difference between the amount of love emoticons and sad emoticons (Z = 0.89, ns), but the surprise emoticons occurred less frequently than either love emoticons (Z = 11.80, p < .001) or sad emoticons (Z = 11.08, p < .001).

The third research question asked about the relationship between emotional expressions and mentions of having skin cancer or knowing a friend or family member with the disease. Chi-square analyses found that comments expressing hope, χ²(df = 1) = 17.02, p < .001; love, χ²(df = 1) = 4.26, p < .05; or love emoticons, χ²(df = 1) = 10.56, p < .01, were significantly less likely to mention a skin cancer diagnosis. However, comments expressing sadness were significantly more likely to mention a diagnosis (see Table 2).

Table 2.

Relationships between personal involvement with skin cancer being mentioned and emotional expressions.

Variable present	Skin cancer (self or family/friend) mentioned		χ² value (Phi value)
Variable present	No	Yes	χ² value (Phi value)
Surprise			2.50 (−.03)
No	3,426	307
Yes	209	11
Hope/inspiration			17.02^*** (−.07)
No	2,064	219
Yes	1,571	99
Love			4.26^* (−.04)
No	3,411	308
Yes	224	10
Empathy			0.01 (.00)
No	3,175	279
Yes	460	39
Happy emoticon			1.51 (−.02)
No	3,252	292
Yes	383	26
Love emoticon			10.56^** (−.05)^a
No	3,465	316
Yes	170	2
Sad emoticon			2.54 (−.03)
No	3,482	311
Yes	153	7
Surprise emoticon			0.45 (−.02)^a
No	3,620	318
Yes	15	0

At least one cell had an expected count less than 5; ^*p < .05, ^**p < .01, ^***p < .001 for Pearson two-sided χ² test with Yates continuity correction, df = 1.

H1 posited that emotional expressions would differ by gender, with female users more like to elicit emotion in response to Jackman’s cancer announcement. Chi-square analyses (which did not include the 18 individuals with unreported genders) showed that women were significantly more likely to express love and empathy than were men. Women were also more likely to use love and sadness emoticons than were men (see Table 3), providing support for the first hypothesis. H2 suggested that users of different genders would discuss skin cancer differently; specifically that women would address it more frequently. Chi-square analyses revealed women were significantly more likely to mention detection behaviours than were men, but there were no gender differences in mentions of prevention or skin cancer diagnoses (see Table 3). As only one of the three tests demonstrated women were more likely to mention the behaviour than men, this hypothesis was not supported.

Table 3.

Relationships between user gender and content variables.

Variable present	User gender		χ² value (Phi)
Variable present	Female	Male	χ² value (Phi)
Surprise			1.55 (−.02)
No	2,898	817
Yes	180	40
Hope/inspiration			0.46 (−.01)
No	1,768	504
Yes	1,310	353
Love			23.15^*** (−.08)
No	2,865	836
Yes	213	21
Empathy			23.55^*** (−.08)
No	2,647	791
Yes	431	66
Happy Emoticon			0.34 (−.01)
No	2,753	773
Yes	325	84
Love Emoticon			27.62*** (−.09)
No	2,916	848
Yes	162	9
Sad Emoticon			4.77* (−.04)
No	2,942	834
Yes	136	23
Surprise Emoticon			0.23^a (−.01)
No	3,065	855
Yes	13	2
Skin cancer (self or family/friend) mentioned			2.59 (−.03)
No	2,819	800
Yes	259	57
Prevention mention			2.31 (−.03)
No	2,906	821
Yes	172	36
Detection mention			4.52^* (−.04)
No	2,896	823
Yes	182	34

At least one cell had an expected count less than 5; ^*p < .05, ^***p < .001 for Pearson two-sided χ² test with Yates continuity correction, df = 1.

These analyses of mentions of prevention and detection behaviours motivated a post-hoc analysis of the frequency of mentions of specific types of these behaviours. Two coders reviewed the data set and separated the prevention and detection behaviours into more specific categories. Consensus was reached between the two on each comment. Of the post-hoc prevention categories, the largest were mentions of sunscreen and SPF (123 out of 209 total prevention comments). The second highest category of comments was the 51 responses that discussed the need to stay safe or be more careful while in the sun. Another 19 mentioned that awareness can help with skin cancer prevention, while 16 comments discussed making a preventative appointment with a healthcare provider. There were 16 other comments that discussed the need to wear a hat. Less often mentioned were changing one’s diet (6) and avoiding tanning (3). Of the 217 comments that mentioned detection, 104 emphasised regular check-ups with a medical provider, 85 mentioned general ‘early detection’, and 46 discussed the need to heed warning signs and any changes in skin.

The final hypothesis (H3) proposed that positive emotional expressions and mentions of skin cancer would garner more likes on Facebook. The number of likes variable (M = 0.15, SD = 0.50) was first checked for distribution. High skew (7.27) and kurtosis (96.58) indicated that it was not normally distributed. Therefore, the variable was dichotomised such that posts with no likes were coded as 0 and posts with at least one like were coded as 1. Chi-square analyses revealed that comments with sad emoticons were significantly less likely to receive a like than those without a sad emoticon (see Table 4). No other emotional expression was significantly associated with receiving likes, and H3 was not supported. However, the mention of detection behaviours was significantly associated with a greater likelihood of receiving likes.

Table 4.

Relationships between number of post likes and content variables.

Variable present	Post Likes		χ² value (Phi)
Variable present	None	1 or more	χ² value (Phi)
Surprise			2.63 (−.03)
No	3,283	450
Yes	202	18
Hope/inspiration			2.17 (.02)
No	2,028	255
Yes	1,457	213
Love			0.00 (.00)
No	3,279	440
Yes	206	28
Empathy			3.48 (−.03)
No	3,032	422
Yes	453	46
Happy emoticon			0.67 (.01)
No	3,130	414
Yes	335	54
Love emoticon			0.00 (.00)
No	3,334	447
Yes	151	21
Sad emoticon			4.45* (−.04)
No	3,335	458
Yes	150	10
Surprise emoticon			0.05 ^a (−.01)
No	3,471	467
Yes	14	1
Skin cancer (self or family/friend) mentioned			1.54 (.02)
No	3,212	423
Yes	273	45
Prevention mentioned			1.10 (.02)
No	3,306	438
Yes	179	30
Detection mentioned			6.52* (.04)
No	3,306	430
Yes	179	38

At least one cell had an expected count less than 5; ^*p < .05, ^**p < .01, ^***p < .001 for Pearson two-sided χ² test with Yates continuity correction, df = 1.

Discussion

The results presented here offer a number of interesting findings with implications for health educators and communication scholars as well as for campaign message designers to consider. First, the data revealed that social media users did mention the disease itself – cancer – in response to a celebrity’s social media disclosure of cancer. However, the rate of mentions of cancer out of all comments is not high (between 2.7% and 5.5% of comments, depending on the specific type of mention). Regardless, given that thousands of people comment and likely tens of thousands of other users saw these comments, they still serve as a type of virtual cancer awareness campaign. Social media and popular culture websites, such as celebritydiagnosis.com – a site dedicated to educating the public about the diseases that appear under the media spotlight once they are associated with a celebrity, can therefore serve as quality platforms for health education efforts. Additionally, strategic health communication practitioners can themselves examine user comments to celebrity posts about disease when crafting campaigns to leverage the mediated coverage of celebrity health announcements. Moreover, existing research shows that interpersonal sources of cancer information can be particularly important for motivating individuals to take action to address their own cancer risks (Ogata Jones et al., 2006).

As such, if social media users perceive a celebrity to be a friend – as often happens through repeated media exposure that results in parasocial relationships (Dibble et al., 2016) – then this cancer information coming from a celebrity’s social media page may likewise motivate many users to take action. Equally possible, social media users who see in their Facebook timeline or homepage that other members of their online network have commented under a celebrity’s post about cancer may likewise consider it an interpersonal conversation or source of information about cancer, which in turn could motivate information seeking and possibly behaviour. While the present study was a content analysis that cannot assess causality of the potential effects of the content on users, future work could test these ideas using longitudinal survey or experimental designs.

The results about the most common specific emotional expressions found in response to Jackman’s online cancer disclosure offer insights regarding the interplay of online celebrity messaging, social media culture and health communication processes. The results revealed that hope was the most common emotional expression in user comments under a celebrity cancer disclosure via Facebook. This suggests that any coordinating strategic campaign may benefit from matching this upbeat emotional tone. Understanding emotional expressions in response to a celebrity health–related post is valuable because, according to appraisal theory, emotional states often predict behavioural intentions and behaviours, with different emotions motivating people to approach goals (e.g. hope) and others motivating them to avoid threats (e.g. fear) (Lazarus, 1991). The frequent occurrence of hope in user comments regarding Jackman’s cancer suggests that these users may be open to additional cancer information that could help them achieve goals of individual and communal health. Message designers could place targeted ads on social media that match this state of a positive outlet for the future and readiness to work toward goals.

Additionally, the fact that hope was a highly prevalent expression in the user comments overlaps with the literature on face-to-face and computer-mediated social support regarding cancer, where hope is often the dominant emotional expression (Helgeson and Cohen, 1996; Kim et al., 2011; Klemm and Wheeler, 2005). Social media users who took the time to respond to Jackman’s post about cancer treated him like a friend, offering positive thoughts and support while avoiding expressions of negativity. Social media appears to allow users to actively respond to health news and to come to the aid of a ‘friend’, even if that individual is a celebrity.

Another contribution of this study was to examine gender differences within social media users’ responses to a celebrity cancer disclosure. Women were more likely to express empathy and love with regard to Jackman’s post, even though statistically skin cancer is more prevalent in men (Centers for Disease Control and Prevention, 2013). One possible explanation for this finding is that gender stereotypes of women as caring and the family caretaker replicated themselves in the social media environment (Annandale and Hunt, 1990), which would also account for the findings that women were more likely to mention skin cancer detection in their posts. Another potential explanation is that copious media coverage of women who tan and then get skin cancer could be leading to misperceptions of actual gender-based susceptibility to the disease. Additional research is needed to test such suppositions.

The results also revealed that comments mentioning detection behaviours were more likely to be liked by other Facebook users than were those that did not mention detection. This suggests that social media users are receptive to reading health promotion–related information from other users and that celebrity cancer disclosures can result in temporary spike in the health-related information viewed by social media users. The post-hoc analysis also demonstrated that wearing sunscreen and seeing a doctor were the most frequent prevention and detection behaviours mentioned, respectively, suggesting that these are the two behaviours social media users are more knowledgeable about. Strategic campaigns could aim to educate social media users about lesser-known but likewise important behaviours, too, like wearing a hat and self-checks of the skin.

Additionally, the results suggest that posts with sad emoticons are less likely to be liked by other users. This finding implies that cancer-related organisations may want to emphasise something other than their sadness/sad emoticons when discussing celebrity cancer disclosures if they want users to positively engage with their online content.

While this study provides valuable empirical insights, these results come with limitations. Because the focus was solely on Jackman and skin cancer, this study took the form of a case study and the findings are not generalisable to all celebrities with any health condition. Furthermore, content analyses can only present a picture of what is present or absent in content. This methodology, unlike experimental work, is not appropriate for ascertaining effects on audiences. Nonetheless, these data are valuable initial steps in supporting the potential for theoretical foundations, like audience involvement and subsequent emotional reactions rooted in appraisal theory, to be useful tools in predicting how social media users interact with celebrity illness disclosures via social media, and how those interactions may influence later discussions, interactions with future health messages, and/or changes in actual health behaviours.

In summary, the data presented here reveal that social media users discuss cancer and express multiple types of emotions in response to a celebrity cancer disclosure. Moreover, they provide guidance for tailoring future cancer-related social media campaigns based on differences in how users of different genders respond to news of celebrity cancer. These data add to existing work on the importance of news coverage of celebrity cancer incidents by analysing social media data that represent the public’s immediate, authentic responses to the situation. Future work can build on these findings to take advantage of the huge wave of attention given to the disease immediately following celebrity cancer disclosures to develop more effective awareness campaigns and even interventions.

Footnotes

Funding

The author(s) received no financial support for the research, authorship and/or publication of this article.

References

Annandale

Hunt

(1990) Masculinity, femininity and sex: An exploration of their relative contribution to explaining gender differences in health. Sociology of Health & Illness 12(1): 24–46.

Bandura

(2004) Health promotion by social cognitive means. Health Education & Behavior 31(2): 143–164.

Basil

(1996) Identification as a mediator of celebrity effects. Journal of Broadcasting & Electronic Media 40(4): 478–495.

Berger

Milkman

(2012) What makes online content viral? Journal of Marketing Research 49(2): 192–205.

Berry

Witcher

Holt

. (2010) A qualitative examination of perceptions of physical activity guidelines and preferences for format. Health Promotion Practice 11: 908–916.

Brown

(2015) Examining four processes of audience involvement with media personae: Transportation, parasocial interaction, identification, and worship. Communication Theory 25(3): 259–283.

Brown

Basil

(1995) Media celebrities and public health: Responses to ‘Magic’ Johnson’s HIV disclosure and its impact on AIDS risk and high-risk behaviors. Health Communication 7(4): 345–370.

Brown

Basil

(2010) Parasocial interaction and identification: Social change processes for effective health interventions. Health Communication 25(6–7): 601–602.

Brown

Fraser

(2004) Celebrity identification in entertainment-education. In: Singhal

Cody

Rogers

. (eds) Entertainment-Education and Social Change: History, Research, and Practice. New York: Routledge, pp. 97–115.

10.

Centers for Disease Control and Prevention (2013) Skin cancer. Available at: http://www.cdc.gov/cancer/skin/ (accessed 1 September 2013).

11.

Choi

Toma

(2014) Social sharing through interpersonal media: Patterns and effects on emotional well-being. Computers in Human Behavior 36: 530–541.

12.

Cohen

Tyler

(2016) Cyberpsychology, Behavior, and Social Networking 19(5): 342–346. doi:10.1089/cyber.2015.0657.

13.

Cohen

(2001) Defining identification: A theoretical look at the identification of audiences with media characters. Mass Communication and Society 4(3): 245–264.

14.

Dibble

Hartmann

Rosaen

(2016) Parasocial interaction and parasocial relationship: Conceptual clarification and a critical assessment of measures. Human Communication Research 41: 21–44.

15.

Eagly

Wood

(1991) Explaining sex differences in social behavior: A meta-analytic perspective. Personality and Social Psychology Bulletin 17: 306–315.

16.

Fox

(2014) The Social Life of Health Information. Washington, DC: Pew Research Center’s Internet & American Life Project.

17.

Giles

(2003) Media Psychology. Mahwah, NJ: Lawrence Erlbaum Associates.

18.

Helgeson

Cohen

(1996) Social support and adjustment to cancer: Reconciling descriptive, correlational, and intervention research. Health Psychology 15(2): 135–148.

19.

Holton

Lee

Coleman

(2014) Commenting on health: A framing analysis of user comments in response to health articles online. Journal of Health Communication 19(7): 825–837.

20.

Horton

Wohl

(1956) Mass communication and para-social interaction. Psychiatry: Journal for the Study of Interpersonal Processes 19: 215–229.

21.

Kim

Han

Shah

. (2011) Predictors of supportive message expression and reception in an interactive cancer communication system. Journal of Health Communication 16(10): 1106–1121.

22.

Klemm

Wheeler

(2005) Cancer caregivers online: Hope, emotional roller coaster, and physical/emotional/psychological responses. Computers, Informatics, Nursing 23(1): 38–45.

23.

Kosenko

Binder

Hurley

(2016) Celebrity influence and identification: A test of the Angelina effect. Journal of Health Communication 21(3): 318–326.

24.

Lazarus

(1991) Emotion and Adaptation. New York: Oxford University Press.

25.

Lerner

(2006) When Illness Goes Public: Celebrity Patients and How We Look at Medicine. Baltimore, MD: Johns Hopkins University Press.

26.

Marwick

boyd

(2011) To see and be seen: Celebrity practice on Twitter. Convergence: The International Journal of Research into New Media Technologies 17(2): 139–158.

27.

Mukherjee

(2010) The Emperor of All Maladies: A Biography of Cancer. New York: Scribner.

28.

Myrick

(2015) The Role of Emotions in Preventative Health Communication. Lanham, MD: Lexington Books.

29.

Myrick

Holton

Himelboim

. (2016) #Stupidcancer: Exploring a typology of social support and the role of emotional expression in a social media community. Health Communication 31: 596–605.

30.

Myrick

Noar

Willoughby

. (2014) Public reaction to the death of Steve Jobs: Implications for cancer communication. Journal of Health Communication 19(11): 1–18.

31.

Myrick

Willoughby

Noar

. (2013) Reactions of young adults to the death of Apple CEO Steve Jobs: Implications for cancer communication. Communication Research Reports 30(2): 115–126.

32.

Noar

Ribisl

Althouse

. (2013) Using digital surveillance to examine the impact of public figure pancreatic cancer announcements on media and search query outcomes. JNCI Monographs 2013(47): 188–194.

33.

Noar

Willoughby

Myrick

. (2014) Public figure announcements about cancer and opportunities for cancer communication: A review and research agenda. Health Communication 29(5): 445–461.

34.

Ogata Jones

Denham

Springston

(2006) Effects of mass and interpersonal communication on breast cancer screening: Advancing agenda-setting theory in health contexts. Journal of Applied Communication Research 34: 94–113.

35.

Peek

Ferguson

Roberson

. (2014) Putting theory into practice: A case study of diabetes-related behavioral change interventions on Chicago’s south side. Health Promotion Practice 15: 40S–50S.

36.

Regan

Shan

McConnon

. (2014) Strategies for dismissing dietary risks: Insights from user-generated comments online. Health, Risk & Society 16(4): 308–322.

37.

Roseman

(2011) Emotional behaviors, emotivational goals, emotion strategies: Multiple levels of organization integrate variable and consistent responses. Emotion Review 3(4): 434–443.

38.

Rubin

McHugh

(1987) Development of parasocial interaction relationships. Journal of Broadcasting & Electronic Media 31(3): 279–292.

39.

Shi

Messaris

Cappella

(2014) Effects of online comments on smokers’ perception of antismoking public service announcements. Journals of Computer-Mediated Communication 19: 975–990.

40.

Sloan

Gough

Conner

(2010) Healthy masculinities? How ostensibly healthy men talk about lifestyle, health and gender. Psychology & Health 25: 783–803.