Sharing beyond Slacktivism: the effect of socially observable prosocial media sharing on subsequent offline helping behavior

ABSTRACT New forms of youth social and political participation have been termed ‘Slacktivism’ – low-cost online forms of social engagement that decrease subsequent offline participation. Previous experimental work has provided support for a ‘Slacktivism effect,’ but it is unclear if this theoretical model applies to youth media sharing on social networking sites. This study uses a novel sharing simulation paradigm to test the effect of publicly vs. anonymously sharing a social cause video on subsequent willingness to engage in offline helping behavior. Results show that publicly (as compared to anonymously) sharing a selected video on one’s own Facebook wall led to a greater willingness to volunteer for an issue-related cause. Participants’ existing use of social media for engagement in social issues/causes moderated the effect, such that only participants low in use of social media for social engagement were susceptible to the sharing manipulation. Implications for reconceptualizing media sharing as a unique form of online participation beyond ‘Slacktivism’ are discussed.

Researchers exploring the psychological underpinnings of a possible 'Slacktivism effect' use impression management (Kristofferson, White, & Peloza, 2014) as a mechanism through which online action may inhibit offline action. There is evidence that some low-cost online actions (e.g., signing petitions or joining Facebook groups) are token displays of support that inhibit subsequent 'meaningful action' (e.g., Kristofferson et al.). However, it is less clear if this inhibitory process applies to online acts that appear 'easy' and tokenistic, but in fact entail a great deal of social investment. For example, the ability to share links, videos, articles, and memes is a key affordance of popular social networking sites (SNS) and has been theorized as a communicative practice with its own social and political purposes (Halupka, 2014). A piece of publicly shared media not only provides 'token' support, but also expresses a specific point of view for others to scrutinize (Thorson, 2013).
Uses and gratification theory, widely used in the study of SNS (e.g., LaRose & Eastin, 2004), also suggests that individual motivations and usage patterns play an important role in the way users navigate online social contexts. We contest the narrative that all forms of online activism reflect self-indulgent 'Slacktivism' by testing the possibility that publicly (as compared to anonymously) sharing a social cause video within a heterogeneous online social network may actually increase subsequent offline helping behavior.

Paradigms of youth (dis)engagement
Despite evidence of a positive relationship between social media use and the development of social capital (Brandtzaeg, 2012;Steinfield, Ellison, & Lampe, 2008) and political engagement (Loader, 2007;, concerns remain about the antisocial consequences of technology (Buckingham, 2006;Turkle, 2011). Accordingly, Bennett (2008) identifies two paradigms for understanding youth civic engagement in the digital age: The Engaged Youth Paradigm, in which young people respond to the erosion in the credibility of government and social institutions with new creative acts of expression and symbolic participation, and The Disengaged Youth Paradigm, which emphasizes the role of the Internet in the privatization of the public sphere and the decline of traditional forms of engagement (e.g., voting and joining civic groups).
Slacktivism emerges as a label from this latter paradigm, which coincides with a popular discourse portraying Millennials as self-centered and morally lazy (Poindexter, 2012;Robertson, 2014). Though there is strong empirical evidence of a positive relationship between youth social media use and political engagement , these findings must be considered in the broader 'civic context' in which young people encounter opportunities for engagement (Zaff, Malanchuk, & Eccles, 2008). Much of youth civic life is shaped by interactions with peers, parent-modeling, and cultural context. Social background variables (education in particular) emerge as strong determinants of volunteering (Smith, 1994).
Of particular relevance to the Slacktivism debate is recent work on rising generational levels of narcissism, a personality characteristic negatively associated with prosocial behavior (Twenge & Campbell, 2009). Twenge, Campbell, and Freeman (2012) find that a shift toward extrinsic values (fame and image) for Millenials correlates with declining interest in learning about and engaging in social problems. Contemporary theorizing of older forms of offline youth engagement must now consider the extent to which prosocial actions are observable by others online. Work on a Slacktivism effect has subsequently focused on social observability as a key mechanism through which online action may inhibit offline action.

Psychological models of Slacktivism
Slacktivism has been defined as, 'a willingness to perform a relatively costless, token display of support for a social cause, with an accompanying lack of willingness to devote significant effort to enact meaningful change' (Kristofferson et al., 2014(Kristofferson et al., , p. 1149. Kristofferson et al. test a theoretical model in which token displays of support (e.g., signing a petition or joining a Facebook group) that are public result in decreased helping behavior (via impression management), while token displays that are private result in increased helping behavior (via internal self-consistency). Impression management theory suggests that behavior is often driven by the discrepancy between one's current and desired images (Leary & Kowalski, 1990). Token displays of support may reduce such a discrepancy, thereby rendering subsequent action unneccssary. Theories of moral balancing predict that performing a prosocial action can free individuals from feeling responsible for future helping, because past moral behavior provides 'license' to behave less morally in the present (Merritt, Effron, & Monin, 2010). Both of these theories suggest that engaging in socially observable displays of support may temporarily dampen normal moral or social pressure to engage in future prosocial behavior.
Both cognitive dissonance and the foot-in-the-door technique research suggest that gaining an initial commitment to help increases the likelihood of subsequent helping (Festinger, 1962;Freedman & Fraser, 1966). When compared to publicly observable token displays, private displays of support lead to greater willingness to help by increasing pressure for participants to remain consistent with their private actions and beliefs (Kristofferson et al., 2014). Of central importance to the present study is Kristofferson et al.'s finding that regardless of their observability, token acts of support for a cause increased subesquent willingness to help when participants felt value alignmentcognitive congruency between a cause's values and their own.
Early studies on Slacktivism suggest two distinct sources of cognitive dissonance that individuals are motivated to reduce: external dissonance based on the need to manage one's social impression and internal dissonance based on the need to remain true to one's values. What remains unclear is how these sources of dissonance play out in online prosocial media sharing, where impression management concerns are ever-present and self-expression (through sharing) is a personally meaningful act. For many social media users, sharing content online may involve concerns about both social observability and value alignment with a social cause. In such situations, sharing content publicly may do little to relieve cognitive dissonance or inhibit the social pressure to engage in subsequent helping behavior.

Sharing on SNS
The ability to share multimedia content with one's social networks is a key affordance provided by SNS such as Facebook, Twitter, and Instagram (boyd & Ellison, 2007;Brandtzaeg, Lüders, & Skjetne, 2010). Digital sharing is used for information seeking (Lee & Ma, 2012), social interaction (Papathanassopoulos, 2011), and political engagement (Vitak et al., 2010). Users have diverse motivations for sharing (see Goh, Ang, Chua, & Lee, 2009), suggesting that sharing is an act that is beyond tokenistic. When young people share a video from a cause they care about on their Facebook wall, they are engaging in both impression management and a complex performance of identity with perceived social consequences. Online sharing is often viewed as low-cost participation (because it is fast and easy to accomplish), but sharing social and political views online is in fact a highly delicate matter. College students approach sharing with concerns about self-presentation and the social repercussions of their posts (Thorson, 2013) and young people balance their desire to express themselves with the knowledge that they are under constant scrutiny (Davis, 2012), a process Xenos et al. (2014) call a 'personalized politics of self-actualization and expressive engagement' (p. 155). On the other hand, online environments can often provide less socially threatening contexts for opinion expression (Ho & McLeod, 2008). Little empirical work has investigated the psychological factors underlying the interplay between self-presentation concerns and public media sharing, despite the latter being a central affordance of many prominent SNS.

The present research
We suggest that the consequences of publicly sharing media in heterogeneous online social networks are quite different from those of wearing a button or signing a petition. In dynamic online environments, individuals are constantly monitoring social feedback regarding their self-presentation (Rosenberg & Egbert, 2011) and are highly aware that its consistency will be evaluated (particularly in the political realm; Davis, 2012). Specifically, media sharing on SNS entails implicitly claiming a position or expressing an opinion that is at once publicly visible and easily stored for comparison to past and future behaviors. Under such conditions, users are never able to substantially reduce the need to manage their impression. Therefore, the mechanism that motivates willingness to engage in future offline helping behavior should be the desire to reduce cognitive dissonance by remaining consistent with one's values as publicly stated on SNS.
Thus, we propose that sharing prosocial media in online social networks is an act of self-expression, which can increase the willingness to take offline action. We test this proposition experimentally using a novel sharing simulation paradigm, in which participants are randomly assigned to share a social cause video, either publicly (with their personal Facebook network) or anonymously (with all of Facebook). 1 We then measure willingness to help an issue-related cause offline. Participants are undergraduates at a large public Midwestern university, a demographic that is among the most likely to engage in volunteering (Smith, 1994). To examine how individual differences beyond education relate to public sharing and subsequent offline support, we control for variables known to influence willingness to engage in prosocial activity (e.g., previous prosocial behavior, and self-efficacy). We hypothesize the following: H1: Participants who publicly share a social cause video will report greater willingness to volunteer for the issue-related cause than participants who share anonymously.
H2: Participants who publicly share a social cause video will report greater willingness to donate to the issue-related cause than participants who share anonymously.
If Slacktivism primarily involves online behavior, our theoretical model must account for individual differences in social media usage patterns and motivations, in particular, the extent to which one uses social media to engage in social/political causes. The present study explores the possible relevance of social media use for social engagement with the following research question: RQ1: Will reporting the use of social media for engaging in social causes moderate the effect of public sharing on willingness to volunteer?

Video stimuli
To identify social cause videos most likely to be shared by participants in our sample, we reviewed the 100 most viewed nonprofit videos on YouTube (Author, in preparation) and selected 3 videos from organizations working in areas previously identified as engaging young people in prosocial activity (Ho, Clarke, & Dougherty, 2015;Kristofferson et al., 2014). Each video featured a personal story of a single affected individual (e.g., child or animal) and had received more than 1 million views (M = 5,901,085, SD = 4,235,587). 2 The videos were similar in length (M = 2.6 min, SD = 0.39 min) and were presented in a neutral player without title, viewing statistics, or comments.

Procedure
Sharing simulation paradigm. To maximize ecological validity while maintaining experimental control and participant privacy, we devised a novel sharing simulation paradigm. Prior to arrival at the laboratory, participants were randomly assigned to one of two conditions: anonymous sharing (n = 91), in which they ostensibly shared a video anonymously on a third-party's Facebook wall, or public sharing (n = 87), in which they ostensibly shared a video publicly on their own Facebook wall.
Participants entered the laboratory in groups of 6-10 and were seated at individual computer stations equipped with headphones. They were told that the goal of the study was to evaluate responses to social change media messages and that they would be required to log into their Facebook account. After providing informed consent, participants were notified that they had been entered into a Participant Pool drawing to win $50, and provided their name to register their participation in the drawing. They were then asked to log into Facebook using a third-party Facebook interface (this task was a simulation and no login information was collected), which then showed a graphic from a Facebook application called 'Research Study App,' requesting permission to share on their Facebook wall. 3 Participants were informed during the consent process that they were free to deny permission to the application and would receive credit for participation regardless.
Impression management motives were assessed both before and after the sharing manipulation using measures from Kristofferson et al. (2014). Six questions (e.g., 'I care about how positively others view me,' 'I want to present myself in a positive way'; 1: strongly disagree to 7: strongly agree) were combined into indices for baseline impression management (Cronbach's α = .80) and post-sharing impression management (Cronbach's α = .82). Higher values indicate greater impression management concerns Video evaluation and selection. The participants were asked to review some videos for a national effort (called NSIP) to help charitable organizations improve their promotional materials, and then viewed the three non-profit videos (on Cancer Research, Farm Animal Welfare, and Global Water Access) in random order. After each video, they rated how informative, interesting, absorbing, boring, engaging, and inspiring the video was (1: not at all to 7: extremely). The ratings were combined into a single index (with boring reversed coded; Cronbach's α = .74). After viewing all three videos, participants were asked to select the one that they would most like to support. The 'Research Study App' then told participants in the anonymous sharing condition that their selected video would be anonymously shared on NSIP's Facebook wall, while participants in the public sharing condition were told that their selected video would be shared publicly on their own Facebook wall. Thus, participants were able to select the video that best aligned with their own values, while maintaining random assignment to a sharing condition.
Affect plays an important role in decisions to help other people (Slovic, Finucane, Peters, & MacGregor, 2007) and was assessed immediately after the sharing manipulation.
Participants were asked to rate how sad, happy, upset, inspired, sympathetic, and guilty they felt, using a set of six-point semantic differentials (e.g., 1: not at all sad to 6: sad, with mid-points unlabeled). Two negative emotions (sad and upset) previously associated with donation behavior (Slovic, 2007) were combined to form an index of negative affect (Cronbach's α = .75). Consistent with prior research, inspiration and happiness did not significantly correlate (Oliver, Hartmann, & Woolley, 2012); therefore, happy was used as a measure of hedonic positive affect. Inspired, sympathetic, and guilty were uncorrelated with any outcome measures and were not included in further analyses.

Dependent variables
Willingness to volunteer. To assess the primary dependent variablewillingness to engage in offline helping behavior after sharingparticipants reported how willing they would be to staff an information table for the organization whose video they shared (1: very unwilling to 6: very willing).
Donation behavior. Participants were asked if they were willing to donate winnings from the Participant Pool drawing to the organization whose video they selected. Those willing to donate were asked how much, using a slider ranging from 0 to 50, anchored at zero.
Additional helping measures. Three measures asked about participants' willingness to engage in other forms of offline volunteering (e.g., attending a meeting) and four measures asked about participants' willingness to engage in forms of online helping (e.g., signing an online petition; 1: very unwilling to 6: very willing).

Prosocial behavior control measures
Several measures were included to control for pre-existing individual differences in helping behavior. To avoid biasing the measurements of impression management, these were administered following the dependent variables. 4 Importance of participating. To assess general predisposition to engage in social causes, participants were asked to rate the importance of being involved in a good cause by donating to or volunteering with a nonprofit organization (1: not at all important to 6: very important).
Perceived efficacy. The perception that helping will be effective is a key moderator in studies of helping behavior (Bandura, 1982). Perceived efficacy was measured with two discrete items asking how effective participants' volunteering or donation would be in making a difference in the lives of the animal/child featured in the video they shared (1: not at all effective to 6: very effective).

Social media usage and motives
Users' sharing behavior is influenced by their frequency of social media use and individual motivations to share (Goh et al., 2009). A battery of items captured how often participants used various social media platforms (Facebook, Instagram, Snapchat, Twitter, and Yik Yak) on 9-point scales (1: never to 5: about every ten minutes). A second battery asked about the likelihood that participants would use these platforms to post about or discuss social issues they care about, on a 4-point scale (1: very unlikely to 4: very likely). Finally, four items (1: not at all like me to 5: just like me) assessed motivations for using social media in regards to social causes/issues. Three of these items ('Social media is an important way for me to participate in causes/issues that I care about,' 'I use social media to learn about what people in my network think about social issues,' and 'Posting/sharing about social issues on social media helps me express who I am') were significantly correlated and combined into an index of Social Media Use for Engagement (SMUE; Cronbach α = .74).

Manipulation checks
Sharing manipulation check. To verify that the experimental sharing manipulation simulated 'public' sharing of content on an individual's own Facebook wall, participants rated the publicness of their sharing (1: very private to 6: very public).
Sharing satisfaction check.
A key concern in this experimental design was creating an ecologically valid scenario while retaining the ability to randomly assign participants to a condition. Thorson (2013) found that the expression of political and social views on Facebook was a complicated social act governed by a variety of self-presentation concerns. Despite the fact that participants were allowed to select the video/cause that they most wanted to support, there remained the possibility that forced assignment to a sharing condition would produce reactance. To control for this, participants reported how satisfied they were with their decision to share the video they selected (1: very dissatisfied to 6: very satisfied).

Manipulation checks
Participants in the public sharing condition rated their sharing as significantly more public than participants in the anonymous sharing condition (M Public = 4.74, SD = 0.87, vs. M Anonymous = 3.48, SD = 1.37; t(155) = 7.36, p < .0001), confirming that the sharing manipulation was successful. Participants in the public sharing condition were less satisfied with their decision to share than were participants in the anonymous sharing condition (M Public = 4.41, SD = 1.03, vs. M Anonymous = 5.13, SD = 0.76; t(158) = −5.27, p < .0001), indicating that some participants may have reacted to the experimental manipulation. In total, there were 16 participants who expressed dissatisfaction (sharing satisfaction < 3). Rather than removing these participants from the data set (as per Kristofferson et al., 2014), sharing satisfaction was included as a covariate in subsequent analyses. 5

Videos
There was no difference in video selection between the experimental conditions, χ 2 (2, n = 175) = 1.67, ns (Table 1). A repeated-measures ANOVA showed that evaluations of the three videos differed significantly, F(2, 177) = 120.8, p < .0001, but this effect was not moderated by condition (Table 1). A dummy variable was created for whether or not participants had previously seen the video they chose (8 participants in total); this and the dummy variables for video choice were included as control variables.

Impression management
Baseline impression management was high and strongly correlated with post-sharing impression management (M Baseline = 6.01, SD = 0.66, M Post-sharing = 5.89, SD = 0.76; r(176) = .84, p < .001). Change in impression management was calculated by subtracting each participant's impression management score after sharing from their score at baseline. Change in impression management did not differ by the sharing condition (M Public = .08, SD = 0.37, vs. M Anonymous = .14, SD = 0.45, t(172) = −1.04, ns) and thus cannot mediate any effects of sharing on intention to help. Baseline impression management was included in the analyses to control for trait-level impression management concern.

Preliminary analysis
A series of independent t-tests revealed that control variables (gender, video selection, video seen, negative affect, importance of participating, general Facebook (FB) use, and FB for social issue discussion) did not significantly vary by experimental condition. A second series of independent t-tests were conducted to determine the main effect of experimental condition on each dependent variable. Consistent with our first hypothesis, willingness to volunteer (i.e., staff an information table) was higher for those sharing publicly than those sharing anonymously (M Public = 4.41, SD = 0.98, vs. M Anonymous = 4.05, SD = 1.30; t(175) = 2.08, p < .05). There was no significant difference between sharing conditions in the amount of time participants were willing to volunteer (M Public = 59 min, SD = 37.75 min, vs. M Anonymous = 56.91 min, SD = 34 min; t(156) = 0.37, ns). Contrary to our second hypothesis, sharing condition did not predict donation decisions, χ 2 (1, N = 178) = .06, ns. Willingness to donate was high (only 10 participants chose not to donate), suggesting a ceiling effect for this outcome variable. There was no significant main effect of condition on donation amount or on the additional offline and online outcome measures.

Stepwise regression analysis
A four-stage stepwise multiple regression was conducted to test possible moderators of the effect of condition on willingness to volunteer, while controlling for impression management motives and individual differences. There were no highly significant correlations among the independent variables (see Table 2), and variance inflation factors were all within acceptable limits, satisfying the assumption of the absence of multicollinearity (Tabachnick & Fidell, 2012). All variables were mean-centered and standardized prior to analysis and standardized coefficients are reported in all regression models. Sharing condition (reference = anonymous) was entered at step one of the regression. Sharing satisfaction, baseline impression management, video selection (reference = Water assistance) and video seen (reference = not seen) were entered as control variables at step two. Previous literature has indicated that gender (Eagly, 2009) and negative affect (Slovic, 2007) influence prosocial behavior; these variables were entered as control variables at step three, along with additional volunteering related attitudes used in prior research on volunteering motivation (importance of participating, perceived efficacy; Hustinx, Cnaan, & Handy, 2010). General FB use, FB for social issue discussion, and social media use for engagement (SMUE) were entered in the final step, as was the interaction of sharing condition and SMUE. Results of this analysis are reported in Table 3. In step one, sharing condition was a significant predictor of willingness to volunteer (β = .15 (.07), p < .05) and accounted for 1.8% of the variance, F(1, 176) = 4.27, p < .01, Adj R 2 = .018. In step two, sharing satisfaction (β = .39, SE = .08, p < .001) and video chosen (Water vs. Cancer; β = −.19, SE = .07, p < .001) were significant predictorsthose who selected the cancer research video were less willing to volunteer than those who selected the global water access video (M Cancer = 4.03, SD = 1.23, vs. M Water = 4.48, SD = 1.08). There was no significant difference in willingness to volunteer between individuals who chose the global water access video and the farm animal health video (M Animal = 4.21, SD = 0.97). Baseline impression management motives and video seen were not significant predictors.
Step two increased the variance explained to a total of 16.2%, F(6, 171) = 6.72, p < .001, Adj R 2 = .162. In step three, all variables previously associated with individual differences in likelihood to engage in prosocial behavior were added, including negative affect, gender, importance of participating, and perceived efficacy. This increased the variance explained to a total of 33.4%, F(10, 167) = 9.88, p < .001, Adj R 2 = .334. In the final step, general FB use and FB social issue discussion were non-significant; however, there was a significant interaction between sharing condition and SMUE (β = −0.14, SE = .06, p < .01). The effect of sharing condition is moderated by participants' use of social media for purposes of social engagement (RQ1). The final model accounted for 36.6% of the variance in Willingness to Volunteer, F(14, 163) = 8.29, p < .001, Adj R 2 = .37. An analysis of simple slopes of SMUE revealed that among participants high in SMUE (+1 SD), there was no effect of Sharing Condition on Willingness to Volunteer, t(163) = 0.40, ns, whereas among participants low in SMUE (-1 SD) sharing publicly (vs. anonymously) increased willingness to volunteer, t(163) = 3.14, p < .01; b = .65, SE = .20, p < .01. Figure 1 plots the simple slopes for this interaction with continuous 95% CIs.

Post hoc analysis
In order to explore the possibility that the interactive pattern between condition and SMUE might also emerge for the other dependent variables (for which no main effect was found), the final regression model was run with other willingness to help measures. The same interactive pattern was found for attend a meeting (β = −.12, SE = .08, p < .06), visit the organization's website (β = −.13, SE = .08, p < .07), and share other organizational   videos (β = −.16, SE = .09, p < .05). There was no significant interaction for the remaining willingness to help measures.

Discussion
The present study offers initial evidence that public sharing of prosocial media does not lead to a decrease in subsequent willingness to engage in prosocial activity. Instead, in comparison to anonymous sharing, there was a significant positive effect of public sharing on willingness to volunteer, despite controlling for numerous variables known to influence prosocial behavior. This small but measurable effect could translate to a large impact on support for social causes at the population level, and has implications for understanding the social dynamics of prosocial media sharing. The discrepancy between the results of this study and previous studies demonstrating a Slacktivism effect suggests that not all forms of online participation are created equalparticularly in terms of (a) the nature of public observability on platforms such as Facebook and (b) the implications of sharing as a prosocial act.
Facebook users in this age group tend to be keenly aware of the diverse nature of their network and are highly concerned with issues of privacy and surveillance (Brandtzaeg, 2012). A constant flow of information means self-presentation on Facebook is a continual process not easily resolved by a single message, post, or share. Indeed, we found that impression management motives were consistently high both at baseline and after sharing, and were unaffected by the manipulation. It is possible that the laboratory setting induced persistent impression management concerns in participants, but in prior studies using similar laboratory contexts, impression management was successfully manipulated (Kristofferson et al., 2014;White & Peloza, 2009). More plausibly, our realistically simulated SNS environment elicited persistent impression management concerns.
The negative effect of baseline impression management in our final model is consistent with Twenge et al.'s (2012) finding that Millenials are more self-focused and less likely to participate in specific forms of engagement. Higher baseline impression management may reflect a predisposition against engagement in social causes, but the need to manage one's impression in the moment may still operate as a powerful incentive to engage in offline helping.
Although we did not explicitly test it as a mechanism for the observed effects, our results are consistent with the possibility that among individuals in the public sharing condition, the desire to remain self-consistent drives the willingness to engage in offline action. Under conditions of chronically high impression management concerns, as observed in the present study, individuals are motivated to reduce any discrepancy between their current and desired image (Leary & Kowalski, 1990). Publicly sharing a social cause video appears to be an act that publicly presents a desired image (supporting a selected cause) without substantially decreasing impression management motives. Anonymously sharing a social cause video may establish moral motives, but entails no public commitment to a desired image. Thus, the drive to remain self-consistent is strongest for individuals in the public sharing condition, because they may experience both internal moral and external social pressure to remain consistent with their statement of support.
Our findings may also be explained by the unique nature of sharing as a social activity. We argue that sharing a social cause video requires a greater social investment than either signing an online petition (Lee & Hsieh, 2013) or joining a Facebook group (Kristofferson et al., 2014). Participants in the present study were able to select the video they most supported before being assigned to a sharing condition, approximating proactive endorsement of a specific cause. In networks with diverse contacts, sharing prosocial media not only serves as a cathartic means of showing that one is a morally righteous person, but also contributes to ongoing self-presentation efforts (Schau & Gilly, 2003). Indeed, individuals are motivated to share information to improve their reputations (Wasko & Faraj, 2005) and are concerned with remaining consistent in their online expressions of identity (Davis, 2012). Together, the social context of Facebook and the social nature of SNS sharing may account for the discrepancy between our results and past work finding that token displays of support lead to Slacktivism (e.g., Kristofferson et al., 2014).
Our results also indicate that the effects of public media sharing are moderated by how individuals use social media to participate, learn, and express themselves. Participants who reported using social media to engage with social issues were willing to volunteer regardless of the observability of their sharing. Social media use is positively associated with political engagement ; it is unsurprising that existing internal motivations to help and past online public prosocial activity would outweigh the effects of our manipulation, consistent with Kristofferson et al.'s (2014) finding that value alignment mitigated the Slacktivism effect.
The story is different for individuals who typically do not use social media to engage in social issues. Low SMUE participants reported greater willingness to volunteer after publicly sharingfor them, public sharing may be a novel public expression of a prosocial image, which motivates them to remain consistent by reporting willingness to volunteer. Sharing a video anonymously should create no such need to remain publicly consistent, and thus fail to increase willingness to volunteer. Public observability in environments that induce impression management concerns may be most important for those who do not regularly engage with social issues or causes online.
Our findings are strengthened by post hoc analyses revealing the same interactive pattern between sharing condition and SMUE for helping outcomes other than willingness to volunteer. In particular, for those low in SMUE, publicly sharing the video increased the willingness to engage in further sharing of social cause content. This suggests a possible virtuous cycle in which an initial act of socially observable sharing makes future online participation easier for disengaged individuals. This finding also has practical significance. On the one hand, it highlights the futility of 'preaching to the choir'the social observability of sharing did little to increasing helping behavior for those who already use social media to engage in social causes. On the other, it suggests that public media sharing can help nudge disengaged individuals toward engagement. Theories of social change emphasize the need for organizations and movements to mobilize the disengaged (Klandermans, 1984), and the persuasion literature indicates that social pressure can be a powerful lever for changing behavior (Wood, 2000). If prosocial media is designed to be shareable enough for disengaged individuals, post-exposure follow-up requests for offline support may be successful.
Though our conclusions are limited by the artificial nature of our sharing manipulation, recent online prosocial media campaigns have been designed to cleverly push users to share content. The ALS Ice Bucket Challenge, launched by the Amyotrophic Lateral Sclerosis (ALS) Association in 2014, utilized user-generated videos to spread a viral 'challenge' in which individuals dumped buckets of ice water on their heads to raise money for ALS. The campaign engaged a wide range of individuals, eliciting donations from over 739,000 new donors (Steel, 2014). In this case, it is highly unlikely that the video-sharing campaign only promoted offline behavior for individuals already using social media for engagement. Our results suggest that the act of sharing may have played a role in engaging those who were not previously engaged in online prosocial activity, by making their commitment to the cause publicly observable.
This study makes a theoretical contribution to the literature on Slacktivism by (1) demonstrating that publicly (vs. anonymously) sharing a social cause video can increase willingness to engage in offline helping behavior (a reverse Slacktivism effect), and (2) providing preliminary evidence that the extant usage of social media for engagement is a boundary condition for this effect. It is important to note that while we found no evidence of 'Slacktivism,' we also did not observe a change in impression management motives (which remained high after sharing). Thus, our results are consistent with Kristofferson et al.'s (2014) theoretical model, which predicts that no Slacktivism effect will occur in contexts where impression management motives do not change. When a media message is shared under conditions of chronically high impression management, it is likely that self-consistency motives will instead be the key driver of willingness to engage in offline helping.
Sonya Dal Cin is associate professor (Communication Studies) and research associate professor (Research Center for Group Dynamics, Institute for Social Research) at the University of Michigan. Dr Dal Cin's research concerns the effects of mediated and unmediated social influence on health and well-being among adolescents and emerging adults [email: sdalcin@umich.edu].