Microblogging and the Value of Undirected

Eva C. Buechel Jonah Berger University of South Carolina University of Pennsylvania

Accepted by Amna Kirmani, Editor; Associate Editor, Aimee Drolet

Online social networks have become extremely popular, but what drives sharing through these channels? We demonstrate that one of the most popular features of online social networks, microblogging (e.g., tweeting or sharing status updates), is driven in part by its undirected nature. Microblogging allows people to simultaneously express themselves to a large number of potential communication partners without having to address anyone in particular. As a result, this communication channel is particularly valued when people feel socially apprehensive; it allows them to reach out without having to impose communication and potentially bother anyone in particular. These findings shed light on one reason why people use online social networks and provide insight into the value of undirected communication. Keywords Online social networks; Social transmission; Microblogging; Social apprehension; Emotion

Online social networks have become a pervasive Walasek, Bhatia, & Brown, 2018). Microblogging is part of everyday life. They are revolutionizing the immensely popular. Each day there are over way we spend our time, communicate with others, 500 million tweets (Internetlivestats, 2017) and more and maintain social relationships. But while these than 125 million Facebook status updates networks allow users to keep in touch with friends, (Hampton, Goulet, Rainie, & Purcell, 2011). increasing feelings of social capital (Ellison, Stein- The current research identifies one reason why field, & Lampe, 2007; Hoffman & Novak, 2012), people microblog. Specifically, we argue that researchers and cultural critics suggest that these microblogging is driven in part by its undirected sites have detrimental effects on users’ welfare and nature, making it a unique channel of communica- relationships (Buffardi & Campbell, 2008; DiSalvo, tion. Most communication methods (e.g., face-to- 2010; Forest & Wood, 2012). Specifically, some have face and phone calls) require reaching out directly argued that online social networks reduce face-to- to a specific person who may feel obligated to face interaction, detracting from meaningful com- respond. Microblogs, on the other hand, allow peo- munication and leaving people depressed, anxious, ple to reach out without addressing anyone in par- and lonely (Kraut et al., 1998; Tonioni et al., 2012; ticular. Such undirected communication, we Turkle, 2015). suggest, allows people to reach out and elicit social Contributing to this debate, we suggest that a interactions without having to worry about impos- unique feature of these networks can actually offer ing unwanted communication on a particular per- a valuable outlet for expression and social sharing. son. Thus, while microblogs may be less personal One of the most popular features of Facebook, and than face-to-face interaction, their undirected nature the hallmark of , is the broadcasted sharing may make this communication channel particularly of short messages (i.e., status updates or tweets) desirable in some instances. about thoughts, feelings, or actions with other users If one value of microblogs lies in their undirected who can read them and respond. While Facebook nature, as we suggest, then microblogging should status updates and tweets vary in a number of be particularly valuable when people want to avoid ways (see General Discussion), academics and prac- more directed communication, such as when they titioners alike collectively call such short messages feel socially apprehensive. Social apprehension “microblogs” (Kaplan & Haenlein, 2011; Lee, 2011; involves anxiety about communicating with others and concerns about being rejected (e.g., being at a Received 8 July 2016; accepted 13 October 2017 party and not having anyone to talk to or feeling Available online 30 October 2017 Correspondence concerning this article should be addressed to Eva C. Buechel, Moore School of Business, University of South © 2017 Society for Consumer Psychology Carolina, 1014 Greene Street, Columbia, SC 29208, USA. Elec- All rights reserved. 1057-7408/2018/1532-7663/28(1)/40–55 tronic mail may be sent to [email protected]. DOI: 10.1002/jcpy.1013 Social Apprehension and Microblogging 41 unwelcome; Horwitz, 2002; Schroeder, Wormworth, friends are said to provide increased social capital & Livesley, 1992; Wrench, Brogan, McCroskey, & (Ellison et al., 2007; Wilson et al., 2012). On the Jowi, 2008). Importantly, the experience of social other, these sites have also been criticized for reduc- apprehension generates two contradictory impulses. ing meaningful face-to-face interaction (Forest & First, it makes people feel inadequate or nervous, Wood, 2012; Kraut et al., 1998; Tonioni et al., 2012; and these negative feelings heighten the need to Turkle, 2015) decreasing and subjective well-being reach out to others for comfort and social support (Kross et al., 2013; Verduyn et al., 2015). (Rime, 2009). Second, it makes people feel tentative Third, the relationship between online and offline about reaching out for fear of rejection. As a result, behavior is unclear. Some researchers have argued people who feel socially apprehensive may find that online behavior is merely an extension of behav- microblogging a particularly useful channel of com- ior offline (Back et al., 2010). Narcissists share more munication because it allows them to reach out self-indulgent content online (Buffardi & Campbell, broadly to elicit social interaction, while avoiding 2008), and extraverts have more friends on Facebook potentially bothering any given individual. and spend more time connecting with them (Gosling, Six studies investigate how microblogging’s Augustine, Vazire, Holtzman, & Gaddis, 2011; Ross undirected nature drives its use. We examine if and et al., 2009). Other data, however, suggest that the when the experience of social apprehension online environment may encourage representations increases individuals’ likelihood of microblogging. that differ from offline ones by allowing people to We also examine whether this increase in sharing is present themselves overly positively (Berger & Iyen- specific to microblogging (i.e., does not extend to gar, 2012; Gonzales & Hancock, 2011; Wilcox & other communication channels). Finally, we exam- Stephen, 2013). ine whether, as predicted, these effects are driven One reason for these differing findings may be by the desire to reach out to others without impos- that most online social networks provide not just ing communication and potentially bothering any- one, but multiple methods of communication. Prior one in particular. work often treats social networks as homogeneous entities, looking at the effect of Facebook use or why people use Twitter. But in reality, each of these sites encompasses multiple features and multiple Online Social Networks communication methods (Wilson et al., 2012). Face- Online social networks allow people to quickly and book users, for example, can post status updates or easily connect with a broad set of social ties, even share on a friend’s wall. Twitter users can send when those ties are geographically distant. A large tweets or direct messages to particular social ties. body of research (Buffardi & Campbell, 2008; Ellison These different sub-channels vary in important et al., 2007; Kraut et al., 1998; Tonioni et al., 2012; ways. While direct messages on Twitter or sharing Wang et al., 2011; Verduyn et al., 2015) suggests that on a friend’s Facebook wall are targeted to a partic- online social networks cater to a variety of needs ular user, microblogging features (i.e., status such as affiliation, self-expression, signaling social updates or tweets) allow people to share broadly status, and information dissemination (Back et al., without directing their communication at any one 2010; Berger, 2014; Gosling, 2009; Hoffman & Novak, person in particular. 2012; Naylor, Lamberton, & West, 2012; Toubia & Consequently, to more deeply understand the Stephen, 2014; Wilson, Gosling, & Graham, 2012), for utility of these online channels, we focus on the an overview see Buechel and Berger (2015). popular, common, and unique feature of online That said, many more specific aspects of online social networks: microblogging. We investigate social networking are still relatively poorly under- how microblogs differ from other communication stood. First, most research in this area has been cor- channels and how this difference can explain when relational. Online social network use has been and why people microblog. linked to a variety of personality factors (e.g., Cor- rea, Hinsley, & De Zuniga, 2010), but relatively lit- tle experimental work has provided evidence for Interpersonal Sharing and Social Apprehension causal relationships between personality variables and behaviors. People have a high need for social interaction, and Second, there is little agreement on how and social relationships are critical for well-being why communication on online social networks (Bowlby, 1977; Burleson, 1998; Harlow, 1961). Social impacts users’ welfare. On the one hand, online interactions are especially important during 42 Buechel and Berger stressful and negative times (Taylor, 2011) because & Burger, 2005; Tamir & Mitchell, 2012). Hence, attention, affection, and comfort from others can feelings of social apprehension place people in a supply needed social support (Zech & Rime, 2005). bind. It is a negative experience that makes people This can provide a so-called “socio-affective buffer,” want to seek social interaction and reach out to which is said to temporarily reduce negative affect others. At the same time, the experience of social and enhance well-being by reducing feelings of anx- apprehension makes them concerned about bother- iety or loneliness (Rime, 2009). ing those whom they reach out to. How can people Yet, social sharing is not always easy. Sometimes resolve this conundrum? people feel “socially apprehensive,” a negative state that is marked by the experience of fear and/or anxiety due to real or anticipated communicative Microblogging: The Value of Undirected social interactions (McCroskey, 2001; Schroeder Communication et al., 1992; see also Buss, 1980; Henderson, Gilbert, & Zimbardo, 2014). The state of social apprehension We suggest that one reason people microblog is is often brought on by the situation. Being at a because it alleviates this problem. Microblogging party and not having anyone to talk to, for exam- increases opportunities for desired social interaction ple, can make people feel socially anxious and while not having to impose communication and apprehensive. Some people are also more prone potentially bother anyone in particular. to feeling socially apprehensive than others Microblogs vary from other communication (McCroskey, 1970; Kessler, Stein, & Berglund, channels in important ways. They differ from face- 1998). Individuals who score high on Social Com- to-face interactions in that they are written and in munication Apprehension (Wrench et al., 2008), for that they do not require making eye contact. Most example, are particularly likely to feel anxious important for our conceptualization, microblogs when communication is expected or desired. That also differ from face-to-face and other forms of is, while highly socially apprehensive individuals (e.g., texting, emailing) in that they do not always experience social apprehension, the are undirected. Initiating social interaction usually experience of state social apprehension is more involves directed communication: reaching out to easily triggered in these individuals, be it by exter- particular individuals and thus imposing social nal or internal factors. These individuals may be interactions onto them. Imagine you’re at work and particularly likely to experience social apprehension want to connect with others. Before online social when they feel external pressure to interact with networks, you’d have to stop by someone’soffice others (e.g., at a party) but are apprehensive to do or call them on the phone, both of which involve so. These individuals might also be more likely to reaching out to a specific person who might feel experience social apprehension when they desire obligated to respond. Similarly, sending or social interaction, such as when they are experienc- messages to one or more people usually requires ing negative affect and feel an internal need to directing communication at selected recipients who interact with others (i.e., they desire a “socio-affec- might feel they should respond. All of this can tive buffer”), but are apprehensive to do so. feel difficult, especially when people feel socially When triggered, the state of social apprehension apprehensive. generates two contradictory impulses. Experiencing In contrast, by offering undirected communica- social apprehension makes people feel inadequate tion, microblogs reduce the burden of bothering and anxious. This negative feeling of distress any particular person while still offering the possi- increases the need to reach out and connect with bility for social interaction. Rather than addressing others (Berger, 2011; Berger & Milkman, 2012; a person (or people) in particular, microblogs (e.g., Peters, Kashima, & Clark, 2009) in an attempt to status updates or tweets) are broadcast to all (or reduce their anxiety via the aforementioned socio- many) users in one’s network who may or may not affective buffer (Rime, 2009). At the same time, see the post in their newsfeed. Thus, while different however, feeling socially apprehensive may make online social networking platforms differ in many people wary of bothering others (Jones & Carpen- ways (see General Discussion) both status updates ter, 1986; Pilkonis, 1977). This should decrease their and tweets go out to a large number of online ties willingness to reach out to anyone in particular for in an undirected manner. This undirectedness fear that they will seem burdensome or be rejected, makes responding more voluntary. Because com- thus depriving themselves of the rewards of disclo- munication is not directed at them specifically, each sure and social interaction/support (Calsyn, Winter, recipient feels less obligated to respond. This, in Social Apprehension and Microblogging 43 turn, should reduce a sender’s concern about both- for more directed communication channels, and ering others by imposing unwanted communication (c) these effects are driven by microblogging’s (i.e., presumably, only people who want to interact undirected nature: Across platforms, consumers will respond). Somewhat ironically then, by com- value microblogging because it allows them to con- municating and reaching out to more people, nect with others without imposing unwanted com- microblogs reduce concerns about bothering each munication and potentially bothering anyone in individual, thus making reaching out less threaten- particular. ing. If the undirected nature of microblogs can facili- Study 1: Sharing Via Microblogs tate sharing, as we suggest, then microblogs should be particularly valuable and useful when people Study 1 examines actual sharing. We manipulate want to reach out but feel socially apprehensive social apprehension and give people the opportu- about doing so. The experience of social apprehen- nity to use online social networks. We predict that sion should increase sharing via microblogs social apprehension will increase the number of because their undirected nature enables desired people who microblog. social interaction while minimizing anxiety about burdening others. The same is not true for directed Method communication. For directed channels, the experi- ence of social apprehension should not increase In total, 145 undergraduates (87 females and = sharing; indeed, it may even decrease willingness 57 males [one missing]; Mage 20.82 years, to share. SD = 2.34) were randomly assigned to one of two Note that we are not suggesting social apprehen- between-subjects conditions (control vs. social sion to be the only driver of microblogging. Like apprehension). any consumer behavior, microblogging is complex First, we manipulated social apprehension, a and driven by many factors. We use social appre- negative state marked by feelings of anxiety about hension to isolate and test microblogging’s undi- reaching out. In the social apprehension condition, rected nature, a feature specific to microblogs that participants wrote about a time they felt socially has not been examined and that may lead to com- apprehensive. Specifically, they were asked to think munication that differs in its antecedents from other and elaborate about a time when they had attended communication channels. Indeed, pilot data high- a party or social gathering and had not had anyone light the importance of social apprehension in to talk to. In the control condition, participants microblogging. Participants were asked whether wrote about office products. A pretest (see Appen- they preferred sharing face-to-face or via microblog dices S1 and S2) confirmed that the social appre- and completed a variety of personality scales (see hension manipulation increased feelings of social Appendix for details). Results indicate that higher apprehension. levels of social apprehension were associated with Second, participants who indicated having an greater preference for microblogging (b = 0.35, online social network account were asked to open a t(99) = 3.75, p < .001; R2 = 0.13). Importantly, this new tab, log into their preferred online social net- preference was better explained by social apprehen- work account (e.g., Facebook or Twitter) and spend sion than other personality variables often linked to the next 2 min on their online social network. They other online social networking behaviors (e.g., were allowed to choose which online social net- extraversion, narcissism, or self-esteem). Thus, work to visit, but were asked to only visit one. while it is clear that use is driven by a After 2 min, they would hear an alarm, at which number of factors, we examine how social appre- point they should return to the survey. hension shapes microblogging. Upon returning to the survey, participants were asked whether they had microblogged (e.g., tweeted or posted a status update). Specifically, we asked, “Microblogs are short messages shared Overview of Studies on online social networks (tweets, status updates, Five studies investigate the value of undirected etc.) Did you share a microblog while you were communication. Supporting our theorizing, they on your online social network ?” [Yes, No]. demonstrate that (a) social apprehension increases Finally, in this and all reported studies, partici- microblogging, (b) this increase does not hold pants indicated their age and gender. 44 Buechel and Berger

moderated mediation. Social apprehension should Results increase concerns about not bothering anyone in As predicted, the experience of social apprehen- particular, which should increase microblogging sion increased microblogging, v2 (1, N = 142) = 5.05, but not face-to-face communication. p = .02, r = 0.18. Compared to the control condition (2.8%), people made to feel socially apprehensive Method were more likely to microblog (12.9%). Logistic regressions including dummy variables for visited In total, 140 Mturkers (78 females and 60 males fi = = online social network did not reveal any signi cant [two missing]; Mage 36.86 years, SD 14.16) were effects or interactions due to platform (all ps < .28), randomly assigned to a condition in a 2 (Social indicating that similar results held across platforms apprehension: control vs. social apprehension) 9 2 (see Appendices S1 and S2 for ancillary analyses and (Channel: face-to-face vs. microblog) between-sub- robustness checks). jects design. First, we manipulated social apprehension. Par- ticipants either wrote about a time they had felt Discussion socially apprehensive (social apprehension condi- Study 1 provides preliminary support for our tion) or about office products (control condition), suggestion regarding the link between social appre- see Study 1. hension and microblogging. Participants made to Second, we manipulated communication channel. feel socially apprehensive were more likely to Participants were asked how likely they were to microblog. By directly manipulating social appre- express themselves to others (1 = Very Unlikely; hension and measuring real behavior, the present 9 = Very Likely). In the microblog condition, they study offers internal and external validity for our were asked about their likelihood to microblog (i.e., proposed effect. tweet or post a status update) if they had the opportunity to do so. In the face-to-face condition, they were asked about their likelihood to express Study 2: Underlying Process themselves in person if they had the opportunity to Study 2 compares microblogging to offline shar- do so. ing and begins to test the hypothesized process Third, we measured our hypothesized process. through mediation. Similar to Study 1, we make To test whether the effects are driven by the half of the participants feel social apprehensive. We undirected nature of microblogging, we asked then ask everyone about their likelihood of express- participants how much they agreed with the state- ing themselves to others, either in person (i.e., face- ment “I don’t want to bother or burden anyone to-face) or via microblog (i.e., status update or in particular” (1 = Strongly Disagree;7= Strongly tweet). Finally, we measure the degree to which Agree). participants did not want to bother anyone in particular. Results We expect the experience of social apprehension to increase willingness to microblog but not to In addition to a main effect of Channel, F (1, = < 2 = increase willingness to share face-to-face. Note that 136) 27.65, p .01, gp 0.17, a 2 (Social appre- we are not suggesting that socially apprehensive hension) 9 2 (Channel) between-subjects ANOVA individuals should be more likely to express them- revealed only the predicted Social apprehen- selves via microblog than face-to-face. Over 80% of sion 9 Channel interaction, F(1, 136) = 6.20, = 2 = word of mouth is face-to-face (Keller & Fay, 2012), p .01, gp 0.05, see Figure 1. Note that in this and overall people are much more likely to express and all remaining studies, we only report margin- themselves face-to-face than online (Berger & Iyen- ally significant (p < .1) and significant effects gar, 2012; Keller & Libai, 2009). If our theorizing is (p < .05) for the omnibus tests. correct about the value of undirected communica- Simple effects revealed that, compared to control tion, however, then experiencing social apprehen- participants (M = 2.60, SD = 1.88), inducing social sion should increase willingness to share via apprehension increased participants’ likelihood of microblog. sharing via microblog (M = 4.02, SD = 2.66), F(1, = = 2 = Furthermore, if our theorizing is correct that not 136) 5.60, p .02, gp 0.05. The same effect was wanting to bother anyone in particular is driving not observed for face-to-face communication these effects, we should find evidence for (F = 1.60, p > .20). Social Apprehension and Microblogging 45

F2F MB appealing because it does not require making eye 6 5.5 contact or because it involves written (rather than 5 oral) expression. To rule out these accounts, we had 4.5 also asked participants how much they agreed with 4 the statement “I’d rather not have to make eye con- 3.5 tact” and “I’d rather express myself in writing.” 3 Contrary to these accounts, neither preference for 2.5

Sharing likelihood “ ” 2 non-face-to-face communication (95% = À = À 1.5 CIF2F 0.15 to 07 and CIMB 0.39 to 0.05) nor 1 “written expression” mediated the effects (95% Control Social apprehension = À = À CIF2F 0.06 to 0.55 and CIMB 0.10 to 0.13). Figure 1. State social apprehension and sharing across channels. Furthermore, the fact that social apprehension did not increase face-to-face sharing rules out the possi- bility that the manipulation increased sharing in Moderated Mediation. We expected the modera- general, independent of channel. tion to occur for the IV–DV path and the Mediator– Ancillary Study. A follow-up study demon- DV path, but not on the IV–Mediator path, leading strates that these results generalize across the two us to test moderated mediation using Model 15 main social media platforms (i.e., Facebook and (Hayes, 2012). The analysis revealed that the effect Twitter). Using the same procedure as Study 2, we of social apprehension on likelihood of sharing was first manipulated social apprehension and then driven by these individuals’ concern about “not manipulated communication channel. Rather than wanting to bothering anyone in particular.” The asking about microblogging in general, this study direction of these effects, however, depended on asked about sharing via microblog on Facebook or communication channel. Social apprehension Twitter specifically, in a 2 (Social apprehension: always increased the desire not to bother anyone in control vs. social apprehension) 9 3 (Channel: face- particular, B = À2.18, SE = 0.98; t(136) = À2.23, to-face vs. Facebook vs. Twitter) between-subject p = .03, but while this sentiment marginally design. Social apprehension again had no effect on decreased the likelihood of sharing face-to-face willingness to share face-to-face, F < 1, but (negative indirect effect, B = À0.80, SE = 0.56; 95% increased willingness to microblog on both Twitter, CI = 0.00 to 0.43), as predicted, it increased the like- F(1, 460) = 4.48, p = .03, g2 = 0.01, and Facebook, F lihood of sharing via microblog (positive indirect (1, 460) = 3.23, p = .07, g2 = 0.01. Note that, while effect, B = 1.24, SE = 0.59; 95% CI = 0.02 to 0.77). directionally stronger for Twitter, there was no dif- ference in willingness to share between Facebook and Twitter (F < 1), suggesting that the effect of Discussion social apprehension on microblogging is similar Study 2 again demonstrates how social appre- across the two platforms. hension affects microblogging and provides prelimi- nary evidence for underlying process. First, social Study 3: Trait Social Apprehension and Sharing Across apprehension only increased the likelihood of shar- Channels ing via microblogs. The fact that it did not have the same effect on face-to-face communication suggests To further understand when and why the undi- that social apprehension driven sharing may be rected nature of microblogging drives sharing, unique to microblogs. Study 3 uses individual differences in social appre- Second, mediation analysis supports the hypoth- hension and manipulates negative affect to activate esized process underlying these effects. As pre- the (internal) need to share (i.e., we activate the dicted, the effect of social apprehension on experience of social apprehension, see Rime, 2009). microblogging was driven by feelings of not want- If the value of microblogging lies in its undi- ing to bother anyone in particular. rected nature, as we suggest, then individuals who Third, ancillary analyses cast doubt on other are prone to feeling socially apprehensive should be potential alternative explanations. We suggested more inclined to share via microblogs. Importantly, that social apprehension increases microblogging however, this does not necessarily mean that these because it does not require addressing and poten- individuals will be more likely to microblog at any tially bothering one individual in particular. That given point in time. Instead, individual differences said, one could argue that microblogging is in social apprehension should be particularly likely 46 Buechel and Berger to impact the likelihood of microblogging when socially apprehensive individuals’ likelihood to that trait is activated, for example via negative share via microblog. While they may feel apprehen- affect (Andersen & Guerrero, 1997; Rime, 2009). As sive about sharing face-to-face or direct message, discussed earlier, the undirected nature of the undirected nature of microblogging should pro- microblogging should make this channel particu- vide a reduced risk way to reach out to others. larly valuable when people (a) want or need to reach out (e.g., sharing is expected or beneficial due Method to the “socio-affective buffer”) but (b) feel appre- hensive about doing so offline or in a more directed In total, 139 participants recruited from a paid manner. Individual differences in social apprehen- subject pool (70 females and 68 males [one miss- = = sion generate the second condition, but not neces- ing]; Mage 33.28 years, SD 12.67) were ran- sarily the first. While socially apprehensive domly assigned to a condition in a 2 (Affect: individuals might generally be more hesitant about control vs. negative) 9 3 (Channel: face-to-face vs. reaching out in person and thus prefer microblogs direct message vs. microblog) between-subjects (as the pilot data suggest), they do not always feel design with social apprehension as a measured the need or desire to share. Negative affect, how- variable. ever, should activate socially apprehensive individ- First, we manipulated affect. Participants uals’ desire to share and reach out. Negative affect watched video clips used in prior work to reliably elicits the desire to generate social interaction via evoke either negative affect or no affect. In the neg- the “socio-affective buffer” (i.e., to receive attention ative affect condition, participants watched a short and social support to buffer their negative experi- clip from a scene in Silence of the Lambs (Hewig ence; Rime, 2009). At the same time, reaching out is et al., 2005); in the control condition, participants particularly difficult for socially apprehensive indi- watched a short clip with images from space viduals; they might worry about imposing commu- (Buechel, Zhang, Morewedge, & Vosgerau, 2014). nication and bothering anyone in particular. Thus, Second, we manipulated communication channel. highly socially apprehensive individuals’ increased Similar to Study 2, participants were asked how sharing via microblog should be more likely to likely they were to express themselves to others manifest itself when they feel negative. Negative (1 = Very Unlikely; 9 = Very Likely), and depending affect alone, however, should not increase on condition, they indicated their likelihood of shar- microblogging and should therefore not increase ing via microblog (e.g., status update or tweet), sharing for low socially apprehensive individuals direct message (e.g., private message on online (i.e., the effect of social apprehension on sharing is social network), or in person (i.e., face-to-face). driven by feeling both, negatively and socially Finally, we measured social apprehension (Social apprehensive about reaching out). Communication Apprehension Scale; Wrench et al., In addition to examining the effect with indi- 2008), which measures anxiety associated with (of- vidual differences in social apprehension, this fline) communication. Typical scale items include “I study further tests whether our effect is unique to can communicate with people in social settings microblogging. We examine an additional commu- without experiencing anxiety (R),”“I am usually nication channel (i.e., direct messaging) that is anxious when talking to people in a bar,” and directed but not face-to-face or spoken. Sending a “Social interaction is the best part of my day (R).” message to a particular person by direct message Scale responses were not impacted by the manipu- on online social networks, chat, , or text is lations, Fs < 1. similar to microblogging in that it is written and not face-to-face, but different in that it is Results directed at someone specific. If the preference for microblogging is driven by its undirected Given that the channel condition has three levels, nature, as we suggest, then we should not we used two dummy variables (face-to-face and observe the same results for direct messaging, direct message) to compare how they differed from and direct messaging should resemble face-to-face the microblog condition. Multiple regression exam- communication. ined how affect (control vs. negative), communica- Overall then, if part of the value of microblog- tion channel, measured social apprehension, and ging comes from its undirected nature, as we sug- their interactions influenced sharing likelihood. gest, then negative affect should activate the In addition to a main effect of face-to-face, experience of social apprehension and increase b = 0.28, t(128) = 2.14, p=.03, and a Affect 9 Social Apprehension and Microblogging 47

Channel interaction, b = 0.59, t(128) = 2.78, respectively, making microblogging the least likely p=.006, the analysis revealed the predicted three- channel for sharing. The pattern reversed, however, way Affect 9 Channel 9 Social Apprehension inter- for highly socially apprehensive individuals. Highly action for both microblogging versus face-to-face, socially apprehensive individuals (+1 SD) were b = À0.44, t(128) = À2.36, p=.02, and microblog- more likely to share via microblog (M = 6.53) than ging versus direct messaging, b = À0.39, t(128) = face-to-face (M = 5.16), b = À0.43, t(61) = À2.16, À2.46, p=.02, see Figure 2. To shed more light on p = .03, or direct message (M = 5.11), b = À0.42, t the pattern of results, we discuss each affect condi- (61) = À1.93, p = .06. tion separately. Looked at another way, negative affect did not Control Condition. For control participants, change highly socially apprehensive (+1 SD) partici- ’ = there was only a main effect of face-to-face, pants likelihood of sharing face-to-face (Mnegative = = = = < b 0.27, t(67) 2.01, p .05. As in Study 2 and 5.16; Mcontrol 5.32; t 1) or via direct messaging = = < prior work, participants were more likely to share (Mnegative 5.11; Mcontrol 4.88; t 1), but, consis- in person (M = 5.92) than via microblog (M = 4.58). tent with our theorizing, it did increase their likeli- = Control participants were no more likely to share hood of sharing via microblog (Mnegative 6.53; = = = = = via private message (M 4.90) than via microblog, Mcontrol 4.77), b 0.51, t(45) 2.34, p .02. t<1. No other effects were significant. Negative Affect Condition. When participants Discussion experienced negative affect (which should activate the need to share), however, the pattern differed. In Study 3 provides further evidence that the undi- addition to a main effect of Social Apprehension, rected nature of microblogging is one reason why b = 0.92, t(61) = 3.86, p < .001, and a marginal main people use this communication channel. Again, con- effect of face-to-face, b = 0.22, t(61) = 1.72, p = .09, sistent with prior work (Berger & Iyengar, 2012; this condition revealed the predicted Chan- Keller & Libai, 2009; Keller & Fay, 2012), people nel 9 Social Apprehension interactions for both generally reported being more likely to share face- microblogging versus face-to-face, b = À0.84, to-face than via microblog. More importantly, how- t(61) = À4.22, p<.001 and microblogging versus ever, for socially apprehensive individuals whose direct messaging, b = À0.61, t(61) = À3.47, p=.001. trait had been activated through negative affect, Slope analysis (Aiken & West, 1991) showed that this pattern reversed. Negative affect increased their low socially apprehensive individuals (À1 SD) likelihood of sharing via microblog and did so to exhibited a similar pattern to the control condition. such a degree that they even preferred microblog- They were more likely to share face-to-face ging to face-to-face sharing. (M = 6.85) or via direct message (M = 5.88) than Moreover, these effects were only observed for via microblog (M = 2.78), b = 0.86, t(61) = 4.51, the microblogging condition and did not emerge p < .001 and b = 0.64, t(61) = 3.70, p < .001, for direct messaging. The fact that the increase

8 Low social apprehension High social apprehension

7

6

5

4

Sharing likelihood 3

2

1 F2F DM MB F2F DM MB Control Negative Affect

Figure 2. Trait social apprehension and sharing across channels. 48 Buechel and Berger in sharing was specific to microblogs underscores between-subjects design with social apprehension the notion that part of the appeal of microblogging as a measured variable. lies in its undirected nature. While neither microblogs First, to activate social apprehension, similar to nor direct messages involve face-to-face communi- Study 3, we induced negative affect, this time by cation and both involve written communication, having all participants write about a negative cus- inducing negative affect only increased socially appre- tomer experience. For 4 min, they elaborated on the hensive people’s sharing via microblogs. This pattern circumstances and how it made them feel in detail. of results is consistent with our suggestion that the Second, we manipulated concerns about impos- undirected nature of microblogging (and not its writ- ing unwanted communication. We told participants ten or non-face-to-face nature) makes it particularly to think about someone in their broader friend useful when people want to reach out but feel tenta- group (e.g., a classmate or co-worker). In the “com- tive about bothering others. munication encouraged” condition, they were told Ancillary Analyses. Ancillary analyses (Appen- that this person had recently opened the door for dices S1 and S2) cast doubt on alternative explana- (offline) communication (i.e., they told them that tions based on generalized (i.e., positive) affect and they liked listening and would happily “lend an provide further evidence that the main study ear”). In the control condition, there were no addi- results are driven by a desire for social interaction tional instructions. when reaching out is difficult. Third, we manipulated communication channel (see face-to-face and microblogging conditions in Study 3). In a seemingly unrelated study about Study 4: Testing the Value of Undirected communication, participants were asked how likely Communication via Moderation they were to express themselves to others in gen- The first three studies demonstrate that social eral (1 = Very Unlikely; 9 = Very Likely), and apprehension, whether measured or manipulated, depending on condition, they indicated their likeli- increases microblogging. Unlike face-to-face interac- hood of sharing via microblog (e.g., status update tions or other types of directed communication or tweet) or in person (i.e., face-to-face). (e.g., direct messages), we have argued that micro- Finally, we measured social apprehension (Social allow people to reach out without potentially Communication Apprehension; Wrench et al., bothering others by imposing unwanted communi- 2008). Scale responses were not impacted by the cation. To further test this suggestion, we next manipulations, Fs < 1.8, p > .18. examine whether diminishing socially apprehensive individuals’ fears of imposing unwanted communi- Results cation decreases microblogging. If microblogging is driven by not wanting to Multiple regression analyses examined how the impose on others, as we suggest, then the appeal of encouraged communication manipulation, commu- microblogs should diminish when socially apprehen- nication channel manipulation, social apprehension sive individuals feel that their offline communication measure (continuous), and their interactions influ- is welcome. Thus, in Study 4, in addition to inducing enced sharing likelihood. negative affect and manipulating communication Results revealed a main effect of Social Apprehen- channel, we manipulate concern of imposing sion, b = 0.42, t(121) = 4.03, p < .001, a marginal unwanted communication by telling half the partici- Communication encouraged 9 Social Apprehension pants that their communication is welcome. If our interaction, b = À0.30, t(121) = 1.80, p = .08, a Com- theorizing is correct, this should moderate our effect. munication encouraged 9 Channel interaction, Among participants told that their (offline) commu- b = 0.44, t(121) = 3.37, p < .001, and a Channel 9 nication is welcome, social apprehension should no Social Apprehension interaction, b = À0.71, t(121) = longer increase microblogging. À3.87, p < .001. Most important, as expected, the analysis revealed a three-way Communication encouraged 9 Channel 9 Social Apprehension Method interaction, b = 0.48, t(121) = 2.77, p=.006, see In total, 128 participants recruited from an online Figure 3. For clarity and to shed more light on the = panel (83 females and 45 males; Mage 35.74 years, pattern of results, we analyze and discuss each shar- SD = 11.64) were randomly assigned to a condition ing condition separately. The results hold and are in a 2 (Communication encouraged: control vs. stronger when analyzed with the full sample (using yes) 9 2 (Channel: face-to-face vs. microblog) Model 3; Hayes, 2012). Social Apprehension and Microblogging 49

Low SA High SA 9

7

5

3 Sharing likelihood

1 F2F MB F2F MB Control Communication encouraged

Figure 3. Trait social apprehension and sharing across channels.

Control Condition. The control condition communication for socially apprehensive individu- revealed a main effect of Social Apprehension, als decreased microblogging and made them b = 0.50, t(59) = 2.38, p = .02, and a marginal main behave like low socially apprehensive individuals. effect of Channel, b = 0.19, t(59) = 1.74, p = .09. These results also cast doubt on an alternative Consistent with the negative affect condition in explanation for the effect. While one could argue Study 2, there was a Channel 9 Social Apprehension that highly socially apprehensive individuals might interaction, b = 0.51, t(59) = 4.03, p < .001. Slope microblog simply to self-express as opposed to analyses (Aiken & West, 1991) illustrate that while seeking social interaction (i.e., the value lies in low socially apprehensive individuals (À1 SD) were microblogs’ unidirectional [vs. undirected] commu- more likely to share face-to-face (M = 7.01) than via nication), this cannot explain why encouraging microblog (M = 4.14), b = À0.66, t(59) = À4.62, communication shifts them toward face-to-face p < .001, the pattern reversed for highly socially communication. Instead, this pattern suggests that apprehensive individuals (+1 SD). Highly socially microblogs are appealing because they allow for apprehensive individuals were more likely to share interaction (or at least potential interaction) without via microblog (M = 6.30) than face-to-face having to worry about imposing unwanted commu- (M = 5.03), b = 0.49, t(59) = 2.23, p = .03. nication and potentially bothering anyone in Communication Encouraged Condition. When particular. participants were told that communication was wel- come, however, the Channel 9 Social Apprehension Study 5: Undirectedness or Audience Size? disappeared and there was only a main effect of Channel, b = À0.64, t(62) = 6.47, p < .001, suggest- The studies so far have provided consistent evi- ing that, consistent with prior work, participants dence that social apprehension increases microblog- were more likely to share in person (M = 7.96) than ging, and that this is driven by microblogging’s via microblog (M = 4.35). undirected nature. That said, one could wonder Looked at another way, eliminating fear of impos- whether microblogs reaching a greater audience ing unwanted communication decreased highly might explain the effect. socially apprehensive participants’ likelihood of shar- While such an alternative has difficulty explain- = = ing via microblog (Mcontrol 6.30; Mencouraged 4.18), ing the mediation via concerns of bothering others b = À0.48, t(69) = À1.75, p = .08. Eliminating fear of in Study 2 and the moderation by welcome com- unwanted communication did not affect low socially munication in Study 4, Study 5 further tests this apprehensive individuals’ likelihood of sharing face- possibility by manipulating both the directedness of to-face or via microblog, Fs < 1. communication channel (i.e., microblog vs. face-to- face) and the audience size. If audience size is driv- ing the effect, it should increase sharing, regardless Discussion of whether the communication is directed or undi- Study 4 extends the prior studies by further test- rected. In contrast, if the effects are driven by the ing the underlying process. When people wanted to undirected nature of microblogging, as we suggest, reach out but felt apprehensive about bothering then socially apprehensive individuals should pre- others, they found microblogs particularly attrac- fer microblogging over directed channels (as in tive. However, eliminating the fear of unwanted Study 1 and Study 2), and audience size should not 50 Buechel and Berger increase sharing via direct messages. In fact, direct SD = 1.92) than in the 20 social connections condi- messaging more people requires targeting and tion (M = 3.69, SD = 1.88), F (1, 388) = 13.46, < 2 = potentially bothering more individuals. Therefore, if p .001, gp 0.03. our theorizing is correct that socially apprehensive Sharing Likelihood. Consistent with our sugges- individuals want to avoid bothering others, direct tion that social apprehension increases microblog- messaging may actually decrease as a function of ging, a 2 (Audience size) 9 2 (Channel) ANOVA audience size. revealed the predicted main effect of Channel, such that participants were more likely to microblog than direct message, F (1, 396) = 4.96, p = .03, Method 2 = gp 0.01. Interestingly, there was also a margin- In total, 400 Mturkers (233 females and 141 males ally significant Audience size 9 Channel interac- = = = = 2 = [28 missing or other]; Mage 27.78 years, SD 4.60) tion, F (1, 396) 3.19, p .07, gp 0.01, see were randomly assigned to a condition in a 2 (Audi- Figure 4. While microblogging did not differ as a ence size: 20 vs. 200) 9 2 (Channel: direct message function of audience size (F < 1), participants were vs. microblog) between-subjects design. less likely to direct message when the audience size First, we induced social apprehension. Similar to was 200 (M = 3.19, SD = 2.41) than when it was 20 Study 1 and Study 2, all participants wrote about a (M = 4.01, SD = 2.64), F(1, 396) = 4.85, p = .03, 2 = time when they had felt socially apprehensive. gp 0.01. Thus, when larger audience size means Second, we manipulated audience size. Partici- bothering more people, as it does in direct messag- pants were asked to imagine they were using a ing, it actually decreases socially apprehensive indi- new online social network and, depending on viduals’ willingness to share, consistent with our condition, were told that they either had 20 or 200 suggestion that not bothering others is part of what connections. drives microblogging among these individuals. Third, we manipulated communication channel. In the microblogging condition, participants were Discussion told that the new online social network allowed for easy sharing of undirected messages. Messages Study 5 provides further evidence that microblog- would be pushed to their online social network ging among socially apprehensive individuals is connections, who would see the message in a News driven by its undirected nature and casts doubt on Feed and who would be able to respond. In the an alternative explanation based on audience size. direct message condition, participants were told First, as theorized and consistent with the first that the new online social network allowed for easy four studies, socially apprehensive participants sharing of directed messages. Messages would be were more likely to share on a hypothetical new sent to some or all of their social connections, who online social network when sharing was undirected would be notified about the message in their inbox (i.e., microblog) than when it was directed (i.e., and who would be able to respond. direct message). Second, while an explanation based Participants were then asked to think about shar- on wanting to share with more others would sug- ing a message with all of their connections, and they gest audience size should increase sharing across were asked how likely they were to express them- selves to their social connections (1 = Very Unlikely; 20 200 9 = Very Likely), either via microblog (microblog con- 4.5 dition) or direct message (direct message condition). 4 Finally, as manipulation checks, participants 3.5 were asked about their perceived number of indi- 3 viduals who would be recipients of their message (1 = Very Few;7= Very Many). 2.5 2 1.5

Results Sharing likelihood 1 Manipulation Check. As expected, a 2 (Audi- 0.5 ence size) 9 2 (Channel) ANOVA on perceived 0 audience size revealed only a main effect of Audi- MB DM ence size; perceived audience size was greater in the 200 social connections condition (M = 4.40, Figure 4. Sharing across channels as a function of audience size. Social Apprehension and Microblogging 51 channels, that was not the case. In fact, audience size Tonioni et al., 2012). By demonstrating how online actually had a negative effect on willingness to share social networks can foster connections for people via direct message, consistent with the notion that who otherwise feel apprehensive about sharing, we socially apprehensive individuals want to avoid illustrate how online social network use can encour- bothering others. age communication because of how it differs from face-to-face interaction. Unlike personal interaction, microblogs allow people to foster social connections without having to reach out to anyone in particular. General Discussion Second, we shed more light on when and why This paper examines when and why people micro- people share online. Specifically, while microblogs (e.g., tweet or share status updates). We certainly share antecedents of other communication demonstrate that part of microblogging’s appeal lies channels/online social network features and are in its undirected nature. By allowing users to send surely used to share information (Hoffman & undirected communication to multiple others, some Novak, 2012), communicate identity (Nadkarni & people value microblogging when they feel socially Hofmann, 2012), and establish social status (Toubia apprehensive because it allows them to reach out & Stephen, 2014), we isolate a new and unique fea- when they desire social interaction and support ture that provides additional benefits. In particular, without having to bother anyone in particular. the undirected nature of microblogging allows peo- Five studies support this theorizing. Whether ple to reach out without feeling that they are measured (Studies 3 and 4) or manipulated (Studies imposing or bothering anyone. Thus, while previ- 1, 2, and 5), social apprehension increases ous research suggests that users share online microblogging. Consistent with the notion that because it allows them to avoid eye contact, share these effects are driven by microblogging’s undi- anonymously, and write rather than engage in oral rected nature, the same effects were not observed expression (and thus have more time to think about for more directed communication channels such as what to say; Berger & Iyengar, 2012; Forest & face-to-face sharing (Study 2) or direct messaging Wood, 2012), the present research suggests that (Study 3). Supporting the idea that microblogging some online environments can provide a valuable is driven by not wanting to impose unwanted com- communication channel because the communication munication, the preference for microblogging disap- is undirected. pears when communication is welcome (Study 4), Third, the present work contributes to prior and it is mediated by feelings of not wanting to work investigating how personality factors predict bother anyone in particular (Study 2). The studies online and offline behavior. Some work suggests rule out a host of alternative explanations, includ- that what people do online is merely an extension ing a preference for written communication, avoid- of their offline personality (Back et al., 2010). We ing eye contact, a general desire to share, negative show an important instance where online and off- affect, the need to self-express, and audience size. line behaviors diverge. We find that while social The present research thus reveals one reason why apprehension decreases willingness to share face-to- people value communicating through online social face (see also Wrench et al., 2008), it increases will- networks. By allowing people to reach out to others ingness to share via microblog. without having to impose and potentially bother Overall then, the present research suggests that anyone in particular, undirected communication online channels can provide important communica- channels may provide an easily accessible social tion opportunities for people who might not other- support system when such support is needed. wise feel comfortable sharing (see also McKenna & Bargh, 1998). By isolating and highlighting the ben- efits of undirected communication, this work can Theoretical Implications help explain the underlying mechanism of related The current research contributes to understand- findings, such as why anxiety is related to increased ing online transmission in three key ways. First, it use (Caplan, 2006) or why individuals who speaks to the upsides and downsides of online score low on self-esteem (a personality variable social networks. Although online social networks associated with social apprehension; Leary & can encourage social connections by enabling users MacDonald, 2003) are more likely to gain social to conveniently stay in touch with more friends capital from Facebook and find Facebook a safe (Ellison et al., 2007), they have also been criticized environment to express themselves (Forest & Wood, for reducing personal interaction (Kraut et al., 1998; 2012; Steinfield, Ellison, & Lampe, 2008). 52 Buechel and Berger

forces behind socially apprehensive driven Limitations and Future Directions microblogging. As with any new topic, many important ques- tions remain. One important question is what Implications for Well-Being and Practice socially apprehensive individuals share when they microblog. While one might assume that people The present research suggests that microblogs talk about their social apprehension or negative may provide valuable benefits for some individuals emotions they are experiencing, the main goal of and populations (e.g., clinical populations with con- sharing in these situations is likely to gain social ditions that are associated with social apprehension). support via attention, affection, and comfort. This Their undirected nature allows people to reach out suggests that savvy posters may not necessarily and potentially receive social support from online post negative emotion, but they might share what- friends when they might not feel comfortable doing ever is most likely to generate a response from so offline. This could deepen social bonds and gener- the audience, even if that has nothing to do with ate future offline interactions. It could also bolster how the sharers themselves are feeling at the well-being after negative affective experiences moment. Consequently, more work is needed to because it generates real and anticipated social sup- thoroughly investigate what people share and port (Buechel & Berger, 2012; Rime, 2009). Impor- how socially apprehensive individuals use online tantly, there is evidence that these benefits accrue social networks to meet their social desires and even before social support is provided. Initial goals. research shows that anticipating social support alone Another important question is to what extent can boosts well-being (Buechel & Berger, 2012). It is our findings generalize across various platforms possible, then, that the anticipation of responses from and types of post. As mentioned in the introduc- online friends might outperform offline social sup- tion, the term “microblogging” is used to describe port. The number of responses is less finite and less the sharing of short messages with other online certain in an online environment, and the prolonged social network users, independent of the platform (optimistic) anticipation of potential responses might used to share (i.e., Twitter, Facebook, etc.). In line help mitigate negative feelings associated with social with this platform independent terminology, our apprehension (Tormala, Jia, & Norton, 2012). studies reveal that the effect of social apprehension In conclusion, while online social networks have on microblogging manifests itself similarly across changed the way people communicate, we still the two major networks, Facebook and Twitter, and know relatively little about their appeal and their extends to hypothetical online social networks. Still, impact. Rather than being homogenous, sites such it is important to keep in mind that the various as Facebook and Twitter provide multiple sub- online social networks are dynamic and that they channels or methods of communication. Each of differ on many dimensions such as the number of these may have different benefits and downsides. network ties, tie strength, norms, character restric- We have shown that the microblogging feature on tions, settings, etc. (Hampton et al., 2011). online social networks provides an avenue for peo- Thus, future conceptual and empirical work is ple to connect with others even when they might needed to further understand how the microblog- feel apprehensive about doing so via traditional ging feature differs across platforms and how these channels. Microblogs therefore provide a unique differences affect socially apprehensive driven and valuable channel of communication. microblogging. The content of socially apprehensive driven microblogging—as well as the resulting interpersonal and psychological effects—may differ Appendix for Twitter, Facebook, and other online social net- Pilot Study work, for example. Future work should also test whether our effect generalizes to posts that do not In total, 100 participants recruited from an online = fall into the narrow definition of text-based micro- panel (52 females and 48 males; Mage 35.78 years, blogs, namely whether it increases posting of stories SD = 12.51) completed the study. Participants filled on or photos on Instagram. Discovering out a short survey asking them how they preferred similarities and differences in sharing across plat- reaching out to friends (1 = Much prefer face-to- forms could provide important insights into the use face; 7 = Much prefer microblogging). of the microblogging feature, and it may provide a We also measured individual differences in social more nuanced understanding of the motivational apprehension (Wrench et al., 2008). Social Apprehension and Microblogging 53

Lastly, to test whether other constructs can idealization. Psychological Science, 21, 372–374. explain any observed relationship between social https://doi.org/10.1177/0956797609360756 apprehension and microblogging, participants com- Berger, J. (2011). Arousal increases social transmission of – pleted the Ten Item Personality Inventory (Gosling, information. Psychological Science, 22, 891 893. Rentfrow, & Swann, 2003), the Narcissistic Person- https://doi.org/10.1177/0956797611413294 Berger, J. (2014). Word-of-mouth and interpersonal com- ality Inventory (Raskin & Terry, 1988), and the munication: A review and directions for future Rosenberg Self-Esteem Scale (Rosenberg, 1965). research. Journal of Consumer Psychology, 24, 586–607. https://doi.org/10.1016/j.jcps.2014.05.002 Berger, J., & Iyengar, R. (2012). How interest shapes Results word-of-mouth over different channels. Unpublished working paper, The Wharton School of the University As predicted, higher levels of social apprehension of Pennsylvania, Philadelphia, PA. were associated with higher relative preference for Berger, J., & Milkman, K. L. (2012). What makes online microblogging (as opposed to personal communica- content viral? Journal of Research, 49(2), 192– tion), b = 0.35, t(99) = 3.75, p < .001; R2 = 0.13. 205. Bowlby, J. (1977). The making and breaking of affectional This relationship was not driven by extraversion, bonds. II. Some principles of psychotherapy. The fifti- narcissism, or self-esteem. Even when measures of = < eth Maudsley Lecture. The British Journal of Psychiatry, extraversion (Gosling et al., 2003; b 0.04, t 1), 130, 421–431. https://doi.org/10.1192/bjp.130.5.421 = narcissism (Raskin & Terry, 1988; b 0.18, t Buechel, E. C., & Berger, J. (2012). Facebook therapy? (95) = 1.76, p = .08), and self-esteem (Rosenberg, Why people share self-relevant content online. Unpub- 1965; b = À0.02, t < 1) were included in the model lished Manuscript, University of South Carolina, (R2 = 0.14), the link between social apprehension Columbia, SC. and microblogging still persisted, b = 0.40, t Buechel, E. C., & Berger, J. (2015). Motivations for con- (95) = 2.55, p = .01. This indicates that while sumer engagement with social media. In C. Dimofte, C. C. Haugtvedt & R. Yalch (Eds.), Consumer psychology in extraversion, narcissism, and self-worth may – explain certain aspects of online behavior, a social media world (pp. 3 22). New York: Routledge. Buechel, E. C., Zhang, J., Morewedge, C. K., & Vosgerau, microblogging is most closely linked to social J. (2014). More intense experiences, less intense fore- apprehension, pointing to a unique link between casts: Why people overweight probability specifications social apprehension and microblogging. In fact, in affective forecasts. Journal of Personality and Social Psy- looked at individually, extraversion alone nega- chology, 106,20–36. https://doi.org/10.1037/a0034478 tively predicted a preference for microblogging Buffardi, L. E., & Campbell, W. K. (2008). Narcissism and (b = À0.21, t(99) = À2.12, p = .04), and narcissism social networking sites. Personality and Social Psychology alone was not related to microblogging (b = 0.09, Bulletin, 34, 1303–1314. https://doi.org/10.1177/ t < 1,p> .37). Thus, while extraverts may certainly 0146167208320061 use online social networks to maintain social ties Burleson, B. R. (1998). Similarities in social skills, interper- (Gosling et al., 2011; Ross et al., 2009), and narcis- sonal attraction, and the development of personal rela- sists may use them for self-promotion (Buffardi & tionships. In J. S. Trent (Ed.), Communication: Views from the helm for the twenty-first century, (pp. 77–84). Boston, Campbell, 2008), the microblogging feature of MA: Allyn and Bacon. online social networks does not seem to provide Buss, A. H. (1980). Self-consciousness and social anxiety. unique appeal to these individuals above and New York: Freeman. fl beyond the of ine environment. Instead, Calsyn, R. J., Winter, J. P., & Burger, G. K. (2005). The microblogging seems to be driven by social appre- relationship between social anxiety and social support hension. in adolescents: A test of competing causal models. Ado- lescence, 40, 103–113. Caplan, S. E. (2006). Relations among loneliness, social References anxiety, and problematic Internet use. Cyber Psychology & Behavior, 10, 234–242. Aiken, L., & West, S. G. (1991). Multiple regression: Testing Correa, T., Hinsley, A. W., & De Zuniga, H. G. (2010). and interpreting interactions. Newbury Park, CA: Sage. Who interacts on the Web?: The intersection of users’ Andersen, P. A., & Guerrero, L. K. (1997). Handbook of personality and social media use. Computers in Human communication and emotion: Research, theory, applications, Behavior, 26, 247–253. https://doi.org/10.1016/j.chb. and contexts. San Diego, CA: Academic Press. 2009.09.003 Back, M. D., Stopfer, J. M., Vazire, S., Gaddis, S., Sch- DiSalvo, D. (2010). Are social networks messing with mukle, S. C., Egloff, B., & Gosling, S. D. (2010). Face- your head? Scientific American Mind, 20,48–55. book profiles reflect actual personality, not self- https://doi.org/10.1038/scientificamericanmind0110-48 54 Buechel and Berger

Ellison, N. B., Steinfield, C., & Lampe, C. (2007). The ben- Internetlivestats (2017). Twitter Usage Statistics. Retrieved efits of Facebook ‘friends:’ Social capital and college from http://www.internetlivestats.com/twitter-statistic students’ use of online social network sites. Journal of s/. Computer-Mediated Communication, 12, 1143–1168. Jones, W. H., & Carpenter, B. N. (1986). Shyness, social https://doi.org/10.1111/jcmc.2007.12.issue-4 behavior, and relationships. In W. H. Jones, J. M. Forest, A. L., & Wood, J. (2012). When social networking Cheek, & S. R. Briggs (Eds.), Shyness: Perspectives on is not working: Individuals with low self-esteem recog- research and treatment (pp. 227–238). New York: Plenum nize but do not reap the benefits of self-disclosure on Press. https://doi.org/10.1007/978-1-4899-0525-3 Facebook. Psychological Science, 23, 296–305. Kaplan, A. M., & Haenlein, M. (2011). The early bird Gonzales, A. L., & Hancock, J. T. (2011). Mirror, mirror catches the news: Nine things you should know about on my Facebook wall: Effects of exposure to Facebook micro-blogging. Business Horizons, 54, 105–113. on self-esteem. Cyberpsychology, Behavior, and Social Net- https://doi.org/10.1016/j.bushor.2010.09.004 working, 14,79–83. https://doi.org/10.1089/cyber.2009. Keller, E., & Fay, B. (2012). Word-of-mouth advocacy: A 0411 new key to advertising effectiveness. Journal of Advertis- Gosling, S. (2009). The ancient psychological roots of ing Research, 52, 459–464. https://doi.org/10.2501/JAR- Facebook behavior. The Harvard Business Review. 52-4-459-464 Retrieved from http://www.blogs.hbr.org/now-new- Keller, E., & Libai, B. (2009). A holistic approach to the mea- next/2009/03/the-ancient-psycho- logical-root.html. surement of WOM. Paper presented at the ESOMAR Gosling, S. D., Augustine, A. A., Vazire, S., Holtzman, N. Worldwide Media Measurement Conference, Stock- S., & Gaddis, S. (2011). Manifestations of personality in holm. online social networks: Self-reported Facebook-related Kessler, R. C., Stein, M. B., & Berglund, P. (1998). Social behaviors and observable profile information. Cyberpsy- phobia subtypes in the national comorbidity survey. chology, Behavior, and Social Networking, 14, 438–488. American Journal of Psychiatry, 155, 613–619. https://d Gosling, S. D., Rentfrow, P. J., & Swann Jr, W. B. (2003). oi.org/10.1176/ajp.155.5.613 A very brief measure of the Big-Five personality Kraut, R., Patterson, M., Lundmark, V., Kiesler, S., domains. Journal of Research in Personality, 37, 504–528. Mukophadhyay, T., & Scherlis, W. (1998). Internet https://doi.org/10.1016/S0092-6566(03)00046-1 paradox: A social technology that reduces social Hampton, K., Goulet, L. S., Rainie, L., & Purcell, K. involvement and psychological well-being? American (2011). Social networking sites and our lives. Retrieved Psychologist, 53, 1017–1031. https://doi.org/10.1037/ from http://pewinternet.org/Reports/2011/Technol 0003-066X.53.9.1017 ogy-and-social-networks/Summary.aspx. Kross, E., Verduyn, P., Demiralp, E., Park, J., Lee, D. S., Lin, Harlow, H. (1961). The development of affectional behav- N., . . . Ybarra, O. (2013). Facebook use predicts declines ior patterns in infant monkeys, in determinants of in subjective well-being in young adults. PLoS ONE, 8, infants’ behavior. In B. M. Foss (Ed.), Tavistock seminar e69841. https://doi.org/10.1371/journal.pone.0069841 on mother and infant interaction (pp. 75–97). London: Leary, M. R., & MacDonald, G. (2003). Individual differ- Wiley. ences in self-esteem: A review and theoretical integra- Hayes, A. F. (2012). PROCESS: A versatile computa- tion. In M. R. Leary, & J. P. Tangney (Eds.), Handbook of tional tool for observed variable mediation, modera- self and identity. New York: Guilford Press. tion, and conditional process modeling [White paper]. Lee, C. K. (2011). Micro-blogging and status updates on Retrieved from http://www.afhayes.com/public/pro Facebook: Texts and practices. In C. Thurlow, & K. cess2012.pdf. Mroczek (Eds.), Digital discourse: Language in the new Henderson, L., Gilbert, P., & Zimbardo, P. (2014). Shy- media, (pp. 110–128). New York, NY: Oxford University ness, social anxiety, and social phobia. In S. G. Hoff- Press Inc. man, & P. M. DiBartolo (Eds.), Social anxiety: Clinical, McCroskey, J. C. (1970). Measures of communication- developmental, and social perspectives (pp. 95–115). Wal- bound anxiety. Speech Monographs, 37, 269–277. tham, MA: Elsevier Inc. https://doi.org/10.1016/B978- https://doi.org/10.1080/03637757009375677 0-12-394427-6.00004-2 McCroskey, J. C. (2001). An introduction to rhetorical com- Hewig, J., Hagemann, D., Seifert, J., Gollwitzer, M., Nau- munication. Boston: Allyn and Bacon. mann, E., & Bartussek, D. (2005). A revised film set for McKenna, K. Y. A., & Bargh, J. A. (1998). Coming out in the induction of basic emotion. Cognition & Emotion, 19, the age of the internet: Identity ‘demarginalization’ 1095–1109. https://doi.org/10.1080/02699930541000084 through virtual group participation. Journal of Personal- Hoffman, D. L., & Novak, T. P. (2012). Why do people use ity and Social Psychology, 75, 681–694. https://doi.org/ social media? Empirical findings and a new theoretical 10.1037/0022-3514.75.3.681 framework for social media goal pursuit. Unpublished Nadkarni, A., & Hofmann, S. G. (2012). Why do people working paper, George Washington School of Business, use Facebook? Personality and Individual Differences, 52, Washington, DC. 243–249. https://doi.org/10.1016/j.paid.2011.11.007 Horwitz, B. (2002). Communication apprehension: Origins Naylor, R. W., Lamberton, C. P., & West, P. M. (2012). and management. Albany, NY: Singular. Beyond the “like” button: The impact of mere virtual Social Apprehension and Microblogging 55

presence on brand evaluations and purchase intentions Psychology, 103, 567–583. https://doi.org/10.1037/ in social media settings. Journal of Marketing, 76, 105– a0029227 120. https://doi.org/10.1509/jm.11.0105 Toubia, O., & Stephen, A. T. (2014). Intrinsic vs. image- Peters, K., Kashima, Y., & Clark, A. (2009). Talking about related utility in social media: Why do people con- others: Emotionality and the dissemination of social tribute content to twitter? Marketing Science, 32, 368– information. European Journal of Social Psychology, 39, 392. 207–222. https://doi.org/10.1002/ejsp.v39:2 Turkle, S. (2015). Reclaiming conversation: The power of talk Pilkonis, P. A. (1977). The behavioral consequences of in a digital age. New York: Penguin Books. shyness. Journal of Personality, 45, 596–611. https://doi. Verduyn, P., Lee, D., Park, J., Shablack, H., Orvell, A., org/10.1111/jopy.1977.45.issue-4 Bayer, J., . . . Kross, E. (2015). Passive Facebook usage Raskin, R., & Terry, H. (1988). A principal-components undermines affective well-being: Experimental and lon- analysis of the narcissistic personality inventory and gitudinal evidence. Journal of Experimental Psychology: further evidence of its construct validity. Journal of Per- General, 144, 480–488. https://doi.org/10.1037/xge sonality and Social Psychology, 54, 890–902. https://doi. 0000057 org/10.1037/0022-3514.54.5.890 Walasek, L., Bhatia, S., & Brown, G. D. A. (2018). Posi- Rime, B. (2009). Emotion elicits the social sharing of emo- tional goods and the social rank hypothesis: Income tion: Theory and empirical review. Emotion Review, 1, inequality affects online chatter about high- and low- 60–85. https://doi.org/10.1177/1754073908097189 status brands on Twitter. Journal of Consumer Psychol- Rosenberg, M. (1965). Rosenberg self-esteem scale (RSE). ogy, 28,40–55. https://doi.org/10.1002/jcpy.1012 Acceptance and Commitment Therapy. Measures Pack- Wang, Y., Komanduri, S., Leo, P. G., Norcie, G., Acquisti, age, 61. A., & Cranor, L. (2011). ‘I regretted the minute I pressed Ross, C., Orr, E. S., Sisic, M., Arseneault, J. M., Simmer- share’: A qualitative study of regrets on Facebook. Paper ing, M. G., & Orr, R. R. (2009). Personality and motiva- presented in Symposium on Usable Privacy and Secu- tions associated with Facebook use. Computers in rity (SOUPS), Pittsburgh, PA. Human Behavior, 25, 578–586. https://doi.org/10.1016/ Wilcox, K., & Stephen, A. T. (2013). Are close friends the j.chb.2008.12.024 enemy? Online social networks, narcissism, and self- Schroeder, M. L., Wormworth, J. A., & Livesley, W. J. control. Journal of Consumer Research, 40,90–103. (1992). Dimensions of personality disorder and their https://doi.org/10.1086/668794 relationships to the big five dimensions of personality. Wilson, R. E., Gosling, S. D., & Graham, L. T. (2012). A Psychological Assessment, 4,47–53. https://doi.org/10. review of Facebook research in the social sciences. Per- 1037/1040-3590.4.1.47 spectives on Psychological Science, 7, 203–220. https://d Steinfield, C., Ellison, N. B., & Lampe, C. (2008). Social oi.org/10.1177/1745691612442904 capital, self-esteem, and use of online social network Wrench, J. S., Brogan, S. M., McCroskey, J. C., & Jowi, D. sites: A longitudinal analysis. Journal of Applied Develop- (2008). Social communication apprehension: The inter- mental Psychology, 29, 434–445. https://doi.org/10. section of communication apprehension and social pho- 1016/j.appdev.2008.07.002 bia. Human Communication, 11, 409–430. Tamir, D. I., & Mitchell, J. P. (2012). Disclosing informa- Zech, E., & Rime, B. (2005). Is talking about an emotional tion about the self is intrinsically rewarding. Proceedings experience helpful? Effects on emotional recovery and of the National Academy of Sciences, 109, 8038–8043. perceived benefits. Clinical Psychology & Psychotherapy, https://doi.org/10.1073/pnas.1202129109 12, 260–287. Taylor, S. E. (2011). Social support: A review. In H. S. Friedman (Ed.), The handbook of health psychology (pp. 189–214). New York: Oxford University Press. Supporting Information Tonioni, F., D’Alessandris, L., Lai, C., Martinelli, D., Cor- vino, S., Vasale, M., . . . Bria, P. (2012). Internet addic- Additional supporting information may be found in tion: Hours spent online, behaviors and psychological the online version of this article at the publisher’s symptoms. General Hospital Psychiatry, 34,80–87. : https://doi.org/10.1016/j.genhosppsych.2011.09.013 Appendix S1. Auxiliary Study Details. Tormala, Z. L., Jia, J. S., & Norton, M. I. (2012). The pref- Appendix S2. Methodological Details. erence for potential. Journal of Personality and Social