Computers in Human Behavior 78 (2018) 298e305 Contents lists available at ScienceDirect Computers in Human Behavior journal homepage: www.elsevier.com/locate/comphumbeh Recovering from social exclusion: The interplay of subtle Facebook reminders and collectivistic orientation Judith Knausenberger*, Gerald Echterhoff Department of Psychology, University of Münster, Fliednerstr. 21, 48149, Münster, Germany a r t i c l e i n f o a b s t r a c t Article history: Received 28 October 2016 Received in revised form 29 June 2017 Accepted 10 October 2017 Available online 12 October 2017 The sense of being connected with others through Social Networking Sites (SNS) can counteract feelings of loneliness. We examined whether subtle reminders of Facebook, the largest SNS, mitigate people's responses to ostracism (i.e., being ignored and excluded), taking into account individual differences in collectivistic orientation. We examined two typical responses to ostracism: interest in social contact (Study 1) and the need to belong (Study 2). After being included or ostracized in Cyberball, a Facebook (vs. Word) icon was displayed on the margin of a computer screen while participants focused on a primary task. We found three-way interactions between icon presentation, ostracism, and participants' horizontal collectivism. In the Facebook-icon (vs. Word-icon) condition, ostracized (vs. included) participants with higher (vs. lower) collectivistic orientation exhibited less compensatory interest in social contact (Study 1) and a greater recovery of their need to belong (Study 2). A meta-analytical synthesis of the effect sizes from the two studies suggests a signiﬁcant, medium-sized effect of the moderation by collectivistic orientation. The ﬁndings suggest that thoughts about being connected through SNS can help people cope with threats to belonging, especially those who cherish relationships with equal others. © 2017 Elsevier Ltd. All rights reserved. Keywords: Facebook Social networking sites Ostracism Belonging Collectivism 1. Introduction More than a billion people use Facebook daily, making it the world's largest social networking site (Facebook, 2016). A common reason to use Facebook is to connect with friends (Bonds-Raacke & Raacke, 2010), and people can compensate feelings of loneliness by actively using Facebook (Sheldon, Abad, & Hinsch, 2011). We reasoned that Facebook may help people cope with adverse social experiences that threaten their sense of belonging (Chiou, Lee, & Liao, 2015; Knausenberger, Hellmann, & Echterhoff, 2015). More precisely, we hypothesized that reminding people of Facebook can attenuate typical responses to ostracism, that is, interest in social contact and the threat of the fundamental psychological need to belong (Williams & Nida, 2011; Williams, 2009). Ostracism (i.e., being ignored and excluded) is a common adverse experience, which imperils fundamental needs (Williams, 2009). According to Williams' temporal need-threat model, ostracism thus threatens the need for self-esteem, meaningful existence, control, and the need to belong. Responses to ostracism can occur at * Corresponding author. E-mail address: firstname.lastname@example.org (J. Knausenberger). https://doi.org/10.1016/j.chb.2017.10.012 0747-5632/© 2017 Elsevier Ltd. All rights reserved. either of three stages: an immediate, reﬂexive stage; a delayed, reﬂective stage; and a resignation stage, which encompasses responses to long-term, repeated ostracism. Here we focus on the reﬂexive and the reﬂective stage. Right after being ostracized, that is, in the reﬂexive stage, victims experience an immediate painful reaction. With the passage of time, that is, in the reﬂective stage, victims can muster resources for coping with the exclusion experience, such as thoughts of existing friendships (McConnell, Brown, Shoda, Stayton, & Martin, 2011). To the extent that victims of exclusion invoke such resources, the negative effects of ostracism are attenuated. Indeed, a recent meta-analysis suggests that attenuating factors, such as reminders of existing friendships, affect fundamental needs to a greater extent at the reﬂective (vs. reﬂexive) stage (Hartgerink, van Beest, Wicherts, & Williams, 2015). During the reﬂective stage, targets of ostracism can replenish their threatened needs mainly by exhibiting aggression (Leary, Twenge, & Quinlivan, 2006; Warburton, Williams, & Cairns, 2006) or reafﬁliative behavior (Derﬂer-Rozin, Pillutla, & Thau, 2010; Maner, DeWall, Baumeister, & Schaller, 2007; Warburton et al., 2006). Aggression can restore the feeling of power and control, whereas reafﬁliation can primarily restore the need to belong (Williams & Nida, 2011). Compensatory reafﬁliation becomes less necessary when ostracized people recall existing social J. Knausenberger, G. Echterhoff / Computers in Human Behavior 78 (2018) 298e305 connections. Common social connections are those with a closerelationship partner, family members, or friends (Kaplan et al., 1988). Indeed, it has been found that rejected participants experienced a reduced threat to need fulﬁllment when they wrote about their best friend (McConnell et al., 2011). Nowadays, social connectedness can be experienced via Facebook (Grieve, Indian, Witteveen, Tolan, & Marrington, 2013), and the Facebook icon has come to represent social networks of friends. The icon is ubiquitous in today's digitalized world, for instance in commercial ads, in contact information of professionals and companies, and on tablet and phone screens. Initial evidence suggests that the impact of ostracism can be alleviated by the incidental or subtle activation of thoughts about Facebook. Chiou et al. (2015) found that participants who were primed with the brieﬂy ﬂashed words “Facebook” and “Googleþ” (vs. control words) reported lower social distress after ostracism. Furthermore, a study by Knausenberger et al. (2015) suggests that subtle reminders of Facebook may temper ostracized participants' compensatory interest in social contact restoration. However, these studies provide limited evidence because they did not assess a crucial process, that is, restoration of threatened needs.1 Ostracism threatens the need to belong (Williams, 2009), and a high need to belong increases afﬁliative behavior (Baumeister & Leary, 1995). We argue that the presentation of a Facebook icon should fulﬁll the need to belong, which leads to a reduced interest in social contact. In our research, we assessed the need to belong to examine whether Facebook reminders can help victims of ostracism restore this need. Another purpose of our studies was to remedy the shortcomings of previous research regarding control conditions. The study by Chiou et al. (2015) lacked a standard control condition (i.e., inclusion), and the reminder manipulation by Knausenberger et al. (2015) faced potential confounds (i.e., color effects). To examine the Facebook-reminder hypothesis more rigorously, we included a control condition for the icon presentation that eliminated these possible confounds. In so doing, we could also check the robustness of the initial ﬁndings by Knausenberger et al. Furthermore, we wanted to examine the role of an individualdifference factor that is related to social connectedness with close others, namely collectivism (Oyserman & Lee, 2008). Collectivism is relevant to the present research because people scoring high (vs. low) on collectivism are more likely to activate knowledge of social relationships (Pfundmair, Aydin, Frey, & Echterhoff, 2014; Ren, Wesselmann, & Williams, 2013). Consistent with this notion, a recent study by Over and Uskul (2016) shows that children from an interdependent (vs. independent) culture found it more likely that an ostracized person would interact with a friend to cope with the ostracism experience. We focused on the horizontal (vs. vertical) type of collectivism. Whereas vertical collectivism reﬂects social obligations from hierarchies and traditional roles, horizontal collectivism encompasses interpersonal connectedness and cooperation among equals (Triandis & Gelfand, 2012). Thus, the horizontal type of collectivism is more pertinent to the present rationale. According to previous research (Pfundmair et al., 2014; Ren et al., 2013), reminders of Facebook should more easily activate a sense of social connectedness in people with a greater (vs. lower) collectivistic orientation. Put differently, a high (vs. low) 1 We note that Knowles et al. (2015, Study 3) examined need satisfaction as a function of an adverse social experience (gaze aversion by an imagined interaction partner) and exposure to photos of Facebook friends (vs. images of trees on Flickr). However, the study authors did not assess the restoration of fundamental needs. Also, the empirical evidence was inconclusive due to a small, non-signiﬁcant effect, and the manipulations are open to various alternative explanations (e.g., familiarity of the stimuli or the presence of faces on the Facebook photos). 299 collectivistic orientation should allow ostracized people to proﬁt to a greater extent from Facebook reminders in coping with the ostracism experience. In this respect, we need to consider a potential caveat suggested by cross-cultural research. It has been argued that people from a collectivistic (vs. individualistic) culture perceive ostracism as less threatening without the need for social reminders. Indeed, in the absence of social reminders, participants from a collectivistic (vs. individualistic) culture have been found to respond with less negative emotions (Pfundmair, Graupmann, Frey, & Aydin, 2015b) and less need threat (Pfundmair, Aydin, Du, Yeung, Frey, & Graupmann, 2015a). Also, children from an interdependent (vs. independent) culture experience less negative emotions after ostracism (Over & Uskul, 2016). All of these ﬁndings were obtained with cross-cultural comparisons. However, there are reasons to suspect that individual differences in collectivism play a different role in a culturally homogeneous setting. Speciﬁcally, for people from a generally individualistic culture a more (vs. less) collectivistic orientation per se may not be sufﬁcient to reduce the effects of ostracism. Thus, we assume that within an individualistic culture, people with a relatively more collectivistic orientation still need to recruit additional resources to cope with ostracism. Consistent with this view, a study by Pfundmair et al. (2014), conducted with participants from Germany, found that a collectivistic orientation shielded against negative ostracism effects only in conjunction with the administration of oxytocin, a neuropeptide that is known to mitigate social distress. Germany is located at the individualistic end of a crosscultural comparison scale (Hofstede, 2001). Consistent with our rationale, Pfundmair et al. found that responses to ostracism were less negative in participants with a horizontal collectivistic (vs. individualistic) orientation only under oxytocin (vs. placebo). Similar to Pfundmair et al. (2014), we examined the inﬂuence of individual differences in horizontal collectivism with participants from the same, individualistic culture, that is, Germany. The above considerations suggest that reminders of one's social connectedness facilitate the activation of resources for coping with ostracism to a greater extent among participants with a higher (vs. lower) collectivistic orientation. We thus predicted that the subtle presentation of the Facebook (vs. a control) icon is more likely to mitigate responses to ostracism in participants with a greater (vs. lower) collectivistic orientation. Two typical responses to ostracism were assessed: seeking new contact (Study 1), and the recovery of need fulﬁllment from the reﬂexive (immediate) stage to the reﬂective (delayed) stage (Study 2). 2. Study 1 The purpose of Study 1was to examine whether Facebook reminders mitigate interest in social contact after ostracism (vs. inclusion), especially in participants with a higher (vs. lower) collectivistic orientation. We extended the experiment by Knausenberger et al. (2015) in two important ways. First, we included an established individual-difference measure of individualism-collectivism (INDCOL, Singelis, Triandis, Bhawuk, & Gelfand, 1995). We expected that the mitigating effect of Facebook reminders would be stronger for participants with a higher (vs. lower) collectivistic orientation. Second, we eliminated a possible confound in the experiment by Knausenberger et al. In this earlier study, the control icon (Flash Player) was red, and exposure to the color red may impact motivation and cognition (Tanaka & Tokuno, 2011). Hence, we used an icon of the same color (Microsoft Word) as the Facebook icon. 300 J. Knausenberger, G. Echterhoff / Computers in Human Behavior 78 (2018) 298e305 2.1. Method 2.1.1. Participants and design The study was based on a 2 (Cyberball experience: ostracism vs. inclusion) x 2 (Icon presentation: Facebook vs. Word) betweenparticipants design with horizontal collectivism as a continuous moderator variable. Participants were randomly assigned to the conditions. The main dependent variable was interest in social contact. We report all measures, manipulations, and excluded participant data. We calculated the sample size needed to detect an effect size of f2 ¼ 0.075 for the interaction effect of ostracism and icon presentation in a multiple regression model (based on the effect size found by Knausenberger et al., 2015). For a power of 0.80, a of 0.05 and a two-tailed test, the optimal sample size is N ¼ 107 (calculated with G*Power 3.1; Faul, Erdfelder, Buchner, & Lang, 2009). To allow sample shrinkage of about 10%, we collected data from 119 students of a German university. We advertised the study in the local Psychology Department and in an e-mail newsletter circulated to all students. There were no a priori selection criteria for participants apart from being a student. After data collection, 13 participants who reported not using Facebook were excluded from the analyses, resulting in a ﬁnal sample size of N ¼ 106 (52 male, 54 female; mean age ¼ 23.59, SD ¼ 2.98). Facebook use was not deﬁned as a prior selection criterion because we did not want to prime Facebook use for all participants. Instead, Facebook use was probed at the end of the study. Participants received course credit or 6V for participation. 2.1.2. Procedure and materials The study was ostensibly about mental visualization. After arriving at the lab, participants ﬁrst played Cyberball, an online ball tossing game (Williams, Cheung, & Choi, 2000), described as an alleged training for mental visualization. Cyberball is a wellestablished procedure for manipulating ostracism. A recent metaanalysis found that Cyberball has robust and large effects on standard outcome measures such as need threat (Hartgerink et al., 2015). Participants played Cyberball with two ostensible other participants. Unbeknownst to the participants, the ball tosses of the two co-players followed a pre-programmed pattern: In the ostracism condition, participants received the ball only twice at the beginning of the game, and then no more. In the inclusion condition, participants received the ball in one third of 30 tosses. Afterwards, the icon manipulation was administered by presenting either a Facebook or Microsoft Word icon in the left corner of the computer screen while participants completed rating items from two scales. These two scales were the horizontal subscale of the Individualism and Collectivism Scale (INDCOL; Singelis et al., 1995) and the Gießen-Test II, which assesses four personality traits including social openness and social approach orientation (Beckmann, Br€ ahler, & Richter, 2012). For the sake of parsimony, vertical INDCOL items were omitted. INDCOL is an established and validated questionnaire for the assessment of individualism and collectivism (Triandis & Gelfand, 1998). A key advantage of the icon presentation on the margin of the computer screen is its high ecological validity: Nowadays, people often encounter Facebook icons in similar ways, for instance, on the screens of user devices or in commercials. Also, a similar type of icon manipulation has been successfully employed in the earlier study by Knausenberger et al. (2015). Next, participants completed a ﬁller task on visual perspective taking to support the plausibility of the ostensible topic and to create an interval between the icon manipulation and the measurement of the dependent measures. Next, participants rated their mood on two items (“Are you in a good or a bad mood right now?” and “How is your mood at the moment?”, on 7-point scales from 1 ¼ very bad to 7 ¼ very good). We then measured the dependent variable, interest in social contact. Interest in social contact was operationalized as in Knausenberger et al. (2015): Participants were told they could receive tickets for a guided city tour, alone or together with two friends. They were asked to rate how much they would like to win the tickets and how important it was to them (on scales from 1 ¼ “not at all” to 7 ¼ “very much”). Additionally, they rated their interest in participating in a new social network at the university on four rating items (from 1 ¼ “not at all” to 7 ¼ “very much”). The six items from both scales were sufﬁciently intercorrelated (Cronbach's a ¼ 0.84) and thus averaged to form an integrated score of interest in social contact. We then administered the Reﬂective Needs Questionnaire (Williams, 2009) to assess the need to belong and three other fundamental needs (self-esteem, control, and meaningful existence). The Reﬂective Needs Questionnaire is established as a valid measure (Hartgerink et al., 2015). Finally, participants answered manipulation checks regarding the effectiveness of the Cyberball game and icon recognition. Cyberball effectiveness was examined with two questions („To what extent were you excluded by the participants during the game?” and „What percentage of the tosses, do you think, were thrown to you?“). Icon recognition was measured by asking participants whether they noticed an icon and if so whether it was a Facebook, Word, Flash Player or Microsoft Excel icon. After participants had reported if they were Facebook users, they indicated the extent to which they used Facebook to afﬁliate with friends (relational Facebook use; four items; Cronbach's a ¼ 0.72). They also rated the importance of Facebook to its users and to themselves (importance of Facebook, three items; Cronbach's a ¼ 0.60) and the extent to which the Facebook and Word icons represent socioemotional needs (Cronbach's a ¼ 0.82 and 0.84 for Facebook and Word, respectively). In the end, participants were probed for suspicion, debriefed, and compensated for their participation (for details, see SOM). 2.2. Results and discussion 2.2.1. Manipulation checks Participants in the ostracism condition felt signiﬁcantly more excluded (M ¼ 4.10, SD ¼ 0.60) than did participants in the inclusion condition (M ¼ 2.28, SD ¼ 1.07), F(1,102) ¼ 113.62, p < 0.001, h2p ¼ 0.53. Ostracized participants felt that they received a smaller percentage of tosses (M ¼ 11.85, SD ¼ 12.79) than did included participants (M ¼ 32.35, SD ¼ 12.03), F(1,102) ¼ 71.91, p < 0.001, h2p ¼ 0.41. The main effects of icon presentation and the interactions were non-signiﬁcant, Fs < 1, ps > 0.352. Participants indicated that the Facebook (M ¼ 2.78, SD ¼ 1.25) icon represented socioemotional needs to a greater extent than Word (M ¼ 1.28, SD ¼ 0.58), t(105) ¼ 11.533, p < 0.001, d ¼ 1.52. All four reﬂective needs were more threatened after ostracism than inclusion, Fs (1,102) > 8.54, ps < 0.005 (see Table 1). Other Table 1 Means (with standard deviations) for reﬂective need fulﬁllment as a function of ostracism experience. Need Need Need Need to belong for control for self-esteem for meaningful existence Ostracism Inclusion 2.28 1.72 3.43 3.71 4.12 2.95 3.81 4.24 (0.77) (0.42) (0.71) (0.90) (0.84) (0.89) (0.64) (0.57) J. Knausenberger, G. Echterhoff / Computers in Human Behavior 78 (2018) 298e305 main effects of icon presentation and interaction effects were not signiﬁcant, Fs < 3.00, ps > 0.086. 2.2.2. Interest in social contact The data depicted in Fig. 1 suggest that compensatory interest in social contact after ostracism (vs. inclusion) was reduced by the Facebook-icon presentation primarily in participants high on horizontal collectivism. For a statistical test of this effect, we conducted a General Linear Model (GLM) analysis (Field, Miles, & Field, 2012) to predict interest in social contact as a function of Cyberball experience, icon presentation, horizontal collectivism, and the respective interaction effects of the three variables. The analysis revealed a signiﬁcant three-way interaction between Cyberball experience, icon presentation, and horizontal collectivism, b ¼ 1.17, t(98) ¼ 2.30, p ¼ 0.024, f2 ¼ 0.048, 95% CI [-2.18; 0.16] (Fig. 1). A signiﬁcant Cyberball Experience x Icon Presentation interaction emerged for participants high in horizontal collectivism (M þ 1 SD), b ¼ 1.72, t(98) ¼ 2.73, p ¼ 0.008, f2 ¼ 0.069, 95% CI [-2.98; 0.46]. In contrast, no such interaction was found for participants low (M e 1 SD) in horizontal collectivism, b ¼ 0.35, t(98) ¼ 0.55, p ¼ 0.581, f2 ¼ 0.003, 95% CI [-0.92; 1.62]. There was no moderating effect of horizontal individualism, t(98) ¼ 0.14, p ¼ 0.886 (see SOM for an ANOVA without the moderator variables). 2.2.3. Additional analyses Forty-one percent of participants correctly recognized the presented icon. For interest in social contact, the four-way interaction of correct icon identiﬁcation, icon presentation, Cyberball experience, and horizontal collectivism was non-signiﬁcant, t(90) ¼ 0.71, p ¼ 0.477. Thus, there was no evidence that correct icon identiﬁcation moderated the main ﬁnding reported above. Also, correct icon identiﬁcation had no main effect or other interaction effects on interest in social contact, ts < 1.48, ps > 0.144. None of the four scales from the Gießen-Test II (including social openness and social approach orientation) had signiﬁcant main or interaction effects on 301 interest in social contact, ts < 1.43, ps > 0.158. There was no signiﬁcant main or interaction effect of participants' mood, ts < 1.85, ps > 0.067, relational Facebook use, ts < 1.62, ps > 0.109, or Facebook importance, ts < 1.17, ps > 0.245. We also examined whether the moderation by horizontal collectivism might be driven by differences in Facebook importance and Facebook use between high and low collectivists. However, horizontal collectivism was not signiﬁcantly correlated with Facebook importance, r(104) ¼ -0.15, p ¼ 0.119, and relational Facebook use, r(104) ¼ -0.11, p ¼ 0.248. 2.2.4. Discussion In sum, we found a mitigating effect of Facebook reminders on ostracismeinduced interest in social contact in participants with a higher (vs. lower) collectivistic orientation. This ﬁnding replicates the main result from Knausenberger et al. (2015) with an improved procedure, which eliminated a possible confound with item color. The observed effects did not depend on whether participants recognized the icon presented on the margin of the computer screen. Additional analyses revealed no evidence that the effect of horizontal collectivism could be driven by social openness, social approach orientation, the subjective importance of Facebook, or the use of Facebook for afﬁliating with friends. However, to close an important lacuna in the literature, it is imperative to demonstrate effects on another outcome variable, that is, the restoration of the belongingness needs after ostracism. 3. Study 2 We designed Study 2 to examine effects of subtle Facebook reminders on the restoration of the need to belong, a basic response to ostracism. Investigating this outcome remedies a major shortcoming in previous research. According to extant research (e.g., Sheldon et al., 2011), Facebook use can increase the sense of social connection. Thus, social networking sites are a means for fulﬁlling the need to belong (Gangadharbatla, 2008). Fig. 1. Interest in social contact as a function of icon presentation (Word vs. Facebook), Cyberball experience (ostracism vs. inclusion) and horizontal collectivism. Grey areas represent 95% CIs, dark grey areas represent overlapping CIs. 302 J. Knausenberger, G. Echterhoff / Computers in Human Behavior 78 (2018) 298e305 To assess need recovery, we included measures at both the reﬂexive (immediate) stage and the reﬂective (delayed) stage. Study 1 did not include the reﬂexive-stage measure, and reﬂective needs were measured only after the main dependent variable (interest in social contact). Learning about a new opportunity for social contact may fulﬁll the need to belong, which would prevent us from detecting effects of icon presentation. Given the focus on need recovery, we thus omitted interest in social contact in Study 2. We predicted that the presentation of the Facebook icon would facilitate recovery of the ostracism-induced need to belong, especially among participants scoring high (vs. low) in horizontal collectivism. 3.1. Method 3.1.1. Participants and design The study was based on a 2 (Cyberball experience: ostracism vs. inclusion) x 2 (Icon presentation: Facebook vs. Word) betweenparticipants design with horizontal collectivism as a continuous moderator variable. Again, participants were randomly assigned to the conditions. The main dependent variable was the restoration of the need to belong. We report all measures, manipulations, and excluded participant data. Based on the same parameter values used for Study 1, the optimal sample size was N ¼ 107 (calculated with G*Power 3.1, Faul et al., 2009). Because more than 10% of participants in Study 1 did not use Facebook, we allowed for sample shrinkage of up to 20% and thus collected data from 128 students of a German university. We advertised for the study as in Study 1 and again had no further selection criteria apart from being a student. Eleven participants did not use Facebook and were excluded from the analyses, resulting in the ﬁnal sample size of N ¼ 117 (33 male, 84 female; mean age ¼ 23.22, SD ¼ 5.30). Participants received course credit or 6V for participation. 3.1.2. Procedure and materials Procedure and Materials of Study 2 were similar to those of Study 1 with the following exceptions: After the Cyberball game, reﬂexive needs were measured using the Reﬂexive Needs Questionnaire by Williams (2009), a widely used and validated measure for immediate responses to ostracism (see Hartgerink et al., 2015). During the icon presentation, we administered the General Regulatory Focus Measure (GRFM; Lockwood, Jordan, & Kunda, 2002) instead of the Gießen-Test II. After the ﬁller task, we measured reﬂective needs (again with the Reﬂective Needs Questionnaire, Williams, 2009) instead of interest in social contact. To obtain a score for need recovery, we subtracted the reﬂexive-need scores from the reﬂective-need scores for each of the four fundamental needs (i.e., need to belong, for self-esteem, for control, and for meaningful existence). Larger positive differences between the reﬂective and reﬂexive stage indicate greater need recovery from the reﬂexive to the reﬂective stage. We could thus analyze the experimental and moderator effects with a four-level dependent variable measured at two time points (Judd, Kenny, & McClelland, 2001). To reduce participants' workload, we omitted Facebook importance and mood, which had no moderating effects in Study 1. 3.2. Results and discussion 3.2.1. Manipulation checks Participants in the ostracism condition felt signiﬁcantly more excluded (M ¼ 6.36, SD ¼ 0.69) than participants in the inclusion condition, (M ¼ 3.33, SD ¼ 1.68), F(1,113) ¼ 165.37, p < 0.001, h2p ¼ 0.59. Neither the main effect of icon presentation nor the interaction reached signiﬁcance. Also, ostracized (vs. included) participants felt that they received a smaller percentage of tosses, F(1,113) ¼ 286.59, p < 0.001, h2p ¼ 0.72. There was a main effect of the icon presentation with participants in the Facebook (vs. Word) condition reporting a higher percentage of received tosses, F(1,113) ¼ 5.13, p ¼ 0.025, h2p ¼ 0.04. These main effects were qualiﬁed by an interaction of Cyberball experience and icon presentation, F(1,113) ¼ 4.13, p ¼ 0.045, h2p ¼ 0.04, with included participants reporting a higher number of received tosses after seeing a Facebook icon (M ¼ 36.10, SD ¼ 11.70) than a Word icon (M ¼ 29.90, SD ¼ 7.10), t(57) ¼ 46.16, p ¼ 0.018. Ostracized participants reported receiving a lower number of tosses both after seeing a Facebook icon (M ¼ 8.83, SD ¼ 4.99) and a Word icon (M ¼ 8.47, SD ¼ 5.58), t(58) ¼ 0.26, p ¼ 0.794. 3.2.2. Effects on fundamental needs We conducted a GLM for repeated-measures data to predict need recovery as a function of Cyberball experience, icon presentation, type of need (to belong, for self-esteem, for control, and for meaningful existence), horizontal collectivism, and the interaction effects of the four variables. Type of need was the repeatedmeasures factor. The GLM revealed a marginally signiﬁcant fourway interaction between Cyberball experience, icon presentation, type of need, and horizontal collectivism on need recovery (reﬂective minus reﬂexive scores), F(3, 327) ¼ 2.58, p ¼ 0.054, h2p ¼ 0.023. There were also signiﬁcant main effects of Cyberball experience and type of need on need recovery, F(1,109) ¼ 58.39, p < 0.001, h2p ¼ 0.349, and F(3,327) ¼ 21.19, p < 0.001, h2p ¼ 0.163, respectively. Additionally, there were signiﬁcant interaction effects of Cyberball experience and icon presentation, F(1,109) ¼ 3.97, p ¼ 0.049, h2p ¼ 0.035, and Cyberball experience and type of need, F(3,327) ¼ 8.34, p < 0.001, h2p ¼ 0.071. To decompose the four-way interaction, we conducted GLM analyses for the four fundamental needs separately, thus predicting need recovery for each type of need as a function of Cyberball experience, icon presentation, horizontal collectivism, and the interaction effects of the three variables. Regarding the recovery of the need to belong, there was a signiﬁcant three-way interaction between Cyberball experience, icon presentation, and horizontal collectivism, b ¼ 1.23, t(109) ¼ 2.07, p ¼ 0.041, f2 ¼ 0.023, 95% CI [0.05; 2.40] (Fig. 2). This result is an extended replication of the three-way interaction found in Study 1. For people low in horizontal collectivism (M e 1 SD), no interaction effect of Cyberball experience and icon presentation emerged, b ¼ 0.12, t(109) ¼ 0.22, p ¼ 0.825, f2 < 0.001, 95% CI [-0.96; 1.20]. There was however, a signiﬁcant interaction effect for people high in horizontal collectivism (M þ1 SD), b ¼ 1.81, t(109) ¼ 3.41, p < 0.001, f2 ¼ 0.066, 95% CI [0.76; 2.87]. Thus, recovery from ostracism was spurred by the Facebook-icon presentation in participants higher on horizontal collectivism. This ﬁnding demonstrates for the ﬁrst time an effect of Facebook reminders on the restoration of threatened needs. There were also signiﬁcant main effects of the three predictors and Icon Presentation x Cyberball Experience and Icon Presentation x Horizontal Collectivism interactions. These results and the results of an ANOVA without moderator variables are described in more detail in the SOM. Regarding the recovery of the three other needs, there were no effects of the icon or interaction effects of the icon (for details, see SOM). There was no moderating effect of horizontal individualism for any of the needs, ts < 0.74, ps > 0.463. 3.2.3. Additional analyses Forty-four percent of participants correctly identiﬁed the icon. There was no four-way interaction effect of correct icon identiﬁcation, icon presentation, ostracism and horizontal collectivism on J. Knausenberger, G. Echterhoff / Computers in Human Behavior 78 (2018) 298e305 303 Fig. 2. Restoration of the need to belong (reﬂective e reﬂexive scores) as a function of icon presentation (Word vs. Facebook), Cyberball experience (ostracism vs. inclusion) and horizontal collectivism. Grey areas represent 95% CIs with dark grey areas representing overlapping CIs. the restoration of the need to belong, t(101) ¼ 1.68, p ¼ 0.097 (for details, see SOM). There was no moderating effect of relational Facebook use or regulatory focus in the main analysis, ts < 1.70, ps > 0.094. Horizontal collectivism was unrelated to relational Facebook use, r(115) ¼ -0.05, p ¼ 0.612, prevention focus, r(115) ¼ 0.03, p ¼ 0.734, and promotion focus, r(115) ¼ 0.15, p ¼ 0.116. 3.2.4. Discussion In sum, after ostracized participants had been presented with the subtle Facebook (vs. Word) icons, their need to belong recovered more from the reﬂexive to the reﬂective stage, especially in participants scoring high (vs. low) in horizontal collectivism. This result supports the hypothesis that the presentation of the Facebook icon facilitates recovery of the ostracism-induced need to belong in conjunction with participants' higher (vs. lower) collectivistic orientation. Again, the effects did not depend on whether participants recognized the previously presented icon. Additional analyses revealed no evidence that the effect of horizontal collectivism was driven by individual differences in regulatory focus or relational-afﬁliative use of Facebook. These data are consistent with the notion that differences in collectivistic orientation rather than the alternative individual differences are critical for the observed moderation effect. 4. Meta-analytical estimation of moderation effects by relational Facebook use and horizontal collectivism We note that the moderation of the Cyberball Experience x Icon Presentation interaction by relational Facebook use found by Knausenberger et al. (2015) was not replicated in either of the two present studies. We thus estimated the average effect size across these three studies (Knausenberger et al., 2015; Study 1 and Study 2) with a meta-analytical procedure. The meta-analytical synthesis was conducted with the tool Meta-Essentials (Van Rhee, Suurmond, & Hak, 2015) based on the effect sizes for the threeway interaction between Cyberball experience, icon presentation, and relational Facebook use. For the data input, we used the standardized regression coefﬁcients b and calculated the meta-analysis for partial correlation coefﬁcients as suggested by Van Rhee et al. The analysis revealed an average effect size of r ¼ 0.10, p ¼ 0.168, 95% CI [-0.21; 0.40]. These ﬁndings suggest the existence of a small to non-existing effect of the moderation by relational Facebook use. We also conducted an analogous meta-analytical synthesis for the moderation effect by horizontal collectivism observed in Study 1 and Study 2. The analysis revealed an integrated effect size of r ¼ 0.21, p < 0.001, 95% CI [0.14; 0.27]. This integration of effect sizes across the two studies thus suggests a medium moderation effect by horizontal collectivism. 5. General discussion Our studies demonstrate for the ﬁrst time the interplay between reminders of social connectedness via online networks, on the one hand, and individual differences in collectivistic orientation, on the other hand, in responses to adverse social events, speciﬁcally, ostracism. It is reassuring that the three-way interaction between ostracism, icon presentation, and horizontal collectivism found in Study 1 was conceptually replicated in Study 2. It has been known that Facebook is associated with feelings of connectedness (Sheldon et al., 2011). Our studies suggest that the feelings of connectedness evoked by Facebook can, in turn, inﬂuence reactions to adverse social experiences, particularly in conjunction with a collectivistic orientation. These ﬁndings resonate well with the notion that interdependent people have more accessible social resources that they can muster in the face of adverse social events (Pfundmair et al., 2014; Ren et al., 2013). When presented with a Facebook icon, participants with collectivistic orientation can more easily activate thoughts of their Facebook connections and thus can cope better with ostracism. In Study 1, we replicated the key ﬁnding by Knausenberger et al. (2015), which showed that a subtle reminder of Facebook can attenuate interest in social contact after ostracism. Interest in social 304 J. Knausenberger, G. Echterhoff / Computers in Human Behavior 78 (2018) 298e305 contact and reafﬁliation is a common compensatory response to ostracism (Williams, 2009). Importantly, we also obtained evidence regarding the underlying motivational processes. In Study 2, Facebook reminders led to a greater recovery of the need to belong that is threatened by the previous experience of ostracism. The recovery of this fundamental need is an indispensable element in social exclusion research; moreover, it represents an important missing link between our novel manipulation of Facebook reminders and reduced reafﬁliation attempts (shown in Study 1 and Knausenberger et al., 2015). Although the majority of participants did not recognize the icon in subsequent probing, the presentation of the Facebook icon affected responses to ostracism. Also, there was no evidence that the pattern of results differed between participants who recognized (vs. did not recognize) the icon. This ﬁnding, which is consistent with the ﬁndings by Knausenberger et al. (2015), supports the notion that even a subtle and incidentally perceived Facebook reminder can attenuate the negative impact of ostracism. We assessed variations of collectivistic orientation within the same culture. Arguably, differences in initial responses to ostracism might be stronger in cross-cultural comparisons, that is, comparisons of participants from an individualistic vs. a collectivistic culture (Pfundmair, Aydin, et al., 2015a; Pfundmair, Graupmann et al., 2015b). We suspect that in a cross-cultural comparison, collectivism could mitigate the effects of ostracism without the need for additional reminders of social connectedness. Still, it is possible that Facebook reminders provide additional support for recovery from ostracism in a more collectivistic (vs. individualistic) culture. It would be exciting to see the present studies replicated with cross-cultural variations in collectivism. There was no evidence that factors other than horizontal collectivism drove the observed moderation effects. Participants with collectivistic orientation may have reacted differently because they use Facebook to a greater extent to afﬁliate or because Facebook is more important to them. However, subjective Facebook importance and Facebook use were uncorrelated with collectivism and did not interact with icon presentation. Also, there were no interaction effects involving social openness, social approach orientation, mood, and regulatory focus in the main analyses. Knausenberger et al. (2015) found that Facebook reminders diminished interest in social contact after ostracism especially in those participants who used Facebook to a greater extent for relational purposes (i.e., maintaining and strengthening friendships). This moderation by relational Facebook use was not replicated in Study 1. Also, relational Facebook use did not moderate the restoration of the need to belong in Study 2. As described above, our meta-analysis for the moderation by relational Facebook use across the three studies suggested a small to non-existent effect size. Future studies would help increase the reliability of this estimate. Previous studies suggest that effects of Facebook use on wellbeing depend on the type of Facebook use: Passive Facebook use entails lower wellbeing (Verduyn et al., 2015), and active posting on Facebook decreases loneliness (grobe Deters & Mehl, 2013). Thus, active (vs. passive) Facebook use may be more likely to shield against threats to the need to belong. However, there were no measures in our study that would distinguish between passive and active Facebook use. Future research should employ such measures to assess the role of Facebook use in a more differentiated manner. The need-threat model encompasses four fundamental needs threatened by ostracism (i.e., the need to belong, control, selfesteem, and meaningful existence; Williams, 2009), and ostracism effects are often found on a combination of these four needs (e.g. Knowles, Haycock, & Shaikh, 2015; McConnell et al., 2011; Zadro, Williams, & Richardson, 2004). In our studies, Facebook reminders affected the recovery of the need to belong, but they did not affect the recovery of the other needs. This distinctive effect supports our hypothesis that Facebook serves as a source of belongingness and can therefore speciﬁcally restore the threatened need to belong. 5.1. Conclusion Previous research examining the link between Facebook use and feelings of belongingness has been mostly correlational (e.g. BondsRaacke & Raacke, 2010; Park, Kee, & Valenzuela, 2009). In our studies, the activation of thoughts about Facebook were experimentally manipulated via subtle reminders of Facebook. Thus, our study provides more direct and more valid evidence on causal effects of Facebook on feelings of belonging. We could demonstrate that Facebook reminders trigger an increase in feelings of belonging and a reduction of a typical reaction to ostracism, that is, interest in social contact. Beyond this scientiﬁc merit, the ﬁndings also have interesting practical implications. Icons of social networks such as Facebook have become omnipresent in our everyday lives, for instance, on the screens of tablet computers and smart phone screens. Our results suggest that this omnipresence might have both positive and negative consequences. On the one hand, icons of social networks might serve as an easily accessible resource to cope with ostracism and dispense with harmful compensatory reactions to ostracism such as aggression (Leary et al., 2006). On the other hand, the ﬁndings from Study 1 suggest that subtle reminders of Facebook could diminish attempts to afﬁliate with other people who would otherwise be motivated to do so. Thus, Facebook reminders could also contribute to social withdrawal and isolation. However, the results from Study 2 also showed that Facebook reminders led to a greater recovery of the need to belong. This novel ﬁnding suggests that Facebook can actually help people cope with the experiences of social exclusion and rejection. Thus, Facebook may afford new opportunities for increasing people's sense of belonging when they need it. We are conﬁdent that a better understanding of such Facebook effects can further advance a critical and scientiﬁcally informed use of social online networks. Declaration of conﬂicting interests The authors declare that they have no conﬂicts of interest with respect to the authorship or the publication of this article. This research was conducted without support from a grant from funding agencies in the public, commercial, or not-for-proﬁt sectors. Acknowledgements We thank Elina Kisselenko and Verena Rettig for their contribution to this research and Michaela Pfundmair for her valuable comments on the manuscript. Appendix A. Supplementary data Supplementary data related to this article can be found at https://doi.org/10.1016/j.chb.2017.10.012. References Baumeister, R. F., & Leary, M. R. (1995). The need to belong: Desire for interpersonal attachments as a fundamental human motivation. Psychological Bulletin, 117(3), 497e529. https://doi.org/10.1037/0033-2909.117.3.497. €hler, E., & Richter, H.-E. (2012). Gt-II Der Gießen-test II. Bern: Huber. Beckmann, D., Bra Bonds-Raacke, J., & Raacke, J. (2010). MySpace and Facebook: Identifying dimensions of uses and gratiﬁcations for friend networking sites. Individual Differences Research, 8, 27e33. https://doi.org/10.1089/cpb.2007.0056. J. Knausenberger, G. Echterhoff / Computers in Human Behavior 78 (2018) 298e305 Chiou, W.-B., Lee, C.-C., & Liao, D.-C. (2015). Facebook effects on social distress: Priming with online social networking thoughts can alter the perceived distress due to social exclusion. Computers in Human Behavior, 49, 230e236. https:// doi.org/10.1016/j.chb.2015.02.064. Derﬂer-Rozin, R., Pillutla, M., & Thau, S. (2010). Social reconnection revisited: The effects of social exclusion risk on reciprocity, trust, and general risk-taking. Organizational Behavior and Human Decision Processes, 112, 140e150. https:// doi.org/10.1016/j.obhdp.2010.02.005. große Deters, F., & Mehl, M. R. (2013). Does posting Facebook status updates increase or decrease loneliness? An online social networking experiment. Social Psychological and Personality Science, 4(5), 579e586. https://doi.org/10.1177/ 1948550612469233. Facebook. (2016). Statistics. Retrieved from: http://newsroom.fb.com/company-info (Accessed on April 27 2016). Faul, F., Erdfelder, E., Buchner, A., & Lang, A.-G. (2009). Statistical power analyses using G*Power 3.1: Tests for correlation and regression analyses. Behavior Research Methods, 41, 1149e1160. https://doi.org/10.3758/BRM.41.4.1149. Field, A., Miles, J., & Field, Z. (2012). Discovering statistics using R. Thousand Oaks CA: Sage. Gangadharbatla, H. (2008). Facebook me: Collective self-esteem, need to belong, and internet self-efﬁcacy as predictors of the iGeneration's attitudes toward social networking sites. Journal of Interactive Advertising, 8, 5e15. https:// doi.org/10.1080/15252019.2008.10722138. Grieve, R., Indian, M., Witteveen, K., Tolan, G. A., & Marrington, J. (2013). Face-toface or Facebook: Can social connectedness be derived online? Computers in Human Behavior, 29(3), 604e609. https://doi.org/10.1016/j.chb.2012.11.017. Hartgerink, C. H., van Beest, I., Wicherts, J. M., & Williams, K. D. (2015). The ordinal effects of ostracism: A meta-analysis of 120 Cyberball studies. PloS One, 10(5), e0127002. https://doi.org/10.1371/journal.pone.0127002. Hofstede, G. (2001). Culture's Consequences: Comparing values, behaviors, institutions, and organizations across nations (2nd ed.). Thousand Oaks CA: Sage. Judd, C. M., Kenny, D. A., & McClelland, G. H. (2001). Estimating and testing mediation and moderation in within-subject designs. Psychological Methods, 6(2), 115e134. https://doi.org/10.1037//1082-989X.6.2.115. Kaplan, G. A., Salonen, J. T., Cohen, R. D., Brand, R. J., Syme, S. L., & Puska, P. (1988). Social connections and mortality from all causes and from cardiovascular disease: Prospective evidence from eastern Finland. American Journal of Epidemiology, 128(2), 370e380. Knausenberger, J., Hellmann, J. H., & Echterhoff, G. (2015). When virtual contact is all you need: Subtle reminders of Facebook preempt social-contact restoration after exclusion. European Journal of Social Psychology, 45, 279e284. https:// doi.org/10.1002/ejsp.2035. Knowles, M. L., Haycock, N., & Shaikh, I. (2015). Does Facebook magnify or mitigate threats to belonging? Social Psychology, 46, 313e324. https://doi.org/10.1027/ 1864-9335/a000246. Leary, M. R., Twenge, J. M., & Quinlivan, E. (2006). Interpersonal rejection as a determinant of anger and aggression. Personality and Social Psychology Review, 10(2), 111e132. Lockwood, P., Jordan, C. H., & Kunda, Z. (2002). Motivation by positive or negative role models: Regulatory focus determines who will best inspire us. Journal of Personality and Social Psychology, 83, 854e864. https://doi.org/10.1037//00223518.104.22.1684. Maner, J. K., DeWall, C. N., Baumeister, R. F., & Schaller, M. (2007). Does social exclusion motivate withdrawal or reconnection? Resolving the “porcupine problem”. Journal of Personality and Social Psychology, 92, 42e55. https:// doi.org/10.1037/0022-3522.214.171.124. McConnell, A. R., Brown, C. M., Shoda, T. M., Stayton, L. E., & Martin, C. E. (2011). Friends with beneﬁts: On the positive consequences of pet ownership. Journal of Personality and Social Psychology, 101, 1239e1252. https://doi.org/10.1037/ a0024506. Over, H., & Uskul, A. K. (2016). Culture moderates children's responses to ostracism situations. Journal of Personality and Social Psychology, 110, 710e724. https:// doi.org/10.1037/pspi0000050. Oyserman, D., & Lee, S. W. S. (2008). Does culture inﬂuence what and how we 305 think? Effects of priming individualism and collectivism. Psychological Bulletin, 134, 311e342. https://doi.org/10.1037/0033-2909.134.2.311. Park, N., Kee, K. F., & Valenzuela, S. N. (2009). Being immersed in social networking environment: Facebook groups, uses and gratiﬁcations, and social outcomes. CyberPsychology and Behavior, 12(6), 729e733. https://doi.org/10.1089/ cpb.2009.0003. Pfundmair, M., Aydin, N., Frey, D., & Echterhoff, G. (2014). The interplay of oxytocin and collectivistic orientation shields against negative effects of ostracism. Journal of Experimental Social Psychology, 55, 246e251. https://doi.org/10.1016/ j.jesp.2014.07.016. Pfundmair, M., Aydin, N., Du, H., Yeung, S., Frey, D., & Graupmann, V. (2015a). Exclude me if you can e cultural effects on the outcomes of social exclusion. Journal of Cross-cultural Psychology, 46, 579e596. https://doi.org/10.1177/ 0022022115571203. Pfundmair, M., Graupmann, V., Frey, D., & Aydin, N. (2015b). The different behavioral intentions of collectivists and individualists in response to social exclusion. Personality and Social Psychology Bulletin, 41, 363e378. https://doi.org/10.1177/ 0146167214566186. Ren, D., Wesselmann, E. D., & Williams, K. D. (2013). Interdependent self-construal moderates coping with (but not the initial pain of) ostracism. Asian Journal of Social Psychology, 16, 320e326. https://doi.org/10.1111/ajsp.12037. Sheldon, K. M., Abad, N., & Hinsch, C. (2011). A two-process view of Facebook use and relatedness need-satisfaction: Disconnection drives use, and connection rewards it. Journal of Personality and Social Psychology, 100, 766e775. https:// doi.org/10.1037/a0022407. Singelis, T. M., Triandis, H. C., Bhawuk, D. P. S., & Gelfand, M. J. (1995). Horizontal and vertical dimensions of individualism and collectivism: A theoretical and measurement reﬁnement. Cross-cultural Research, 29, 240e275. https://doi.org/ 10.1177/106939719502900302. Tanaka, A., & Tokuno, Y. (2011). The effect of the color red on avoidance motivation. Social Behavior and Personality: An International Journal, 39(2), 287e289. https:// doi.org/10.2224/sbp.2011.39.2.287. Triandis, H. C., & Gelfand, M. J. (1998). Converging measurement of horizontal and vertical individualism and collectivism. Journal of Personality and Social Psychology, 74, 118e128. Triandis, H. C., & Gelfand, M. J. (2012). A theory of individualism and collectivism. Handbook of Theories of Social Psychology, 2, 498e520. https://doi.org/10.4135/ 9781446249222.n51. Van Rhee, H. J., Suurmond, R., & Hak, T. (2015). User manual for meta-essentials: Workbooks for meta-analysis (version 1.0). Rotterdam, The Netherlands: Erasmus Research Institute of Management. Retrieved from:www.erim.eur.nl/ research-support/meta-essentials. Verduyn, P., Lee, D. S., Park, J., Shablack, H., Orvell, A., Bayer, J., … Kross, E. (2015). Passive Facebook usage undermines affective well-being: Experimental and longitudinal evidence. Journal of Experimental Psychology: General, 144(2), 480e488. https://doi.org/10.1037/xge0000057. Warburton, W. A., Williams, K. D., & Cairns, D. R. (2006). When ostracism leads to aggression: The moderating effects of control deprivation. Journal of Experimental Social Psychology, 42(2), 213e220. https://doi.org/10.1016/ j.jesp.2005.03.005. Williams, K. D. (2009). Ostracism: A temporal need-threat model. In M. Zanna (Ed.), Advances in experimental social psychology (Vol. 41, pp. 279e314). NY: Academic Press. https://doi.org/10.1016/S0065-2601(08)00406-1. Williams, K. D., Cheung, C. K. T., & Choi, W. (2000). Cyberostracism: Effects of being ignored over the internet. Journal of Personality and Social Psychology, 79, 748e762. https://doi.org/10.1037/0022-35126.96.36.1998. Williams, K. D., & Nida, S. A. (2011). Ostracism: Consequences and coping. Current Directions in Psychological Science, 20, 71e75. https://doi.org/10.1177/ 09637214114024800963721411402480. Zadro, L., Williams, K. D., & Richardson, R. (2004). How low can you go? Ostracism by a computer is sufﬁcient to lower self-reported levels of belonging, control, self-esteem, and meaningful existence. Journal of Experimental Social Psychology, 40(4), 560e567. https://doi.org/10.1016/j.jesp.2005.10.007.