In this study, we theorize that the way consumers communicate on social media (“liking”2 vs. posting) leads to consequential preferences for luxury products. Specifically, for light users, who spend less than one hour on social media, “liking” (vs. posting) strengthens (vs. weakens) preference for social value-framed luxury products (i.e., creates a good impression to others) compared to functional value-framed products (i.e., superior quality). This contrasts with heavy users, who spend more than two hours on social media, where posting (vs. “liking”) strengthens (vs. weakens) preference for social value-framed luxury products compared to functional value-framed products. Thus, the relationship between social media interaction (“liking” vs. posting) and preference for luxury products is conditionally mediated by communication expectation with others. Both the direct effect and indirect effect are moderated by the time spent on social media and luxury value type.

When social media users post their opinion on social media, they might expect to receive a favorable evaluation from others (e.g., ‘like’ on Facebook and ‘heart’ on Instagram). On the contrary, when users read and evaluate other’s posts, they are less likely to care about others. What’s more, if users spend more time on social media, posting would make them more care about others. This research answers the question ‘Does social media interaction alter the way we perceive value and affect different choice? Three studies show 1) how social media interaction affects consumers’ luxury value perception and the mediating role of social media self-view (interdependent vs. independent self), 2) how main effect and mediated relationship are different between two groups divided by participants’ time spent on social media (high vs. low: ±1 SD) and 3) how the type of social media interaction (post vs. “like”) affects handbag choice between social and functional luxury-superior option.

Introduction Luxury industry, in addition to feelings, started to provide reasons for consumers to justify their consumption. Also, it helps them to avoid social risks by offering inconspicuous luxury products. For example, ‘Mandarin Oriental Barcelona’ suggests luxury holiday package for people running Barcelona marathon, London cocktail bar ‘Bart’ serves cocktails designed to 'facilitate focus and encourage relaxation', ‘Tiffany’ has dropped the spelled-out brand name from its fashion jewelry line in favor of simple “T”. (Trend watching 2017) Then, what changed consumers' interest? Interestingly, the current expectations of luxury consumers have a lot in common with the features of interdependent self-construal. The "we" effect of social media can explain this trend. Literature review and Conceptual framework Social Media usage and Interdependent Self-Construal Although earlier research in cultural psychology conceptualized self-construal as culturally determined (Markus and Kitayama 1991), more recent research suggests that independent and interdependent self-construal can also be made temporarily accessible. The most frequently adopted priming procedure to induce self-construal is the procedure by stories about we or I. (Gardner, Gabriel, and Lee 1999; Mandel 2003; Krishna, Zhou, and Zhang 2008; Hong and Chang 2015). We suggest that Social media usage can induce interdependent self-construal because social media users read stories and watch videos of others. Prior findings indicated that participants in interdependent self-construal condition have an equivalent number of thoughts about themselves and others. (Lee, Aaker, and Gardner 2000). Social Media, Interdependent Self-construal, and Consumers’ Luxury Value Perception Consumers’ perceptions of luxury value appear to be determined mainly by Functional, Individual, and Social aspects (Wiedmann, Hennigs, and Siebels 2009). Recent luxury marketing researchers compare conspicuous consumption and inconspicuous consumption, two types of social aspect consumption (Han, Nunes, and Dreze 2010; Eckhardt, Belk, and Wilson 2015). They are interested in Functional luxury value, added to Individual luxury value because small utilitarian additions to a hedonic luxury are often valued more than their stand-alone value (Keinan, Kivetz, & Netzer 2016). These two trends seem to have nothing in common. But when the concept of self-construal is adopted, they have something in common. Namely, interdependent selves, which focuses on others, are the one who cares about both inconspicuousness and utilitarian aspects. We suggest that because of the features of interdependent self-construal, consumers more using social media tend to have a different priority regarding luxury value perception than consumers less exposed to social media. We explore the mediating role of relational interdependent self-construal in this model (Triandis 1989; Cross, Bacon, and Morris 2000). Also, luxury brand company’s social media marketing efforts (Kim and Ko 2012; Godey et al. 2016) have different features. We expected that each social media marketing activity has a different impact on luxury value perception. Interdependent Self-Construal, Reason, Functional luxury value We propose that because of interdependence priming effect of social media, participants who are exposed to social media (social media priming group) tend to rely on both reason and feeling, which in turn set an equivalent value on Functional luxury value (cognitively superior) and Individual luxury value (affectively superior) (Hong and Chang 2015). Interdependent Self-Construal, Social Risk, Inconspicuousness The authors propose that because of interdependence priming effect of social media, participants who are exposed to social media (social media priming group) tend to avoid social risk than participants who are not exposed to social media (no priming group). Study 1: The effect of social media on the balance between self and others. A total of 709 respondents (389 females) participated in this study via Amazon Mechanical Turk (MTurk) They were randomly assigned to one of the two conditions (with social media manipulation vs. No manipulation) The impact of ‘one-time social media use’ Following the previous study (Hong and Chung 2015), we conducted a 2 (social media manipulation: No vs. Yes) x 2 (thought type: self vs other) mixed ANOVA with social media manipulation as a between-subject factor and thought type as a repeated measure. The main effect of thought type is significant (F(1,707) =293.54, p<.000). However, the interaction between social media manipulation and thought type was not significant. (F (1,707) = 0.128, p<.721). So, the result shows that one-time social media use doesn’t manipulate interdependent self-construal. The impacts of ‘chronic social media use’. To check whether chronic social media use can manipulate interdependent self-construal, we conducted a 2 (time spent on social media: Low vs. High) x 2 (thought type: self vs. other) mixed ANOVA. The main effect of thought type is significant (F(1,707) =293.81, p<.000). The interaction between time spent on social media and thought type was significant. (F (1,703) = 6.65, p<.000). Also, we conducted a 2 (communication on social media: Low vs. High) x 2 (thought type: self vs. other) mixed ANOVA. The main effect of thought type is significant (F(1,707) =300.76, p<.000). The interaction between communication and thought type was significant. (F (1,704) = 5.764, p<.000). So, the result shows that chronic social media use manipulates interdependent self-construal. Also, we computed three types of groups, the others-dominant, the balanced, and the self-dominant by modifying past research (Kitayama et al. 2009; Wu, Cutright, and Fitzsimos 2011, Hong and Chang2015). A chi-square test revealed that participants who spent more time and had more communication on social media had more balanced thought type (32.6%, 31.0%) than participants who spent less time and had less communication on social media (24.8%, 24.0%) Study 2: The effect of using social media on luxury value perceptions (SEM) A total of 255 respondents (85 females) participated in this study through Amazon Mechanical Turk (MTurk). The participants were between 18 and 65 years old and lived in the U.S., and the modal age group was 25-34 (43.9 %), followed by 35-44 (25.5%). We conducted exploratory factor analysis (EFA) using maximum likelihood estimation with Promax rotation. The goodness of fit statistics is: 𝒙𝟐=707.14, df =377, P<0.0005, 𝒙𝟐/df =1.876, CFI=0.95, and RMSEA =0.059. The final CFA model fits well with the data (Hu and Bentler, 1999; Hair et al., 2010). After confirming the measurement model, SEM with maximum likelihood extraction was estimated to test Hypothesis. The goodness of fit criteria of the model meets the proposed thresholds in previous studies: 𝒙𝟐=677.97, df =374, P<0.0005, 𝒙𝟐/df =1.813, CFI=0.95, and RMSEA =0.057 (Hu and Bentler, 1999; Hair et al., 2010). The SEM model showed each social media activity has a different effect on each luxury value. Interaction positively impacts the functional luxury value (β=0.38, p<.000) but negatively impacts the individual (β= - 0.27, p<.000) and social luxury value (β= - 0.34, p<.000). Entertainment/WOM has a positive effect on individual luxury value (β=0.39, p<.000) and on social luxury value (β=0.32, p<.000). Customization positively impacts the social luxury value (β=0.26, p<.000). We could also find Entertainment/WOM and Customization did not influence the functional luxury value. Only Entertainment/WOM positively impacts relational interdependent self-construal (β=0.32, p<.000) Relational interdependent self-construal positively impacts on individual value (β=0.24, p<.000) and social value (β=0.27, p<.000). We conducted a mediation analysis. Among social media activities, only Entertainment/WOM had a significant positive effect on Relational Interdependent Self-Construal. Study 3: the effect of social media on self-construal and choice between functional value and individual value superior options 454 Responses (304 females) collected via Mturk to stimulate the environment of social media. Respondents are primed on interdependence by social media usage while control group respondents are not primed. The authors checked the validity of the self-construal manipulation effect of social media. Respondents were given a choice task, designed by adopting previous method (Hong and Chang 2015). They were told to imagine that they were going to buy a hand bag. Handbag X is superior on functional (cognitive) dimensions whereas handbag Y is superior on all individual (affective) dimensions. Participants are asked to indicate their choice between two Handbags Choice. A chi-square test revealed that the effect of social media manipulation is not significant on the handbag choice (𝑥2(l) 0.371, p<.309). However, the effect of interdependent self-construal on the participant's’ handbag choice was significant (𝑥2(l) 5.85, p<0.01). As predicted, participants with a higher Interdependent self-construal primed by social media have a smaller preference gap between Functional luxury value and Individual luxury value (F:73.2%-I:26.8%=46.4%), than participants with lower Interdependent self-construal participants (F:82.7%-I:17.3%=65.4%) Study3 supports the results of previous two studies. As individuals communicate more and spend more time on social media, they are more likely to prefer equally for Functional luxury value and Individual luxury. Study 4: the effect of social media on self-construal and choice between conspicuous and inconspicuous options. After the self-construal manipulation with social media, respondents were given a choice task. The choice was designed by adopting previous method (Han, Nunes, and Dreze 2010). Respondents were told to imagine that they were going to buy a handbag. One option has an inconspicuous design whereas the other option is conspicuous design. Participants are asked to indicate their choice between two handbags. Social media manipulation and Handbags Choice. All three chi-square tests revealed a significant or marginal effect of social media manipulation on participants handbag choice with the same pattern. ( 𝑥2(1) = 6.328, p > .032; 𝑥2(1) = 3.086, p > .079; 𝑥2(1) = 1.933, p > .098) As predicted, participants in social media manipulation group prefer the inconspicuous option to the conspicuous option (59.3%>40.7%; 58.9%>41.1%; 60.4%>39.6%) than participants in control group (47.3%, 52.7%; 50.5%, 49.5%, 53.8%, 46.2%). Interdependent-self construal and Handbags Choice. All three chi-square tests revealed a significant effect of interdependent self-construal on participants handbag choice with the same pattern. ( 𝑥2(1) = 4.2, p > .047; 𝑥2(1) = 4.79, p > .029; 𝑥2(1) = 6.11, p > .013) As predicted, participants with higher Interdependent self-construal primed by social media have a equivalent preference about the inconspicuous option and conspicuous option (50%, 50%; 50.8%, 49.2%; 52.4%, 47.6%) than participants with lower Interdependent self-construal(59.6%, 40.4%; 61.1%, 38.9%, 63.9%, 36.1%). Study4 shows that one-time social media exposure can reduce conspicuous consumption. However, as individuals communicate more and spend more time on social media, they become more likely to prefer equally for conspicuous option and inconspicuous option. Contribution In luxury brand marketing, this study sheds light on the balance between two motivations, luxury for self and luxury for others (Kapferer, Jean-Noël, and Vincent 2009) using the concept of self-construal. This research also contributes to the literature on consumers’ luxury value perception (Hennigs et al. 2012; Wiedmann et al. 2009). In general marketing subjects, it adds to the increasing literature on consumer risk and choices between affective and cognitive options. For the luxury brand industry, this study can give marketers a practical idea about choosing social media marketing activities and designing new products. For example, if luxury brand companies promote customers posting opinions on websites, they should know that their activities might advertently affect their sales because interaction negatively influences on individual and social luxury value.