Determinants of mobile social media use, customer heterogeneity, and international microsegmentation

How to effectively examine the heterogeneous nature of consumer preferences across borders and cultures is a challenge for firms that use mobile social media. This study identifies the determinants of individuals’ mobile social media use behavior and profiles the resulting international microsegments. We propose a model that integrates different theoretical perspectives and sets of factors to explain mobile social media use behavior and test it on a sample of users in China and the United States via online surveys. We estimate a global model (GM) based on all respondents and three local models created by post hoc international microsegmentation. The three local models reveal the existence of three unobserved user segments: “usage goal experts,” “determined pragmatists,” and “pressured hedonists.” Perceived usefulness is the most influential factor in the GM, while users in the three segments significantly differ in their behavioral patterns, cultural value orientations, and demographic characteristics. This study is the first to examine unobserved heterogeneity and international microsegmentation in the mobile social media domain. It provides insights into the factors explaining use behavior and international microsegmentation for scholars and marketers operating in the global marketplace.


JU et al.
social media can play a relevant role.The ubiquity of social media influences individuals' behavior (Lamberton & Stephen, 2016) and provides new opportunities for researchers and firms that operate in global markets (Rosado-Serrano & Paul, 2017).Particularly, mobile social media use behavior and international segmentation are important research topics that deserve more attention from marketing scholars, as argued below.
Since firms cannot serve the world's heterogeneous population using standardized marketing strategies, researchers and marketers have shown an increasing interest in assessing the heterogeneous nature of customers, and in segmenting them (Steenkamp & Hofstede, 2002).
However, little research exists on heterogeneity and the international segmentation of mobile social media users, and two limitations affecting prior mobile social media research and international microsegmentation are particularly relevant to the current study.First, from the few studies on the actual use of mobile social media, researchers have mainly shed light on the effect of hedonic and integrative gratifications on users' attitudes toward the platforms, as well as their actual use behavior (Ha et al., 2015).However, users' attitudes may be affected not only by perceived hedonic (intrinsic motivation) and integrative gratifications but also by factors such as extrinsic motivations and the perceived influence of their social environment.Thus, identifying the determinants of users' attitudes and actual use behavior from multiple theoretical perspectives is expected to provide a more comprehensive explanation of mobile social media use behavior and mitigate the problem of utilizing more simplistic frameworks that do not integrate some of the relevant factors that drive their usage.
Second, when researchers have conducted cross-cultural analyses to explore users' behavioral differences, the country has been the most frequently used criterion for classifying users (Hoehle et al., 2015).Users in a country are typically assumed to display similar behavioral patterns and cultural values (Hoehle et al., 2015).However, when there are minimal differences in users' behavioral patterns across countries, the country becomes less important as a segmentation variable (Marreiros et al., 2020).
A priori segmentation, as a segmentation approach that groups consumers based on predefined segmentation variables such as demographics (e.g., nationality) rather than on market research, may hide differences in consumer behavior when heterogeneous behavioral patterns exist (Wedel & Kamakura, 2002).Therefore, post hoc segmentation, as a method in which respondents are clustered according to consumer data and the similarity of multivariate individual profiles (Kalafatis & Cheston, 1997), is a more suitable approach.
Against this background, this study's objectives are to bridge these gaps by integrating from several theoretical perspectives the factors that drive actual mobile social media use behavior and by uncovering the different behavioral patterns, cultural value orientations, and demographic characteristics across international microsegments.More specifically, this study seeks to answer the following research questions: (a) How do intrinsic and extrinsic motivations, technology acceptance, and social influence explain the attitude toward and mobile social media use?(b) How can post hoc international microsegmentation of behavioral patterns be effectively applied to identify users' heterogeneity in mobile social media use?
To answer these questions, we propose a hybrid model that integrates the technology acceptance model (TAM) (Davis, 1986;Davis et al., 1989), motivation theory (Deci & Ryan, 1987), and social influence theory (Kelman, 1958) as the main foundations for examining individuals' attitude and actual use behavior.We apply partial least squares path modeling (PLS-SEM) (Hair Jr et al., 2016) to estimate the model using a sample of mobile social media users collected in China and the United States via online questionnaires.Then, we adopt Response Based Unit Segmentation in PLS path modeling (REBUS-PLS) (Vinzi et al., 2008) to detect unobserved heterogeneity and segment users with similar behavioral patterns.Finally, we explain the behavioral differences across segments through the lens of cross-cultural variation theory (Hofstede et al., 2010) and demographic variables.
Our study offers two contributions to the literature on mobile social media and international microsegmentation.First, it identifies and integrates, from multiple theoretical perspectives, different sets of factors that ultimately drive users' actual use behavior, which can
Various theories have been adopted to explain mobile social media use behavior.The TAM (Davis, 1986;Davis et al., 1989) is the most widely used theory, and studies on mobile social media have highlighted the significant role of the perceived usefulness and perceived ease of use in using technology (Leong et al., 2018;Oghuma et al., 2016;Tan et al., 2018;Zhao et al., 2016).In addition, gratification theory (Katz et al., 1973) has also been widely used.
Researchers have found several gratifications perceived from mobile social media use that positively affect individuals' use behavior, as follows.Social gratification satisfies users' social connection needs (Han et al., 2015), hedonic gratification satisfies users' entertainment and enjoyment needs (Ha et al., 2015;Hsiao et al., 2016;Yang & Lin, 2019), utilitarian gratification satisfies users' needs to contact others and obtain information, and integrative gratification satisfies users' needs to form a sense of belonging to a group and enhance personal values (Ha et al., 2015).In addition, researchers have highlighted motivation theory (Deci & Ryan, 1987) to explain users' behavior.Specifically, perceived usefulness is classified as an extrinsic motivation (Lin & Lu, 2011;Wamba et al., 2017), whereas hedonic gratification is an intrinsic motivation (Zhang et al., 2017).Finally, the perceived social influence of other users also plays an important role in driving users' behavior (Gao & Bai, 2014;Lin & Lu, 2015;Zhang et al., 2017;Zhao et al., 2016).
All the studies were quantitative and conducted data collection via online surveys.The most frequent country for data collection was Mainland China (see Table 1, column four); however, for our purposes, we only found one cross-cultural study (Hoehle et al., 2015) that explained use.All studies applied structural equation modeling (SEM) to analyze the hypothesized relationships.By exploring the antecedents of mobile social media use, these studies deepen our understanding of individuals' behavior and provide useful insights to mobile social media developers and marketing practitioners (see Very little research has considered the role of attitudes toward mobile social media, or the corresponding antecedents that form such attitudes (e.g., Ha et al., 2015).Attitude is an important component in the behavioral process of using information systems because it constitutes a key affective response that connects cognitive and behavioral responses (Davis, 1986;Davis et al., 1989).The current study considers the role of attitude to connect actual use with several cognitive components.

| International market segmentation
Globalization of the business world has turned international market segmentation into an essential concept in both marketing theory and practice (Wedel & Kamakura, 2002).Market segmentation is the process of dividing a heterogeneous market into smaller and relatively homogeneous segments (Kale & Sudharshan, 1987).It can help identify groups of consumers that are separated by national or cultural boundaries but have similar needs and wants (Grunert, 2019).
Segmentation helps companies address consumers' specific needs and target potential customers at the international segment level (Steenkamp & Hofstede, 2002).However, how to effectively deal with both the structure of heterogeneity in customers' needs and wants and segment consumers across borders are among the main challenges for scholars and international marketers (Steenkamp & Hofstede, 2002).International segmentation is a challenging issue because it involves obtaining comparative information pertaining to not only countries, industries, products, and consumers in a given market, but also to the cultures of the markets (Papadopoulos & Martín Martín, 2011).
The evolution of the Internet toward the more interactive Web 2.0, the new online platforms for marketing research, and the growth of social media have resulted in both an increase in the availability of data at the consumer level and opportunities for new studies on international (micro)segmentation.Most international segmentation studies have ignored individual heterogeneity by using the country as the unit of analysis (for exceptions, see Cleveland et al., 2011) and they have not focused on segmenting mobile social media users.
Taking the country as the unit of analysis, Gaston-Breton and Martín Martín (2011) proposed a two-stage market selection and segmentation model that integrated market attractiveness and consumer values to help institutions and market-seeking multinational enterprises segment European macro-regions and groups of consumers.Gao and Bai (2014)   Although globalization may increase the similarity in needs and wants across national boundaries, countries and mobile social media users remain largely heterogeneous in their preferences and behaviors, and researchers have not paid attention to the heterogeneous nature and international microsegmentation of mobile social media users.

| THEORE TI C AL FR AME WORK AND HYP OTHE S E S
This section proposes a hybrid model of actual mobile social media use based on the TAM, motivation theory, and social influence theory (see Figure 1), and explains the impact of heterogeneity in mobile social media use.The TAM was built based on the theory of reasoned action (TRA) (Ajzen & Fishbein, 1980) to explain information system acceptance.The TAM posits that users' attitude toward use is a predictor of behavioral intention, which, in turn has a direct effect on actual use.Users' attitudes are jointly determined by perceived usefulness and perceived ease of use (Davis, 1986;Davis et al., 1989).Motivation theory (Deci & Ryan, 1987) connects the cognitive responses arising from perceived usefulness, ease of use, and enjoyment with the affective response of attitude toward use.
Social influence theory was proposed by Deutsch and Gerard (1955) and Kelman (1958), and indicates that referent others influence an individual's attitudes, beliefs, and subsequent actions or behaviors.
Building on social influence theory, we complete the model with subjective norms, referent network size, and attachment motivation as other sets of determinants of attitude toward use.

| Perceived enjoyment
We define perceived enjoyment as the extent to which the activity of using mobile social media is perceived to be enjoyable.It is an intrinsic and hedonic motivation driven by internal rewards such as enjoyment (Davis et al., 1992;Lee et al., 2005;Venkatesh, 2000;Venkatesh & Bala, 2008).Users may adopt mobile social media not only for communication or interaction (Ju et al., 2021), but also for the enjoyment that is obtained from the usage process (Zhou & Lu, 2011).Thus, perceived enjoyment can be seen as a crucial factor that affects users' adoption behavior (Ha et al., 2015;Hsiao et al., 2016;Omoush et al., 2012;Pagani & Malacarne, 2017;Zhou & Lu, 2011).
People may form beliefs about an object by associating it with specific attributes, and may form attitudes from beliefs (Fishbein & Ajzen , 1975).Social psychologists have proposed that attitude formation can occur through direct experience and emotional reactions (Bem, 1970).Thus, when users experience enjoyment from the use of a given technology, they may form a positive attitude toward it.
Consistent with this logic, in the social media sphere, Chen (2014) found that the more pleasure users obtained from using Facebook, the more positive their attitudes toward the social network were.
Similarly, Bailey et al. (2018) explored social media use behavior among young users and showed that perceived enjoyment was positively related to users' attitudes toward social media.Thus, we suggest that the perceived enjoyment of mobile social media also leads users to form a favorable attitude toward it.Therefore, we propose the following hypothesis: H1. Perceived enjoyment is positively related to attitude toward mobile social media.

| Perceived usefulness
Based on Davis et al. (1992), we define perceived usefulness as user expectations that their use of mobile social media will improve the effectiveness of achieving their goals.Motivation can also be extrinsic (Deci & Ryan, 1987).Extrinsic motivation refers to user behavior that is driven by external rewards, such as improved job performance, salary, or promotion (Davis et al., 1992;Venkatesh & Bala, 2008).Perceived usefulness is an extrinsic motivator (Davis et al., 1992;Kwon et al., 2021;Venkatesh & Bala, 2008;Wamba et al., 2017;Zhou & Lu, 2011) because it represents a user's utilitarian need for social media platforms (Hong et al., 2017;Hsiao et al., 2016).
Previous social media studies have shown that users may form a positive feeling about platforms when their usage experience matches their expectations (Bailey et al., 2018;Gao et al., 2013;Im & Ha, 2013;Lee et al., 2005;Lin & Kim, 2016).Gao et al. (2013) found that consumers in the United States, mainland China, and Europe formed a positive attitude toward mobile marketing when they increased their perceived utility in participating in marketing activities using mobile technology.Notably, perceived usefulness is the most influential factor affecting American consumers' attitudes toward mobile marketing (Gao et al., 2013).Therefore, we propose the following hypothesis: H2. Perceived usefulness is positively related to attitude toward mobile social media.

| Perceived ease of use
We define perceived ease of use as the extent to which a person believes that using mobile social media will be free from effort.A given technology's ease of use may be inferred from the effort put into the process of using the technology to accomplish a task (Davis et al., 1992).Less effort would imply a higher value for the user and, therefore, a positive attitude toward the technology.
Previous social media studies have indicated that users may have a positive attitude toward technology when they have had a good experience and have made less effort previously to complete personal goals (Bailey et al., 2018;Gao et al., 2013;Im & Ha, 2013;Lee et al., 2005).Since social media constitutes important advertising channels (Sreejesh et al., 2020), Lin and Kim (2016) examined undergraduate students' acceptance of sponsored advertising on Facebook.They showed that the easier use of Facebook advertising resulted in a positive attitude toward Facebook advertising.Thus, we suggest that users will form a positive attitude toward mobile social media when less effort is required to use their functions.Therefore, we propose the following hypothesis: H3. Perceived ease of use is positively related to attitude toward mobile social media.

| Subjective norms
Subjective norms are one of the root constructs of social influence (Venkatesh et al., 2003).This influence means that a user's decision to use an information system is influenced by people who are important to them, including family, friends, and colleagues (Venkatesh & Bala, 2008).In the information system context, subjective norms can be considered as the psychological pressure perceived by peers regarding technology adoption (Davis, 1986), and they play an essential role in influencing users' attitudes and decision-making processes.Generally, people may choose to perform a behavior if they believe that one or more essential referents think they should (Venkatesh & Davis, 2000).Subjective norms may influence behavioral intentions via attitude (Davis et al., 1989).
We suggest that a previous positive influence on the use of mobile social media from relevant referents may lead to a positive attitude toward this technology.Prior studies have shown that subjective norms influence users' attitudes toward a given technology.Specifically, Im and Ha (2013) examined American consumers' adoption decisions regarding mobile coupons, and found a positive relationship between subjective norms and attitude.That is, individuals were more likely to form a positive attitude toward the given technology when their peers believed that the technology should be used (Im & Ha, 2013).In the context of social media, Chen (2014) found that subjective norms positively affected user attitudes toward Facebook.We posit that users may form a favorable attitude toward mobile social media under social pressure.Therefore, we propose the following hypothesis: H4. Subjective norms are positively related to attitude toward mobile social media.

| Referent network size
The referent network size refers to the number of participants in a social network.This aspect represents the perceived value of direct network externalities (Lin & Bhattacherjee, 2008).The value of technology increases to a person when others use it (Shepherd & Lane, 2019), and because of this "herd effect" (Hong et al., 2017), users' usage decisions regarding interactive technologies are often influenced by the number of friends and colleagues who use them in their social circle.The number of people with whom users can communicate increases with the number of people who use IT (Lin & Bhattacherjee, 2008).In contrast, when the referent network size is small, users may have less interest in a technology and stop using it (Zhou & Lu, 2011).
Although empirical studies have verified the impact of the referent network size on network benefits (Lin & Bhattacherjee, 2008) and perceptual outcomes (Hong et al., 2017), no study has yet examined its impact on users' attitudes toward mobile social media.Under the influence of the "herd effect," people are more likely to believe in the evaluation of and feelings toward the mobile social media used by their reference groups, and imitate their adoption decisions (Hong et al., 2017).Thus, users' attitudes toward mobile social media may be more positive as the number of users increases.Therefore, we propose the following hypothesis: H5.The referent network size is positively related to attitude toward mobile social media.

| Attachment motivation
Attachment motivation is the degree to which individuals believe that they can improve their social interactions and interpersonal relationships via mobile social media (Ma & Chan, 2014).In the current study, attachment reflects the emotional bond connecting an individual to mobile social media platforms.Social media serve not only to communicate, create, and share messages with others (Yannopoulou et al., 2019), but also shape people's thoughts and attitudes (Hwang & Kim, 2015) and empower them (Tajurahim et al., 2020).People desire to form relationships with others.Being accepted by others and receiving positive feedback from members of mobile social media may increase a user's sense of belonging to groups, and may stimulate more interaction with others and the development of new relationships (Ma & Chan, 2014).Fulfilled needs influence users' emotional and psychological states regarding social media platforms, which consequently affects users' allocation of cognition and attitude toward these platforms (Fedorikhin et al., 2008).
The attachment motivation construct has received some attention in the marketing literature (e.g., Chen, 2014; Ma & Chan, 2014).
For example, Chen (2014) found that Facebook use could fulfill users' needs for interpersonal relationships and increase users' attachment to the platform.Consequently, users had a positive attitude toward Facebook and had intentions to continue using it in the future (Chen, 2014).In the current study's mobile social media context, we posit that a high level of attachment to mobile social media will motivate users to form a positive attitude toward it.Therefore, we propose the following hypothesis: H6. Attachment motivation is positively related to attitude toward mobile social media.

| Attitude and outcome behavior
We adapt Ajzen's (1991) definition of attitude to our study's context and define it as an individual's positive or negative feelings about using mobile social media.According to the TAM (Davis, 1986;Davis et al., 1989) and theory of planned behavior (Fishbein & Ajzen, 1975), attitudes exert a positive effect on behavioral use intention and users' consequent actions.
The literature on social media is consistent with these views.Ha et al. (2015) focused on Facebook and KakaoTalk use in Korea to shed light on the significant impact of users' attitudes toward mobile social media on their actual use behavior.Bailey et al. (2018) suggested that young Latin American users' attitudes toward social media were positively related to their active social media behaviors.
In our study, we suggest that users' attitudes toward mobile social media will prompt them to spend more time on mobile social media.
Therefore, we propose the following hypothesis: H7. Attitude toward mobile social media is positively related to the actual use of mobile social media.

| Heterogeneity and international market segmentation
The heterogeneity of customers' needs and preferences is the driving force behind market segmentation (Floh et al., 2014;Kim et al., 2020).Considerable heterogeneity characterizes international markets, so it is critical to identify similar underlying behavioral patterns in this context.Global users may have similar needs and preferences for mobile social media, regardless of the country in which they live.Thus, a post hoc segmentation method can be effectively used to conduct international microsegmentation (Wedel & Kamakura, 2000).Segmentation is particularly important in the development and implementation of successful global marketing strategies (Olsen et al., 2009;Walters, 1997).
Previous studies have highlighted the role of customer heterogeneity in the marketing domain (Chocarro et al., 2015;Floh et al., 2014;Wamba et al., 2017).For example, Wamba et al. (2017) assessed unobserved heterogeneity in the social media market, and segmented users from five countries based on their responses.Their findings demonstrated that users' adoption behaviors and demographic characteristics were different in distinct segments.Thus, explaining users' behavioral differences in connection with their specific profiles, such as their demographics and culture, can help determine the characteristics that describe each of the revealed classes and manage users effectively (Chocarro et al., 2015).
Users' motives and behavioral patterns differ across segments, which is associated with users' demographic characteristics and cultural values.Cultural differences reflect the variation in values, attitudes, and behaviors that people share in society (Bond, 2004).

| Sampling and respondents' demographic characteristics
We selected mainland China and the United States for three reasons.First, both countries have a high penetration of mobile social media (71% in China and 61% in the United States) (We are Social & Hootsuite, 2019).Second, these two countries are profoundly different in their cultural value orientation (Hofstede, 2001).Third, some of their characteristics, such as their diverse and multicultural populations, indicate a high probability of the existence of unobserved customer heterogeneity.Quota sampling was applied to select survey participants in both countries because a sampling frame was unavailable, and this approach can provide a representative sample of various subgroups within a population (Babin & Zikmund, 2015).We controlled the proportions of age and sex intervals to resemble the countries' populations.We targeted Chinese and American users who had used mobile social media in the last six months.
Table 2 presents the demographic information of the 844 respondents who completed the questionnaire.The sample was quite balanced in terms of sex (53% males, 47% females) and nationality (49.76%Chinese, 50.24%Americans).Approximately 81% of the respondents were young, and 87% had higher education.Regarding location size, there are different definitions of city sizes and boundaries between cities and rural areas.To enhance comparability between Chinese and American locations, we defined the size of the location using the percentage of a location's population compared to the country's total population.Locations containing the highest 5% of the country's population were defined as large cities, locations containing the top 5% to the top 30% of the country's population were defined as medium cities, and the remaining locations were considered as small cities and rural areas.

| Questionnaire development, pretest, and data collection
We designed a four-section questionnaire to collect data from mainland China and the United States.This instrument was first developed in English and translated into simplified Chinese.A backtranslation technique (from English to Chinese and back to English) was applied by third-party professionals to avoid linguistic biases (Jarvis et al., 2003).The first section inquired about aspects of actual use behavior, such as whether users had used social media on a smartphone during the last six months, the most used types of mobile social media, frequency of use, and the purpose of use.The second section explored the factors that drove users' actual use behavior.The third section measured users' cultural values, and the last section asked for users' demographic information.To check the content validity and accuracy of the questions and its measures, we conducted qualitative and quantitative pretests before formally collecting data.We first asked for qualitative feedback and suggestions from various marketing professors, and then pretested the administration of the pretest with 30 mobile social media users.After some modifications, we conducted a quantitative pretest among 200 users by distributing a survey on several popular mobile social media applications in China, such as WeChat, QQ, and Sina Weibo.
After the pretests, we contracted an online panel company, Wenjuanxing, to create and distribute the questionnaire among Chinese respondents.Data collection began on January 30, 2019, and 420 valid responses were received in one week.All respondents were of Chinese nationality and, as requested, had used social media on a smartphone during the last six months.To collect the American sample, we created a questionnaire in Qualtrics and distributed it through Amazon Mturk.Data collection began on February 18, 2019, and we received 424 valid responses over two months.All respondents were of American nationality and had also accessed social media on smartphones during the last six months.
We used chi-square tests to verify whether there were significant differences with respect to age and sex between the two country groups (Hair et al., 2013).The p-values of the age interval and sex of the two countries were 0.719 and 0.484, respectively, implying no significant differences.

| Measurement of the variables
All constructs, except for actual use, which was measured with a single item, were operationalized using three items.Seven-point Likert scales ranging from "strongly disagree" (1) to "strongly agree" (7) were developed to measure each item.To measure perceived enjoyment (see Appendix 1 for the specific wording of the items in each construct), we focused on the pleasure that the users' obtained from mobile social media.Two of the items were adapted from Zhou and Lu (2011).Consistent with Rauniar et al. (2014) and Strader et al. (2007), the perceived usefulness items explored the consumers' perceptions of the effectiveness of mobile social media.
Perceived ease of use items measured the extent to which it was easy to use mobile social media, and two of the items were adapted from Wamba et al. (2017).The fourth item in the scale (PEOU4) was reverse-coded in the database.
Subjective norms measured the influence received from influential people, and two items were adapted from Al-Debei et al. (2013).
Referent network size captured the influence of social circle size, and we adapted two items from Zhou and Lu (2011).Attachment motivation reflected the extent to which users were emotionally connected with their social relationships through mobile social media, and one of the items was adapted from Chen (2014), while the other two were self-developed from the construct's definition.
The three attitude items examined users' feelings about mobile social media applications, and two of the items were adapted from Chen (2014).Finally, actual use was a manifest variable adapted from Rauniar et al. (2014) to measure the hours that users spent daily on mobile social media.
Concerning the variables used to profile the international segments, collectivism examined the degree of the users' feeling of belongingness to a group, and two of the items were adapted from To prevent and detect common method bias, we applied ex ante approaches in the research design stage (procedural remedies in designing and administering the questionnaire), while ex post approaches (statistical analyses) were implemented after the research was conducted (Chang et al., 2010;Podsakoff et al., 2003Podsakoff et al., , 2012)).Regarding the procedural remedies, to diminish social desirability, threats to self-esteem, and defensiveness (MacKenzie & Podsakoff, 2012;Podsakoff et al., 2003Podsakoff et al., , 2012)), we first assured that our survey was anonymous and there were no correct/incorrect answers.Second, to enhance the feeling of altruism (MacKenzie & Podsakoff, 2012;Podsakoff et al., 2012), we emphasized the importance of the participants' opinions.Third, we randomized the order of appearance of the items in the questionnaire (Podsakoff et al., 2003(Podsakoff et al., , 2012)).Regarding the statistical analyses, empirical evidence has indicated that this approach can detect biasing levels of common method variance under conditions usually found in survey-based marketing research (Fuller et al., 2016).We performed two Harman's single factor tests for the Chinese and American data.The unrotated main factor from all the substantive items explained 21% of the total variance in the Chinese dataset and 34% in the American dataset; therefore, common method bias was unlikely to be present in our data (Fuller et al., 2016).

| Data analysis
We applied PLS-SEM to empirically test our model (R package plspm).PLS-SEM has several benefits, such as its high statistical power and flexible assumptions regarding model specification and data (Hair et al., 2011).Moreover, as PLS-SEM integrates the response-based unit segmentation REBUS-PLS algorithm (Vinzi et al., 2008), this allowed us to detect local models with better performance and to conduct post hoc segmentation.The REBUS algorithm was first used to estimate a global model (GM) with all observations (n = 844).Then, after the communality and structural residuals of each class from the GM were obtained, a hierarchical cluster analysis of the computed residuals from both the measurement and structural models was performed to obtain the number of classes (Vinzi et al., 2010).We obtained three local models (LM1, LM2, and LM3) and estimated their parameters.The differences between the three classes were used to profile the resulting international segments.

| Reliability and validity
Before testing our research hypotheses, we studied the measurement model in terms of item and construct reliability and convergent and discriminant validity.First, the factor loadings and descriptive statistics for all items are as shown in Table 3, and the factor loading of most items reaches the 0.7 threshold (Sanchez, 2013).The exceptions are PEOU2 in GM, LM1, LM2, and LM3; PEOU4 in LM3; and SN3 in GM, LM1, and LM3, although they are still over 0.5 (Sanchez, 2013) and have suitable construct reliability and validity, as explained below.Second, construct reliability was assessed in terms of composite reliability (CR; see Table 3).CR is greater than the 0.7 threshold (Hair et al., 2013) for all constructs and models except for perceived usefulness (PU) (0.65) and attachment motivation (AM) (0.69) in one of the local models (LM1).Third, with respect to convergent validity, we checked the average variance extracted (AVE) of the constructs.This is well above the cutoff point of 0.5 (Hair et al., 2013) for all constructs (see Table 3).Finally, when testing discriminant validity, the square root of the AVE of each construct should be higher than its correlations with other constructs (Hair et al., 2013), as is the case in the GM and the three local models (see Table 4).
Overall, the results show that the items and constructs used in the GM and local models have been measured with an acceptable degree of reliability and validity, similar to most of the empirical research on international marketing.

| Local model results
We applied the REBUS-PLS (Vinzi et al., 2008) response-based clustering technique to group users automatically into segments, and obtained three local models, which we named LM1, LM2, and LM3.
The results from the local models reveal that the determinants of attitude toward mobile social media vary across segments (see Table 5, columns four-six).
Thus, H7 is supported.Respondents in this segment consider both the practicality and enjoyment of using social media platforms.
However, the regression weight of perceived enjoyment peaks.In addition, these users are the only users who are influenced by subjective norms.Influential people exert psychological pressure on attitudes toward the use of new technologies, such as mobile social media.Consequently, we labeled this segment "pressured hedonists."
WeChat is the most used application in this segment (n = 144:59%).

| Cultural characteristics
We performed statistical tests to examine whether the three segments had significantly different cultural orientations (see Table 6).
We first estimated the mean and standard deviation of the standardized factor scores of each cultural dimension in each of the three segments.Then, given that the variables were not normally distributed, we conducted the Kruskal-Wallis test and found heteroscedasticity across the groups.The results show that, except for power distance (p = 0.115), the cultural dimensions are significantly different across the three segments: collectivism (p < 0.01), masculinity (p < 0.05), and uncertainty avoidance (p < 0.05).According to the mean of the cultural values, respondents in segment 1 ("usage goal experts") have the highest collectivism and masculinity scores.Therefore, they may perceive a higher sense of belonging to a group and may pay more attention to competition and achievements (Hofstede, 2001).In segment 2 ("determined pragmatists"), respondents are more individualistic, feminine-oriented, and tolerant toward uncertainty.
Consequently, they may only look after themselves and their direct families, pay substantial attention to their quality of life, and be less influenced by ambiguous or unknown situations (Hofstede, 2001).
Respondents in segment 3 ("pressured hedonists") have the highest degree of uncertainty avoidance, meaning that ambiguity or unknown situations may cause them anxiety (Hofstede, 2001).

| General discussion
This study identifies the determinants of attitude toward mobile social media, confirms the significant role played by attitude as a driver of actual use behavior, assesses users' heterogeneity, and conducts an international microsegmentation of mobile social media users.This finding is in line with the earlier correlational studies by Bailey et al. (2018) and Lin and Kim (2016), who showed that users' positive attitude toward social media (Bailey et al., 2018) and social media advertisements (Lin & Kim, 2016) were mainly due to users' beliefs in their usefulness.Moreover, users have other significant motivations, and aim to enjoy using mobile social platforms.This finding is in line with Bailey et al. (2018) and Chen (2014), who pointed out users' need for entertainment on social media platforms.The more relaxed and pleasant the user experience is on these platforms, the more positive their attitude toward the platforms will be (Bailey et al., 2018;Chen, 2014).Users also pay attention to whether these platforms are easy to use (in the GM and two local models), which is also in line with Lin and Kim (2016).
Second, users' attitude toward mobile social media platforms are also affected by social influence factors.However, subjective norms are not a significant determinant of users' attitudes toward platforms (in the GM and two local models).This result is inconsistent with the studies that suggest that users' attitudes toward technology will be positively influenced by people who are important to them (Chen, 2014;Im & Ha, 2013).This may be because users currently have more channels (e.g., online forums, online discussion groups, and online brand communities) from which to gather information about technological applications.The increased number of online reference groups may weaken the significance of the opinions from peers or important persons during the behavioral process.
Users will form a positive attitude toward mobile social media platforms if they are popular in their social circle (in the GM and one local model).Regarding the referent network size, we found limited empirical evidence on the relationship between users' social circle size and attitude (in the GM and one of the local models).In relation to attachment motivation, this is the most important determinant of attitude toward mobile social media within this second group of factors.Users expect to improve their relationships with others through mobile social media, which is consistent with Ma and Chan (2014) and Chen (2014).
Third, users' attitude toward mobile social media is a critical factor that stimulates usage.This result is in line with Omoush et al. (2012) and Bailey et al. (2018).The more favorable users' attitude toward a social media platform is, the more likely they are to employ it.Fourth, when taken together, the varying results for the local models suggest that international users with different demographics and cultural characteristics may have different use motivations and technology adoption behaviors.
Finally, our results also show that the three segments based on respondents' behavioral patterns ("usage goal experts," "determined pragmatists," and "pressured hedonists") are heterogeneous.Thus, assuming that mobile social media users are homogeneous may provide a misleading view of their real behavior.Overall, our findings are consistent with Wamba et al. (2017), who suggest that social media users are not homogenous and that the effects of motivators differ in distinct user segments that have different demographic characteristics.

| Implications for theory
Our findings have important implications for theory.First, this study enhances the understanding of attitude toward mobile social media and actual use behavior by integrating into a hybrid model factor related to the TAM, motivation theory, and social influence theory.The previous literature (a) has not accounted, in a single model, for all the theoretical perspectives and factors that we combine herein (e.g., Bailey et al., 2018;Ha et al., 2015); (b) has not paid much attention to the role of attitude toward mobile social media to explain actual use behavior (e.g., Assimakopoulos et al., 2017;Guenzi & Nijssen, 2020); and (c) neglects the referent network size as a relevant driver of users' attitudes toward mobile social media (e.g., Hong et al., 2017;Zhang et al., 2017).Our study is also unique in revealing that motivation theory may have more explanatory power on attitude and mobile social media use than social influence theory.In other words, based on our findings, we can state that although both motivation and social influences affect attitude toward mobile social media, motivation plays a more important role.In addition, we can confirm, across different populations, that attitude is an important component in the behavioral process of mobile social media use.
Second, our study also contributes to the literature on international market segmentation by highlighting the importance of heterogeneity in international microsegmentation.This study provides strong support for the necessity of conducting international microsegmentation at the individual level in the research domain of mobile social media, and paves the way by developing new items to operationalize some of our constructs (in particular, some dimensions of cultural value orientation) and by using them at the individual level.Most of the previous research on international marketing and social media has used Hofstede's scores at the country level to operationalize culture (e.g., Amaro & Duarte, 2017;Sheldon et al., 2017), which is misleading in research contexts such as social media, where there are more users who are both of younger age and frequent users of ICT (Bailey et al., 2018).

| Implications for practice
From a practical perspective, our findings have important implications for mobile social media application developers and international marketing practitioners.Given that perceived enjoyment, perceived usefulness, and attachment motivation are three significant drivers of attitude toward mobile social media in the GM and three local models, application developers can enhance users' attitude by improving the entertainment, practical, and social functionality of their applications.Moreover, our results also show that generally, ease of use positively impacts users' attitude toward mobile social media.Thus, mobile social media developers should pay attention to applications' interface design to adapt to aspects such as the limited screen size of mobile devices to better integrate social media applications with smartphone technology.
Our empirical findings revealed three user segments with significant differences in behavioral patterns, demographic characteristics, and cultural values.Therefore, international marketing practitioners should avoid designing international marketing strategies that rely exclusively on a priori segmentation at the country level.Instead, they should conduct international segmentation based on users' behavioral patterns and offer customized mobile social media services.
In the "usage goal experts" segment, considering the significant impact of perceived usefulness, perceived enjoyment, and attachment motivation, marketers should strengthen the entertainment, practical, and social functionality of their mobile social media applications.Concerning the "determined pragmatists" segment, since perceived usefulness is the most influential factor, and that the coefficients of perceived usefulness and referent network size are the highest among the three segments, the most effective marketing strategies would be to improve the practicality of applications and make use of the herd effect.Finally, regarding the "pressured hedonists" segment, perceived enjoyment is the most important factor, while the impact of attachment motivation and subjective norms are the highest among the three segments.Therefore, firms targeting this segment should consider increasing the entertainment functionality of the application, apart from relying on influencers to strengthen users' attitude.

| FUTURE RE S E ARCH AND LIMITATI ON S
We follow the theory-context-characteristics-methodology framework (Paul & Rosado-Serrano, 2019) to propose an agenda for fu- and WhatsApp for messages).Therefore, there is also heterogeneity across mobile social media platforms (Ju et al., 2021).
help scholars and firms to understand users' preferences and behaviors more comprehensively.Second, our study also contributes to the under-researched topics of unobserved heterogeneity and international microsegmentation in mobile social media by effectively detecting unobserved heterogeneity and segmenting users from different countries who share behavioral patterns.Knowledge regarding different user groups, including the most influential factors, demographic characteristics, and cultural values at the individual level, can help firms more effectively tailor their market offerings toward each segment.
This study provides insights into increasing experiential engagement and mitigating the negative influence of privacy concerns on the active use of location based SNSs.Zhang et al. (2017) To examine the effects of network externalities on users' perceived values and continuance intention.Network externalities, social interaction ties, Perceived values Structural equation modeling, 240 responses of WeChat users were collected via survey in Mainland China.This study provides insights into users' postadoption behavior of SNS by testing the mediating role of social interaction ties in the effect of network externalities on different perceived values.Gong et al. (2018)To explore the reasons why experienced mobile social apps users are likely to continue using the app.Experiential learning theoryStructural equation modeling, 295 responses of WeChat users were collected via survey in Mainland China.This study enriches our understanding of users' postadoption behavior of mobile social media and the role of levels of experience in increasing their continuous usage intention.Leong et al. (2018)To explore the determinants of students' usage intention of mobile SNSs for their pedagogical purposes.Technology Acceptance ModelStructural equation modeling, 600 students from public universities of Malaysia filled out the survey.The study extended TAM by confirming the moderating effect of experience on the intention to mobile social media adoption for pedagogical purposes.Tan et al. (2018Tan et al. ( (2014) )  focused on the economic and cultural factors that affected international country segments, and their stability over time, in 30 countries.Cleveland et al. (2011) studied urban consumers from eight countries and found that ethnic identity and demographics impacted consumer behavior.Wamba et al. (2017) assessed unobserved heterogeneity in the information system field by segmenting user social media acceptance within the workspace in five countries.
Omoush et al. (2012) andHoehle et al. (2015).Based on Hofstede's conceptualizations, power distance items examined the degree of acceptance of inequalities among people; the items' specific wording relied onLi et al. (2009).Masculinity examined whether competition, achievement, and success motivated people, while uncertainty avoidance measured users' reactions to ambiguous or unknown situations.The last items of these two variables (MF3 and UA3 in Appendix 1) were adapted fromLi et al. (2009), while the rest were self-developed.
nificant differences in the demographic variables among the three TA B L E 3 Mean, standard deviations (SD) and factor loadings Our hybrid research model integrates factors based on the TAM, motivation theory, and social influence theory.We tested the hypothesized relationships in the GM and the three local models.The findings show that users' attitude toward mobile social media is influenced by motivational and social influence factors.As expected, users' attitude results in actual use behavior.Nevertheless, some determinants differ across the segments because of the users' different demographic characteristics and cultural value orientations.First, regarding motivational factors, the results highlight perceived usefulness as the most influential factor (in the GM and two local models) that affects users' attitude toward mobile social media.
ture research.First, in terms of theory, our study included several factors from three relevant theoretical lenses (the TAM, motivation theory, and social influence theory).However, other theories and factors could explain attitude toward mobile social media and actual use and complete our model specification.In particular, perceived playfulness(Wamba et al., 2017) and perceived utility (Lacka & Chong, 2016) may be additional motivational factors, and informational social influence and normative social influence (Li, 2013) could be considered from a social influence theory perspective.In addition, attitude is the mechanism connecting motivational and social influence factors to mobile social media use.We explored its mediating role and found mixed results (the mediating role of attitude toward mobile social media varies across segments).Therefore, future research should study its mediating role in detail, and consider other potential mediators such as behavioral intention.Second, in terms of context, our study discusses mobile social media use behavior in general.However, there are different types of mobile social media platforms that are used based on their distinct functionalities (e.g., YouTube for videos, Instagram for photos, Future studies could explore attitudes toward mobile social media and use behavior across different types of platforms.In addition, despite the emphasis on the impact of the COVID-19 pandemic on customers' online behavior (Karpen & Conduit, 2020), few efforts have been made to study mobile social media use during the pandemic.Since people may have different motivations for using mobile social media in unusual situations, future research could replicate our study in light of the COVID-19 context to explore the potential differences in users' behavioral patterns.
Table 1 summarizes their objectives, theories, methodologies, and contributions.Studies focusing on mobile social media use behavior can be traced back to 2014 (see Table 1, column one), and have investigated the antecedents that drive individuals' use behavior (see Table 1, column two).Researchers have primarily studied continuous usage intention
Discriminant validity a , correlations, and square roots of the average variance extracted (AVE) (n = 132:66%), female (n = 108:54%), aged between 25-30 years TA B L E 4 Descriptive statistics and significance of local model categorical and continuous variables