A Test of the Mobile Phone Appropriation Model: A Comparison between Chinese and US Samples

The

O ver 60% of the world's population uses mobile phones, and these portable devices are now even more important for Internet accessibility than computers (DataReportal, 2021).For many, checking mobile phones is the first thing they do after waking up every day and the last action to take before going to bed.Mobile phones are not only multifunctional tools assisting various professional and personal activities but also symbolic devices embedded with identity and cultural values (Goggin, 2008).While people favor mobile phones for their usefulness and positive effects on life satisfaction, social cohesion, and conveniences (Wei et al., 2022), they also fear potentially negative outcomes brought by over-reliance on such technologies (Thomée, 2018).For example, intensive use of mobile phones could be associated with degraded physical fitness (Lepp et al., 2013), depression and anxiety (Coyne et al., 2019), low sleep quality (Exelmans & van den Bulck, 2016), and daytime dysfunction (Derks & Bakker, 2014).
In recent years, mobile phone adoption and usage have attracted great scholarly interests across disciplines, and everyday uses of mobile phones have been the most frequently investigated context (Kim et al., 2017).With the nearly universal adoption of cellular phones in such countries as the United States, Australia, China, and South Korea, the question goes above and beyond what kind of people choose to use them.More urgently, researchers seek to scrutinize how mobile phones are used differently across distinct socio-cultural groups and what outcomes such differences entail.Kim et al. (2017) reviewed the most prominent theoretical frameworks employed in mobile communication research are general theories of technology adoption (e.g., Rogers'[2003] diffusion of innovation theory, Davis' [1989] technology acceptance model), which were not specifically invented for explaining mobile communication per se.Omnipresence, portability, availability, locatability, and multimediality fundamentally distinguish mobile media from other communication technologies (Schrock, 2015), so the field calls for more specialized theorization.Therefore, some scholars have developed theoretical frameworks particularly dedicated to mobile communication (e.g., Bayer et al., 2016;Katz & Aakhus, 2002;Ling, 2012).
Among all, the mobile phone appropriation (MPA) model is one of the few theoretical frameworks seeking to comprehensively explain mobile phone appropriation (Wirth et al., 2007(Wirth et al., , 2008)).Rooted in the perspective of structuration theory, appropriation emphasizes the duality of technology structures (Orlikowski, 1992) and describes how people take an active role in selecting and determining the adoption and practices of technology, which may or may not be aligned with the spirits embedded by designers (DeSanctis & Poole, 1994).This perspective therefore overcomes the limitations of simple technological determinism or social constructionism (Leonardi, 2013) by recognizing that the variation in technology adoption and usage is jointly shaped by technology and social action.With the recognition that the meaning of media products is co-constructed by users, producers, and mass media, the term appropriation in the MPA model refers to "the process by which people adopt and adapt technologies, fitting them into their working practices" (Dourish, 2003, p. 465) and stresses that users achieve full ownership of products by fusing them into everyday life (Lee et al., 2016).Appropriation research differs from the adoption research and examines mobile communication above and beyond the binary choice of adoption (i.e., use and non-use).
In prior studies, the MPA scale was tested in its validity and reliability with a German and a US sample (Lee et al., 2016;Wirth et al., 2007), and the proposed model structure was later tested with a larger US sample (Lee & Cionea, 2023).However, it remains unclear whether the theoretical constructs and relationships stay equally applicable in non-Western contexts.Taking an approach of multigroup structural equation modeling, the current research examines the factor structure and relationships proposed by the MPA model with a Chinese sample to assess its compatibility with non-Western culture.With established measurement invariance, the study further discusses the similarities and differences in mobile phone appropriation across cultures by comparing path coefficients between the Chinese sample and the US sample collected by Lee and Cionea (2023).
X. Zhang et al.

The MPA Model
The MPA model developed by Wirth et al. (2007Wirth et al. ( , 2008) ) proffers an integrative framework for explaining individuals' mobile phone appropriation patterns from a communication perspective.This model synthesizes factors from five extant frameworks: diffusion of innovation (Rogers, 2003), theory of planned behavior (Ajzen, 1985), cultural studies (Silverstone & Haddon, 1996), frame analysis (Höflich, 2003), and uses and gratifications theory (Katz et al., 1973).Building upon previous frameworks, the MPA model conceptualizes mobile phone appropriation as a dynamic and creative process rather than binary choices and attempts to merge the divergence between qualitative and quantitative perspectives.The model contains three major components (see Figure 1): metacommunication, usage beliefs, and actual usage behavior.
The first dimension, metacommunication (i.e., communication about mobile phones), originally included three subdimensions: two parts: functional (i.e., control, distraction/ pastime, management of daily schedule, and relationship maintenance) and symbolic (i.e., social and psychological) usage.Starting from metacommunication, the MPA model posits individuals develop beliefs about different usage dimensions, which eventually impact their usage behavior, both functionally and symbolically (Wirth et al., 2008).Presenting mobile phone appropriation as a constant circulation, the model strives to capture the complex dynamics of everyday mobile phone usage.
The MPA model has made an impact in the field of mobile communication research.The 89-items scale accompanying this model was originally developed in German by Wirth et al. (2007) and later translated into English by Lee et al. (2016).Most subscales achieved good reliability after modification with the English scales via confirmatory factor analyses (CFAs).On the qualitative side, Humphreys et al. (2013) conducted interviews guided by MPA in the US and Germany to explore the perceptions of those characteristics of mobile Internet as well as extractive versus immersive usage, surprisingly finding little cross-cultural difference in metacommunication and handling of mobile Internet.Research by Aricat et al. (2015) simplified the circular model and focused on the interactions amongst three components: metacommunication, pragmatic usage, and symbolic usage (usage prestige and social identity).The findings revealed four types of mobile phone users (i.e., convenience seekers, experimenters, group communicators, and tabula rasa) in an Indian Malayali community in Singapore, related with three mindsets toward migrant acculturation (i.e., culture campaigner, culture connoisseur, and culturally petrified).On the quantitative side, Lee and Cionea (2023) further examined the relationships between the proposed factors with a US sample and found the overall MPA model received partial support via structural equation modeling.
These studies established MPA to be a viable theoretical framework for assessing mobile phone appropriation across cultures.
As a follow-up study to Lee and Cionea (2023), the present study first analyzes the MPA model's compatibility with a Chinese sample using a translated (and back-translated) Chinese scale.Based on the original MPA model proposed by Wirth et al. (2008), the initial hypotheses are proposed as below, which are later revised based on the results of the multigroup CFA in the current paper.
H1: Levels of metacommunication will be positively associated with levels of (a) functional, (b) symbolic, (c) normative, and (d) restrictive evaluations.H2: Levels of functional evaluations will be positively associated with levels of (a) functional and (b) symbolic aspects of usage.H3: Levels of symbolic evaluations will be positively associated with levels of (a) functional and (b) symbolic aspects of usage.H4: Levels of normative evaluations will be positively associated with levels of (a) functional and (b) symbolic aspects of usage.H5: Levels of restrictive evaluations will be negatively associated with levels of (a) functional and (b) symbolic aspects of usage.countries.Before the COVID-19 pandemic, Chinese adults spent 2.5 hours on mobile phones every day (Liu, 2020) whereas Americans spent 2.9 hours on average (Wurmser, 2019).Because of quarantines, people's usage amount significantly grew from 2020 to 2021 (China Internet Watch, 2021;O'Dea, 2021).The large number of users and considerable usage duration make mobile phone appropriation an essential research topic for both countries, along with the expansion of international business and intercultural contacts.

Cultural Influences on Mobile Phone Appropriation
Yet scholarly attention has been paid to China unproportionally, compared with the US and South Korea, where most of the mobile communication research has been conducted (Wei et al., 2022).By reviewing the publications on mobile communication from major communication journals, Wei et al. (2022) revealed the number of studies in the US context is seven times greater than those taking China as the study site, although China has three times as many users.More scholarly attention needs to be paid to the unique context of Chinese mobile communication, and our research can provide some important insights into how it differs from the American context (Ang & Zhou, 2023).More broadly, research findings can provide a more in-depth understanding of both universal and culturally specific factors that influence mobile phone appropriation.

The Individualism-Collectivism Dimension
The current study further explores psychometric equivalence and factor structures of the MPA model to examine its applicability above and beyond cultural heterogeneity, and the large cultural distance between China and the US makes them good reference points for each other (Shi & Wang, 2011).One factor contributing to cultural differences in mobile phone appropriation is the individualismcollectivism (IC) dimension.While Chinese culture is typically considered collectivistic, where individuals are tightly linked to their social units, the culture of the US is considered highly individualistic, where kin connections are much weaker (Hofstede, 1980).In fact, the metaanalysis conducted by Oyserman et al. (2002) showed that China and the US make an ideal contrast because China is the only Asian country that consistently exhibits large effects of both low individualism and high collectivism, as opposed to the US, a country with high individualism and low collectivism.
Whereas mobile phones are considered devices promoting individualism in the West, mobile communication research has illustrated they can be reaffirming collectivism under collectivistic cultures such as South Korea (Yoon, 2003) and China (Holmes et al., 2015).Previous studies have also shown the IC dimension has significant influences on mobile phone use.For example, Arpaci (2019) found that horizontal collectivism, which emphasizes ingroup equality and social harmony, is positively associated with anxious attachment to mobile phones in Turkey, and that collectivism reduces the impacts of perceived ease of use on perceived usefulness of mobile phones in Jordan (Faqih & Jaradat, 2015).In sum, cultural values and norms can moderate people's mobile appropriation behavior.

Observed Differences in Beliefs and Usage
As proposed by van Biljon and Kotzé (2008), mobile phone usage is greatly determined by the sociocultural context beyond technological usability and functionality.An ongoing line of research has focused on how new media are utilized differently across cultures (Shuter, 2012).On the macro level, numerous studies (e.g., Stump et al., 2008;Vimalkumar et al., 2020) have explored national mobile phone diffusion and use with a broad focus on social, economic, and political factors affecting global digital divides and national information technology infrastructures (Qiu, 2010).Besides, some studies have revealed how cultural backgrounds, in addition to micro-level individual characteristics (e.g., income, education, age, gender), influence mobile phone usage despite the pressing uniformity of mobile communication across countries (Katz & Aakhus, 2002).Past literature contains three prominent themes revealing significant cultural differences associated with individuals' attitudes towards mobile phones and their use behavior: mobile communication norms, purchase intentions, and usage patterns.
For communication norms, Mante (2002) noted that Dutch participants sensed less responsibility of being socially reachable and more repulsion against work interfering with personal lives.Similarly, Caporael and Xie (2003) found that Chinese and American users had different ideas about whether it was acceptable to receive phone calls after work.Campbell (2007) collected college student samples from Hawaii, Japan, Sweden, Taiwan, and the US, and showed that the Japanese were less tolerant of mobile phone use in public and that the Swedish attached more importance to security.Khattab and Love (2009) showed Sundanese participants were more willing to turn off phones in certain public spheres than British people.Shuter and Chattopadhyay (2014) found Danish and American people have different attitudes toward using mobile phones during faceto-face conversations in different settings with different relational partners.Altogether, these studies indicate mobile phones are appropriated differently across cultures regarding when to connect and whom to connect with.
For purchase intentions, Lee et al. (2013) found innovation factors had stronger effects on Americans' adoption decisions, while Koreans relied more on evaluations from other likeminded individuals.Concerning the symbolic meanings associated with mobile phones, Katz et al. (2003) observed Korean youth perceived mobile phones to be more expensive, stylish, and necessary than Americans, and Cui et al. (2007) identified Asians were more likely to decorate their phones than Europeans and Americans for self-display.Hoang (2015) asserted there are significant differences between Vietnamese and Finnish consumers in terms of shopping intentions and consciousness.Overall, past research indicates different cultures attach distinct identity values and symbolic meanings to mobile phones, with collectivistic cultures demonstrating greater social influences.
For usage patterns, Peltonen et al. (2018) pointed out low power distance and high individualism were related to greater usage of leisure-type apps, and Lee et al. (2002) found that Koreans perceived more emotional values in mobile communication while Japanese paid more attention to functionality.Worthington et al. (2012) found Germans were more likely to consider business and emergency as primary purposes of mobile phone usage beyond keeping in touch with family/ friends than people from Finland, Korea, and the US, and different cultures had distinct norms for privacy management.In general, it appears that collectivistic cultures attach more importance to the social dimension of mobile phone usage compared to the cases of individualistic cultures.
The extant literature primarily explores specific user behavior instead of systematically investigating multiple aspects of mobile phone appropriation.Thus, this study takes the integrative approach induced by MPA and simultaneously focuses on the connections amongst metacommunication (i.e., communication about mobile communication), user beliefs, and usage behavior.The study also fills some gaps in US-China comparison research, as prior cross-cultural comparisons of mobile phone appropriation between the two countries are scant.
The generalization made by Katz and Aakhus (2002) through a series of empirical studies is that different cultures exhibit striking similarities in mobile phone appropriation, so we expect the MPA model to be tenable across cultures.Before further exploring cultural influences on mobile phone appropriation, we need to first investigate whether different cultures interpret different components of the MPA scale in similar manners.Therefore, we requested the US data from Lee and Cionea (2023), collected a Chinese dataset using similar sampling parameters, and conducted a multigroup CFA to examine scale compatibility across two cultures' samples.
RQ1a: Is the MPA scale comparable between a Chinese and an English version?

Cross-Cultural Comparisons of MPA
As discussed above, we expect to observe some nuances in mobile phone appropriation despite overarching cross-cultural similarities.Given the large IC difference, Chinese people's beliefs and usage may be more closely linked to what is considered socially normative in the cultural context than Americans' (Meng & Kim, 2020).
With high collectivistic values, Chinese people may pay more attention to other community members' opinions on social occasions and attach greater importance to communication regarding social norms of mobile phone usage than Americans.As the metacommunication component of the MPA model spotlights how personal beliefs and evaluations of mobile phones can be jointly shaped by personal connections, social environment, and mass media and highlights the social discourse of mobile phone appropriation the effects of metacommunication may be stronger in China.
It is also possible that mobile phones are appropriated with more collectivistic purposes (e.g., social obligations, cohesion, support) in China than in the US due to the cultural emphasis on kinship and relationship maintenance.Predictably, Chinese people may draw greater connections between metacommunication and relationship maintenance as well as control.Furthermore, Chinese people may attach greater symbolic values to mobile phones, encompassing both social status display and psychological attachment, as indicated by past research on purchase intentions (Cui et al., 2007;Hoang, 2015;Lee et al., 2013).
The patterns of mobile phone appropriation may differ across cultures owing to distinct cultural norms and values, and the strength of cultural norms and social influence may also vary across cultures.Thus, we consider whether the associations between different components significantly differ by culture with established sample comparability.Although we could speculate over the potential differences, we propose the following research questions to maximize parsimoniousness in multigroup comparisons: Most Chinese participants were Han Chinese (95.9%), followed by Hui Chinese (0.8%), Uighurs (8.5%), Mongolians (0.6%), Tibetans (0.2%), and others (2.5%).Many American participants were non-Hispanic white (73.5%), followed by Hispanic (9.0%), African American (8.8%),Asian American (6.6%), native American (0.6%), and others (1.4%).For educational backgrounds, 75.7% of Chinese participants had college education, with 17.5 % high school education or below and 6.8% graduate-level education; whereas 61.8 % of American participants had college education, with 22.5 % high school education or below and 15.7% graduate-level education.The medium monthly income of Chinese participants was between 7,000 and 9,999 RMB (between approximately 1000 and 1,500 USD), and the medium monthly income for US participants was between 4,000 and 4,999 USD.

Procedures
The data for both samples were collected through a professional survey company, SurveyMonkey, during summer 2016.As compensation, the researchers paid $5 per response to the company for collecting data from their Chinese and US panelists.Two bilinguals between Chinese and English participated in translation and backtranslation of the Chinese MPA scales.Before filling out the survey, participants were informed of the general study purpose and asked to report their demographic information.See Appendix A for descriptive statistics of the key variables across cultures and scale reliability scores and Appendix B for the Chinese and English measures.The reliability scores reported below are calculated with two samples combined.
Restrictive Evaluations.Four items were used to capture participants' restrictive evaluations (RE) of mobile phones, including economic, temporal, and technical factors (Lee & Cionea, 2023).Three items were retained in the RE scale (α = .74).
Symbolic Aspects of Usage.Eight items measured the two subdimensions of symbolic aspects of usage (SA) (Lee & Cionea, 2023).One item was dropped from social (SASO; α = .87)and two items were dropped from psychological usage (SAPS; r = .70).

Multigroup Confirmatory Factor Analysis of MPA Scales and Invariance Testing
Before running structural equation modeling (SEM), a multigroup confirmatory factor analysis (MGCFA) was conducted with the R package, lavaan 0.6-14 (Rosseel, 2012) software with maximum likelihood (ML) estimation.Hu and Bentler (1999) emphasize that "it is difficult to designate a specific cutoff value for each fit index," but nonetheless suggest values for a "relatively good fit" (RMSEA ≤ .06,CFI ≥ .95,and SRMR ≤ .08;p. 449).High correlations (Tabachnick & Fidell, 2019) were found between three subconstructs of metacommunication (r between .94 and .96across samples).Likewise, the FEDIS and FEORG were highly correlated in the Chinese sample (r China = .97,r US = .75),and FECONT and FECONN were also highly correlated (r China = .94,r US = .98).SESO and NE were also highly correlated (r US = .92and r China = .99).Each of these values indicated multicollinearity, and the subdimensions were collapsed for (a) metacommunication and (b) FECONN and FECONT, as advised by Brown (2015).Because FEDIS and FEORG are more varied constructs, and because FEDIS had two poorly loading indicators, only FEORG was retained.Further, given the construct overlap between NE and SESO and the poor loadings for six NE indicators across samples (as in Lee & Cionea, 2023), we chose to drop NE.
DeVellis (2016) suggests that latent variables and indicators should share a moderate relationship (r 2 > .30,see Appendix A), we used this criterion for retaining indicators in both sampled groups.In line with measures reported above, five indicators with low loadings were iteratively removed and three indicators were removed as modification indices demonstrated cross-loading on multiple latent constructs.Error covariances were allowed between measurement items sharing similar phrasing or meaning within any given latent construct, but not between constructs.The final model met some, but not all, of Hu and Bentler's (1999) criteria: RMSEA = .051,CFI = .904,and SRMR = .064.Because this combined sample exceeds 1,000 participants and because CFI is definitionally contingent upon unique variances and is harmed in models with high levels of shared variance among latent constructs (Moshagen & Auerswald, 2018), we consider the fit acceptable.Given that normative evaluations and distract/ pastime dimension of functional evaluations were excluded and that subdimensions of metacommunication as well as functional evaluations of control and connections were combined, some parts of the initial hypotheses were not testable.In the following, we only report results of those hypotheses that were testable after these modifications.MGCFA facilitates measurement invariance testing to assess the psychometric equivalence of the models across the US and Chinese samples (Putnick & Bornstein, 2016).The model proceeded through three steps.First, establishing a baseline configural model which assesses model fit given multiple groups.This model assesses if the latent constructs have "the same meaning across groups" (Kühne, 2013, p. 155).The second model tests for metric invariance.Metric invariance implies "each item contributes to the latent construct to a similar degree across groups" (Putnick & Borenstein, 2016, p. 75).To test this assumption, we constrained the loadings across groups and computed a nested version of the configural model.The third, more stringent model tests for scalar invariance.Scalar invariance is constrained to both loadings and intercepts, which is also called strong factorial invariance (Kühne, 2013).When present, scalar invariance "generally supports cross-group comparisons of manifest (or latent) variable means on the latent variable of interest" (Rutkowski & Svetina, 2014, p. 35).Fit statistics across each invariance test are presented in Table 1.
Given the complexity of the models, it was unlikely that changes in χ 2 between the nested models which follow would be non-significant.This is because "χ 2 is overly sensitive to small, unimportant deviations from a 'perfect' model in large samples" (Putnick & Bornstein, 2016, p. 78).Thus, we used the alternative fit indices (i.e., ΔRMSEA, ΔCFI, and ΔSRMR) to assess fit.Specifically, we followed the guidance suggested by Rutkowski and Svetina (2014), ΔRMSEA < .03,ΔCFI < -.02, and ΔSRMR ought not exceed .03for metric invariance (Putnick & Bornstein, 2016).For scalar invariance, these criteria are more stringent: ΔRMSEA < .01,ΔCFI < -.01, and ΔSRMR < .015(Putnick & Bornstein, 2016).The models met assumptions for configural and metric invariance, signaling the latent constructs are psychometrically similar and comparable across cultural contexts.But the models did not quite meet the criteria for scalar invariance, signaling that it was inappropriate to compare means between the two countries.The baseline configural model and constrained comparison model statistics are presented in Table 1.

Analysis Plan
With metric invariance established, we utilized multi-group SEM (MGSEM) for hypothesis testing, we used maximum likelihood estimation.Though the results are similar to Lee and Cionea's (2023), our results differ slightly because we use the multigroup approach to revise the model, including scale item inclusion.MGSEM yields separate χ 2 values but a combined set of fit indices.For the Chinese data, χ 2 = 4713.83,df = 1828.For the US data, χ 2 = 4824.48,df = 1828.The model fit reasonably well, given the complexity: RMSEA = .06,CFI = .88,and SRMR = .09. Figure 2 presents the results from the US and Chinese sample.

Hypotheses Testing
Results for H1 For H1, metacommunication was positively predicted the functional evaluations on daily organization of life (β = .72,p < .001)and the combined factor of staying connected and control (β = .61,p < .001),which supported H1a.Metacommunication was significantly associated with social symbolic evaluations (β = .79,p < .001)and psychological symbolic evaluations (β = .36,p < .001),which supported H1b.It was also positively associated with restrictive evaluations (β = .15,p = .005),which supported H1d.In all, H1 was fully supported with the exception that H1c was not testable due to the removal of normative evaluations.Metacommunication had great explanatory power on functional and symbolic evaluations, and its effects on restrictive evaluations were comparatively smaller (see Figure 2).
The findings also partially supported H2b.Functional evaluations of organization of daily lives were positively associated with the social dimension (β = .13,p = .013)but not with the psychological dimension (β = .01,p = .813) of symbolic mobile usage.While functional evaluations of staying connected and control were negatively associated with the social aspect of symbolic usage (β = -.17,p < .001),this combined factor was positively associated with the psychological aspect of symbolic usage (β = .34,p < .001). Figure 2 shows significant paths and amount of variance explained.
For H3b, social symbolic evaluations were positively associated with both the social (β = .71,p < .001)and psychological (β = .24,p < .001)dimension of symbolic usage.Furthermore, the psychological symbolic evaluations were positively associated with the psychological dimension (β = .23,p < .001),but negatively with the social dimension (β = -.37,p < .001) of symbolic usage.Thus, H3b was partially supported with the positive associations between social symbolic evaluations and two dimensions of symbolic usage as well as a positive association between psychological evaluations and psychological usage, but the analysis also indicated a negative relationship between psychological evaluations and the social aspect of usage, which was unexpected.H4 was not testable due to the removal of the normative evaluations dimension from CFA.For H5a results showed restrictive evaluations were not associated with functional evaluations on control (β = .004,p = .912),but the paths from restrictive evaluations to distraction/pastime (β = -.13,p = .006),management of daily schedule (β = .09,p = .048),and relationship maintenance (β = .10,p = .028)were significant.Overall, H5a was supported for functional usage for distraction/pastime but not for other dimensions.Contradicting H5a, there were positive associations between restrictive evaluations and usage for managing daily schedules and relational maintenance.Regarding H5b, restrictive evaluations were negatively related to the psychological aspect (β = -.11,p = .022)but were positively associated with the social aspect of symbolic usage (β = .24,p < .001) in the opposite direction of the hypothesized.In summary, H5b was not supported for social aspects but was for psychological aspects of symbolic usage.

Exploring RQ1
For RQ1a, the results of MGCFA demonstrated that these two models were metric invariant but did not meet the criteria for scalar invariance.Thus, it was appropriate to compare paths but not means between the Chinese and US data.It was worth noting these results varied slightly from published models from Lee and Cionea (2023) because during the MGCFA, we retained several different indicators and relationships to improve model fit in both US and Chinese results.RQ1b asked what differences emerged when comparing the significant paths in the Chinese and US models, and RQ1c asked if there were significant differences of path magnitude in the two models.Figure 2 showed the contrast.
In both samples, there were positive relationships between metacommunication and all evaluation dimensions.Based on the z-score differences across the samples, we could conclude that the path coefficient from metacommunication to the psychological dimension of symbolic evaluations was greater in the US sample (z = 13.62)than that of the Chinese sample (z = 7.10), p < .001,and the path from metacommunication to restrictive evaluations was stronger in the US sample (z = 8.65) than it was in the Chinese sample (z = 2.83), p < .001.The differences imply that metacommunication was more closely related to these two factors in the US sample in comparison to the Chinese sample.
The associations between functional evaluations and two aspects of usage varied in a few ways between the Chinese and US participants.For functional usage, there was no significant relationship between functional evaluations of connection and control and functional usage for relationship maintenance in the US data, while the relationship was significant in the Chinese sample (β = .16,p < .001).Although the combined dimension (functional evaluations of connection and control) was significantly associated with functional usage for distraction/pastime, the relationship was positive in the US sample (β = X.Zhang et al. .21, p < .001)while being negative in the Chinese sample (β = -.11,p = .003).
For symbolic usage, the association between the functional evaluations on organization of daily lives and social aspect of usage was negative in the US sample (β = -.19,p < .001)but positive in the Chinese sample (β = .13,p = .013).In addition, the association between the functional evaluations on connection and control and social aspect of symbolic usage was non-significant in the US sample, but negative in the Chinese sample (β = -.17,p < .001).The z-score differences indicated that the remaining significant paths did not significantly differ across samples.
For symbolic evaluations, there were several variations.In terms of functional usage, the social evaluations were positively related to usage for control in the US sample (β = .15,p = .028),in the opposite direction of the Chinese findings (β = -.13,p = .016).Further, there was a significant relationship between social evaluations and functional usage for relationship maintenance in the US sample (β = .35,p < .001),while there was not one in the Chinese sample.
For the psychological evaluations dimension, it was significantly associated with usage for control only in the Chinese sample (β = .61,p < .001)but not in the US sample.The dimension was negatively associated with usage for distraction/ pastime in the US sample (β = -.19,p = .002)but positively in the Chinese one (β = .21,p < .001),and it was positively associated with usage for daily schedule management in the US sample (β = .16,p = .005)but negatively in the Chinese sample (β = -.21,p < .001).The z-score difference indicated that the path from the psychological evaluations to relational usage was stronger in the Chinese sample (z = 6.82) than it was in the US sample (z = 2.15), p < .001.
In terms of social usage, a reversed relationship between psychological evaluations and social usage was observed in the Chinese sample (β = -.37,p < .001)while the relationship was positive in the US sample (β = .26,p < .001).The analysis also revealed a positive association between psychological evaluations and psychological usage in the Chinese sample (β = .23,p < .001),which was non-significant in the US sample.The z-score differences indicated that the remaining significant paths did not significantly differ across samples.
Finally, restrictive evaluations had many changes from the US to the Chinese data.We found restrictive evaluations to be positively associated with social usage in both the US (β = .37,p < .001)and Chinese (β = .24,p < .001)sample.There was a negative relationship to usage for control in the US data (β = -.20,p < .001),while there was no such relationship in the Chinese data.Restrictive evaluations were not significantly associated with usage for management of daily schedule or relationship maintenance and psychological usage in the US sample but were positively associated with usage for daily schedule management (β = .09,p = .048)and relationship maintenance (β = .09,p < .001) as well as psychological usage (β = -.11,p = .022)in the Chinese sample.By comparing the z-score differences, we observed a stronger association between restrictive evaluations and social usage in the US sample (z = 7.82) than that of the Chinese sample (z = 5.12), p = .007.In all, contrasting the Chinese and US models, there were 15 paths with varied signs, new, or missing relationships, out of 35 total possible paths.That means there were 20 paths of the same valence and similar significance, four of which had different magnitude across cultures.We discuss these differences and similarities in greater detail below.

DISCUSSION
The current study tested the MPA model (Wirth et al., 2007(Wirth et al., , 2008) ) with a Chinese sample and compared the mobile phone appropriation patterns of Chinese participants with those of a US sample collected by Lee and Cionea (2023) post hoc.Overall, the test results bolstered the core logic of MPA (i.e., "metacommunicationevaluations-usage") within the Chinese sample, and multiple strong path coefficients that indicated the MPA model had satisfactory explanation power, with the model explaining more than 30% of factor variances for eight latent factors out of 10.The findings also suggested it was edifying to discriminate between functional, symbolic, and restrictive aspects of mobile phone evaluations as well as the functional and symbolic aspects of usage.In the following section, we discuss several issues of the MPA model revealed by MGCFA and MGSEM and highlight the potential influences of cultural contexts.

Measurement of the MPA across Cultures
The CFA results revealed some problems of this 89-item MPA scale.First, the high correlations in the study suggested that the factor structures of metacommunication and symbolic evaluations should be reconsidered.Corresponding with Lee and Cionea (2023), the present study found three subconstructs of metacommunication (i.e., interpersonal, mass-mediated, and observational) had poor discriminant validity and issues with multicollinearity.Different aspects of metacommunication might have been measured without enough distinction, or the participants might have processed metacommunication holistically.
As media technology rapidly advances, it is no longer easy to make clear distinctions between interpersonal and mass communication via any technology, as described by O'Sullivan and Carr's (2018) concept of masspersonal communication.The participants might not have precisely identified information sources (e.g., a person, or a person shown in the mass media) when asked to recall metacommunication related to mobile phones.
Additionally, the present study also found empirical overlaps between functional evaluations for control and connectivity.As argued by Lee and Cionea (2023), the dimension of control evaluations seemed to include two types of control: accessibility (e.g., "It is important for me to be available 24/7") and environmental surveillance (e.g., "It is important for me to be in control of my surroundings").The former emphasizes control exerted on self-behavior, which also overlaps with the relationship maintenance need, while the latter focuses on control exerted on the external environments, which may have negatively impacted the internal consistency of this subscale.Also, the connectivity dimension appears to overlap with control by asking participants to indicate how important it was for them to always know what was going on with significant others (environmental surveillance) and to always stay in touch with friends and family (accessibility), which can explain why the two factors were highly correlated in both samples.Moreover, the evaluation subscales differed from the corresponding usage subscales for connectivity and control in the way that control usage subscale only focuses on accessibility, and the relational usage subscale only focuses on relationship maintenance.Thus, the conceptualization and operationalization of control and connectivity evaluations should be reconsidered in their future measurement and to align better with the ones for usage.
Noticeably, two dimensions were dropped from the scale: functional beliefs on distraction/ pastime and normative evaluations.The distraction subscale was excluded due to its high correlation with the organization of daily lives subscale, indicating the dimensions highly overlapped for the participants.Two items from the distraction/ pastime subscales were closely related with management of daily schedules and chronemic expectations, such as "It is important to me that I do not waste my time with anything during the day, " which offers an explanation why their factor loadings were poor.The subscale for normative evaluations was excluded partially due to the lack of internal consistency.This subscale contained both X. Zhang et al.
positively and negatively worded statements (see Appendix B), encompassing both communication norms (when and where to use mobile phones) and social implications of usage (how other people think of certain use behavior), which makes the subscale problematic.Herein, we advocate for further revisions of these two subscales.

Model Structure
Given that only H1 was fully supported, with other hypotheses being partially supported, the findings revealed some practice of mobile phone appropriation in the Chinese sample that deviated from the predictions made by the MPA model.The MGCFA and MGSEM results further unmasked some interesting cultural differences.In this section, we discuss the missing paths after comparing two samples, highlight those paths with reversed relationships, and elucidate the observed differences in path magnitude.
First, the analysis revealed seven significant paths in the Chinese sample that the US sample did not contain: from functional evaluations of connectivity and control to (a) functional usage for relationship maintenance and (b) social usage; from psychological evaluations to (a) usage for control and (b) psychological usage; and from restrictive evaluations to (a) usage for schedule management, (b) usage for relationship maintenance, and (c) psychological usage.The results potentially indicated that for Chinese people, evaluations for relational usage penetrate more aspects of mobile phone use.Possibly, relationship maintenance behavior is incorporated more into people's everyday agenda and is motivated more by their needs for environmental surveillance and accessibility in the Chinese context, given the emphasis on social connections by collectivistic cultures.The perceived usefulness of mobile phones as a tool for relationship maintenance reinforces its symbolic usage for socializing, which means Chinese people convey social messages by mobile phone usage not only to the social environment in general but also to their significant others.Such messages can encompass both identity display and relational messages (e.g., importance, closeness, intimacy).
That Chinese people's collectivism and interdependence emphasizes the social environment more could be a reason behind the association between psychological importance and usage for control.Indeed, in past work the fear of missing out was greater among Asians than Americans (Karimkhan, & Chapa, 2021), so the importance of mobile phones might have been boosted by their stronger desires to stay in the loop through mobile phone use.The absence of the predicted association between psychological evaluations and psychological usage in the US sample could be explained by habitual attachment: living in a more techsavvy environment than Chinese participants, American participants might have been psychologically attached to their mobile phones regardless of how they evaluated such attachment.
It was particularly interesting that restrictive evaluations positively predicted usage for relationship maintenance and daily schedule management in the Chinese context, which contradicted the model.This implies the more costly Chinese participants perceived mobile phones usage to be, the more they used it for relationship maintenance and daily schedule management, which distinguished them from American participants.The logic behind this finding might be that the more financial, technical, temporal, or cognitive difficulties Chinese people perceived, the more they perceived mobile phones usage to be a form of privilege or luxury.As reflected by Aricat et al. (2015), even though mobile phones and mobile services were considered expensive, many migrant workers still chose to use them because such use could not only display their financial resources but also boost their social status as they were treated as technical experts with respect in the community.In the same sense, Chinese participants who perceived more restrictions and barriers might have paradoxically increased their usage as a means to gain both social respect and relational communication with their significant others.Therefore, it is important to consider the inverse relationship identified between affordability and functional usage when examining mobile phone appropriation under the Chinese context.
The US sample also contained some paths that were missing from the Chinese sample: (a) from social evaluations to functional usage for relationship maintenance and (b) from restrictive evaluations to usage for control.The reason why this proposed relationship between social evaluations and relational usage was not observed in the Chinese sample was potentially due to social evaluations being constructed with individualistic values.The items from the subscale emphasized uniqueness and independence, which deviates from the collectivistic self-construal of harmony and interdependence (Markus & Kitayama, 1991).To pursue cross-cultural applicability, the may need to be revised to reflect more universal values accompanying mobile phone appropriation.Moreover, the absent path from restrictive evaluations to usage for control in the Chinese sample may be explained by the fact that using mobile phones for control is less of a social display in general.Because staying reachable (accessibility) is less socially observable as a constant state than calling friends (relationship maintenance) as an episodic activity, the paradox between affordability and usage may be less salient in mobile usage for control.
Besides, the analysis revealed several negative relationships supported by the Chinese sample, which were positive in the US sample: between the functional evaluations of connection and control and functional usage for distraction/pastime, between social evaluations and usage for control, as well as between psychological evaluations and (a) usage for daily schedule management and (b) social symbolic usage.The negative association between the functional evaluations of connection/control and distraction/pastime usage in the Chinese sample may reflect the particularly negative connotations for distracted usage under the cultural context.Chinese users who valued functions of relational maintenance and control might have considered themselves to be serious users, which made them despise usage for distraction and entertainment and perceive such uses to be frivolous.Therefore, the more importance they attached to the pragmatic value of mobile phones, the less they used them to kill time.That the Chinese data revealed a negative association between social evaluations and usage for control may be explained by the similar division that Chinese people draw between serious users and non-serious users and the notion that control usage is less observable in the public sphere.The less they considered mobile phones to be objects for social display, the more they pursued pragmatic uses of mobile phones for accessibility, and vice versa.
By contrast, there were two paths with positive valence in the Chinese sample but negative valence in the US sample: (a) the functional evaluations on organization of daily lives and social aspect of usage, and (b) psychological evaluations to usage for distraction/pastime.The first difference indicates the more US participants valued mobile phones for facilitating management of daily activities, the less they used them for social display.One reason for this phenomenon could arguably be that Americans draw more clear boundaries between the professional and the personal when it comes to technology usage than Chinese people (Caporael & Xie, 2003).Because of this, the more US participants considered mobile phones to be a tool for (work) schedule management, the less they were willing to use it symbolically for identity or status display.
As for the negative path from psychological evaluations to distraction/pastime, one candidate explanation is that distraction/pastime usage is socially stigmatized so there were more errors in self-reporting caused by social desirability bias and cognitive dissonance.American participants might not have been very honest about how often they used mobile phones for distraction/ pastime because of their perception of those negative images associated with it.Another possibility is that usage for distraction/pastime has become so common that people habitually kill time with mobile phones usage, no matter what they think of psychological attachment (similar to how people keep smoking or driving inefficient vehicles, despite knowing the risks or harms).In this case, they may play with mobile phones due to habits and customs even if they deem psychological attachment to be undesirable.Under either condition, these dynamics require further exploration.
Last, there were four paths with significantly different magnitude across the two samples.The path from metacommunication to the psychological evaluations was stronger for American participants than for the Chinese, indicating that Americans may communicate more about mobile communication interpersonally, through mass media, or by observation of others, and this active metacommunication aligns with their psychological evaluations more.It was also observed that metacommunication better predicted restrictive evaluations in the US sample, and our guess is that Chinese people prefer not to talk about restrictions as much interpersonally and that it is not discussed as often by mass media, given the social implications attached to it.Acknowledging ignorance or inability may be more embarrassing and face-threatening for Chinese users.Another guess is that because Chinese people recognize the symbolic values of mobile phones, they tend to disregard metacommunication concerning use restrictions and believe the claims to be less relevant relying more on their own experience to form their perspectives.
Next, the path from psychological evaluations to relational usage was found to be stronger in the Chinese participants, potentially because their psychological attachment to social networks (Shi & Wang, 2011) was a bigger part of their attachment to mobile phones, under the collectivistic appropriation of mobile phones (Arpaci, 2019;Holmes et al., 2015;Yoon, 2003).

Limitations and Future Directions
One limitation of the present study is that the data was collected in 2016, and mobile technology has greatly advanced since then.Thus, future studies may scrutinize how mobile phone appropriation has evolved over the years including its usage for health aspects (e.g., walking, heart rate, connection with smart watches) and short form videos consumption (e.g., TikTok and Instagram Reels).Another limitation is that two dimensions (i.e., functional evaluations on distraction/ pastime and normative evaluations) were dropped from the model testing, which inhibited some of the initial hypotheses and comparisons.Furthermore, the causality of the hypothesized relationships could not be determined with our cross-sectional data.Additionally, self-report data might have increased measurement errors especially for dimensions of usage behavior.Future researchers should improve ways to capture mobile users' actual behavior, not based on their recall or perception.The data will be more accurate and reliable if collected directly from tracking mobile devices instead of selfreporting (Kobayashi & Boase, 2012).

CONCLUSION
In conclusion, the MPA model provides a useful framework for examining the complex reality of mobile phone appropriation, and most of the proposed relationships were supported under different cultural contexts (i.e., the US and China).Although the kernel logic of MPA (i.e., metacommunication-evaluationsusage) seemed to be tenable, the MGCFA results suggested potential avenues for model revisions, and a more parsimonious model can be attained given many subdimensions overlapped.Further comparisons between the Chinese and US samples indicated the social contexts were more influential in Chinese people's MPA, and socio-cultural differences greatly shaped how people understood different usage purposes.The analysis also revealed an interesting paradox between affordability and usage, given the social implications of mobile phones in the Chinese context.Overall, more unexpected paths were found in the Chinese model, indicating that Chinese people's MPA deviated more from the hypothesized initially developed in the German context, than Americans' .Therefore, more research efforts are needed to illuminate how highly distinct cultural contexts of mobile phone appropriation impact metacommunication, evaluations, and usage of mobile phones.It is important to me to always be reachable, whenever and wherever.
FECONT2 对我而言，随时关注身边发生的事情很 重要。 For me, it's important to be aware of what is going on around me.
It is important for me to be in control of my surroundings.
It is important for me to be available 24/7.

SESO1
我认为，手机能够把我从尴尬的情景中 解救出来(如，通过电话)是一件好 事。 I think [it is good] that with a mobile phone one has the ability to pull back from uncomfortable situations (for example, by pretending to answer a phone call).
SESO2 因为手机，一个人变得独立，我认为这 是一件好事。 I think it is good that one can be independent thanks to their mobile phone(s). SESO3

Figure 2 .
Figure 2. Paths Coefficients and Variances Explained in the Chinese (left) and US Sample

FAC3
my cellphone with me to be able to reach my family all the time.FAC4 通常情况下我会随身带着手机，以便与 朋友随时联系。 I always have my cellphone with me to be able to reach my friends all the time.FAC5 通常情况下我会随身带着手机，以便家 人与我随时联系。 I always have my cellphone with me so that my family can reach me all the time.FAC6 通常情况下我会随身带着手机，以便朋 友与我随时联系。 I always have my cellphone with me so that my friends can reach me all the time.phone when there's nothing better to do.FAM1 我使用手机来安排每天行程。 I use my phone to set up my daily schedule.FAM2 我使用手机来有序安排我的行程、联系 人以及需要完成的事情。 I use my phone to keep my schedule, my contacts and my responsibilities in order.FAM3 我使用手机来安排工作相关的会见。 I use phone to coordinate work-related appointments.FAM4 我使用手机来安排与朋友的活动。 I use phone to plan activities with my friends.FAM5 我使用手机来规划家庭活动。 I use phone to plan activities with my family.Label Chinese item English item any, are present in the factor structure of the MPA in the Chinese and US sample?RQ1c: What differences, if any, are present in the path magnitude of the MPA in the Chinese and US sample?
METHOD ParticipantsBoth Chinese and US samples were drawn from general populations, and the sampling parameters for the Chinese sample were set in the way to maximize its comparability with the US sample, which was collected by hiring the same online survey panel service.There were 510 completed surveys in total by Chinese and 502 by US participants.For the Chinese sample, approximately half of the participants were males (n = 260), and half were females (n = 250).For the US sample, there were 251 male participants, 250 female participants, and 1 non-binary participant.For the Chinese sample, age and 39 (19.9%); 40 to 49 (20.1%);50 and 59 (19.9%); and above 60 (20.1%).

Table 1 .
MGCFA Testing Model Invariance X. Zhang et al.
X. Zhang et al.Final Scale Reliability Scores in the U.S. SampleIt is important to me that I have an organized schedule.It is important to me that I adapt to any change of schedules quickly enough.It is important to me to always know what is going on with my significant others.
From a mobile phone's accessories, such as its cover and decorations, one can tell what kind of person the owner is and what values s/he has.There are functions I would like to use but are not available in my mobile phone because it is too old.You would need a lot more time to be able to learn fully what functions a mobile phone has to offer and be able to operate it.I always have my cellphone with me to be able to call for help in emergencies.