Baixe o app para aproveitar ainda mais
Prévia do material em texto
lable at ScienceDirect Computers in Human Behavior 71 (2017) 196e208 Contents lists avai Computers in Human Behavior journal homepage: www.elsevier .com/locate/comphumbeh Understanding factors influencing information communication technology adoption behavior: The moderators of information literacy and digital skills Tai-Kuei Yu a, *, Mei-Lan Lin b, Ying-Kai Liao c a Department of Business Administration, National Quemoy University, One University Road, Jinning Township, Kinmen Hsien, Taiwan b Department of Hospitality Management, Southern Taiwan University of Science and Technology, Taiwan c Bachelor's Program of International Business, Nanhua University, Taiwan a r t i c l e i n f o Article history: Received 17 July 2016 Received in revised form 9 January 2017 Accepted 1 February 2017 Available online 1 February 2017 Keywords: Media richness theory Digital divide Information literacy Digital skill Moderating effect * Corresponding author. E-mail addresses: yutk2000@gmail.com (T.-K (M.-L. Lin), yksuper889@nhu.edu.tw (Y.-K. Liao). http://dx.doi.org/10.1016/j.chb.2017.02.005 0747-5632/© 2017 Elsevier Ltd. All rights reserved. a b s t r a c t Digital inequality is one of the most critical issues in the “information age”, few studies have examined the social inequality in information resources and digital use patterns. In the rural areas, such infor- mation communication technology (ICT) facilities could not guarantee that users can easily access in- formation technology and overcome the so-called “digital divide.” This research aims to discover the psychological factors that influence information and communication technology (ICT) adoption behavior, as well as confirm whether “information literacy” and “digital skills” have moderator effects in the research model. Using a survey of 875 participants and a structural equation modeling approach, we find that task characteristics and social interaction improve media richness, media experience, and media technostress, which in turn enhance ICT adoption behavior. The proposed theoretical model shows that the impact of ICT adoption behavior is moderated by information literacy and digital skills. The findings of this research can offer guidelines for policy makers and educators who evaluate a community's ICT adoption behavior so as to provide proper access to ICT and promote its visibility by incorporating ICT in educational activities. © 2017 Elsevier Ltd. All rights reserved. 1. Introduction According to Taiwan's internet users behavior investigation (Taiwan Network Information Center, 2013), the number of regular Internet users in Taiwan reached 17.98 million in 2013, comprising 77.10% of the entire population. We are in the midst of globalization and the so-called “Information Age”. Although the cost of infor- mation and communication technology (ICT) devices has dramat- ically decreased in recent years, their acquisition and maintenance still require financial commitment for households in the rural areas. According to Venkatesh and Sykes (2013), social inequality infor- mational resourcemaymanifest itself not only in differential digital access but also in differential digital use patterns. The “digital divide” generally implies differences in access based on socio- economic divisions (van Deursen & van Dijk, 2015). In many . Yu), mllin@stust.edu.tw countries, residents of urban areas have better information literacy than those in the rural areas, making the existence of the digital divide a common rural phenomenon. Compared with the signifi- cant penetration of information technologies in urban areas, the application of ICT is less prevalent in rural areas (Gerpott& Ahmadi, 2015a). Hence, individuals who have been reared within more privileged socio-economic environments acquire more opportu- nities to access ICT. ICTadoption behavior is a vital topic because ICT has become ubiquitous, pervading our daily lives in various ways. This is especially true for the Taiwanese youth, whose exposure to digital media shapes the way they communicate and interact with the rest of the world. Some scholars (Friemel, 2016; Ghobadi & Ghobadi, 2015; van Deursen & van Dijk, 2015; Venkatesh & Sykes, 2013) have suggested that the high failure rates of projects meant to reduce the digital divide are due to a lack of under- standing of different ICT choice behaviors. In fact, the acceptance/ adoption of the ICT approach has been identified as a critical issue in improving digital divide in its successful usage. Many theoretical frameworks have been used to measure technology usage and adoption; however, few have been used in the context of improving mailto:yutk2000@gmail.com mailto:mllin@stust.edu.tw mailto:yksuper889@nhu.edu.tw http://crossmark.crossref.org/dialog/?doi=10.1016/j.chb.2017.02.005&domain=pdf www.sciencedirect.com/science/journal/07475632 www.elsevier.com/locate/comphumbeh http://dx.doi.org/10.1016/j.chb.2017.02.005 http://dx.doi.org/10.1016/j.chb.2017.02.005 http://dx.doi.org/10.1016/j.chb.2017.02.005 T.-K. Yu et al. / Computers in Human Behavior 71 (2017) 196e208 197 the ICT digital divide. Recent researches (Brown, Dennis, & Venkatesh, 2010; Weber & Kauffman, 2011) define ICT as medium that support data and in- formation processing, storage, transmission, and communication via the Internet and other means. The flexibility of technology- enabled work has not only brought many advantages to busi- nesses and various organizations, but has also become embedded in the fabric of every activity. The willingness of people to use ICT needs to be considered together with technology, products or ser- vices, social interaction, and human factors (Goldhammer, Naumann, & Kebel, 2013; Zylka, Christoph, Kroehne, Hartig, & Goldhammer, 2015). ICT conveys forms of knowledge and literacy as well as integrate places (e.g., home, school, work, and commu- nity) of learning (Moore, 2013). The primary reason is the greater expansion of completed Internet infrastructures in urban than in rural areas. Therefore, in recent years, the utilization of ICT in rural areas has gradually received more attention because, from both of economic and social concerns, those ICT drive rural regional economies and improve the quality of life. Therefore, reducing the digital divide between urban and rural areas has become main- stream public policies in many countries (Bruno, Esposito, Genovese, & Gwebu, 2010; Gerpott & Ahmadi, 2015b; Ghobadi & Ghobadi, 2015). van Deursen, van Dijk, and Peter (2015) argued that digital technologies and media have influenced not only the intensified social connections but also human daily activities as well. ICT use can reduce the digital divide across the curriculum, as revealed by examinations of information technology skill and standardizedmeasures of information literacy. Consequently, while ICT appears to motivate some people to use new information technologies, measures have hardly capitalized on the intrinsic properties of information literacy as well as interactive and online technology. The digital divide debate has centered on the acquisi- tion of necessary skills and literacy to use ICT efficiently and effectively. These changes challenge our understanding of the digital divide, information literacy, and information technology skills in rural area communities. In the contemporary knowledge-based economy and the Internet era, personal wealth is no longer the sole attribute that gives a person a rich or poor social status; at present, knowledge and information have also become important determinants of wealth (Ayanso, Cho,& Lertwachara, 2014; Van Deursen& Van Dijk, 2011) under the premise that accumulated knowledge is conveyed worldwide through the Internet-based media. Thus, those who have no digital skills or no access to computers and the Internet at home have become the minority groups. Therefore, the digital divide has lead to a new era of social welfare issue. In order to understand the usage of technology and ICT adoption behavior,this study aims to address the following research questions: Research question 1. We follow the stream of research that focuses on individual acceptance of technology by using intention as the dependent variable. What are the psychological factors and the characteristics of ICT media impacts of rural residents' ICT adoption behavior. Research question 2. How does the information literacy and digital skill mix moderate the impacts of media experience and media technostress on ICT adoption behavior. 2. Theoretical foundations The medium of information technology has rapidly shortened the communication speed between the communicators and re- ceivers, thus improving knowledge communication diffusion and collection. Zhong (2011) and Aesaert, van Braak, Van Nijlen, and Vanderlinde (2015) argued that the role of information in social differentiation lies in the information environment based on a society's resources, relationships, and ICT. Furthermore, ICT itself plays the role of being the communicative bridge between in- dividuals and others in a society (De Wit, Heerwegh, & Verhoeven, 2012). The ICT medium could provide service interaction to people and enable them to access peer opinions; hence, people can see the potential for additional benefits (e.g., more choices, lower prices, better quality of goods) through additional effort. Meanwhile, people all want to eliminate the incompatible barriers within different media form in ICT, and expect ICT to greatly increase the collection capability of information and knowledge (King & Xia, 1997; Wu, Chang, & Sha, 2016; Zhang, Li, Ge, & Yen, 2012). In the process of communication, users have to use their own under- standing of information structure or content to express knowledge; the process of information transfer is influenced by individual or group capabilities, which involve individual factors (e.g., core ca- pabilities of information and information literacy, social pressure, common beliefs, and incentives motivation), social network struc- ture, physical proximity, recipient availability, and media charac- teristics (presence of symbols and accessibility to media) (Aesaert et al., 2015; Gerpott & Ahmadi, 2015a; Wang, Guo, & Jou, 2015). According to Burns, O’Connor, and Stocklmayer (2003), the objec- tive of effective communication is to achieve the five English characters of AEIOU: present awareness, enjoyment, interest, public opinion, and understanding; these comprise the core of a so- called “net-society.” 2.1. Media richness theory MRT was first proposed by Daft and Lengel (1986). This theory posits that different platforms of communication have different levels of media richness (medium richness/information richness). Carlson and Zmud (1999) integrated MRT, the symbolic media, communication patterns, and so on, and proposed channel expansion theory to clarify the relationship between group communication and the channel of communication. The term “media richness” refers to the richness of communication media that can pass through a number of information channels (Ayanso et al., 2014; Friemel, 2016; Purdy & Nye, 2000). MRT initially focused on information in organizations and the process by which media selection affects organizational structure. MRT posits that communication consists of a variety of cues, such as verbal tones, facial expressions, body language, dress, appearance, and settings, which convey information to receivers, that in turn, creates or elicits cognitive interpretations and concomitant social, emotional, and characteristic states (Kahai & Cooper, 2003; Van Deursen, van Dijk,& Peters, 2011; Kishi, 2008; Purdy&Nye, 2000; Xu, Ma,& See- To, 2010). Furthermore, the richness of a communication medium refers to its capacity for immediate feedback, multiple cues, lan- guage variety, and personalization. On the one hand, the lower the level of richness, the more difficult it is to quickly communicate, develop an understanding of each other's viewpoints, and resolve differences (Bertot, Jaeger,&Hansen, 2012; Liu, Liao,& Pratt, 2009). On the other hand, the higher the level of richness, the easier it is to convey information to receivers, thus facilitating better communi- cation and social interaction because the information receiver could rapidly and correctly understand the other party (Dennis & Kinney, 1998; Purdy & Nye, 2000; Liu et al., 2009; Anandarajan, Zaman, Dai, & Arinze, 2010; Van Deursen et al., 2011). A number of researches about online classes (e.g. Baehr, 2012; Giesbers, Rienties, Tempelaar, & Gijselaers, 2013; Lin, Wen, Jou, & Wu, 2014; Liu et al., 2009; Lu, Kim, Dou, & Kumar, 2014; Wang et al., 2015), workplace (e.g. Badger, Kaminsky, & Behrend, 2014; King & Xia, 1997; Sheer & Chen, 2004; Wu et al., 2015) and social networking service (e.g. Lee, Son, & Kim, 2016; Ogara, Koh, & T.-K. Yu et al. / Computers in Human Behavior 71 (2017) 196e208198 Prybutok, 2014) areas were also found positive effects of rich media on users' evaluations toward ICT platforms. 2.2. Social interactions Different levels of media richnessmainly depend on four factors, namely, multiple cues, immediacy of feedback, language diversity, and participation (Bertot et al., 2012; Click & Petit, 2010; Daft, Lengel, & Trevino, 1987). According to the previous studies (Brown et al., 2010; Weber & Kauffman, 2011), a significant rela- tionship exists between groups of social factors and the technology adoption intention. This finding suggests that the ICT adoption behavior of a member within a group is affected by others in the same group who have already adopted the technology; this phe- nomenon highlights the importance of “social interaction” (Ayanso et al., 2014; Hernandez, Montaner, Sese, & Urquizu, 2011). Furthermore, the Internet-based CMC interaction has a significant influence on various dimensions of daily life, including work, ed- ucation, family life, business modes, and so on. Social interactions are conceptualized based on the affective social and advice network relationships among individuals. In using social interaction media platforms (e.g., Facebook, Twitter, Wikis, YouTube), users' behavior are often affected by their workplace superiors, colleagues, or friends (i.e., social interaction). The quick adoption of relevant ICTs increases the media richness perception, which in turn, promotes communication and the further expansion of the social network. Social ties influence who can access new information and the speed by which they receive such information. When an individual finds that people around him/her use an interactive tool in communication environments, social influence may also serve as a strong social motivator to use ICT interactive tools. In this study, we propose that individuals may want to use an ICT interactive tool because their friends are using it. 2.3. Media technostress People prefer user-friendly technologies. Hence, technological complexity, overload and stress are major barriers that can affect ICT adoption behavior. Researchers (Ragu-Nathan, Tarafdar, Ragu- Nathan, & Tu, 2008; Shu, Tu, & Wang, 2011; Tarafdar, Tu, Ragu- Nathan, & Ragu-Nathan, 2007, 2010, 2015) have started to explore the various user, task and ICT characteristics that lead to ICT usage related stress, misuse and information overload. “Computer technostress” refers to the negative expected or actual process of individuals interacting with computers or ICT devices (Heinssen, Glass, & Knight, 1987; Shu et al., 2011). Tarafdar et al. (2007) developed a measurement scale to measure organizational envi- ronment and self-efficacy and evaluate the impact of technostress on the use of ICT media. Meanwhile, RagueNathan et al. (2008) focused on technostress amongst organizations resulting from the implementation of successive computer-based business processes that force employees to constantly adapt to new applications and workflows. Moreover, Tarafdar et al. (2010) demonstrated the negative effects of technostress on ICT usage, and established a negativeassociation between technostress and overall perfor- mance (Tarafdar et al., 2015). In other words, media technostress is a form of psychosomatic stress of adaptation caused by one's inability to cope with the new ICT devices (Ragu-Nathan et al., 2008). ICT media technostress can be viewed as a result of the negative consequences on one's attitudes, beliefs, and thoughts by the direct or indirect use of ICT, also be seen as an antecedent to one's adoption behavior. In this research, the definition of ICT media technostress emerges in the negative individual experience that it imposes on person, who are passive victims, trapped in user interface or software applications. 2.4. Information literacy and digital skill Through these theoretical findings on human behavior and the impact they have on an individual's willingness to adopt ICT, we can identify the effectiveness of using such technologies. Briefly, such communication consists of cues that convey information and use of appropriate skills through appropriate devices that promote ac- tivities and effective dialogues. Hence, users need to go through the information medium to obtain the required information (knowl- edge) and services, thus enabling them to convey messages using the AEIOU framework. According to the American Library Association (2000), information literacy, which is the basis of life- long learning in the Information Age, can be classified into two kinds: understanding information contents and competency in using ICT. Information literacy enables people to master informa- tion content and extend their investigations, recognize when in- formation is needed, and possess the ability to effectively locate, evaluate, and use the needed information. Information literacy also refers to one's capacity to use ICT as means to reach particular personal and professional goals. ICT use requires a distinct set of skills that allow one to effectively cope with a medium that is generally deprived of verbal cues and visual cues. For instance, before you can evaluate the results of a search query, you must be able to perform one, or before you can ask a question in an elec- tronic forum, you need to create a user account and register. Media- related skills consist of technology literacy, which include a basic command of ICT operational skills and the ability to manipulate the complex ICT communication applications. 3. Hypotheses development Most research on information literacy have not been conducted in regions that have a digital divide problem. However, these re- gions play an important role in solving the digital divide issue. Thus, it is necessary to identify the determinants influencing rural peo- ple's ICT adoption behavior. Accordingly, the current research at- tempts to integrate two theories, namely, “media richness theory (MRT) (Daft & Lengel, 1986)” and “information technology accep- tance” (e.g., Cognitive theory, Compeau & Higgins, 1995; Task- technology fit, Goodhue & Thompson, 1995; Technology Accep- tance Model, Davis, 1998; Theory of Planned Behavior, Hsieh, Rai,& Keil, 2008) as the theoretical foundation upon which to establish the research framework for explaining people's ICT adoption behavior. We also identify information literacy as an important moderating variable (information and digital skills), because it has been found to have a significant role in relation to these models. Many of the characteristics that have been ascribed to ICT are not innate physical characteristics; rather these are socially derived characteristics, such as social interaction and immediacy of feed- back. Task has long been recognized as an important factor influ- encing adoption behavior (Brown et al., 2010). Many studies in this field have focused on specific tasks or the complexity level of task characteristics based on the assumption that task characteristics play an important role as an antecedent variable of adoption behavior (Koo, Chung,&Nam, 2015; Zylka et al., 2015). When a task fits better with the ICT characteristics, such characteristics can have a stronger influence on a person's adoption behavior (Brown et al., 2010; Thatcher & Brown, 2010; Han, Hiltz, Fjermestad, & Wang, 2011; Aesaert & van Braak, 2014). When examining user interac- tion and experiences, the manner by which an individual uses the media characteristics is just as critical as the ICT medium chosen (Anandarajan et al., 2010; Giesbers et al., 2013). Regarding MRT, complex tasks lead to higher ICT adoption behavior (Daft et al., 1987; Aesaert et al., 2015; Brooks, 2015; Koo et al., 2015). Individual users determine the richness of ICT media T.-K. Yu et al. / Computers in Human Behavior 71 (2017) 196e208 199 platforms through their experiences and communication trans- actions using such channels (Rice, 1992; D'ambra, Rice, & O'connor, 1998; Dennis, Fuller, & Valacich, 2008). ICT media can further enhance the positive effect of media richness on overall conve- nience by making the information obtained more actionable in terms of task characteristics. In this situation, individuals are likely to use rich media for complex tasks and lean media for relatively simpler topics. Individuals with high willingness to try out a new ICT would consider the tasks delivered through ICT media as innovation, and their desire to try the media comes from their intention to experience the novelty of the ICT. The abovementioned information leads us to the following hypotheses: H1. Users' perceived task characteristics of ICT are positively related to the medium richness of ICT. H2. Users' perceived task characteristics of ICT are positively related to their ICT adoption behavior. Previous studies (Bagchi, Udo, Kirs,& Choden, 2015; Bruno et al., 2010; Gerpott & Ahmadi, 2015b; Giesbers et al., 2013; Tondeur, Sinnaeve, Van Houtte, & van Braak, 2010; Weber & Kauffman, 2011) on computer-mediated communication (CMC) have largely proven that, under the environment of rapidly expanding Internet infrastructures, information technology has provided a variety of rich communication channels, enabling people to explore re- lationships and knowledge through these modern formats at any time. The ICT media have several unique features for facilitating interpersonal communications: they provide nearly synchronous communication, users can simultaneously engage in multiple conversations on a one-to-one basis through separate windows, and they allow users to build communities of interest via the tools of social network (Anandarajan et al., 2010; Giesbers et al., 2013). Those technologies open the world to many people and facilitate communication through the exchange of short messages, which are mostly real-time status updates, as well as facilitate effective divi- sion and allocation of local and global knowledge and labor through the coordination of work via ICT. ICT platforms recognize the growth of many communities and offer tools that allow users within such communities to manage membership. Giving messages with personal focus enhances re- ceivers' involvement and the effectiveness of communication (Kietzmann, Hermkens, McCarthy, & Silvestre, 2011; Kishi, 2008). Moreover, the amount of information is another key construct of efficient communication richness. Generally, the appropriate amount of information guarantees communication effectiveness and richer media have better transmission efficiency. However, information overload can also lead to information inefficiency. For instance, the receivers may not be able to understand the right messages from the excessive information (Bons�on & Flores, 2011; Liu et al., 2009). According to the MRT hypothesis, rational and effective users choose media of appropriate richness for tasks that involve communication (D'ambra et al., 1998; Dennis et al., 2008), and if their choice of communicationmedia is restricted tomedia of lower richness due to accessibility constraints, social interaction and media experience may decrease in rural communities. Com- munitiesperceive a high level of media richness when they are familiar with either their communication partners or the commu- nication equipment. For these social interaction purposes, users often exchange messages and files with subtle meanings and want to reach mutual understanding (Anandarajan et al., 2010; Walther, 1996). In other words, if a user highly thinks of ICT media as communication channels, he or she will also consider these as so- cialization tools, that is, a certain level of ICT medium richness can facilitate social interaction with others. In this sense, perceived ICT media richness is an important factor in social interaction, which leads us to the following hypotheses: H3. Users' perceived ICT media richness is positively related to their ICT media experience. H4. Users' perceived ICTmedia richness is positively related to the level of their social interaction. When an individual first begins to use a new ICT medium, satisfaction often decreases because its use requires new skills and newmanners of interaction. In this sense, well-designed interactive ICT applications could generate higher satisfaction by providing greater control to users who wish to personalize their information search (Van Deursen et al., 2011). “General media experience” is frequently operationalized as the number of years an individual has been using an ICT device or the daily/weekly time spent using an ICT device. “Media experience,” which refers to the ability to use a specific type of ICT, can play a role in one's perceptions of ICT. In some instances, people also need specialized training and opera- tion experience to acquire skills necessary to use these ICTs. Greater control over the surfing experience increases the pleasure and convenience of using ICT, suggesting that people's ICT adoption behavior is positively related to the degree of characteristics of ICT media (De Wit et al., 2012). Research on ICT reported that people are often attracted to the use of ICT to satisfy their curiosity and imagination and to seek out novel experiences. For individuals who are more open to experiences, using ICT media platforms for the purposes of using a new widget and unearthing the world of in- formation via the Internet have substantial intrinsic interest (Venkatesh, Sykes, & Venkatraman, 2014). Van Deursen and Van Dijk (2011) identified frequency of ICT use as a significant predic- tor of individual digital information processing and usage behavior. According toMRT research, various task characteristics of ICT, an experienced user can potentially influence their adoption and use of ICT (Koo, Wati, & Jung, 2011). For comparable levels of media richness and use, the lower effort deployed by ICT media is likely to lead to greater satisfaction. Consequently, an individual's experi- ence with the specific ICT will improve and the ICT gradually be- comes easier to use. This is because, as an individual's perceptions of his/her ability to use ICT media increases, adoption behavior intention also increases. This leads us to the following hypothesis: H5. Users' media experience of the ICT is positively related to their ICT adoption behavior. Some studies (Bons�on & Flores, 2011; Click & Petit, 2010; Ghobadi & Ghobadi, 2015; Liu et al., 2009) have reported that so- cial interaction factors have significant and positive impact on ICT adoption behavior. Individuals' use of ICT and the development of their ICT knowledge and skills are socially embedded and depend, to some context, on feedback and interactionwith other individuals (e.g., peers and teachers) (Zylka et al., 2015). Further, evidence suggests that key members from an individual's social network may have a normative influence upon his/her ICT adoption behavior (Venkatesh & Brown, 2001; van Deursen, Courtois, & van Dijk, 2014). This leads us to the following hypothesis: H6. A users' level of social interaction is positively related to his/ her ICT adoption behavior. Most of the related studies have demonstrated a relationship between technostress and ICT behavior (Ayyagari, Grover, & Purvis, 2011; Maier, Laumer, Weinert, & Weitzel, 2015; Ragu-Nathan et al., 2008; Tarafdar et al., 2015; Tarafdar, Tu, Ragu-Nathan, & Ragu- Nathan, 2011, 2007). The impact of a strong positive belief system is likely to transfer to beliefs related to tasks associated with technology use. Increased technostress commonly leads to lower ICT adoption intention. In the DOC context, media technostress T.-K. Yu et al. / Computers in Human Behavior 71 (2017) 196e208200 describes the stress phenomenon resulting from the use of ICT for personal tasks, which can be attributed to characteristics of modern ICT, such as constant change. Given that ICT use is dependent on technology, it follows that higher levels of media technostress lead to lower levels of ICT usage. This leads us to the following hypothesis: H7. A medium level of technostress is negatively related to a user's ICT adoption behavior of ICT. In this study, we focus on the instrumental “technology” and “skill” that can be applied in finding answers to the questions at hand. In recent years, insufficiency of digital skills is a considerable barrier to one's ability to benefit from the use of ICT media (Bruno et al., 2010; van Deursen et al., 2014). The ICT digital skills refer to a set of basic skills in using ICTs. People with insufficient ICT digital skills might turn to self-directed learning, that is, by learning through trial and error at home and in the workplace. Other people can also turn to helpdesks and other sources. We conceptualized ICT digital skill as self-perceived ability in handling daily ICT media experiences competently. The current research focuses on the density of information literacy and digital skill level, because they have been found to affect users' ICT adoption behavior (Bertot, Jaeger, & Grimes, 2010; Click & Petit, 2010). Levy (2009) noted that people need to match their infor- mation technology capabilities with the information processing capacity of the technology devices they use, including those with new communication media. Zhong (2011) found that the suffi- ciency of ICT facilities determines how ICT is implemented into everyday life and can help people enhance their digital skills. As such, the relationship between media experience and ICT adoption behavior may be moderated by information and digital skills. ICT technostress arises from digital illiteracy and the lack of skills in ICT use. People in rural communities generally think that ICT media are only reserved for those with high educational background and often deny themselves access to computers. Therefore, we present the following hypotheses: M1a. The relationship between media experience and ICT adop- tion behavior is moderated by the level of digital skill, that is, the relationship is weaker under conditions of low digital skill and stronger under conditions of high digital skill. media richness task characteristics social interaction media experience H3 H1 H5 H Fig. 1. Conceptua M1b. The relationship between media technostress and ICT adoption behavior is moderated by the level of digital skill, that is, the relationship is weaker under conditions of low digital skill and stronger under conditions of high digital skill. M2a. The relationship between media experience and ICT adop- tion behavior is moderated by the level of information literacy, that is, the relationship is weaker under conditions of low information literacy and stronger under conditions of high information literacy. M2b. The relationship between media technostress and ICT adoption behavior is moderated by the level of information literacy, that is, the relationship is weaker under conditions of low infor- mation literacy and stronger under conditions of high information literacy. A conceptualized structural model, which demonstrates the moderating role of information and digital skill, is presented in Fig. 1. 4. Methods and descriptive statisticsThe current study is an attempt to construct a theoretical model by which to predict and explain community information literacy and Internet adoption behavior, as well as to test the model empirically. Specifically, prior studies (Gerpott & Ahmadi, 2015b; van Deursen et al., 2015) have reported that using scale measure- ment easily produces unsuitability because extant research have only been conducted in urban regions and developed countries. In order to fill this gap, the current study intends to develop an effi- cient measurement scale for the regions suffering from the digital divide, that is, domestic rural areas. Adopting the questionnaire survey method to better understand how information literacy and digital skills affect rural community residents' willingness to apply ICT in their daily lives. This study adopts a positivist research approach, contributing to the methodological pluralism that is necessary for the complete understanding of a phenomenon. Rigorous statistical testing was possible as the data were collected through a structured questionnaire. A pilot test was conducted before the final questionnaire was distributed to the subjects. To ensure the appropriateness of the research design, the validity and reliability of the itemswere also tested. A self-administered, closed- adoption behavior media technostress information literacy digital skill H2 H6 M2 a M1 a 4 H7 M2b M1b l framework. T.-K. Yu et al. / Computers in Human Behavior 71 (2017) 196e208 201 ended questionnaire with ordered choices was used to survey a sample of Taiwan digital opportunity center communities. 4.1. Measure development and validation New ICT adoption behavior is difficult to observe from an external perspective because of the nature of Internet behavior in relation to practice and information habitus to use the Internet (Schradie, 2011; Van Deursen et al., 2011). Furthermore, users are the best judges on whether behavior intention has changed. In other words, self-reporting is the best way to measure changes in actual behavior intention. Here, themeasurement of intentions was separated from the measurement of actual behavior intention as we tested the research model. This was done to minimize the possibility of the participants reconstructing history to present a consistent and logical picture. The questionnaire comprised previously published multi-item scales with favorable psychometric properties. In addition to the review of preexisting survey instruments from the literature, dur- ing the construction of the survey questionnaire from the pilot study, interviews were reviewed for context-specific details that warranted inclusion. Our scale development followed the recom- mendations of Straub, Boudreau, and Gefen (2004) as well as the standard psychometric scale development procedures of DeVellis (2012). The generation of constructs based on an extensive study of the prior literature in related fields, such as task characteristics, richness, experience, and adoption behavior, were adapted from measurement items validated in previous empirical studies (Brown et al., 2010; Carlson & Zmud, 1999; Koo et al., 2011; De Wit et al., 2012; Van Deursen et al., 2011; van Deursen et al., 2014). The scales for social interaction and media technostress were adapted from prior research (Anandarajan et al., 2010; Bons�on & Flores, 2011; Ragu-Nathan et al., 2008; Tarafdar et al., 2015) where possible. The constructs of information literacy and digital skill were measured using scales adapted from Koltay (2011), Goldhammer et al. (2013), and Bertot et al. (2010). The existing literature was first surveyed for items that were previously used for the constructs employed in the current research. We developed the first version of the questionnaire in English, after which the context of the questionnaire items was translated the measures into Chinese. The double-translation method was used to ensure conceptual equivalence and increased content val- idity. Three researchers, who are experts or have great interest in information technology adoption behavior and information liter- acy, reviewed the initial instrument. Each expert was provided with operational definitions for the eight substrata and the 62 initial items printed on separate index cards. The participants were asked to note ambiguity in item wording (and recommend changes if necessary), categorize individual items into eight substrata, and rank items within each substratum according to their semantic proximity to that of the underlying substratum. Moreover, the face and content validities of the instrument were verified based on the in-depth interviews with these professionals. On the basis of expert comments, we made minor adjustments to refine the questionnaire. Before conducting the final study, the pilot study questionnaire consisted of 50 questions divided into four major areas: (1) de- mographic profile of participants; (2) evaluation of ICT task char- acteristics, richness, experience; and technostress; (3) regarding their ICT adoption behavior; and (4) personal information literacy and digital skill. Participants were asked to rate 42 attributes on a 7- point scale (ranging from “strongly disagree” to “strongly agree”). To ensure the desired balance and randomness of the items in the questionnaire, five items were negated, and all items were randomly arranged to reduce the potential ceiling (or floor) effect that can induce monotonous responses in response to items designed to measure the same construct. Considerable effort was exerted to ensure that each statement in the final instrument captured the intended meaning of the construct under investiga- tion. A pilot test was later conducted with 32 adults residing in Yunlin County, Taiwan, to evaluate whether the revised question- naire was appropriate in terms of readability, ease of understand- ing, formatting, and layout. The focus of the pilot study was to examine the reliability and validity of the scales in the context of ICT media. Upon response from the pilot study, the internal con- sistencies of Cronbach's alphas ranging from 0.724 for information literacy to 0.912 for social interaction imply that the scales used in this study were satisfactory in terms of measuring the constructs of interest. 4.2. Sample and descriptive statistics Since 2005, the National Ministry of Education has invested resources in digital opportunity centers aimed at upgrading the ICT infrastructure and establishing life-long learning environments to narrow the Taiwan's digital divide. This study adopted a quantita- tive survey and face-to-face interviews to collect data. Participation in the study was completely voluntary, but was limited to those subjects who were above the age of 18 and had new ICT adoption experience via the Internet or mobile applications. After connecting the staffs of the digital opportunity centers, who helped explain the content of the questionnaire to the respondents, the duration of data collection took around 6 months. A total of 1500 question- naires were sent at the same time. We also designed three distinct acts that must be completed and two distinct information cues about the attributes of the task-related stimulus object all re- spondents have to process when performing a task characteristic. A total of 1225 questionnaires were returned, of which 350 were considered problematic because of excessive missing data, “don't know” or N/A answers, and response biases. These were excluded. Finally, a total of 875 valid samples, which were collected from a total of 160 digital opportunity centers of the Taiwan Ministry of Education, were collected. The valid response rate was 71.43%. Owing to privacy concerns, the digital opportunity center staffs did not share information about the participants who chose to com- plete the study or the date of response. Therefore, we were unable to calculate coefficient of response bias, using a comparison of early and late respondents or non-respondents.The sample demographic data in Table 1 indicate a diverse cross-section of population. Of the respondents, 41.6% were males and 58.4% were females. The average age of the sample population was 37.61 years old (standard deviation was 13.6 years old). Participant status in terms of formal education is varied: less than grade 11 (19.1%), high school/technical school certificate (41.6%), faculty degree (16.5%), bachelor's degree (19.4%), master's degree or above (3.0%), and missing data (0.5%). 4.3. Common methods bias When the dependent and independent variables are latent and measured by the same method, common method bias may occur (Podsakoff, MacKenzie, Lee, & Podsakoff, 2003). Owing to the na- ture of the data collection method, we tested for common method bias using Harman's one-factor test (Malhotra, Kim, & Patil, 2006; Podsakoff & Organ, 1986) on eight first-order latent variables in our research model. Our findings show that no single factor accounted for the majority of the covariance in the measures, implying that no common method bias occurred. Therefore, CMV was not considered a major concern in our data set. Table 1 Respondent profiles. Demographics Level Count Percentage Gender Male 364 41.6 Female 511 58.4 Formal education Less than grades 11 167 19.1 High school/technical school certificate 364 41.6 Faculty degree 144 16.5 Bachelor degree 170 19.4 Master degree or above 26 3.0 Missing value 4 0.5 T.-K. Yu et al. / Computers in Human Behavior 71 (2017) 196e208202 5. Results In the past, solving the moderating effect of research variables required adopting multiple regression analysis to understand the relationship of interaction effects between variables. However, this method can only handle a single dependent variable (Aiken&West, 1991; Baron & Kenny, 1986; Jaccard & Wan, 1995). Partial least squares (PLS) method is a linear model of exploration or con- struction of statistical methods in a general linear model, which includes moderating effects and multiple dependent variables. With the process of moderating effect test following previous works (Chin, Marcolin, & Newsted, 2003; Goodhue, Lewis, & Thompson, 2007), PLS method was used in the modeling and data analysis. The actual data analysis operations used the SmartPLS 2.0 software of Ringle, Wende, and Will (2005), which was developed to process measurement model and structural model analysis with bootstrapping method sampling analysis to estimate the parameters in the outer and inner model and t-values (Chin, 2010, pp. 655e690; Saad�e & Bahli, 2005). With 875 ques- tionnaires, the research model was over identified; thus, we applied nonparametric bootstrapping procedure with 2500 repli- cations to obtain the standard errors of the estimates. 5.1. Measurement model evaluation By using PLS analysis, the composite reliability (CR) and average variance extracted volume (AVE) can assess the reliability and validity of the structural model, respectively. Accordingly, this study followed Bagozzi and Yi (1988) recommendations and selected the three most commonly used indicators of future eval- uation indicators, which reflected themeasurement mode. The first indicator refers to the reliability of individual items (i.e., factor loading). This indicator is used to assess the measuring variables' factor loading of the latent variables and test the statistical signif- icance of every variable loading. In this study, all factor loadings Table 2 Construct reliability results. Construct Cronbach's Alpha Media richness 0.904 Media experience 0.837 Information technology adoption behavior 0.923 Task characteristics 0.908 Social interaction 0.919 Media technostress 0.805 Information literacy 0.833 Digital skill 0.788 Media experience � information literacy 0.913 Media technostress � information literacy 0.957 Media technostress � digital skill 0.921 Media experience � digital skill 0.947 Discriminant validity ¼ AVE/(Correlation)2 Where (Correlation)2 ¼ highest (Correlation)2 between factors of interest and remaining were higher than the recommended value of 0.6, which indicated significance; Furthermore, the sample factor loadings were be- tween 0.657 and 0.920, consistent with the recommended values of Hair, Black, Babin, and Anderson (2010). The second indicator refers to the composition of latent variables of reliability (CR). This value indicates that the higher the internal consistency reliability of the potential construct variable (the value of CR), the better it is. Fornell and Larcker (1981) recommended that the values should be equal to or greater than 0.6. In the current the CR values ranged from 0.851 to 0.962, thus confirming the reliability of the scales in this study. The third indicator refers to the AVE, which calculates the explanatory power of the latent variables to the measured vari- ables. If the AVE values are higher, then the potential variables have better discriminant validity and convergent validity. According to Fornell and Larcker (1981), the standard value must be greater than 0.5, and in the current study, AVE values ranged from 0.486 to 0.805. Convergent and discriminant validities were evaluated by calculating the AVE value for each factor within each model. Discriminant validity is shown if the square root of the AVE of a measure is larger than its correlation coefficients with other mea- sures, and the overall discriminant validity coefficient of the factors achieves satisfactory level discriminant validity. These three in- dicators suggest that the data support high internal consistency and discriminant validity. These results, presented in Table 2, confirm both the convergent and discriminant validities of the proposed research model (see Table 3). 5.2. Testing the moderating effects Prior to the analysis of themoderating effects, we first tested the assumptions of the measurement variable used in this study and did not find violations of the normality distribution assumption. We used PLS to test the main effect of media experience and media technostress and the moderating effects of information literacy and Composite Reliability AVE Discriminant validity 0.929 0.723 1.322 0.902 0.754 1.819 0.942 0.764 1.342 0.932 0.732 1.749 0.943 0.805 1.472 0.851 0.590 28.547 0.900 0.750 1.318 0.877 0.707 3.105 0.928 0.589 4.091 0.962 0.682 2.413 0.916 0.486 2.382 0.955 0.703 3.665 factors. Table 3 Baron and Kenny test of moderating effect. Latent Sample (n ¼ 875) Model 1 Model 2 Model 3 Task characteristics/ Media richness 0.720** 0.720** 0.720** Task characteristics/ Adoption behavior 0.149** 0.020 0.021 Media richness/ Media experience 0.693** 0.693** 0.693** Media experience/ Adoption behavior 0.142** 0.084** �0.005 Media richness/ Social interaction 0.796** 0.796** 0.796** Social interaction / Adoption behavior 0.143** 0.055* 0.059* Media technostress/ Adoption behavior 0.139** 0.068** 0.081* Moderator main effect Information literacy/ Adoption behavior 0.654** 0.644** Digital skill/ Adoption behavior 0.079** �0.142 Moderator interaction effect Media experience � Information literacy/ Adoption behavior �0.054* Media experience � Digital skill/ Adoption behavior 0.286* Media technostress � Information literacy/ Adoption behavior �0.031 Media technostress � Digital skill/ Adoption behavior �0.045 Adoption behavior R2 0.179 0.595 0.600 △R2 0.416 0.005 T.-K. Yu et al. / Computers in Human Behavior 71 (2017) 196e208 203 digital skills on information technology adoption behavior (Figs. 2 and 3, respectively). For example, to test the moderating effect, media experience (predictor) and information literacy (moderator) were multiplied to create an interaction construct (media experience� information literacy) to predict ICT adoption behavior. All variables were standardized so that each variable was centered at zero and had a standardized score, thus avoiding the potential multi-collinearity problem and making it easier to interpret the results. Following Baron and Kenny (1986), the empirical studydeter- mined the moderating roles of information literacy and digital skill according to the significance of the interaction terms in Model 3. Among the four hypothesized moderating effects, M1b and M2b were not significant, that is, information literacy and digital skill did not havemediating effects betweenmedia technostress and ICT adoption behavior. Interestingly, all of the significant interaction effects were related to media experience. Information literacy negativelymoderated bothmedia experience effect on ICTadoption behavior, on the contrary, digital skill was a positive coefficient (M1a: beta ¼ �0.054, p < 0.05; M2a: beta ¼ 0.286, p < 0.05). These media experience R2=0.481 task characteristics media richness R2=0.518 social interaction R2=0.633 0.6 93 ** 0.796** 0.720** * p<0.05 ** p<0.01 Fig. 2. Path coefficients for the research mo additional analyses provide support for the moderation pattern shown in our model. Meanwhile, Fig. 4 presents the full results of the moderation analysis, including the structural path estimates and explained variances. Contrary to Hypothesis M1a, media experience and information literacy had an opposite direction interaction effect on ICT adoption behavior. Consistent with Hy- pothesis M2a, media experience and digital skill had a positive interaction effect on ICT adoption behavior. Specifically, we found that the effect of media experience on ICT adoption behavior increased with a decrease in information literacy. Our results apparently provide further support for the self-perception experi- ence, that is, the effect of experience on adoption behavior increased as information literacy decreased. However, the effects that media experience had on adoption behavior increased as digital skill increased. It is interesting to the analyzed result in model 2; when indi- vidual ICT media experience is getting richness, the information literacy is also enhanced. Moreover, some important changes occur in model 3 shifted from model 2. When the interaction effects generate by the moderator information literacy, the effect of path media technostress adoption behavior R2=0.179 0.1 43* * 0.149** 0.142** 0.1 39 ** del (excluding moderator main effect). media technostress media experience R2=0.481 task characteristics adoption behavior R2=0.595 media richness R2=0.518 social interaction R2=0.633 0.0 55* 0.020 0.084**0.6 93 ** 0.796** 0.720** 0.0 68 ** digital skill information literacy 0. 456 ** 0. 970 ** * p<0.05 ** p<0.01 Fig. 3. Path coefficients for the research model (including moderator main effect). media technostress media experience R2=0.481 task characteristics adoption behavior R2=0.600 media richness R2=0.518 social interaction R2=0.633 0.0 59* 0.021 -0.0050.6 93 ** 0.796** 0.720** 0.0 81 * digital skill information literacy 0. 64 4* * -0 . 241 media technostress * information literacy media technostress * digital skill media experience *information literacy media experience *digital skill -0.045 -0.031 -0. 05 4* 0.2 86* * p<0.05 ** p<0.01 Fig. 4. Path coefficients for the research model (including interaction effect). T.-K. Yu et al. / Computers in Human Behavior 71 (2017) 196e208204 coefficient on media experience to adoption behavior becomes insignificant and from positive to negative. The major reason is that when individual information literacy is higher than the threshold value of media experience, individual may create and evaluate in- formation and data misfits in an ethical manner yield IT-included overload and stress. In this study, the R2 and Cohen's effect size of the coefficient of statistical power as indicators of the original research model R2 was 0.600. The sequentially eliminated dependency levels of the media, task characteristics, social interaction, and media technostress related to model R2 were 0.594, 0.597, 0.595, and 0.590, respec- tively. Using the Cohen (1988) effect size criteria (small: 0.02, me- dium: 0.15 large: 0.35), we found that the factor effect size was between 0.011 and 0.029, indicating that the moderator effects were very small. 5.3. Testing the structural model Recently, Tenenhaus, Vinzi, Chatelin, and Lauro (2005) have suggested goodness-of-fit (GoF) as a good indicator for measuring the structural model fitness for PLS pathmodeling. For this research model, we obtained a GoF value of 0.652, which exceeded the baseline value of 0.36 for large effect sizes of R2. Thus, with acceptable model fit, our measures were considered to be appro- priate for subsequent tests of the causal model and the research hypotheses. By using the PLS to estimate the path relationship between every research construct, among total seven path re- lationships, we found five assumptions reached the significance level, a ¼ 0.05. The structural model path analysis coefficients for infusing technology information into community residents' behavior were as follows: the Task characteristics / Media rich- ness (0.720); Media richness / Media experience (0.693); Media richness / Social interaction (0.796); Social interaction/ adop- tion behavior (0.059); andMedia technostress/ adoption behavior (0.081). Task characteristics predicted media richness, thus sup- porting H2. Moreover, media richness had two main effects on media experience and social interaction, thus supporting H3 and H4. The results related to the prediction of ICT adoption behavior T.-K. Yu et al. / Computers in Human Behavior 71 (2017) 196e208 205 were consistent with the social interaction and media technostress hypotheses that were adapted to this context, thus supporting H6 and H7. All estimated standardize path coefficients (significant paths indicated with an asterisk) are shown in Fig. 4. The empirical result of interacting effect path relationship be- tween the moderators showed that information literacy and digital skill adjusted the level of medium experience to ICT adoption behavior; as for the variance within variables explanatory power. Fig. 4 also presents the results of this study with explanatory powers. The model explains a substantial portion of the variance in all the endogenous variables: “media richness” for 51.8%, “social interaction” for 63.3%, “media experience” for 48.1%, and adoption behavior for 60.0% (see Fig. 4). We also found that the proposed model explained a significant amount of variation in the endoge- nous variables (i.e., more than 35% on average). Endogenous vari- ables showed good explanatory power of variation, which indicated a construction of model robustness and stability. 6. Discussions This study showed that ICT media facilitated information ICT adoption behavior, and that the proposedmodel can lead to a better theoretical framework by which to understand the intermediation or disintermediation process taking place among the community residents (Venkatesh & Sykes, 2013; Venkatesh et al., 2014). The nature of MRT was theoretically developed and empirically vali- dated, and this research identified five key constructs of electronic information intermediation: task characteristics, media richness, media experience, social interaction, and ICT adoption behavior. Empirical testing in the rural area community ICT adoption behavior context showed that these were key dimensions of ICT media acceptance behavior. On the other hand, the project of reducing the digital divide in rural villages can be seen as a suc- cessful project, demonstrating how ICT can be used to support underprivileged children and the elderly in these areas. Our find- ings reveal that capturing local knowledge is important but not sufficient; we extended this work to the context of Taiwan and find the often studied variables to be fairly predictive even in the new context. There are two ways by which interaction can be seen as a very important characteristic for CMC: ICT serves as a key interactive medium that supports two-way communication and provides feedback rapidly and efficientlywithin a global village. Many ICT devices are designed primarily to facilitate conversations among individuals and groups, thus ensuring ubiquitous connectivity among regions through ICT despite variations in political and cul- tural contexts. Thus, early experiences and social interaction are driven strongly by medium richness, which is crucial in fostering the success of digital divide initiatives. Hence, borders for communication are lessened. ICT devices give people the ability to access information quickly and to communicate with others. Consistent with the prior litera- ture, a higher level of ICT media richness promotes the adoption of ICT and more frequent use (Brown et al., 2010; Bertot et al., 2010, 2012,; Aesaert et al., 2015), and this variable also leads to higher social interaction levels (Dennis et al., 2008; Anandarajan et al., 2010; Van Deursen et al., 2011). In this sense, the important of ICT richness should be emphasized, because media richness is positively associated with media experience and social interaction. In addition, more frequent usage can improve richness of the communicative medium, because media richness also influences individuals' media experience, prompting easier integration into the world of ICT. Consequently, the community behavior of resi- dents is affected by the medium richness and experience of ICT, but the nature of task characteristics can only be directly influenced by media communicative richness. According to Ghobadi and Ghobadi (2015), ICT can lead to hu- man interactions that often result in virtual world participation. In this study, we focus on social interaction because people relate specifically to sharing information and providing support to other members of the community to achieve similar goals. The ICT adoption behavior of peers is influenced by intrinsic motivation; for instance, others who are already adopting ICT can encourage other members to further socialize, gain feedback, and initiate discussion, thus highlighting the importance of “social interaction” in enhancing ICT adopting behavior (Ayanso et al., 2014; Hernandez et al., 2011). Additionally, increased civil social interaction has been identified as a positive determinant of ICT adoption (Brown et al., 2010; Click & Petit, 2010; van Deursen et al., 2014). This study goes beyond established notions of social interaction, which seems highly valuable in the context of ICT-related adoption behavior research. In this study, from the media perspective, the task characteris- tics and ICT media experience have no direct effect on the ICT adoption behavior of residents in rural communities. Such an insignificant relationship failed to support MRT (Daft & Lengel, 1986). Contrary to the prior studies based on task characteristics (Brown et al., 2010; Van Deursen et al., 2011; Purdy&Nye, 2000; Xu et al., 2010), our results demonstrated that ICT adoption behavior had no relationship with task characteristics. The users are more likely to utilize ICT media features when they are better able to do so for their communicative tasks. With complex tasks, more frequent communication occurs; furthermore, the more diverse the ICT media contents, the greater the effect of content on ICT adop- tion behavior. Using the ICT device to form friendships can be a positive means of social interaction. Another explanation is that ICT has been considered to be an efficient communication equipment, a part of routine life; thus, users do not deliberate on ICT usage to accomplish certain tasks. People can nowcommunicatewith others around the world as quickly as possible through ICT media. Such synchronous communication allows participants to engage in real- time conversations even when they are in different locations. Sur- prisingly, this finding indicates that ICT media are chosen by resi- dents when they face a complexity task. Meanwhile, we identify media technostress as a means to in- crease ICT adoption behavior, contrary to the prior literature (Ayyagari et al., 2011; Tarafdar et al., 2011, 2015, 2010). We further identify “information literacy” and “digital skills” as potential means to reduce media technostress. The individual's ability to use ICT with ease is positively associated with their felicity in accom- plishing task and usage behavior. Consistent with Shu et al. (2011), we find that higher dependence on ICT for the completion of routine tasks and lower levels of digital skill are associated with higher levels of media technostress. People who hold self- efficacious digital skills can adapt ICT to challenges more readily and are more persistent in their pursuit of individual's task. In addition to extending knowledge in ICT adoption behavior among rural residents, our study is one of the first to provide a comprehensivemodel explaining themoderator effect of individual differences on information literacy and digital skills on ICT use. We theorized and found support for our model, which demonstrated significant interaction effects of information literacy and digital skills on media experience. For the perspective of individual digital skills, individual media technostress can significantly affect one's subsequent adoption behavior. Owing to the general lack of levels of information literacy and digital skills combined with media ex- periences, people in rural communities often cannot properly assess the benefits of ICT initiatives. Moreover, our research model demonstrates the moderating relationships among media experience, technostress, and ICT T.-K. Yu et al. / Computers in Human Behavior 71 (2017) 196e208206 adoption behavior, thus providing insights that could drive future research about the IT artifact and factors that influence user ICT adoption. However, information literacy and digital skills would have moderating effects between the degree of media experience and the community adoption behavior, which also transform the path of task characteristics and social interaction to community adoption behavior from a significant effect to an insignificant. Particularly, when adding interaction effect into the path, the main effect of individual digital skills to ICT adoption behavior would become insignificant, indicating that information literacy is the mainmoderating effect. In order to clarify the moderating effects of information literacy, we divided the data into three groups: for the low degree of information literacy, their digital skills moderate the degree of media experience and community adoption behavior; individual digital skills can also influence ICT community adoption behavior; for the medium level of information literacy, digital skills would moderate the path of media technostress and community adoption behavior, and moreover, the moderator of information literacy can affect the path relationship of media experience to community behavior into a significant relationship. Our findings revealed that: (1) low information literacy and high media experience affect ICT adoption behavior negatively, and (2) low information literacy and highmedia technostress has a positive influence on ICT adoption behavior. When a digital divide exists, it is important to keep on investing in information literacy develop- ment activities for rural communities to help them develop their ICT competence, although these training initiatives are sensitive to media experience measures. Relevant in this context is the plea of van Deursen et al. (2014) who argued that ICT literacy should reflect the level of ICT competencies of the individuals involved. This finding not only supports the other studies, which argue that in- formation literacy is an important factor in new ICT adoption and increased ICT usage, but also shows that media experiences can be effective tools in enhancing and reinforcing ICT adoption behavior. Finally, further research should focus on the two variables introduced in this study (“information literacy” and “digital skills”), which are measured as self-reported data. Such anindirect mea- surement of ICT literacy or digital competency provides relatively weak self-evaluation of such abilities (Mabe &West, 1982). For this reason, future research on individual ICT literacy and competency should try to measure these abilities in a direct and authentic way, although we attempt to measure individual ICT competences in a direct way by using complete distinct tasks in our questionnaires. Future researchers can combine these two direct and indirect measurement instruments in their new research designs. 7. Conclusion This study extends the literature by identifying the “MRT,” “in- formation literacy,” “digital skill” and “ICT adoption behavior” literature, and proposed a moderating model of the ICT adoption behavior of residents in a rural village in Taiwan by using structural equation model to test an empirical model. The present study provides evidence that adoption of ICT devices behavior and the relationship between ICT device richness, experience and perceived ICT media technostress. It provides improvements to the mea- surement of digital skill and advances understanding of informa- tion literacy mitigation by considering the role of media experience and adoption behavior framing. From a practical perspective, un- derstanding what drives or deters rural community's ICT adoption behavior can better inform decisions made by information literacy educators and e-Taiwan policy managers. We hopeful that the conceptualization model presented in this study serves as an acti- vator for researches on ICTs and digital divide, and serves a guide and a call to attract more researches in this area. Funding The author thank the Ministry of Science and Technology of Taiwan for financially supporting this research under contract MOST: 103-2511-S-507 -001 -MY2. Conflict of interest The authors whose names are listed immediately below certify that they have NO affiliations with or involvement in any organi- zation or entity with any financial interest (such as honoraria; educational grants; participation in speakers' bureaus; member- ship, employment, consultancies, stock ownership, or other equity interest; and expert testimony or patent-licensing arrangements), or non-financial interest (such as personal or professional re- lationships, affiliations, knowledge or beliefs) in the subject matter or materials discussed in this manuscript. References Aesaert, K., & van Braak, J. (2014). Exploring factors related to primary school pupils' ICT self-efficacy: A multilevel approach. Computers in Human Behavior, 41, 327e341. Aesaert, K., van Braak, J., Van Nijlen, D., & Vanderlinde, R. (2015). Primary school pupils' ICT competences: Extensive model and scale development. Computers & Education, 81, 326e344. Aiken, L. A., & West, S. G. (1991). Multiple regression: Testing and interpreting in- teractions. California: Sage. Anandarajan, M., Zaman, M., Dai, Q., & Arinze, B. (2010). Generation Y adoption of instant messaging: An examination of the impact of social usefulness and media richness on use richness. Professional Communication. IEEE Transactions on, 53(2), 132e143. Association of College and Research Libraries. (2000). Information literacy compe- tency standards for higher education. Chicago, IL: The Association of College and Research Libraries. Retrieved October 16, 2015, from: http://www.ala.org/ala/ mgrps/divs/acrl/standards/standards.pdf. Ayanso, A., Cho, D. I., & Lertwachara, K. (2014). Information and communications technology development and the digital divide: A global and regional assess- ment. Information Technology for Development, 20(1), 60e77. Ayyagari, R., Grover, V., & Purvis, R. (2011). Technostress: Technological antecedents and implications. MIS quarterly, 35(4), 831e858. Badger, J. M., Kaminsky, S. E., & Behrend, T. S. (2014). Media richness and infor- mation acquisition in internet recruitment. Journal of Managerial Psychology, 29(7), 866e883. Baehr, C. (2012). Incorporating user appropriation, media richness, and collabora- tive knowledge sharing into blended e-learning training tutorial. IEEE Trans- actions on Professional Communication, 55(2), 175e184. Bagchi, K. K., Udo, G. J., Kirs, P. J., & Choden, K. (2015). Internet use and human values: Analyses of developing and developed countries. Computers in Human Behavior, 50, 76e90. Bagozzi, R. P., & Yi, Y. (1988). On the evaluation for structural equation models. Journal of the Academy of Marketing Science, 16, 74e94. Baron, R. M., & Kenny, D. A. (1986). The moderator-mediator variable distinction in social psychological research: Conceptual, strategic and statistical consider- ations. Journal of Personality and Social Psychology, 51(6), 1173e1182. Bertot, J. C., Jaeger, P. T., & Grimes, J. M. (2010). Using ICTs to create a culture of transparency: E-government and social media as openness and anti-corruption tools for societies. Government Information Quarterly, 27, 264e271. Bertot, J. C., Jaeger, P. T., & Hansen, D. (2012). The impact of polices on government social media usage: Issues, challenges, and recommendations. Government In- formation Quarterly, 29, 30e40. Bons�on, E., & Flores, F. (2011). Social media and corporate dialogue: The response of the global financial institutions. Online Information Review, 35(1), 34e49. Brooks, S. (2015). Does personal social media usage affect efficiency and well-be- ing? Computers in Human Behavior, 46, 26e37. Brown, S. A., Dennis, A. R., & Venkatesh, V. (2010). Predicting collaboration tech- nology use: Integrating technology adoption and collaboration research. Journal of Management Information Systems, 27(2), 9e54. Bruno, G., Esposito, E., Genovese, A., & Gwebu, K. L. (2010). A critical analysis of current indexes for digital divide measurement. The Information Society, 27(1), 16e28. Burns, T. W., O'Connor, D. J., & Stocklmayer, S. M. (2003). Science communication: A contemporary definition. Public Understanding of Science, 12(2), 183e202. Carlson, J. R., & Zmud, R. W. (1999). Channel expansion theory and the experiential nature of medium richness perceptions. Academy of Management Journal, 42(2), 153e170. Chin, W. W. (2010). How to write up and report PLS analyses. In Handbook of partial least squares. Springer Berlin Heidelberg. Chin, W. W., Marcolin, B. L., & Newsted, P. R. (2003). A partial least squares latent http://refhub.elsevier.com/S0747-5632(17)30078-X/sref1 http://refhub.elsevier.com/S0747-5632(17)30078-X/sref1 http://refhub.elsevier.com/S0747-5632(17)30078-X/sref1 http://refhub.elsevier.com/S0747-5632(17)30078-X/sref1 http://refhub.elsevier.com/S0747-5632(17)30078-X/sref2 http://refhub.elsevier.com/S0747-5632(17)30078-X/sref2 http://refhub.elsevier.com/S0747-5632(17)30078-X/sref2 http://refhub.elsevier.com/S0747-5632(17)30078-X/sref2 http://refhub.elsevier.com/S0747-5632(17)30078-X/sref3 http://refhub.elsevier.com/S0747-5632(17)30078-X/sref3 http://refhub.elsevier.com/S0747-5632(17)30078-X/sref4 http://refhub.elsevier.com/S0747-5632(17)30078-X/sref4 http://refhub.elsevier.com/S0747-5632(17)30078-X/sref4 http://refhub.elsevier.com/S0747-5632(17)30078-X/sref4 http://refhub.elsevier.com/S0747-5632(17)30078-X/sref4 http://www.ala.org/ala/mgrps/divs/acrl/standards/standards.pdf http://www.ala.org/ala/mgrps/divs/acrl/standards/standards.pdf http://refhub.elsevier.com/S0747-5632(17)30078-X/sref6 http://refhub.elsevier.com/S0747-5632(17)30078-X/sref6 http://refhub.elsevier.com/S0747-5632(17)30078-X/sref6 http://refhub.elsevier.com/S0747-5632(17)30078-X/sref6 http://refhub.elsevier.com/S0747-5632(17)30078-X/sref7 http://refhub.elsevier.com/S0747-5632(17)30078-X/sref7 http://refhub.elsevier.com/S0747-5632(17)30078-X/sref7 http://refhub.elsevier.com/S0747-5632(17)30078-X/sref8 http://refhub.elsevier.com/S0747-5632(17)30078-X/sref8 http://refhub.elsevier.com/S0747-5632(17)30078-X/sref8 http://refhub.elsevier.com/S0747-5632(17)30078-X/sref8 http://refhub.elsevier.com/S0747-5632(17)30078-X/sref9 http://refhub.elsevier.com/S0747-5632(17)30078-X/sref9http://refhub.elsevier.com/S0747-5632(17)30078-X/sref9 http://refhub.elsevier.com/S0747-5632(17)30078-X/sref9 http://refhub.elsevier.com/S0747-5632(17)30078-X/sref10 http://refhub.elsevier.com/S0747-5632(17)30078-X/sref10 http://refhub.elsevier.com/S0747-5632(17)30078-X/sref10 http://refhub.elsevier.com/S0747-5632(17)30078-X/sref10 http://refhub.elsevier.com/S0747-5632(17)30078-X/sref11 http://refhub.elsevier.com/S0747-5632(17)30078-X/sref11 http://refhub.elsevier.com/S0747-5632(17)30078-X/sref11 http://refhub.elsevier.com/S0747-5632(17)30078-X/sref12 http://refhub.elsevier.com/S0747-5632(17)30078-X/sref12 http://refhub.elsevier.com/S0747-5632(17)30078-X/sref12 http://refhub.elsevier.com/S0747-5632(17)30078-X/sref12 http://refhub.elsevier.com/S0747-5632(17)30078-X/sref14 http://refhub.elsevier.com/S0747-5632(17)30078-X/sref14 http://refhub.elsevier.com/S0747-5632(17)30078-X/sref14 http://refhub.elsevier.com/S0747-5632(17)30078-X/sref14 http://refhub.elsevier.com/S0747-5632(17)30078-X/sref15 http://refhub.elsevier.com/S0747-5632(17)30078-X/sref15 http://refhub.elsevier.com/S0747-5632(17)30078-X/sref15 http://refhub.elsevier.com/S0747-5632(17)30078-X/sref15 http://refhub.elsevier.com/S0747-5632(17)30078-X/sref16 http://refhub.elsevier.com/S0747-5632(17)30078-X/sref16 http://refhub.elsevier.com/S0747-5632(17)30078-X/sref16 http://refhub.elsevier.com/S0747-5632(17)30078-X/sref16 http://refhub.elsevier.com/S0747-5632(17)30078-X/sref17 http://refhub.elsevier.com/S0747-5632(17)30078-X/sref17 http://refhub.elsevier.com/S0747-5632(17)30078-X/sref17 http://refhub.elsevier.com/S0747-5632(17)30078-X/sref18 http://refhub.elsevier.com/S0747-5632(17)30078-X/sref18 http://refhub.elsevier.com/S0747-5632(17)30078-X/sref18 http://refhub.elsevier.com/S0747-5632(17)30078-X/sref18 http://refhub.elsevier.com/S0747-5632(17)30078-X/sref19 http://refhub.elsevier.com/S0747-5632(17)30078-X/sref19 http://refhub.elsevier.com/S0747-5632(17)30078-X/sref19 http://refhub.elsevier.com/S0747-5632(17)30078-X/sref19 http://refhub.elsevier.com/S0747-5632(17)30078-X/sref20 http://refhub.elsevier.com/S0747-5632(17)30078-X/sref20 http://refhub.elsevier.com/S0747-5632(17)30078-X/sref20 http://refhub.elsevier.com/S0747-5632(17)30078-X/sref21 http://refhub.elsevier.com/S0747-5632(17)30078-X/sref21 http://refhub.elsevier.com/S0747-5632(17)30078-X/sref21 http://refhub.elsevier.com/S0747-5632(17)30078-X/sref21 http://refhub.elsevier.com/S0747-5632(17)30078-X/sref22 http://refhub.elsevier.com/S0747-5632(17)30078-X/sref22 http://refhub.elsevier.com/S0747-5632(17)30078-X/sref23 T.-K. Yu et al. / Computers in Human Behavior 71 (2017) 196e208 207 variable modeling approach for measuring interaction effects: Results from a monte carlo simulation study and an electronic-mail emotion/adoption study. Information Systems Research, 14(2), 189e217. Click, A., & Petit, J. (2010). Social networking and Web 2.0 in information literacy. The International Information & Library Review, 42(2), 137e142. Cohen, J. (1998). Statistical power analysis for the behavioral science. NJ: Lawrence Erlbaum Associates. Compeau, D. R., & Higgins, C. A. (1995). Computer self-efficacy: Development of a measure and initial test. MIS Quarterly, 19(2), 189e211. D'ambra, J., Rice, R. E., & O'connor, M. (1998). Computer-mediated communication and media preference: An investigation of the dimensionality of perceived task equivocality and media richness. Behaviour & Information Technology, 17(3), 164e174. Daft, R. L., & Lengel, R. H. (1986). Organizational information requirements, medium richness and structural design. Management Science, 32(5), 554e571. Daft, R. L., Lengel, R. H., & Trevino, L. K. (1987). Message equivocality, media se- lection, and manager performance: Implications for information systems. MIS Quarterly, 11(3), 355e366. Davis, D. R. (1998). Technology, unemployment, and relative wages in a global economy. European Economic Review, 42(9), 1613e1633. De Wit, K., Heerwegh, D., & Verhoeven, J. C. (2012). Changes in the basic ICT skills of freshmen between 2005 and 2009: Who's catching up and who's still behind? Education and Information Technologies, 17(2), 205e231. Dennis, A. R., Fuller, R. M., & Valacich, J. S. (2008). Media, tasks, and communication processes: A theory of media synchronicity. MIS Quarterly, 32(3), 575e600. Dennis, A. R., & Kinney, S. T. (1998). Testing media richness theory in the new media: The effects of cue, feedback, and task equivocality. Information System Research, 9(3), 256e274. van Deursen, A. J., Courtois, C., & van Dijk, J. A. (2014). Internet skills, sources of support, and benefiting from Internet use. International Journal of Human- Computer Interaction, 30(4), 278e290. van Deursen, A. J., & van Dijk, J. A. (2015). Internet skill levels increase, but gaps widen: A longitudinal cross-sectional analysis (2010e2013) among the Dutch population. Information, Communication & Society, 18(7), 782e797. van Deursen, A. J., van Dijk, J. A., & Peter, M. (2015). Increasing inequalities in what we do online: A longitudinal cross sectional analysis of internet activities among the Dutch population (2010 to 2013) over gender, age, education, and income. Telematics and Informatics, 32(2), 259e272. DeVellis, R. F. (2012). Scale development: Theory and applications. 3rd Sage Publications. Fornell, C., & Larcker, D. F. (1981). Evaluating structural equation models with unobservableand measurement errors. Journal of Marketing Research, 18, 39e50. Friemel, T. N. (2016). The digital divide has grown old: Determinants of a digital divide among seniors. New Media & Society, 18(2), 313e331. Gerpott, T. J., & Ahmadi, N. (2015a). Advancement of indices assessing a nation's telecommunications development status: A PLS structural equation analysis of over 100 countries. Telecommunications Policy, 39(2), 93e111. Gerpott, T. J., & Ahmadi, N. (2015b). Composite indices for the evaluation of a country's information technology development level: Extensions of the IDI of the ITU. Technological Forecasting and Social Change, 98, 174e185. Ghobadi, S., & Ghobadi, Z. (2015). How access gaps interact and shape digital divide: A cognitive investigation. Behaviour & Information Technology, 34(4), 330e340. Giesbers, B., Rienties, B., Tempelaar, D., & Gijselaers, W. (2013). Investigating the relations between motivation, tool use, participation, and performance in an e- learning course using web-videoconferencing. Computers in Human Behavior, 29(1), 285e292. Goldhammer, F., Naumann, J., & Keßel, Y. (2013). Assessing individual differences in basic computer skills: Psychometric characteristics of an interactive perfor- mance measure. European Journal of Psychological Assessment, 29(4), 263e275. Goodhue, D., Lewis, W., & Thompson, R. (2007). Statistical power in analysis interaction effects: Questioning the advantage of PLS with product indicators. Information Systems Research, 18(2), 211e227. Goodhue, D. L., & Thompson, R. L. (1995). Task-technology fit and individual per- formance. MIS Quarterly, 19(2), 213e236. Hair, F., Jr., Black, W. C., Babin, B. J., & Anderson, R. E. (2010). Multivariate data analysis: A global perspective (7th ed.). New York: Macmillan. Han, H. J., Hiltz, S. R., Fjermestad, J., & Wang, Y. (2011). Does medium matter? A comparison of initial meeting modes for virtual teams. Professional Communi- cation, IEEE Transactions on, 54(4), 376e391. Heinssen, R. K., Glass, C. R., & Knight, L. A. (1987). Assessing medium technostress: Development and validation of the medium technostress rating scale. Com- puters in Human Behavior, 3, 49e59. Hernandez, B., Montaner, T., Sese, F. J., & Urquizu, P. (2011). The role of social mo- tivations in e-learning: How do they affect usage and success of ICT interactive tools? Computers in Human Behavior, 27(6), 2224e2232. Hsieh, J. J. P.-A., Rai, A., & Keil, M. (2008). Understanding digital inequality: Comparing continued use behavioral models of the socio-economically advantaged and disadvantaged. MIS
Compartilhar