Despite the widespread use of mobile government (m-government) services in developed countries, the adoption and acceptance of m-government services among citizens in developing countries is relatively low. The purpose of this study is to explore the most critical determinants of acceptance and use of m-government services in a developing country context.
The unified theory of acceptance and use of technology (UTAUT) extended with perceived mobility and mobile communication services (MCS) was used as the theoretical framework. Data was collected from 216 m-government users across Bangladesh and analyzed in two stages. First, structural equation modeling (SEM) was used to identify significant determinants affecting users' acceptance of m-government services. In the second stage, a neural network model was used to validate SEM results and determine the relative importance of the determinants of acceptance of m-government services.
The results show that facilitating conditions and performance expectancy are the two important precedents of behavioral intention to use m-government services, and performance expectancy mediates the relationship between MCS, mobility and the intention to use m-government services.
Academically, this study extended and validated the underlying concept of UTAUT to capture the adoption behavior of individuals in a different cultural context. In particular, MCS might be the most critical antecedent towards mobile application studies. From a practical perspective, this study may provide valuable guidelines to government policymakers and system developers towards the development and effective implementation of m-government systems.
This study has contributed to the existing, but limited, literature on m-government service adoption in the context of a developing country. The predictive modeling approach is an innovative approach in the field of technology adoption.
Talukder, S., Chiong, R., Dhakal, S., Sorwar, G. and Bao, Y. (2019), "A two-stage structural equation modeling-neural network approach for understanding and predicting the determinants of m-government service adoption", Journal of Systems and Information Technology, Vol. 21 No. 4, pp. 419-438. https://doi.org/10.1108/JSIT-10-2017-0096Download as .RIS
Emerald Publishing Limited
Copyright © 2019, Emerald Publishing Limited
Mobile technologies are being widely adopted by governments around the world for the delivery of various services to citizens, employees, businesses, and other organizations; this concept is generally known as mobile government (m-government) (Ahmad and Khalid, 2017). The widespread use of m-government services is not surprising, given that approximately seven billion people (97 per cent of the global population) currently live under the mobile-cellular network coverage (ITU, 2016). Globally, the total mobile-broadband penetration exceeds 3.6 billion (47 per cent of the global population), more than the coverage of PC-based internet (3.2 billion) (ITU, 2016). In this context, there is a growing need to understand the effectiveness of the acceptance of services provided through m-government (Rana and Dwivedi, 2015).
Despite the growing mobile penetration, improved design and usability, and increased government support, the adoption of m-government services in developing countries has not yet achieved its objectives, and studies have shown similar e-government projects have either failed or experienced a slow adoption process (Gao et al., 2014; Alshibly and Chiong, 2015; Sharma et al., 2018). For instance, in the context of developing countries, only 15 per cent of e-government projects were successfully implemented, 50 per cent were partially completed, and 35 per cent of e-government projects failed to be implemented (Napitupulu and Sensuse, 2014). Therefore, a key challenge for m-government systems, to date, has not necessarily been their design, but their utilization (Almarashdeh and Alsmadi, 2017). However, the utilization or adoption behaviour of users of m-government systems has not been adequately addressed in the relevant literature, and there is room for more research in this area (Saxena, 2018; Sharma et al., 2018).
Although some recent studies have explored the adoption of m-government services (Ahmad and Khalid, 2017; Gao et al., 2014; Saxena, 2018; Wang, 2014), they do not include context-specific predictors such as quality of mobile communication services (MCS) and perceived mobility. For example, in a developing country context, m-government is still considered a disruptive technology (Chen et al., 2016); and the importance of MCS might be the most critical factor in m-government service adoption (Sultan and Steve, 2015). MCS, therefore, need to be considered in the study of m-government service adoption. This study attempts to bridge this gap in the literature by adding contextual predictors to the unified theory of acceptance and use of technology (UTAUT) model and by observing the effect of some additional contextual constructs, namely quality MCS and mobility, on the adoption of m-government services. It should be noted that, in this study, in order to facilitate the understanding of the topic, m-government is considered as an overall concept, rather than focusing on any one particular service, and Bangladesh – a developing country with a large population – has been selected as the target for this study.
A large majority of prior studies in this field have employed conventional statistical methods such as multiple regression analysis and structural equation modeling (SEM) as their methodological approach. These techniques, however, might not be adequate in explaining complex interaction effects among multiple predictors, since relationships between predictors and system usage behaviors could be asymmetric, and conventional regression techniques are incompatible with asymmetric relationships (Liu et al., 2017). This limitation lends credibility to predictive modeling techniques, such as neural networks, as an alternative approach for exploring information systems (IS) usage behavior (Raymond et al., 2010; Liu et al., 2017; Woodside, 2013). A neural network offers an alternate avenue for deducing causal relationships between the antecedents and outcome of IS adoption behavior by taking into account interdependencies among the former. Therefore, we address the limitation discussed above by employing a two-stage predictive modeling approach suggested by Chong (2013), whereby SEM is used to examine the causal relationships between predictors of m-government service adoption, and a neural network model is used to validate SEM results and rank the independent predictors. This multi-analytical statistical approach provides a fine-grained understanding of the topic, and the two approaches generally offset each other’s weaknesses (Scott and Walczak, 2009).
The rest of this paper is organized as follows. In Section 2, we discuss the research model and develop hypotheses to be tested in this study. We then discuss the research methodology in Section 3. After that, we provide the results and analyze them in Section 4, and discuss their theoretical and practical implications in Section 5. Finally, we conclude the paper in Section 6.
2. Research model and hypotheses development
2.1 The unified theory of acceptance and use of technology model
The UTAUT is a unified model that comprises eight theoretical models of technology adoption (Venkatesh et al., 2003). It consists of four major constructs: performance expectancy, effort expectancy, social influence, and facilitating conditions, which can be used to determine user acceptance and user behavior of any innovative and complex technology (Venkatesh et al., 2003). In contrast to other models, UTAUT is known to explain about 70 per cent of the variance in the behavioral intention to use and accept a technology (Venkatesh et al., 2003).
M-government has been treated as a “technology” because of its inherent dependence on an information technology (IT) platform where information exchange happens between the government (supplier of m-government services) and the citizens (users of m-government services) via mobile technologies. The rationale behind using the UTAUT framework is that it helps to examine social and other facilitating factors that influence the IT linked with m-government systems. Social factors, such as the behavior of m-government users and the influence of others on them, are important for the acceptance and use of m-government services (Zuiderwijk et al., 2015). The importance of examining social factors in technology adoption is well recognized (Gwebu and Wang, 2011), particularly in countries with collectivist culture, such as Bangladesh (Srite and Karahanna, 2006). There are also a number of facilitating conditions (e.g. support, training, infrastructure) that impact the actual use and acceptance of a technology (Sykes et al., 2009). These facilitating conditions are included in the theoretical framework proposed in this study.
Moreover, Venkatesh et al. (2012) argued that there is a need to assess the magnitude, direction, and significance of the relationships among variables; such relationships might also vary according to the context. The UTAUT model is appropriate for capturing end user behavioral intention for complex and sensitive ICT-related services where contextual differences exist (Dwivedi et al., 2016). Therefore, the UTAUT model is employed as the theoretical framework to explore m-government service adoption in the context of Bangladesh.
The UTAUT model is widely used along with additional contextual constructs incorporating specific elements of a required field (Venkatesh et al., 2012). Introduction of additional contextual predictors demonstrates a clearer understanding of consumers’ acceptance of domain-specific technology (Cimperman et al., 2016; Jeong et al., 2017; Venkatesh et al., 2011). Based on previous research and the m-government context, two contextual variables, namely quality of MCS and mobility were added to the existing model. The extended model is shown in Figure 1. As shown in the figure, performance expectancy, effort expectancy, social influence and facilitating conditions are taken as direct antecedents of behavioral intention, while MCS and mobility are used as antecedent of performance expectancy and effort expectancy, respectively. Behavioral intention is taken as the dependent variable in this study. Behavioral intention is defined as an individual's intention, plan, or prediction to use m-government services in the future. The development of hypotheses based on these antecedents is discussed in the following section.
2.2 Hypothesis development
2.2.1 Performance expectancy.
Performance expectancy is defined as “the degree to which an individual believes that using the system will help him or her to attain gains in job performance” (Venkatesh et al., 2003). We assert that the availability of m-government services, such as m-government platforms, application, tools, and interfaces, increases an individual's expectancy to perform better. Thus, consistent with the theoretical arguments underlying UTAUT, we anticipate a direct and positive impact of performance expectancy on the intention to use and accept m-government services.
2.2.2 Effort expectancy.
Effort expectancy is related to the degree of ease associated with the use of a technology (Venkatesh et al., 2003) and the degree to which a person believes that the use of the technology will be free of effort (Gwebu and Wang, 2011). We define effort expectancy as the extent to which a person believes that using m-government services will be free of effort and easy to learn. In the context of this study, we assert that people analyze their expectations of the degree to which m-government services are easy or difficult to use, and that this perceived ease of use influences their intention to use m-government services.
2.2.3 Social influence.
Social influence is defined as a change in an individual’s thoughts, feelings, attitudes, or behaviors that results from communicating with another individual or a group (Rashotte, 2007). Social influence may come from friends, family and other people who influence a person’s behavior or are important to this person (Talukder et al., 2019). Almuraqab et al. (2017), Yfantis et al. (2013) and Sultana and Ahlan (2014) have shown that social influence has a significant impact on the behavioral intention to use m-government services. In contrast, Alshehri et al. (2012) showed that social influence has no positive impact on behavioral intention in Saudi Arabia. Such differences may be the reflection of contextual difference in Saudi Arabia, such as culture and demographic variables. However, in this study, we argue that it is possible that the influence of colleagues, peers, family members, supervisors and others could determine an individual’s usage of m-government services.
2.2.4 Facilitating conditions.
Facilitating conditions can be defined as “the degree to which an individual believes that an organizational and technical infrastructure exists to support use of the system” (Venkatesh et al., 2003). Although prior research has shown that facilitating conditions are not the best predictors for the behavioral intention to use technology (Alshibly et al., 2016), we expect that facilitating conditions influence the intention to adopt m-government services. If facilitating conditions such as mobile networks, connection to the internet, sufficient and appropriate m-government infrastructure are available, the intention to use m-government will be higher.
Based on the above arguments, we propose the folowing hypotheses:
Performance expectancy has a positive impact on behavioral intention to use m-government services.
Effort expectancy has a positive impact on behavioral intention to use m-government services.
Social influence has a positive impact on behavioral intention to use m-government services.
Facilitating conditions have a positive impact on behavioral intention to use m-government services.
2.2.5 Mobile communication services.
In rural Bangladesh, mobile network coverage is not yet widely available, and high-speed data transmission (i.e. 3G, 4G) coverage is unavailable even in many cities. As a consequence, it is difficult to access and use government services by using mobile technologies, anytime and anywhere. Mobile communication characteristics, such as rich content, real-time content update and availability, personalization, comprehensive customer service, service accessibility, network coverage, transmission speed, and signal quality are pre-requisities for any m-government service to be used with less effort and while in motion (Shieh et al., 2014). Widespread and reliable mobile communication facilities allow users to access and use m-government systems from anywhere and anytime (Al-Hubaishi et al., 2017). Previous research has found MCS to be an essential factor for citizens’ acceptance of mobile-based services (Chen et al., 2016; Hung et al., 2013; Shieh et al., 2014; Wu et al., 2009). Strong and sound telecommunication facilities improve the convenience, and usefulness of m-government systems for the citizens (Shieh et al., 2014) along with providing safer and faster access to m-government services. Therefore, it is believed that MCS will make m-government systems more useful, accessible with minimal effort, and able to support myriad needs of the users. Thus, the following hypotheses are proposed:
Mobile communication service has a positive impact on performance expectancy towards the use of m-government services.
Mobile communication service has a positive impact on effort expectancy towards the use of m-government services.
Mobile communication service has a positive impact on mobility towards the use of m-government services.
2.2.6 Perceived mobility.
Perceived mobility is defined as the level of user awareness towards the value of mobile technologies (Huang et al., 2007). In a mobile communication environment, mobility refers to the characteristics of mobile devices to handle ubiquitous use, and access information, communication and business transactions in a state of motion via wireless networks (Yuan et al., 2010). Perceived mobility is one of the crucial determinants of perceived usefulness in m-government service adoption (Alotaibi and Roussinov, 2016; Wang, 2014), in using mobile learning services (Huang et al., 2007), and mobile communication networks (Siau and Shen, 2003). Gunawardana and Ekanayaka (2009) suggested that perceived mobility significantly impacts effectiveness of mobile systems and service quality. Perceived mobility is assumed as an antecedent to performance expectancy because mobility is an essential characteristic of any wireless network service. Thus, we propose the following hypothesis:
Mobility has a positive effect on performance expectancy of m-government systems.
3. Research methodology
3.1 Measurement development
All constructs of the research model were adapted from previous research and modified for this study. The number and sources of items used to measure each construct are summarized in Appendix 1. A structured questionnaire was initially developed in English and then translated into the local language (Bangla) by a professional translator. Next, another professional translator translated the questionnaire back into English. Based on this double translation process (Hoque and Sorwar, 2017), slight corrections were made to the questionnaire to ensure that the meaning of all items remained the same during translation. Two experts were asked to review the questionnaire in both languages to ensure clarity of instructions, content validity, and consistency. In the questionnaire, a five-point Likert scale was used to gather data for the constructs, where 5 indicates “Strongly agree” and 1 indicates, “Strongly disagree”. The survey questionnaire comprises two parts. The first part consists of 23 items related to the nine constructs of the proposed research model. The second part consists of five questions about demographic information of the respondents, including gender, age, education, and experience in using mobile technologies. Prior to data collection, the questionnaire was refined via limited pre-testing with 20 MBA students from University of Dhaka, Bangladesh.
3.2 Data collection
A cross-sectional study was conducted. The data was collected over a period of five months, from October 2017 to February 2018. Non-probability convenience sampling was undertaken due to time, cost and workforce constraints (Sekaran, 2006). In particular, we focused on university students, professionals in government and private sectors, people in business, and retirees from various public places of Dhaka city such as university campuses, parks, playgrounds, and community centers.
Printed questionnaires were personally distributed among the potential participants in the targeted places. Although home and location-based surveys are more time consuming, they provide a higher response rate compared to other instruments such as, telephone or online surveys (Malhotra, 2008). We sought the respondents’ verbal informed consent and confirmed the privacy of their information. Participants were also advised of their right to withdraw from the study at any time without incurring any negative consequences. In total, 230 survey questionnaires were distributed and since it was a self-administered survey, the response rate was 100 per cent. 14 questionnaires with missing data was excluded from further analysis (Joseph et al., 2010) and, thus, 216 valid responses were used for further analysis.
4. Data analysis and results
Data was analyzed using SPSS 23.0 and SmartPLS 3.2.8 version software. In the first stage, SEM was applied to determine the reliability and validity of the constructs, and to evaluate the predictive relevance of the model and the total variance explained (Chin, 1998). In the final stage, neural network analysis was conducted to validate the SEM results and to rank the key predictors of m-government service adoption.
4.1 Demographic information
The demographic characteristics of respondents presented in Appendix 2 show that males and females constituted 62 per cent and 38 per cent of the participants, respectively. A large majority of respondents (89 per cent) were between 20 and 40 years old. Most of the participants (89 per cent) were university students and professionals, with 57 per cent of them having more than 3 years of mobile phone usage experience.
4.2 Measurement model
The measurement model was assessed by examining the internal reliability, convergent validity, and discriminant validity (Hair et al., 2013), and the results are shown in Tables 1 and 2. Table I shows that the composite reliability values range from 0.86 to 0.90, and Cronbach’s alpha values range from 0.75 to 0.86. These indicate strong internal reliability, as all the values are above the acceptable 0.7 threshold (Hair et al., 1998). Next, we can see in Table I that the average variance extracted (AVE) ranges (0.61 to 0.75) and the estimated loading ranges (0.72 to 0.89), are greater than the recommended threshold of 0.50, indicating convergent validity (Anderson and Gerbing, 1988). Therefore, the conditions of internal reliability and convergent validity of the measurement instruments were satisfied in this study.
In addition, since the square root of AVE (shown in the diagonal of the factor correlations in Table II) for each latent variable was greater than the correlation of that variable with other latent variables, the Fornell–Larcker criterion was satisfied (Fornell and Larcker, 1981); thus indicating good discriminant validity (Henseler et al., 2015).
4.3 Structural model
The structural model was constructed to identify the path relationships among the constructs in the research model. The bootstrapping method was used to test the hypotheses at a significance level of 0.05 (p < 0.05) and the path coefficients. The relationship between dependent and independent variables was tested by path coefficient (β) and t-statistics value above 1.96 at 5 per cent level of significance (Hair et al., 2012). The R square value was used to calculate the percentage of the variance explained by the independent variables in the structural model (Klarner et al., 2013). The research model accounts for behavioral intention (BI) R2 = 59 per cent of the variance in response to m-government service adoption in Bangladesh.
The results from the research model show that all the antecedents of behavioral intention were found statistically significant, except for social influence. More specifically, performance expectancy (PE: t = 4.239, β = 0.286), effort expectancy (EE: t = 3.953, β = 0.210), and facilitating conditions (FC: t = 3.965, β = 0.213) were found to have significant effect on behavioral intention to use m-government services, whereas social influence (SI: t = 0.801, β = 0.050) was found to not have a significant effect. Results also indicate that mobile communications services (MCS: t = 10.394, β = 0.571), (MCS: t = 3.162, β = 0.275), and (MCS: t = 6.428, β = 0.464) have significant and direct effect on effort expectancy, performance expectancy and perceived mobility, respectively. Additionally, perceived mobility (PM: t = 4.823, β = 0.433) has a significant positive effect on performance expectancy towards behavioral intention to use m-government services. In contrast, social influence (SI: t = 0.801, β = −0.050) was found to have a non-significant effect on behavioral intention. Therefore, among the hypotheses, H1, H2, H4, H5, H6, H7, and H8 were supported, and H3 was not supported. The detailed results are shown in Table III.
The study also supports the previous findings using Q2 predictive relevance measure (Stone, 1974). The obtained Q2 values, after running the blindfolding procedure (Chin, 1998) with an omission distance D = 7, were 0.23 for performance expectancy, 0.18 for effort expectancy, 0.15 for perceived mobility, and 0.42 for behavioral intention. All of the Q2 values are well above zero; indicating the predictive relevance of the PLS path model. Finally, the f2 values (i.e. effect size) for performance expectancy, effort expectancy, and perceived mobility were measured. As shown in Table IV, according to the standard of Aiken et al. (1991), the effect size for performance expectancy on behavioral intention is medium, effort expectancy on behavioral intention is small, and perceived mobility on performance expectancy is also medium.
4.4 Neural network modeling
While SEM provides us an excellent understanding of the relationships among the constructs in the model, it is often not enough to understand the complexity of decision-making in forming the behavioral intention to use m-government services (Al-Shihi et al., 2018; Sharma and Sharma, 2019). Therefore, in addition to SEM, we have employed an innovative hybrid SEM and neural network approach to help us develop a more fine-grained understanding of the relationships among the various constructs in the research model. Employing a neural network allows the detection of both linear and nonlinear relationships among decision variables (Sharma et al., 2018). Furthermore, an artificial neural network model does not need to test requirements of any data distribution such as linearity, normality and independence (Chong, 2013).
4.4.1 Validations of neural networks.
The back propagation multilayer perceptron neural network model was used to describe the neural networks and the variables that the user can control during training of a neural network (Poulton, 2001). To avoid overfitting of the model, a ten-fold cross validation approach was used. In this study, we have divided data into 10 equal groups, where 90 per cent of the data was used to train the neural network and the remaining 10 per cent were used to test the prediction accuracy of the trained network.
The accuracy of the neural network model was measured using Root Mean Square Error (RMSE) values. The two significant constructs, namely quality of MCS and mobility, were used as input layer and performance expectancy (Figure 2). In the second model, performance expectancy, effort expectancy, and facilitating conditions were used as input layer and behavioral intention was used as the output layer (Figure 3). From Table V, we can see that the average cross-validated RMSE for the training and testing was 0.121 and 0.115, respectively for model 1; and 0.141 and 0.162, respectively for model 2. Given the relatively small RMSE values, we can conclude that the network models were quite reliable in capturing the numerical relations between the predictors and the outputs.
4.4.2 Sensitivity analysis.
The normalized importance ratio is calculated as “the ratio of the importance of each predictor to the highest importance value” (Chong et al., 2015). Table VI summarizes the average relative importance and normalized relative importance (in percentage) reported by neural network models 1 and 2. In neural network model 1, MCS is the most important predictor of satisfaction followed by perceived mobility. This indicates that there is no change in the order of the importance of the predictors of performance expectancy in SEM and neural network results. Next, using neural network model 2, facilitating conditions is the key predictor of intention to use m-government services followed by performance expectancy and effort expectancy. However, performance expectancy is the most important predictor of the intention to use m-government services followed by facilitating conditions, effort expectancy, and social influence in the SEM results. This difference in the results of SEM and neural network modeling may be supported by the higher order of predictive power and non-compensatory nature of the latter model. Despite this difference, this outcome may be considered as the validation of the results.
5. Discussion and implications
According to the results summarized in the previous section, the proposed research model examined in this study was able to achieve a satisfactory level of predictive power extracted by dependent constructs: performance expectancy (39.2 per cent) and intention to use (59.4 per cent). In addition to these satisfactory results, reliability and validity of constructs were within the acceptable limits. Based on the results obtained using neural networks, the most significant predictor of behavioral intention is facilitating conditions, which positively influences citizens’ intentions towards the adoption of m-government services. The result indicates that arranging training programs, providing organizational and technological infrastructures, and making relevant resources available to facilitate the use of m-government services are some of the mechanisms a government can use to significantly improve its citizens' intention to use m-government services.
This study also confirms the significant influences that performance expectancy and effort expectancy have on behavioral intention to adopt m-government services. Results indicate that an individual's intention can be determined by the ease of use of the m-government system (i.e. is less complex) and the degree to which it may prove useful and beneficial (i.e. has greater performance) to the users.
Most interestingly, the study found that MCS have a strong influence on performance expectancy, effort expectancy and perceived mobility. This study considered technological factors like signal quality, network coverage, transmission speed, services content and service assurance to have a substantial impact on m-government service adoption in developing countries. This study is the first approach that considers the effects of MCS from three different perspectives. First, this study found MCS had a significant effect on perceived mobility. This result indicates that while users access and use m-government services according to their convenience (anytime and anywhere) (Anckar and D'incau, 2002), they need suitable and reliable MCS (Huang et al., 2007). Second, this study found MCS (t = 3.162) to have a positive influence on performance expectancy. MCS help users to use m-government services whenever and wherever they are required, thus making such services more convenient and useful. Third, this study found MCS have a significant impact on effort expectancy. Ahn et al. (2007) stated that high-quality systems provide users faster access to information with less effort. A reasonable number of previous empirical studies have found that the quality of MCS significantly influence user satisfaction and attitude towards behavioral intention to use mobile-based technologies (Chen et al., 2016; Hung et al., 2013; Shieh et al., 2014; Wu et al., 2009).
Further, this study identifies that mobility has a significant effect on performance expectancy towards behavioral intention to use m-government services. Park and Joon Kim (2013) stated that perceived mobility is an antecedent of performance expectancy and a key feature for mobile-based service development. This finding is consistent with a number of studies that have found mobility to be a significant determinant of mobile-based technology adoption in general (Huang et al., 2007), and m-government adoption in particular (Alotaibi and Roussinov, 2016; Wang, 2014).
Finally, in contrast with most of the existing studies (Talukder et al., 2019; Thongsri et al., 2019), we found social influence to be a non-significant antecedent of behavioral intention. According to modernization theory (Cowgill and Holmes, 1972), today’s modernized and scientific society can trigger a loss of social status for the elderly because knowledge from the book is now valued more than knowledge acquired through personal experience (Brown, 1996). People tend to disregard the influence of societal pressure, image, and social status and tend to pursue more emotionally meaningful goals (Carstensen et al., 2003). Therefore, and as confirmed by the results, people of Bangladesh no longer place high importance on their previous habits, and are ready to change their modes of receiving m-Government services if necessary. Additionally, this finding is consistent with some of the previous results reported in different IS domains (Cimperman et al., 2016; Sharma et al., 2018).
5.1 Implications for research
This study provides a number of useful insights to academicians and researchers. First, it has extended the existing literature by adding two additional contextual constructs, namely quality of MCS and mobility into the UTAUT model to explain the behavioral intention towards m-government in the context of a developing country. The inclusion of MCS makes this study novel and MCS might be the most significant construct in mobile application literature. MCS and mobility make mobile-based services more effective and efficient. This study will add more knowledge regarding m-government service adoption, specifically in the study of mobile applications.
Second, the application of an innovative two-stage SEM-neural network modeling approach provides a holistic understanding from an analytical point of view. The application of neural network modeling is an attempt to address one of the limitations of many previous studies on technology adoption by developing a non-compensatory model to predict behavioral intention of citizens towards m-government services in developing countries. Neural network modeling is non-compensatory in nature and overcomes the weaknesses of compensatory nature of linear SEM. This study demonstrated the validity of the research model using SEM for understanding the effect of independent constructs on a dependent construct, and neural network modeling was used to validate the SEM results and to rank significant independent constructs.
5.2 Implications for practice
This study has several practical implications for the stakeholders involved in the development and implementation of m-government systems in Bangladesh, such as relevant policymakers, government officials and system developers. Citizens’ adoption and utilization rate of m-government services can be improved in several ways as discussed below.
Results found performance expectancy and effort expectancy to be the most important antecedents of behavioral intention to use m-government services. This implies that individuals attribute substantial importance to the technological efficacy of an m-government system and its ease of use. Therefore, designers, system analysts and developers responsible for the design and development of m-government systems should focus more on minimizing the complexities associated with exploration and use of the system, if there are any, and improving the usefulness of the system.
We also found that facilitating conditions had a direct impact on behavioral intention to use m-government services. This suggests that individuals may place importance on facilitating conditions such as help desks, technological and infrastructural resources, and training programs, as well as the experiences of other individuals in using the m-government system in question. Therefore, concerned government organizations or departments should consider equipping their training program centers with adequate infrastructural and technological facilities for providing the prerequisite training to users. Such actions can positively impact the use and adoption of relatively new and useful m-government systems.
Furthermore, the most critical MCS such as signal quality, network coverage, transmission speed, service content and service assurance strongly influence perceived mobility, performance expectancy, and effort expectancy towards the adoption of m-government services. The government can collaborate with already established private mobile network service providers through Private Public Partnership (PPP) to provide government services over all geographical locations. It could be a win-win situation for both the government (cost reduction) and the private mobile operators (profit sharing). The development of 4G high-speed data transmission throughout the country would further help improve the quality of MCS and enable the government to realize the objective of providing m-government services to its citizens at their doorsteps. Finally, the government should ensure high availability and reliability of m-government systems so users can access and use them anytime and from anywhere. The results of the research model will also be useful to the developers and designers of m-government services in guiding them towards implementing new m-government services and updating existing services to ensure better acceptance by the citizens
5.3 Limitations and future directions
Despite the contributions, this study has a few limitations that need to be considered when interpreting the findings. First, m-government is still relatively new and an emerging trend in developing countries such as Bangladesh. Future studies can consider measuring the diffusion of m-government activities across time, and examine whether the variables in this study change at various m-government diffusion stages. Second, future study should use the longitudinal method to explain the causal relationship between variables over time since the model in this study is cross-sectional, which measures the intention only at a single point in time. Finally, this study only examines behavioral intention to use m-government services. Future studies to examine citizens’ continuous usage intention and to evaluate the results with the intention to use m-government application is highly recommended, as the initial-adoption and continuance usage behaviors are two theoretically distinct concepts and the latter is more important than former (Bhattacherjee, 2001).
This study presented an integrated model to predict the adoption of m-government services among the citizens of Bangladesh. The main research model was divided into two sub models: Model 1 (independent variables: perceived mobility, MCS and dependent variable: performance expectancy) and Model 2 (independent variables: performance expectancy, effort expectancy and facilitating conditions and dependent variable: behavioral intention to use). In this study, it was found that quality of MCS and perceived mobility are the key determinants influencing performance expectancy and behavioral intention to use m-government services. The obtained results also suggest that higher level of facilitating conditions and higher performance expectancy will help in attracting new and potential users. A combination of SEM and neural network modeling was employed, which provides an additional originality to this study. SEM has been used to test and verify the research hypotheses in many prior studies, but has rarely been combined with machine learning models. SEM is primarily used to test linear models and sometimes oversimplifies the complexities involved in the adoption models in business domains. In this study, SEM was used to test research hypotheses and identify statistically significant predictors and the neural network model used these predictors as input and ranked them for better decision-making insights. Thus, this study provides very useful theoretical as well as practical implications for researchers, academics and m-government service providers.
|Mobile Communications Services||MCS1||0.829||0.859||0.671||0.754|
AVE = Average variance extracted; CR = Composite reliability
Correlation matrix and square root of the AVE
BI = Behavioral intention; PE = Performance expectancy; EE= Effort expectancy; SI = Social influence; FC = Facilitating conditions; MCS = Mobile communication services; PM = Perceived mobility
|SL no.||Path||Coefficient (β)||t-statistics||Comments|
|H1||PE -> BI||0.286||4.239||Supported|
|H2||EE -> BI||0.265||4.153||Supported|
|H3||SI -> BI||0.050||0.801||Not Supported|
|H4||FC -> BI||0.203||3.765||Supported|
|H5||MCS -> PE||0.275||3.162||Supported|
|H6||MCS -> EE||0.571||10.394||Supported|
|H7||MCS -> PM||0.464||6.428||Supported|
|H8||PM -> PE||0.433||4.823||Supported|
R2 for BI = 0.594; Significant at p < 0.05
Results of R2, Q2 and f2
|Endogenous latent variables||R2||Q2||F2|
Neural network validations
|Network||Model 1||Model 2|
SD: Standard deviation
Importance of independent variables
|Networks||Model 1: Output PE||Model 2: Output BI|
|Construct||Corresponding items||Items sources|
|Performance expectancy||PE1. I find m-government service useful in my daily life||Talukder et al. (2019), Venkatesh et al. (2003)|
|PE2. Using m-government service helps me accomplish things more quickly|
|PE3. Using m-government service increases my productivity|
|PE4: Using m-government service improves my quality of work|
|Effort expectancy||EE1. Learning how to use m-government service is easy for me||Talukder et al. (2019), Venkatesh et al. (2003)|
|EE2. My interaction with m-government service is understandable|
|EE3. I find m-government service easy to use|
|EE4. It is easy for me to become skillful at using m-government service|
|Social influence||SI1: People who are important to me think that I should use m-government system
SI2: People who influence my behavior think that I should use m-government system
SI3: I would use m-government system if my friends and colleagues used them
|Talukder et al. (2019), Venkatesh et al. (2003)|
|Facilitating conditions||FC1. I have the resources necessary to use m-government service||Talukder et al. (2019), Venkatesh et al. (2003)|
|FC2. I have the support necessary to use m-government service|
|FC3. M-government is compatible with other technologies I use|
|Mobile communication services||QMS1: Our mobile telecommunications services incorporated with many services contents (such as; content richness, real-timeliness, and personalization)
QMS2: Our mobile telecommunications services incorporated with service assurance (such as; comprehensive customer service and service accessibility)
QMS3: Our mobile telecommunications services incorporated with network reliability (such as; transmission speed, signal quality and network coverage)
|Chen et al. (2016), Shieh et al. (2014), Wu et al. (2009)|
|Perceived Mobility||PM1: It is convenient to access m-government systems anywhere at any time
PM2: I expect that mobile government services would be available for use whenever I need it
PM3: I would find mobile government services to be easily accessible and portable
|Alotaibi and Roussinov (2016), Huang et al. (2007)|
|Behavioral Intention||BI1. I intend to use m-government service in the future||Talukder et al. (2019), Venkatesh et al. (2003)|
|BI2. I will always try to use m-government service in my daily life|
|BI3. I plan to use m-government service frequently|
|Less than 20||10||5|
|Private sector employee||56||26|
|Mobile using experience|
|Less than 1 years||2||1|
|More than 10 years||24||11|
Ahmad, S.Z. and Khalid, K. (2017), “The adoption of M-Government services from the user’s perspectives: empirical evidence from the United Arab Emirates”, International Journal of Information Management, Vol. 37 No. 5, pp. 367-379.
Ahn, T., Ryu, S. and Han, I. (2007), “The impact of web quality and playfulness on user acceptance of online retailing”, Information and Management, Vol. 44 No. 3, pp. 263-275.
Aiken, L.S., West, S.G. and Reno, R.R. (1991), Multiple Regression: Testing and Interpreting Interactions, Sage, Thousand Oaks, CA.
Al-Hubaishi, H.S., Al-Hubaishi, H.S., Ahmad, S.Z., Ahmad, S.Z., Hussain, M. and Hussain, M. (2017), “Exploring mobile government from the service quality perspective”, Journal of Enterprise Information Management, Vol. 30 No. 1, pp. 4-16.
Almarashdeh, I. and Alsmadi, M.K. (2017), “How to make them use it? Citizens acceptance of M-government”, Applied Computing and Informatics, Vol. 13 No. 2, pp. 194-199.
Almuraqab, N., Jasimuddin, S. and Mansoor, W. (2017), “Mobile government (mgovernment) adoption factors in the UAE: a conceptual framework based on UTAUT”, International Journal of Engineering Technology, Management and App. lied Sciences, Vol. 5 No. 3, pp. 14-19.
Alotaibi, S. and Roussinov, D. (2016), “Developing and validating an instrument for measuring mobile government adoption in Saudi Arabia”, World Academy of Science, Engineering and Technology, International Journal of Social, Behavioral, Educational, Economic, Business and Industrial Engineering, Vol. 10 No. 3, pp. 710-716.
Alshehri, M. Drew, S. Alhussain, T. and Alghamdi, R. (2012), “The effects of website quality on adoption of E-Government service: AnEmpirical study applying UTAUT model using SEM”, 23rd Australasian Conference on Information Systems, Melbourn, Australia.
Alshibly, H. and Chiong, R. (2015), “Customer empowerment: does it influence electronic government success? A citizen-centric perspective”, Electronic Commerce Research and Applications, Vol. 14 No. 6, pp. 393-404.
Alshibly, H., Chiong, R. and Bao, Y. (2016), “Investigating the critical success factors for implementing electronic document management systems in governments: evidence from Jordan”, Information Systems Management, Vol. 33 No. 4, pp. 287-301.
Al-Shihi, H., Sharma, S.K. and Sarrab, M. (2018), “Neural network approach to predict mobile learning acceptance”, Education and Information Technologies, Vol. 23 No. 5, pp. 1805-1824.
Anckar, B. and D'incau, D. (2002), “Value creation in mobile commerce: findings from a consumer survey”, JITTA: Journal of Information Technology Theory and Application, Vol. 4 No. 1, p. 43.
Anderson, J.C. and Gerbing, D.W. (1988), “Structural equation modeling in practice: a review and recommended two-step approach”, Psychological Bulletin, Vol. 103 No. 3, pp. 411-423.
Bhattacherjee, A. (2001), “Understanding information systems continuance: an expectation-confirmation model”, MIS Quarterly, Vol. 25 No. 3, pp. 351-370.
Brown, A.S. (1996), The Social Processes of Aging and Old Age, Prentice Hall Upper Saddle River, New York, NY.
Carstensen, L.L., Fung, H.H. and Charles, S.T.J.M. (2003), “Socioemotional selectivity theory and the regulation of emotion in the second half of life”, Motivation and Emotion, Vol. 27 No. 2, pp. 103-123.
Chen, Z.-J., Vogel, D. and Wang, Z.-H. (2016), “How to satisfy citizens? Using mobile government to reengineer fair government processes”, Decision Support Systems, Vol. 82, pp. 47-57.
Chin, W.W. (1998), “The partial least squares approach to structural equation modeling”, Modern Methods for Business Research, Vol. 295 No. 2, pp. 295-336.
Chong, A.Y.-L. (2013), “A Two-Staged Sem-Neural network approach for understanding and predicting the determinants of M-Commerce adoption”, Expert Systems with Applications, Vol. 40 No. 4, pp. 1240-1247.
Chong, A.Y.-L., Liu, M.J., Luo, J. and Keng-Boon, O. (2015), “Predicting RFID adoption in healthcare supply chain from the perspectives of users”, International Journal of Production Economics, Vol. 159, pp. 66-75.
Cimperman, M., Brenčič, M.M. and Trkman, P. (2016), “Analyzing older users’ home telehealth services acceptance behavior – applying an extended UTAUT model”, International Journal of Medical Informatics, Vol. 90, pp. 22-31.
Cowgill, D.O. and Holmes, L.D. (1972), Aging and Modernization, Appleton-Century-Crofts and Fleschner Publishing Company, New York, NY.
Dwivedi, Y.K., Shareef, M.A., Simintiras, A.C., Lal, B. and Weerakkody, V. (2016), “A generalised adoption model for services: a cross-country comparison of mobile health (m-health)”, Government Information Quarterly, Vol. 33 No. 1, pp. 174-187.
Fornell, C. and Larcker, D.F. (1981), “Evaluating structural equation models with unobservable variables and measurement error”, Journal of Marketing Research, Vol. 18 No. 1, pp. 39-50.
Gao, S., Krogstie, J. and Siau, K. (2014), “Adoption of mobile information services: an empirical study”, Mobile Information Systems, Vol. 10 No. 2, pp. 147-171.
Gunawardana, K.D. and Ekanayaka, S. (2009), “An empirical study of the factors that impact medical representatives' attitude toward the intention to use M-Learning for career development”, Sasin Journal of Management, Vol. 15 No. 1.
Gwebu, K.L. and Wang, J. (2011), “Adoption of open source software: the role of social identification”, Decision Support Systems, Vol. 51 No. 1, pp. 220-229.
Hair, J., Anderson, R., Tatham, R. and Black, W. (1998), Multivariate Data Analysis with Readings, Prentince-Hall Internacional, Upper Saddle River, NJ.
Hair, J.F., Jr, Hult, G.T.M., Ringle, C. and Sarstedt, M. (2013), A Primer on Partial Least Squares Structural Equation Modeling (Pls-Sem), Sage Publications, NJ.
Hair, J.F., Sarstedt, M., Ringle, C.M. and Mena, J.A. (2012), “An assessment of the use of partial least squares structural equation modeling in marketing research”, Journal of the Academy of Marketing Science, Vol. 40 No. 3, pp. 414-433.
Henseler, J., Ringle, C.M. and Sarstedt, M. (2015), “A new criterion for assessing discriminant validity in variance-based structural equation modeling”, Journal of the Academy of Marketing Science, Vol. 43 No. 1, pp. 115-135.
Hoque, R. and Sorwar, G. (2017), “Understanding factors influencing the adoption of mhealth by the elderly: an extension of the UTAUT model”, International Journal of Medical Informatics, Vol. 101, pp. 75-84.
Huang, J.-H., Lin, Y.-R. and Chuang, S.-T. (2007), “Elucidating user behavior of mobile learning: a perspective of the extended technology acceptance model”, The Electronic Library, Vol. 25 No. 5, pp. 585-598.
Hung, S.-Y., Chang, C.-M. and Kuo, S.-R. (2013), “User acceptance of mobile E-Government services: an empirical study”, Government Information Quarterly, Vol. 30 No. 1, pp. 33-44.
ITU (2016), “Ictfactsfigures2016”, International Telecommunication Union.
Jeong, S.C., Kim, S.-H., Park, J.Y. and Choi, B. (2017), “Domain-Specific innovativeness and new product adoption: a case of wearable devices”, Telematics and Informatics, Vol. 34 No. 5, pp. 399-412.
Joseph, F., Hair, J., Black, W., Babin, B. and Anderson, R. (2010), Multivariate Data Analysis: A Global Perspective, 7th (Global Edition) ed., Prentice Hall, Upper Saddle River, NJ.
Klarner, P., Sarstedt, M., Hoeck, M. and Ringle, C.M. (2013), “Disentangling the effects of team competences, team adaptability, and client communication on the performance of management consulting teams”, Long Range Planning, Vol. 46 No. 3, pp. 258-286.
Liu, Y., Mezei, J., Kostakos, V. and Li, H. (2017), “Applying configurational analysis to is behavioural research: a methodological alternative for modelling combinatorial complexities”, Information Systems Journal, Vol. 27 No. 1, pp. 59-89.
Malhotra, N.K. (2008), Marketing Research: An Applied Orientation, 5th ed., Pearson Education India, Bengaluru, Karnataka.
Napitupulu, D. and Sensuse, D.I. (2014), “The critical success factors study for E-Government implementation”, International Journal of Computer Applications, Vol. 89 No. 16, pp. 23-32.
Park, E. and Joon Kim, K. (2013), “User acceptance of Long-Term Evolution (LTE) Services an application of extended technology acceptance model”, Program, Vol. 47 No. 2, pp. 188-205.
Poulton, M.M. (Ed.) (2001), “Chapter 3 Multi-Layer perceptrons and back-propagation learning”, Handbook of Geophysical Exploration: Seismic Exploration, Elsevier, Pergamon, pp. 27-53.
Rana, N.P. and Dwivedi, Y.K. (2015), “Citizen's adoption of an E-Government system: validating extended social cognitive theory (SCT)”, Government Information Quarterly, Vol. 32 No. 2, pp. 172-181.
Rashotte, L. (2007), “Social influence”, in Ritzer, G. (Ed.), The Blackwell Encyclopedia of Social Psychology, Wiley, Hoboken, NJ.
Raymond, C., Ferrante, N. and Robert, I.M. (2010), “Nature that breeds solutions”, in Raymond, C. (Ed.), Nature-Inspired Informatics for Intelligent Applications and Knowledge Discovery: Implications in Business, Science, and Engineering, IGI Global, Hershey, PA, pp. 1-24.
Saxena, S. (2018), “Role of ‘perceived risks’ in adopting mobile government (M-Government) services in India”, Foresight, Vol. 20 No. 2, pp. 190-205.
Scott, J.E. and Walczak, S. (2009), “Cognitive engagement with a multimedia ERP training tool: assessing computer Self-Efficacy and technology acceptance”, Information and Management, Vol. 46 No. 4, pp. 221-232.
Sekaran, U. (2006), Research Methods for Business: A Skill Building Approach, John Wiley and Sons, Hoboken, NJ.
Sharma, S.K., Al-Badi, A., Rana, N.P. and Al-Azizi, L. (2018), “Mobile applications in government services (Mg-App) from user's perspectives: a predictive modelling approach”, Government Information Quarterly, Vol. 35 No. 4, pp. 557-568.
Sharma, S.K. and Sharma, M. (2019), “Examining the role of trust and quality dimensions in the actual usage of mobile banking services: an empirical investigation”, International Journal of Information Management, Vol. 44, pp. 65-75.
Shieh, L.-F., Chang, T.-H., Fu, H.-P., Lin, S.-W. and Chen, Y.-Y. (2014), “Analyzing the factors that affect the adoption of mobile services in Taiwan”, Technological Forecasting and Social Change, Vol. 87, pp. 80-88.
Siau, K. and Shen, Z. (2003), “Mobile communications and mobile services”, International Journal of Mobile Communications, Vol. 1 Nos 1/2, pp. 3-14.
Srite, M. and Karahanna, E. (2006), “The role of espoused national cultural values in technology acceptance”, MIS Quarterly, Vol. 30 No. 3, pp. 679-704.
Stone, M. (1974), “Cross-validatory choice and assessment of statistical predictions”, Journal of the Royal Statistical Society. Series B (Methodological), pp. 111-147.
Sultan, A-M. and Steve, L. (2015), “Building a user centric success factors model for mobile government”, ”Research Methods: Concepts, Methodologies, Tools, and Applications, A. Information Resources Management, IGI Global, Hershey, PA, pp. 1579-1599.
Sultana, M.R. and Ahlan, A.R. (2014), “Understanding citizen’s intention to use mobile government services in Bangladesh: role of perceived good governance and less corruption”, International Conference on Governance and Innovation, Bard Comilla.
Sykes, T.A., Venkatesh, V. and Gosain, S. (2009), “Model of acceptance with peer support: a social network perspective to understand employees' system use”, MIS Quarterly, Vol. 33 No. 2, pp. 371-393.
Talukder, M.S., Chiong, R., Bao, Y. and Hayat Malik, B. (2019), “Acceptance and use predictors of fitness wearable technology and intention to recommend: an empirical study”, Industrial Management and Data Systems, Vol. 119 No. 1, pp. 170-188.
Thongsri, N., Shen, L. and Bao, Y. (2019), “Investigating academic major differences in perception of computer Self-Efficacy and intention toward E-Learning adoption in China”, Innovations in Education and Teaching International, pp. 1-13, available at: https://doi.org/10.1080/0144929X.2019.1581259
Venkatesh, V., Morris, M.G., Davis, G.B. and Davis, F.D. (2003), “User acceptance of information technology: toward a unified view”, MIS Quarterly, Vol. 27 No. 3, pp. 425-478.
Venkatesh, V., Thong, J.Y., Chan, F.K., Hu, P.J.H. and Brown, S.A. (2011), “Extending the two-stage information systems continuance model: incorporating UTAUT predictors and the role of context”, Information Systems Journal, Vol. 21 No. 6, pp. 527-555.
Venkatesh, V., Thong, J.Y. and Xu, X. (2012), “Consumer acceptance and use of information technology: extending the unified theory of acceptance and use of technology”, MIS Quarterly, Vol. 36 No. 1, pp. 157-178.
Wang, C. (2014), “Antecedents and consequences of perceived value in mobile government continuance use: an empirical research in China”, Computers in Human Behavior, Vol. 34, pp. 140-147.
Woodside, A.G. (2013), “Moving beyond multiple regression analysis to algorithms: calling for adoption of a paradigm shift from symmetric to asymmetric thinking in data analysis and crafting theory”, Journal of Business Research, Vol. 66 No. 4, pp. 463-472.
Wu, H., Ozok, A.A., Gurses, A.P. and Wei, J. (2009), “User aspects of electronic and mobile government: results from a review of current research”, Electronic Government, an International Journal, Vol. 6 No. 3, pp. 233-251.
Yfantis, V., Vassilopoulou, K., Pateli, A. and Usoro, A. (2013), “The influential factors of M-Government’s adoption in the developing countries”, Mobile Web Information Systems, Springer, Berlin, Heidelberg, pp. 157-171.
Yuan, Y., Archer, N., Connelly, C.E. and Zheng, W. (2010), “Identifying the ideal fit between mobile work and mobile work support”, Information and Management, Vol. 47 No. 3, pp. 125-137.
Zuiderwijk, A., Janssen, M. and Dwivedi, Y.K. (2015), “Acceptance and use predictors of open data technologies: drawing upon the unified theory of acceptance and use of technology”, Government Information Quarterly, Vol. 32 No. 4, pp. 429-440.