An empirical validation of a unified model of electronic government adoption (UMEGA)

TL;DR: A unified model of e-government adoption (UMEGA) is developed and validated using data gathered from 377 respondents from seven selected cities in India, indicating that the proposed unified model outperforms all other theoretical models, explaining the highest variance on behavioral intention, acceptable levels of fit indices, and significant relationships for each of the seven hypotheses.

About: This article is published in Government Information Quarterly.The article was published on 2017-01-01 and is currently open access. It has received 376 citations till now. The article focuses on the topics: Context (language use).

Summary

1. Introduction

• Citizens are apt to be more skeptical and cynical toward governments.
• The Online Permanent Account Number (PAN) card registration system is one such e-government system, which provides transactional services to the citizens of India.
• Moreover, another aspect germane to the discussion of trust (or risk) in egovernment emerges from citizens’ strong opinions about political entities.
• Section 6 presents the proposed research model and developed hypotheses to support the interrelationships among the constructs.

2. Review of Existing User Acceptance Models

• IS research has long studied how and why individuals adopt new information technologies.
• While each of these streams makes a significant contribution to the literature on users’ acceptance of IT, the theoretical models to be included in the current review employ intention and/or usage as the key dependent variable(s) (Venkatesh et al., 2003).
• Fishbein and Ajzen (1975) divide the beliefs antecedent into two conceptually different sets: behavioral and normative.
• Diffusion is defined as the process by which the innovation is communicated to members of society using certain channels (Rogers, 1995).
• The similarity of the unified constructs used in the UTAUT model along with the other constructs and their definitions are provided later, alongside an explanation of the hypotheses developed for the proposed research model.

3. Research Data and Methods

• The sample of the study consisted of a wide spectrum of respondents from different cities of India, including Delhi (North India); Pune and Mumbai (West India); Bangalore (South India); and Patna, Siliguri, and Gangtok (East India), covering different demographics in India.
• The final questionnaire consisted of total 66 questions, including ten about respondents’ demographic characteristics and the remaining 56 questions related to seven constructs of the proposed research model.
• To institute the sample frame of citizens, respondents from different backgrounds were included, such as students, publicsector employees, private-sector employees, unemployed individuals, and pensioners.
• Secondly, the survey was designed to consider the responses of only non-adopters of the e-government systems.
• In addition, the basic reasons for selecting the seven cities were largely motivated by the researchers’ acquaintances and the need to represent cities from the east, west, north, and south zones of the country.

4. The Indian Context and the OPCRS

• India is the largest democracy in the world.
• E-government in India has gradually advanced from the computerization of government departments to initiatives supporting citizen centricity, service orientation, and transparency.
• CSCs are front-end service delivery points at the village level for the delivery of government-initiated services to users, where most service delivery takes place.
• The PAN card is compulsory and required by authorities while doing financial transactions with them.
• It must also be quoted when purchasing or selling a vehicle, buying/selling property, and when purchasing highvalue jewelry.

5. Empirical Comparison of Nine Models of Technology Adoption

• Table 2 illustrates nine different theories/models of IS/IT adoption, which have been validated using the data collected for the OPCRS.
• It had the highest variance (67%) among all the models in explaining behavioral intention, and reasonably acceptable fit indices (comparative fit index (CFI)=0.915, goodness-of-fit index (GFI)=0.906, adjusted GFI (AGFI)=0.831) were obtained for the model.
• The UTAUT, which has been considered a recommended model in most of its implementations, was found to have all significant relationships, except the relationship between performance expectancy and behavioral intention; however, its fit indices (χ 2 /DF=9.873, CFI=0.683, GFI=0.703, AGFI=0.661, RMSEA=0.154) significantly underperformed and possibly demonstrated the worst performance among all the models of IS/IT adoption.
• The e-government context differs from e-commerce primarily in three aspects: access, structure, and accountability (Carter & Belanger, 2005; Jorgensen & Cable, 2002).
• Table 2 shows that these models underperformed, as fit indices, path coefficients, and other aspects did not perform as per the recommended levels.

6. Proposed Research Model and Hypothesis Development

• Table 2 presents the different relationships between the constructs in the existing models of IS/IT adoption, based on the data gathered from non-adopters of the OPCRS.
• Moreover, the integrated constructs of the UTAUT model largely characterize the constructs that were used in prior models.
• Specifically, the authors position attitude as a mediating variable between performance expectancy and behavioral intention, between effort expectancy and behavioral intention, between social influence and behavioral intention, and between perceived risk and behavioral intention.
• The theoretical underpinnings of the TAM by Davis et al. (1989) and the DTPB by Taylor and Todd (1995b) indicate that perceived usefulness significantly determines an individual’s attitude in the context of IS/IT adoption.

7. Selection of Most Appropriate Items

• Table 3 presents the items of the proposed research model and their corresponding factor loadings.
• The authors adopted a similar approach, where the higher loading items from similar constructs constituting the variables were selected and used for the UMEGA.
• While choosing the items, the authors ensured that they picked a minimum of three items for a construct in the proposed model either beyond the recommended level of factor loading (i.e. ≥0.40) or some of the highly loading items that enhanced the performance of the proposed model (Field, 2005).

8. Results

• The characteristics of the data gathered from the respondents from various geographical locations indicated that the majority of the population was from a relatively younger generation.
• The education qualification level for close to 87.8% of the overall population was found to be undergraduate or above.
• Table 5 presents the means and standard deviations for the selected items of each construct used for the proposed research model.
• The authors report the goodnessof-fit index (GFI), the adjusted GFI (AGFI), the comparative fit index (CFI), and the Tucker– Lewis index (TLI).

9. Discussion

• The current research examined the alternative models of IS/IT adoption in the perspective of a transactional e-government system called the OPCRS.
• This test provided strong empirical support for UMEGA, which posits four direct determinants (performance expectancy, effort expectancy, social influence, and perceived risk) of attitude, two direct determinants (facilitating conditions and attitude) of behavioral intention, and one direct determinant of effort expectancy (facilitating conditions).
• The findings of this research critically underscore the significance of explicitly modeling individual characteristics through the proposed UMEGA.
• Moreover, the government needs to strengthen its infrastructure as prescribed in the NeGP initiatives, including State-Wise Area Networks , State Data Centers (SDCs), CSCs, and Electronic Service Delivery Gateways (NeGP, 2015).

10. Conclusion

• The adoption literature is fragmented, with many splinters of knowledge, which leaves policymakers in a void about which model is suitable and should be used.
• In the alternative theoretical model, the authors proposed and emphasized the need for the inclusion of attitude as a mediating variable.
• Hence, the empirical investigation shows that the proposed model that reframed the propositions of the UTAUT model is a more meaningful alternative for understanding e-government systems adoption.
• The model should be used by researchers in the egovernment community as a substitute for alternative theoretical models (e.g. the TRA, the TAM, the TPB, the DTPB, the DOI, etc.) of IS/IT adoption, as it effectively includes essential constructs from all such models, including the UTAUT.
• Developments like the mandatory use of services might also influence these models.

Figures

Table 8 Model Fit Summary for the Research Model

Table 4 Respondents’ Demographic Profile

Table 1 Theories, Models, and Constructs of Individual Acceptance

Table 5 Means and Standard Deviations (SD) of Items (N=377)

Table 7 Factor Correlation Matrix

Fig. 1. Proposed UMEGA (adapted from Venkatesh et al., 2003; Rana et al. 2016; 2015a)

Table 3 Item Loadings Using AMOS (N=377)

Table 2 Comparison of Prior Technology Adoption Models

Table 9 Path Coefficients and Hypothesis Testing

Table 6 Results of the Confirmatory Factor Analysis (CFA)

Frequently asked questions

Q1. What have the authors contributed in "An empirical validation of a unified model of electronic government adoption (umega)" ?

In this research, nine well-known theoretical models of information technology adoption are evaluated and 29 different constructs are identified. 

Q2. What have the authors stated for future works in "An empirical validation of a unified model of electronic government adoption (umega)" ?

As further research, the authors recommend a meta-synthesis of existing models to analyze the constructs. 

Q3. What are the independent constructs that positively and significantly influenced attitude?

The independent constructs performance expectancy, effort expectancy, and social influence positively and significantly influenced attitude (H1, H2, and H3), whereas perceived risk negatively and significantly influenced attitude (H6). 

Q4. Why did the authors examine the validity of the UTAUT model without using moderators?

Due to the lack of variation in the moderator for the adoption and use context, the authors examined the validity of the model without using moderators in it. 

Q5. What is the impact of facilitating conditions on effort expectancy?

The significant impact of facilitating conditions on effort expectancy indicates that the technical support and infrastructure provided by the government to its users led to easy access to the system. 

Q6. What is the significance of perceived risk in the proposed model?

The relevance of perceived risk as an external construct and its relationship with attitude indicates the relevance of risk as an important e-government-specific variable in the proposed unified model. 

Q7. Why did Venkatesh et al. drop moderators from the proposed model?

The other reason for dropping moderators from the proposed model was primarily to present a parsimonious model that could be tested for any e-government situation, rather than depending too much on any specific context (such as age, gender, education, income, etc.), asmodels like the UTAUT (Venkatesh et al., 2003) and the UTAUT2 (Venkatesh et al., 2012) do. 

Q8. What was the original argument for removing attitude from the TAM in an organizational setting?

The authors made this argument for removing attitude from the TAM in an organizational setting, which was originally proposed by Davis (1989) with attitude as a mediating variable for the model. 

Q9. How many of the questionnaires were partially completed?

Further scrutiny of the questionnaires revealed that 97 of them were either partially completed or filled in a biased manner (i.e. only one option ticked throughout the questionnaire), so they were rejected from subsequent analysis.