An empirical and comparative analysis of E-government performance measurement models: Model selection via explanation, prediction, and parsimony
Introduction
The performance of services delivered through government information and communication technology (ICT) applications has a substantial impact on the overall performance of a government, on citizens' perceptions of that performance, and even on the success of some of the most important public policies enacted by the government. The validity of this statement is now beyond debate. The high-profile launch of the HealthCare.gov website by the U.S. federal government – and more specifically, the many failings of that website in the eyes of both the experts and the general public – evidences just how important e-government has become, not just to the quality of services offered to citizens, but also to the fundamental and foundational relationship between citizens and governments (Office of Inspector General, 2016). No longer is e-government just one among the many alternative channels of government service delivery; increasingly, it is the single most important point of contact between government and citizens, and successes and failures therein have effects that can be felt across a political system.
Consider, for instance, the significant impact of the trouble-plagued launch of the HealthCare.gov website on the political fortunes of the then U.S. President Barack Obama. The website went live on October 1, 2013 and was designed to be the primary mechanism for the implementation of the Affordable Care Act (ACA) healthcare reform law. Needless to say, it had major shortcomings immediately after its launch. President Obama himself noted: “I am the first to acknowledge that the website that was supposed to do this all in a seamless way has had way more glitches than I think are acceptable” (Schwarz, 2013).
On the day of the launch of the HealthCare.gov website, the Gallup Daily Tracking Poll (of presidential job approval) put Obama at 45% approval and 48% disapproval, a negative gap of three percentage points. On November 3, 2013, a little more than a month after the launch of the website – a month filled with intense media scrutiny regarding the website's shortcomings – that gap had grown to 14 percentage points, at just 39% approval and 53% disapproval. Most if not all of the decline in Obama's job approval ratings, that had sunken to the lowest level of his presidency to that point in most of the reputable polls, was linked to the failed launch of HealthCare.gov (Brandus, 2013; McLean, 2013). In short, the functionality problems encountered by a single e-government website – albeit a website supporting a highly visible and undeniably controversial political initiative – caused a substantial erosion in citizen trust for the U.S. president and his entire political administration. Thus, at least in this one instance, it is fair to conclude that e-government service satisfaction can be a major driver of citizen trust in government above and beyond the traditional factors such as citizen demographics, political ideology, and party affiliation. While we may not have reached this point yet, it is not inconceivable to envision a future where e-government failures contribute to the fall of an elected politician at the highest level.
Driven by events like the troubled HealthCare.gov launch and the increasing importance and prevalence of e-government overall, a substantial and growing number of academic studies investigating the adoption and impact of the performance of this service channel have emerged (Dwivedi et al., 2017). More specifically, a veritable slew of the studies has investigated the links among e-government, performance of e-government, and citizen trust in the government (Cyr, 2008; Im, Cho, Porumbescu, & Park, 2014; Kim & Lee, 2012; Lim, Tan, Cyr, Pan, & Xiao, 2012; Morgeson & Petrescu, 2011; Morgeson, VanAmburg, & Mithas, 2011; Osman et al., 2014; Parent, Vandebeek, & Gemino, 2005; Porumbescu, 2016; Robinson, Liu, Stoutenborough, & Vedlitz, 2012; Teo, Srivastava, & Jiang, 2008; Tolbert & Mossberger, 2006; Van Ryzin, 2007; Welch, Hinnant, & Moon, 2005). Lying at the intersection of the literatures on e-government and citizen-centered administrative performance measurement, many of these studies have investigated not only how e-government impacts citizens' perceptions of different government services, but also how it influences broader perceptions of a government, a political system, and even individual political actors.
Not surprisingly, given the relative “age” of this literature, extant research in this area has used a wide array of different theoretical and conceptual models to measure e-government performance, citizen satisfaction with that performance, and the consequent impact on citizen trust. However, few (or no) studies have empirically explored the tradeoffs among explanation, prediction, and parsimony across these models with the goal of finding an “optimal” model for the key outcome in this context – citizen trust in government (Shmueli & Koppius, 2011). Given the growing importance of these websites to government service delivery of all types and at all levels, the ubiquitous and long-running “crisis of trust” in Western (and other) governments, and the prospect (at least among some) that strong e-government performance might play a role in repairing citizen trust, a critical comparison of the performance of different models for assessing the linkages between e-government, citizen satisfaction, and citizen trust is long overdue.
Our goal in this study is to apply two ubiquitous citizen-centered performance measurement models linking citizen satisfaction to trust in government – the expectancy-disconfirmation model (EDM) and the service quality model (SQM) – to the e-government context, and to critically compare the relative performance of each in explaining and predicting trust. To this end, we test a variety of model specifications to identify the models that best explain and predict trust in e-government, and that do so parsimoniously. Our efforts are guided by the recommendations for “theory-pruning” suggested in recent work (Davis, 2015; Gray & Cooper, 2010; Leavitt, Mitchell, & Peterson, 2010). More specifically, we compare the models for their in-sample explanatory power, out-of-sample predictive accuracy, and parsimony. By staging a direct empirical comparison between EDM and SQM frameworks our work contributes to theoretical development by delineating the strengths and weaknesses of the models vis-à-vis their explanatory and predictive abilities (Clarke, 2007; Edwards, 2010; Gray & Cooper, 2010).
Our results suggest that while the expectancy-disconfirmation paradigm exhibits high levels of explanatory power, a parsimonious model offers the best out-of-sample predictive accuracy, which is the statistical “gold standard” for judging predictive models (Shmueli & Koppius, 2011). In addition, the service quality model offers the best balance between explanatory power and predictive accuracy and thus emerges as an optimal choice for achieving the trade-off between explanation and prediction. In sum, we conduct an exhaustive comparison between the two major e-government citizen satisfaction-trust modeling paradigms and provide relevant implications for researchers and practitioners alike. The implications and the benchmarks established will, at the very least, help inform those deciding upon a modeling approach when analyzing and forecasting with this type of data.
Section snippets
Literature review: E-government and performance measurement modeling
The theorized relationships among e-government performance, citizen satisfaction, and relevant outcome perceptions such as trust (along with loyalty or intention to reuse) have resulted in a large number of studies examining these relationships across a variety of administrative contexts (Clemons et al., 2016; Cyr, 2008; Everard & Galletta, 2005; Lowry, Vance, Moody, Beckman, & Read, 2008; Mithas, Ramasubbu, Krishnan, & Fornell, 2006; Tang, Yu, & Smith, 2008; Vance, Elie-dit-cosaque, & Straub,
Model selection based on explanatory and predictive power
With the ongoing political upheaval across the world, a deeper understanding of the empirical strengths of theories that both explain and predict citizen trust in government is of critical importance. We argue that while the literature is replete with variations of SQM and EDM inspired models, there is lack of coherent work that judges the empirical strength of these models using practical industry-standard dataset. In this vein, Gray and Cooper (2010, pp. 620) remark that “the field of
Results
We use composite-based partial least squares (PLS) path modeling to compare the models, as this method aims to maximize the “explained variability” in all endogenous constructs (Henseler, Ringle, & Sinkovics, 2009) while having superior predictive accuracy than factor or covariance-based structural equation models (CB-SEM) that focus on accuracy of parameter estimates (Evermann & Tate, 2014).8
Main findings
The primary objective of this paper was to critically compare models from two equally ubiquitous modeling paradigms focused on the e-government citizen satisfaction-trust relationship — the expectancy-disconfirmation and the service-quality paradigms — along several dimensions, including their ability to explain trust, their out-of-sample predictive abilities, and their parsimony. To do so, we have analyzed a pooled, cross-agency, random sample of survey data collected by the American Customer
Conclusion
Due to the socio-political and economic upheavals around the world citizen trust in government has remained near historic lows (Pew, 2017). As governments across the globe increasingly interact and offer services to their citizens electronically, a deeper knowledge of the factors that affect citizen trust contingent on their interaction with the e-government services is warranted. The EDM and SQM are the two most widely used frameworks that focus on citizen trust in government. Despite their
Acknowledgments
We thank Dr Yogesh Dwivedi, three anonymous reviewers of Government Information Quarterly, seminar participants at the Frontiers in Service Conference 2015, and the Americas Conference on Information Systems 2015 for their helpful comments.
Pratyush Nidhi Sharma is an Assistant Professor in the Alfred Lerner College of Business & Economics, University of Delaware. He received his PhD from University of Pittsburgh in 2013. His research interests include online collaboration communities and networks, open source software development, and research methods used in Information Systems, particularly Partial Least Squares Path Modeling. In addition, he is also interested in issues surrounding technology use and adoption and human
References (134)
The nature of doubly concrete constructs and how to identify them
Journal of Business Research
(2016)- et al.
An empirical validation of a unified model of electronic government adoption (UMEGA)
Government Information Quarterly
(2017) - et al.
Publication bias in clinical research
The Lancet
(1991) A pragmatic view of accuracy measurement in forecasting
Omega
(1986)- et al.
On the asymmetry of the symmetric MAPE
International Journal of Forecasting
(1999) - et al.
Continued usage intention of multifunctional friend networking services: A test of a dual-process model using Facebook
Decision Support Systems
(2014) - et al.
Understanding continued information technology usage behavior: A comparison of three models in the context of mobile internet
Decision Support Systems
(2006) - et al.
Another look at measures of forecast accuracy
International Journal of Forecasting
(2006) - et al.
The e-government evaluation challenge: A south African Batho Pele-aligned service quality approach
Government Information Quarterly
(2011) Accuracy measures: Theoretical and practical concerns
International Journal of Forecasting
(1993)
The importance of complexity in model selection
Journal of Mathematical Psychology
Model comparison methods
COBRA framework to evaluate e-government services: A citizen-centric perspective
Government Information Quarterly
e-GovQual: A multiple-item scale for assessing e-government service quality
Government Information Quarterly
Building citizen trust through e-government
Government Information Quarterly
Linking public sector social media and e-government website use to trust in government
Government Information Quarterly
Model selection for ecologists: The worldviews of AIC and BIC
Ecology
Trust the federal government
On the evaluation of structural equation models
Journal of the Academy of Marketing Science
Interactive e-government: Evaluating the web site of the UK Inland Revenue
Journal of Electronic Commerce in Organizations (JECO)
The cost disease: Why computers get cheaper and health care doesn't
Predictive validity and formative measurement in structural equation modeling: Embracing practical relevance
Digitizing government interactions with constituents: An historical review of e-government research in information systems
Journal of the Association for Information Systems
Choice of structural model via parsimony: A rationale based on precision
Psychological Bulletin
Appropriate use of single-item measures is here to stay
Marketing Letters
The predictive validity of multiple-item versus single-item measures of the same constructs
Journal of Marketing Research
Potential for comparative public opinion research in public administration
International Review of Administrative Sciences
Just one question: If one question works, why ask several?
Journal of Epidemiology and Community Health
Healthcare.gov, Obama's Cratering Popularity, and the Democrats' Big 2014 Problem. TheWeek.com, December 2, 2013
Technology adoption decisions in the household: A seven-model comparison
Journal of the Association for Information Science and Technology
Multimodel inference understanding AIC and BIC in model selection
Sociological Methods & Research
Information technology acceptance by individual professionals: A model comparison approach
Decision Sciences
Building agent-based decision support systems for word-of-mouth programs. A freemium application
Journal of Marketing Research
The partial least squares approach to structural equation modeling
Modern Methods for Business Research
Controlling for common method variance in PLS analysis: The measured latent marker variable approach
The necessity of being comparative: Theory confirmation in quantitative political science
Comparative Political Studies
Global differences in online shopping behavior: Understanding factors leading to trust
Journal of Management Information Systems
Trends in theory building and theory testing: A five-decade study of the academy of management journal
Academy of Management Journal
Digital by default: A guide to transforming government
Coefficient alpha and the internal structure of tests
Psychometrika
Modeling web site design across cultures: Relationships to trust, satisfaction, and E-loyalty
Journal of Management Information Systems
Editorial essay: What is organizational research for?
Administrative Science Quarterly
Model comparisons using information measures
Journal of Modern Applied Statistical Methods
Comparing predictive accuracy
Journal of Business & Economic Statistics
Reintroducing prediction to explanation
Philosophy of Science
Do we really need multiple-item measures in service research?
Journal of Service Research
Reconsidering theoretical progress in organizational and management research
Organizational Research Methods
How presentation flaws affect perceived site quality, trust, and intention to purchase from an online store
Journal of Management Information Systems
Comparing out-of-sample predictive ability of PLS, covariance and regression models
Evaluating structural equation models with unobservable variables and measurement error
Journal of Marketing Research
Cited by (27)
Responsive E-government in China: A way of gaining public support
2023, Government Information QuarterlyModel selection uncertainty and multimodel inference in partial least squares structural equation modeling (PLS-SEM)
2020, Journal of Business ResearchCitation Excerpt :More precisely, while the differences in the criteria values are useful in ranking and selecting models, such differences can often be small in practice, leading to model selection uncertainty (Preacher & Merkle, 2012). For example, in their comparative analysis of E-government performance models, Sharma, Morgeson, Mithas, and Aljazzaf (2018) empirically compare 26 models using PLS-SEM and Information Theoretic model selection criteria. The authors then rank the models based on the raw criteria values and other model evaluation statistics.
Digitally vulnerable populations’ use of e-government services: inclusivity and access
2024, Asia Pacific Journal of Public AdministrationDo you see my effort? An investigation of the relationship between e-government service quality and trust in government
2023, Transforming Government: People, Process and PolicyAre Business Cycles in the Fashion Industry Affected by the News? -An ARIMAX Time Series Correlation Analysis between the KOSPI Index for Textile & Wearing Apparel and Media Agendas-
2023, Journal of the Korean Society of Clothing and Textiles
Pratyush Nidhi Sharma is an Assistant Professor in the Alfred Lerner College of Business & Economics, University of Delaware. He received his PhD from University of Pittsburgh in 2013. His research interests include online collaboration communities and networks, open source software development, and research methods used in Information Systems, particularly Partial Least Squares Path Modeling. In addition, he is also interested in issues surrounding technology use and adoption and human computer interaction. His research has been published in distinguished journals such as the Journal of the Association for Information Systems, Decision Sciences, Journal of Information Systems, Journal of Business Research, Journal of International Marketing, and International Journal of Accounting Information Systems, in book chapters, and in top conferences such as the International Conference in Information Systems (ICIS), Americas Conference in Information Systems (AMCIS), INFORMS, and PLS.
Forrest V. Morgeson III (Ph.D., University of Pittsburgh) is Director of Research at the American Customer Satisfaction Index (ACSI) in Ann Arbor, Michigan. As Director of Research, Dr. Morgeson is responsible for managing ACSI's academic research, advanced statistical modeling and analysis, and the company's international projects and licensing program (Global CSI).Dr. Morgeson also holds the position of Instructor in the Department of Marketing at Michigan State University. Dr. Morgeson's research focuses on citizen satisfaction with government services, cross-national citizen and consumer satisfaction, and the financial impact of customer satisfaction in the private sector. His research has been published in the leading journals in both administration and marketing, including Public Administration Review, Journal of Public Administration Research & Theory, International Review of Administrative Sciences, Electronic Government, Journal of Marketing, Journal of Marketing Research, Marketing Science, Journal of Service Research, Journal of the Academy of Marketing Science, International Journal of Research in Marketing, and Journal of International Marketing. Dr. Morgeson's book, Citizen Satisfaction: Improving Government Performance, Efficiency, and Citizen Trust (Palgrave Macmillan), was released in 2014. In addition, over the past fifteen years Dr. Morgeson has consulted with dozens of government agencies and corporations on citizen and consumer satisfaction, and has delivered lectures and presentations in dozens of countries around the world.
Sunil Mithas is a World Class Scholar and Professor with the Muma College of Business's Information Systems and Decision Sciences Department. Mithas has taught at the Robert H. Smith School of Business at the University of Maryland, and has held visiting positions at the University of Mannheim in Germany and in the Graduate School of Management at the University of California, Davis. He is the author of the books Dancing Elephants and Leaping Jaguars and Digital Intelligence: What Every Smart Manager Must Have for Success in an Information Age. He earned his PhD from the Ross School of Business at the University of Michigan and an engineering degree from IIT, Roorkee. Before pursuing his PhD, he worked for nearly ten years in engineering, marketing, and general management positions with the Tata group. Identified as an MSI Young Scholar by the Marketing Science Institute, Mithas is a frequent speaker at industry events for senior leaders. He has worked on research or consulting assignments with organizations such as A. T. Kearney, Ernst & Young, Johnson & Johnson, the Social Security Administration, and the Tata Group. Sunil's research focuses on strategies for managing innovation and excellence for corporate transformation, focusing on the role of technology and other intangibles, such as customer satisfaction, human capital, and organizational capabilities. His research has appeared in premier journals, including Management Science, Marketing Science, Information Systems Research, MIS Quarterly, Journal of Marketing, and MIT Sloan Management Review. Some of this work has been featured in business publications such as Bloomberg, CIO, Computerworld, and InformationWeek. His papers have won best-paper awards and have received several best-paper nominations. He has also won the best reviewer award at an INFORMS conference.
Salman Aljazzaf is a PhD candidate at the University of Maryland, College Park. His research interests are Business value of IT, IT governance, online platforms, and online marketing.