3.1 Literature Review
For the purposes of this article, the authors queried three widely used scientific databases (SCOPUS®, Google SCHOLAR® and Science Direct®) progressively in three phases: April 2016, September 2017, and September 2018, with the use of the keywords “smart” AND “government” including all the articles that had been published until the day that the searches took place. The combination of these key words was preferred compared to others (i.e., “smart government” or “smart governance” directly) that returned only a few results, while no clear or broad use of “smart government” could be located at the beginning of this study (in the early 2016). Moreover, the authors wanted to uncover all the potential use of “smartness” in government and these keywords addressed this objective. As such, the authors had to keep this keyword combination during the entire study duration, otherwise the results could be different.
This progressive review was performed due to the following reasons: first, because a concurrent Delphi study was performed, where the contributors participated distantly and asynchronously during the two rounds of the methodology. The overall Delphi study was long, because the appropriate experts had to be located and reserve their time for contribution. The authors respected their availability but it took time to collect their inputs. In parallel, the literature review results were updated, in case for some important changes to the concept of the SG to be collected too. The collected outcomes (Table
1) were not as many as were expected, while many of the articles that the queries returned appeared in all the pools of results.
After the literature review search, the authors then proceeded with screening the results, leaving out irrelevant articles and articles about SC alone or government alone as well as editorial articles. In 2016, out of 1,477 results from the three scientific databases used, only 35 articles were chosen after the screening. In 2017, 52 were chosen from 2,016 results, and in 2018, 13 articles were selected out of 4,290 results. The remaining articles returned useful outcomes. First, this evidence shows that the concept of SG is quite new, since it is being discussed from the early 2000’s as the
expected form of government, which will be based on intelligent and well-trained human resources. Moreover, this evolving pool of articles demonstrate that scholars’ view on SG changes: they have recently initiated discussions about SG as quite a “whole-of-government” approach, many of them hardly separate SG from digital (or e-) government, while others combine smart government and smart city (Table
2).
Another interesting finding is that only a couple of journals participate in this SG debate (i.e., GIQ and Information Polity, etc.), which however was not surprising for the authors, since these leaflets lead the government innovation research agenda. Moreover, results from other journals were excluded from this study since they were focusing on SC alone. Even in early 2021, when Google Scholar® for instance returns more than 2 million records with the same keywords, only a few additional journals host articles with the SG subject (i.e., IJEGR, International Journal on Emerging Technologies, International Journal of Management and Human Science (IJMHS) etc.), while others are still discussing mainly about government and SC.
Moreover, there is a lack of conceptual clarity in the identified definitions and in this regard the approach from Vial [
84] was followed to document corresponding challenges according to conceptual definition rules [
84;
86]:
use of primitive and derived terms; uniqueness; unambiguous and clear terminology; consistency with the field; narrowness; new hypothesis avoidance; and statistical testing of the terms.
More specifically, scholars provided with alternative meanings the SG term, which can initially be located in 2001 (Hope, 2001) as the
type of government that focuses on its strategic roles with an organizational design and managerial setup that permit it to perform its roles in an optimally effective and efficient manner. On the one hand, Mellouli et al. [
55], Cellary [
15] and Puron-Cid [
64] name SG as
the extensive use of technology by governments to perform governmental tasks, while Taylor [
79] and Gil-Garcia et al. [
26,
27] relate the terms “smart city” and “government” demonstrating
innovation and intelligence for local or national governments as the means to increase their efficiency and effectiveness, a definition that adopt Eom et al. [
20], Scholl and Scholl [
77], and Scholl and Dwivedi [
76] too, who interrelate this innovation-mix with an enhanced government ability to deal with problem solving and livability at a local level.
Works like from [
33,
46] claim that
smart governments utilize the power of “data” in their attempt to improve public services; to enable an integrated, seamless service experience; to engage with citizens; to co-develop policies; and to implement solutions for well-being of the community. They adopt Rubel's [
67] definition for smart government according to which,
government smart transformation engages citizen participation, information transparency and service improvement. This definition seems to be followed by Ruhlandt [
68], Scholl and Al Awadhi [
75] and by Gil-Garcia [
24], who see a
“whole-of-government” result behind this information integration, while Maheshwari and Janssen [
53] recognize
the need for public organizations’ interconnection too and discuss corresponding interoperability issues. The power of data for the development of a smarter government is labeled “data-smart government” by Janowski [
41], it is adopted by Cornadie and Choenni [
17] too, while Turle [
80] sees shared service provision to be the outcome of this transformation.
Other SG definitions are given by Gil Garcia et al. [
27] and Gil-Garcia and Sayogo [
31], who conclude that SG
is a creative mix of emerging technologies and innovation in the public sector, which can handle
complexity and uncertainty with coordination, continued engagement, access to open data, and shared information. More specifically, they claim that smart government is a continuous effort and not a specific goal, which is supported by a set of emerging technologies (i.e., big data, open government data, social networking, blogs,
Really Simple Syndication (RSS) feeds, web design and programs, mobile government, smartphone applications, cloud computing, and sensors). Similarly, Criado et al. [
18] determine this creative mix to be the combination of social media, open government and big data or in other terms social technologies, while Cazares [
14] views SG as the result of the technological progress in
Artificial Intelligence (AI). Chatfield and Reddick [
16] and Wirtz et al. [
88] examine the
Internet-of-Things (IoT) – based smart government. Ojo and Mellouli [
61] compare alternative approaches of networks as a means for governments to deal with social challenges and view SG
through a collaborative, cross-boundary governance network.
SG is also defined
as the next step for e-government, with the use of innovation and updated services [
71,
73] or
as the next step of open government [
83,
65]
with citizen engagement, interoperability and accountability. Innovation can lead to the production of new public value, that is “value created by government through services, law regulations and other actions” and in this respect a triangle controls the migration from e-government to smart government, which consists of
politics, values, and
evidence. Similarly, Jimenez et al. [
44] introduce an intelligent model for public organizations, entitled “Smart Government Ecosystem Matrix”. This model is a two-dimensional matrix that combines open government features (transparency, collaboration, participation, and interoperability) and smart city context (organizational and management, technology and infrastructure, governance and policy, social, economy, and natural environment) and defines SG -again- as
the next step of open government. Jun and Chung [
45] agree with this approach and introduce a platform (so-called Government 3.0 platform), which shares open data with citizens, government and enterprises and enables public process co-design. Moreover, Jetzek [
42] uncovers smart government's foundation to be interoperability split in three layers (technical, conceptual, and organizational).
From a quite similar viewpoint, Linders et al. [
50] consider this open government's next step as
the transformation of typical public services to pro-active ones, which will drive smart development, meaning that innovations in ICT can serve as an enabler for national development. Additionally, Jansen and Olnes [
40] adopt the European Commission's approach [
21], which understands SG
as the availability of specific e-government key enablers and more specifically of the smart applications eID, eDocuments, eSafe, Single Sign On (SSO) and authentic sources.
Finally, various scholars also see
SG within the SC nexus and a corresponding research-practice consortium has been structured to investigate this relationship. This discussion started quite early in Telecities (Dai, 2003), where SG was one of the primary aspects and continued with the definition of smart governance dimension for smart city [
22,
23,
58,
74,
90,
91]. Under these lens, Gil Garcia et al. [
27] locate a shift of government innovation from a value-based concept into a concrete goal with specific targets, which is used to measure smartness. In this respect, governments utilize innovation to gain a good understanding of the communities (being percipient); to accurately assess situations or people (being astute); to judge sharply (being shrewd), and to decide and respond quickly or effectively (being quick). According to their approach,
SC is only a subset of smart government, where local governments understand the term “being smart” as their attempts to enhance their efficiency, effectiveness, transparency and collaboration with emerging technologies and innovation.Moreover, Nam and Pardo [
57] see SC
as an area of practice for government innovation, which enhances government effectiveness and efficiency, service delivery, process transparency and collaboration; quite similarly, Anthopoulos and Reddick [
7] see
SC as a means for smart government deployment, as well as utilities for recent government challenges’ management and new policy development (i.e., climate change management); Gil Garcia et al. [
28] identify SG
as the source of smart public service delivery within a SC, of city administration and of public engagement; Scholl and Scholl [
77] view
SG as SC government, where the local government implements policies for smart local development and stakeholders’ engagement; Lee and Lee [
48] view
smartness in city's ability to deliver smart services classified in several typologies; Gil-Garcia and Aldama-Nalda [
25] document
smart governance as the facilitator for local economy via the efforts of local governments to adjust local regulatory frameworks for new business attraction and creation, while Gil-Garcia et al. [
29] claim that smart government in its attempt to develop a SC
uses and integrates ICT and innovation in the activities of governing, including internal operations, but also public services and citizen engagement; Alawadhi et al. [
4] compare alternative definitions to justify
the need for better governance to manage SC initiatives, while from a quite similar point of view, Iwan and Poon [
39] consider
SG as government practices that deal with SC challenges (e.g, climate change); Lv et al. [
52] view
SG via the dissemination of typical e-government services via SC platforms; Aguilera et al. [2; and Barns [
13] view
locally produced and collected data as a key-enabler for a variety of purposes, including transparency, performance monitoring, reporting, planning, and policy-making by local governments in SC, and Matheus et al. [
54]
emphasize on corresponding dashboards that are fueled with this data. All these definitions are depicted and compared in Table
2.
Since both SG and smart governance terms are used in literature, a distinction must be given: “
Government occurs when those with legally and formally derived authority and policing power execute and implement activities” and “
Governance refers to the creation, execution, and implementation of activities backed by the shared goals of citizens and organizations, who may or may not have formal authority or policing power” and in this regard smart governance is “an abbreviation for the ensemble of principles, factors and capacities that constitute a form of governance able to cope with the conditions and exigencies of the knowledge society” [
87]. Therefore, it is concluded that
smart governments implement smart governance initiatives [
27].
The above definitions are aggregated and presented in Table
2 and some important findings can be summarized (Figure
1). More specifically, the figure depicts the number of scholars that mention each of the definitions (directly or ‘closely’) with the size of bubble and the position on the axis-Y; and the year (axis-X) when most of the collected references mention each definition. Results show that the SG was introduced as
the next step of open government, which aims to
transform governments to more open, effective, and attractive to their citizens with the use of smart technology and these approaches are mostly discussed in 2014. Later in 2015, scholars identified both a high relation between the
SG and the digital (e-) government (orange bubble with 7 citations) and the
SG and the SC (green bubble with 13 citations). This shift shows either that “smart” is the evolution of “digital” government or that SG is obtained at a local level (alternatively, that SG affects the local communities). Finally, later in 2018, scholars interrelate the SG with
innovation and technology in the public sector, where
data plays an important role.
All the above definitions could be combined to the following one:
SG concerns the transformation of government to an open, more effective, efficient, and transparent one, with the mix of emerging technologies and innovation, which goes beyond typical digital and open government approaches. Moreover, SG is seen under the lens of local government, where SC government is performed at a specific geographic location and community. This definition partially provides with answer the
RQ1, without however avoiding conceptual clarity mistakes. In this regard and following the process of Vial [
84] for digital transformation's definition, the above SG definition could be revised to
SG concerns the process that aims to improve the government's performance and operations with the combination of digital technologies (data, computing, and networking) and innovation, which is more likely to take place at a local level. This definition does not conceptualize the SG, something that is being performed in the following sub-section.
3.2 The SG Conceptual Structure
The literature evidence that was presented above return significant findings about the SG conceptualization too, but only a few SG conceptual models could be in the beginning of this study (early 2016). The authors were inspired by [
10] and [
30] because they both included more components for the SG conceptualization: Anthopoulos [
10] attempted to recognize the conceptual entities that define the SG in general and to propose a multi-tier model, with components that interrelate and interact; Gil Garcia et al. [
30] define SG conceptual entities, but their definition and analysis is performed with a focus on SC. As such, the combination of these two conceptual models can cover both a generic and a local government approach. However, the model had to be more complicated for the following reasons: first, the SG was met in the literature with the key-terms
smart government, smart governance, and
smart city government, which are being approached with the following labeling:
digital (or e-)
government;
open government; and
smart city (even in the form of smart community). These labels can be considered as the
terms or more likely
the dimensions that synthesize the
first layer of the SG conceptual analysis (outer ring), since they explain better the initial key-terms that were directly excluded from the literature review (Table
2).
Moreover, these three dimensions are explained with quite the same terminology: they deal with technologies (more likely ICT-based) that address economic growth (at local or regional or national levels); government effectiveness and efficiency; citizen engagement in political issues; new forms of policy making; the delivery of modern smart services; innovation development (in or supported by) government; accountability and transparency; the technology itself and more specifically the necessary key enablers for government digital transformation; and smart city government.
These second-level terms can be considered as the entities that comprise the second layer of the SG conceptual analysis (medium ring), since they explain what the terms aim to achieve. Finally, these nine entities are being implemented with the combination of ICT innovation, emerging technologies (e.g., artificial intelligence (AI); autonomous objects; Internet-of-Things (IoT); 5G networks etc.) and data, which concern the elements for all the SG entities’ (and in this regard, even the SG itself) implementation and comprise the third layer of the SG conceptual analysis (inner ring). Some contents of this conceptual model appear to be duplicate (e.g., smart city (outer ring) and smart city government (medium ring); innovation (medium ring) and ICT innovation (inner ring). However, they concern different aspects in the SG definition, they were identified by literature review and they could be interpreted with the following text (with a flow from the outer to the inner ring of the model): (a) SG concerns government performance enhancement within a smart city as well as smart city government, which is enabled with data, ICT innovation and emerging technologies; On the other hand: (b) SG concerns the process of improving government performance with innovation activities, which are enabled by data, emerging technologies and ICT innovation in general. After all, this model came out from literature evidence and it must be clarified and tested with regard to its validity and contents’ mutual strengths.
These two models have similarities and differences, while some associations can be extracted from their contents (Table
3) that can be validated: (1)
Government accountability concerns a topic of
openness in government. (2) SG considers smart public service design around citizen needs and, in this regard,
smart services can be associated with
citizen centricity. (4) SG enhances decision and policy making with the use of -even real time collected data and in this regard,
data is associated with
evidence-based decision making. (5) SG prioritizes and supports economic growth at the local level, with the support of a startup launch and with the simplification of the business-oriented services. As such,
economic growth can be associated with
entrepreneurialism. (7–8)
Emerging technologies is quite an ambiguous term and its context changes in time. For instance, during this article's development emerging technologies evolved mainly to AI, cloud, IoT and blockchain, while during its completion new ones appeared (e.g., edge computing).
Key-enablers on the other hand are the technological artefacts that make Nevertheless, it is a term that describes new and evolving ICT and aligns to
technology savviness context, which deals with corresponding knowledge and competencies in government to utilize emerging technologies. (8–9, 10)
Innovation and
ICT Innovation in government is the process for defining or updating services and internal processes with the use of the ICT, while
creativity deals with human characteristics that enable innovation (managerial leadership, support, and commitment). (9–11)
Interoperability deals with the ability of different -either in technological or organizational terms- government systems to interact and exchange information, while
integration addresses inter-organizational information sharing. Finally, (10–12, 13, 14)
smart city government concerns the governance of a smart city in terms of successful project and program development, as well as of deliverables’ operations management that meet the predefined local challenges; while,
sustainability is a critical challenge that refers to the environmental, economic and social viability;
equality is another challenge that deals with social inclusion; and
resiliency is the ability of government to respond against emergent and disastrous effects. The rest of the elements are the same (3–3: citizen engagement; 6–6, 7 effectiveness and efficiency).
The above outcomes show that the model introduced by [
10] appears to be more generic, since it discusses SG in general and not with a strong focus on SC, while it avoids components’ hierarchical adjustment with the use of 3-rings and in this regard, it is considered appropriate to define the SG (Figure
2). Rings’ contents are “loosely” interrelated and their position in the figure is indicative, which means that any of them influence the upper level's contents, while any of them is also affected by the lower level's contents. For the purposes of this study, the authors preferred to adopt and test an existing conceptualization model, rather than introducing an alternative (e.g., a Venn Diagram with two intersecting circles for “smartness” and “government” that could also be applicable).
However, the authors wanted to test the selected model about its efficiency to discuss the fundamentals of the SG and SC terms, which have been found in literature that some scholars interrelate. As such, the authors defined a three-tier framework (Figure
3) with associated items between the selected model and others that were in literature and associate SG SC. More specifically,
the selected model is located on
layer 1 (the SG framework); an
SG - SC framework that was introduced by Jiménez et al. [
44] and presents the synthesis of the SG and SC is placed on
layer 2; and a
unified development framework for the SC that was introduced by [
9] is located on
layer 3 (the SC framework). Connections between the three models are depicted on Figure
3, which come from the interrelations of Table
4.
Jiménez et al. [
43] consider SG as the highest modernization phase for the public sector in a smart city, which is evolved from open government and in this regard, they introduced the SG-SC framework (layer 2), which they name as Smart Government Matrix. This framework is analyzed in layers that from top to bottom are as follows: they consider SG as
open and smart ecosystem (layer 1) as a system of systems within an urban space, where actors and institutions, systems, environments, energy, citizens, infrastructures, information, policy and technology co-exist and interact. This ecosystem is the outcome of an
open government -in terms of transparency, collaboration, and participation-
and smart city -in terms of the existence ICT networks, intelligence, sensors and software at a local area- (layer 2); that followed an
interoperability strategy, documented on an interoperability framework, which addressed the organizational dimension of the local government (layer 3); which defined the appropriate
ontologies for big and open data collected from the urban area (layer 4); as pieced of
data received from sensors (layer 5).
On the other hand, Anthopoulos et al. [
8] introduced a unified SC model (USCM), which synthesizes several SC conceptual models and it is analyzed in the following layers: (layer 1) physical
facilities located in a city (e.g. for transportation, energy, water, buildings, etc.); (layer 2)
services that are being offered in the city and utilize the above facilities (e.g. health, education, tourism, etc.); (layer 3) SC
governance, which is performed with
planning & management and follows a specific SC
architecture that enables monitoring and benchmarking, with
data that are being collected from the city.
According to the above explanations the open and smart ecosystem in the middle is connected on the left with polices and smart services, whose outcome is -among others- economic growth, while on the right with facilities, services, and people. Similarly, open gov is connected on the right with governance and on the left with accountability, citizen engagement, and effectiveness and efficiency and SC relates to SC government, emerging technologies, key-enablers, and innovation and ICT innovation. Additionally, interoperability of the SG & SC framework has a clear connection with interoperability on the left and with planning and management on the right, which are performed on a specific architecture. Finally, the connection between data and ontologies is clear between all the tested frameworks.
These connections validate that the selected model is efficient to discuss the fundamental theories of SG and can be considered appropriate to lead to a combined SG-SC development, which in turn can lead to SC development. The total schema can be also considered an SG-SC development framework, where SG and SC interact.
Additionally, the selected model must be tested for its efficiency to support the future research evolution of SG and SC. In this respect, the model was checked for interrelations with the common research agenda that was identified by [
7]. More specifically, this two-tier framework consists of the selected model (layer 1) and SC & SG evolution framework (layer 2). The
meta level of layer 2 represents the SG & SC future challenges that are being or expected to be discussed and, in this respect,
key-enablers, data and emerging technologies appear in the identified literature evidence and they are directly associated. About the role of
government and SC, the government interventions for
livability, sustainability and resilience, city management and
city competition are all associated with the following elements on the left:
economic growth, smart city government, effectiveness and efficiency, smart services, and
data. Finally, the topics that deal with e-government and SC can be associated with the model's elements on the left:
citizen engagement, service co-design and
digital neighborhood relate to
citizen engagement, accountability, policy making and
interoperability;
standardization is associated with
emerging technologies, key-enablers, innovation, and
ICT innovation.
The outcomes (Figure
4) (Table
5) return the existence of close connections between the entities of the model and the SG-SC evolution framework, which validate the appropriateness of the selected model, while they introduce an SG-SC research development framework.
The above outcomes validate that the selected model of (Figure
2) is efficient to describe the SG and could be considered
unified, since it has “loosely-coupled” connections between the terms, the entities and the elements that the literature review returned; it is capable to discuss the fundamentals of the SG and SC terms (Figure
3) (Table
4); it contains the theoretical foundations to describe the SG and to lead to SC-SG evolution, while it has the capacity to support SC-SG research evolution (Figure
4) (Table
5). The selected conceptual model, together with the previously given SG definition, provides with answer the RQ1.
Summarizing the current knowledge on the terms, “smart government”, “smart governance”, and “smart city government” to answer research questions. The conceptual entities that define SG from the list of definitions from Table
2 and Figure
1 are: the combination of emerging technologies, data, and innovation which are the elements of the first layer of SG in the conceptual model [
10]. This framework builds upon a relationship that emerged through our analysis across overarching three rings of the SG conceptual model.
A view to the selected and tested conceptual model returns useful findings: the first, outer ring shows that SG utilizes the outcomes from both digital and open government, while it deals with SC due to its easier applicability at a local level. The intermediate, second ring with the SG entities, due to the rapid technological emergence, government must take into consideration the complexity of SG and SC framework when implementing smart governance. In connection with the relative strengths between the SG conceptual entities is that these interrelated entities can be employed by the government that forces the growth and development of a digitally transformed and open government or of a SC. Based on the research conducted, SC is a subset of SG and relies on emerging technologies, data and ICT innovation. The three rings of SG can support the relationship between SG and SC, and multiple elements must cooperate to achieve better, sustainable outcomes with the use of emerging technologies, ICT innovation and data.
3.3 The Delphi Methodology
The previously presented conceptual model consists of three rings that contain several interrelated elements and entities, it defines the SG, and it has the capacity to lead an SG-SC development. Nevertheless, it is important to measure the relevant strengths of the model's components since the collected values can depict the relative importance of each of the component to the SG definition and evolution. In this regard, the model was decided to be tested with the contribution of experts, who would express their opinion on the relative importance of the model's components, via a two-round Delphi method [
60]. The research methods used in this paper have been successfully employed in past research [
60,
72]. For instance, the literature review was broad and systematic, it lasted a long time, and it explored a diverse number of sources. Moreover, the sample of the experts on both the SG and the SC domains was not large (a fact that makes Delphi an optimal choice when having a small sample size) due to the complexity in locating experts with the appropriate combination of skills and availability to participate.
The underlying principle of the Delphi method is that group-based forecasts are more accurate compared to individual forecasts [
60,
72]. The experts were asked to validate the conceptual SG model (Figure
2(b)) with a pair-comparison of the elements of the intermediate ring from the perspective of the outer ring (e.g.,
how important is “Economic Growth” compared to “Policy Making” from the “Open Government” perspective?); and a pair-comparison of the elements of the inner ring, from an element-perspective (e.g.,
how important is “Emerging Technologies” compared to “ICT Innovation” from the “Economic Growth” perspective?).
Numerous combinations are generated (1) for the requested comparison and more specifically, 45 combinations for each of the three perspectives of the outer ring's dimensions (2) that makes 135 comparisons for the intermediate ring; and three combinations for each of the 10 perspectives (3) and for each of the three dimensions, for the inner ring's entities that makes 90 comparisons for the inner ring. For this reason, and in an attempt to simplify the contribution of the experts in terms of time and concentration, the
Analytic Hierarchy Process (AHP) was selected for the definition of the questions. The AHP [
70] is a
Multi Criteria Decision Making (MCDM) Technique that can structure any complex, multi-criterion, multi-person problem hierarchically, and identifies the strength with which one alternative dominates another with respect to a given criterion. Moreover, since the AHP can also handle qualitative inputs and subjective judgements of individuals, it appears to be a suitable process for this kind of mutual comparisons and its application reduced the number of comparisons too (from 225 to 135).
The AHP was used to structure an online questionnaire that was filled-in during the first round of the Delphi process. The choices that were used to the AHP concerned the model's components, which had to be compared in pairs from different perspectives, as it was explained earlier. The Expert Choice® software was used to validate the collected inputs from the participants.
A group of experts was assembled late in 2016, after the end of a relative scientific workshop, where the initial concept of this study was discussed. All the experts were approached during the first Delphi round with an e-mail invitation and a brief explanation of the interview (a process that lasted until the end of 2017) and they were asked to fill in an online structured questionnaire. The expert's names and affiliations were kept confidential in order to get more candidate responses to the questions. This group consisted of 24 SC experts, coming from universities from all around the world. The identification process used the following criteria:
(a)
Being involved in a publication, research or case study that is relative to either the SG or the SC or both.
(b)
Coming from different countries and continents. In this regard, five (5) experts came from the USA, one (1) from Latin America, ten (10) from Europe, five (5) from Asia, two (2) from Australia and one (1) from Africa, while they were in 15 different countries.
(c)
Not all the initially invited participants accepted the invitation from the beginning. In some cases, because of a lack in their availability, it took more than six months for the interview to be performed and validated. As such, the first round took almost one year to be completed, a period where literature was enriched with corresponding articles, which were checked progressively for potential updates to study.
(d)
To secure experts’ relevance with the survey, the model was communicated and explained to the experts before each interview.
(e)
Questions were codified on the questionnaire (
Qi: i = 1 to 135 for the questions;
Ej: j = 1 to 16 for all the model's components), to collect as much information as possible against each topic and issue. The answers were decoded after the completion of the interview. The first interview contained a traditional Delphi round. When all interviews were completed and the results were processed, the second Delphi round was initiated (early February 2018), and the same experts were invited to confirm the first round's outcomes, as the Delphi methodology suggests (Figure
5). The second round was also long and ended in February 2019, a period during which -again- literature was checked for updates.
3.3.1 Round 1.
To identify the components’ relative strengths, they were compared in pairs with the use of an online questionnaire
1 that was provided to the interviewed experts, who were asked to present their opinions for their relative importance. The model's components concerned the AHP choices, which the experts had to compare in pairs, from different perspectives as it was explained earlier.
Participants used Likert scale values (from 1 to 9) (Table
6), which expressed their preference between two alternative components (I and J), according to a followed perspective each time. The registered preference represented the relative importance between the compared components. When all the experts’ completed their contribution, data were consolidated and were statistically analyzed with the use of SPSS® software and several statistical methods. Two limitations were recognized during the statistical analysis of the collected data sample. The first concerns the small sample size, while the second refers to the following observation: Table
7 illustrates the percentage of the collected value one (1) that the experts gave to the comparisons either to each one of the entities (E1, E2, E3) and the Q27 (inner ring) section or in general. This outcome demonstrates that one third of the experts consider all the compared variables (model components) to be equally important.
For the authors to decide the type of the statistical analysis that they had to follow, they defined the primary assumptions for using parametric statistics and tests: first, data are considered independent amongst each other. Second, each of the experts answered without being aware what the other experts answered to the same question. Third, since all the variables were ordinal, one of the normality assumptions does not comply. As such, the non-parametric methods appeared to be more suitable for the analysis.
Two data analysis methods were performed: after all the mutual entities’ test, the authors checked the rest 27 comparisons (for the inner ring). Each value represents the rate for Emerging Technologies, ICT Innovation and Data in the context of the nine main variables (Economic Growth, Policy Making, Citizens Engagement, etc.). The lowest p value was equal to 0,095 for the comparisons of the nine variables in the context of the E1 dimension (Smart City Government). As such, there is not sufficient information to reject the null hypothesis (that W = 0, disagreement). The conclusion is that the experts’ rate for E1 expressed a complete disagreement.
About the dimension E2 (Open Government) the lowest p value was equal to 0,180 for all the comparisons, except from the case where the economic growth versus all other variables was compared (p = 0,004). So, there was not sufficient information to reject the null hypothesis (that W = 0, disagreement). The conclusion is that the experts’ rate for E2 presented no agreement, except from the comparison of the economic growth entity, which is translated as follows: Economic Growth entity was more important than the others, from the Open Government perspective.
Regarding the dimension
E3 (
Digital Government) the lowest
p value was equal to 0,108 for all the comparisons. So, there was not sufficient information to reject the null hypothesis (that W = 0, disagreement). The conclusion is that the experts’ rate for
E3 expressed no agreement. The above analysis returns that
Economic Growth entity was more important than the other components from the
Digital Government perspective. The investigation for agreements about the model's inner ring returned important findings: Table
8 shows that the experts expressed agreement that
ICT Innovation versus
Data comparison is more powerful from the perspectives
Citizens Engagement, Accountability, Key Enablers and
Smart City Government. This part of the analysis returns that
ICT Innovation is the most important element compared to the others (
Emerging Technologies and
Data) from the perspectives of
Citizens Engagement, Accountability, Key Enablers, and the
Smart City Government dimension.
Finally, an examination of the comparisons between all the entities of the intermediate ring was performed, with the use of the
Kendall's W test for the level of the respondents’ agreement to be extracted. Disagreement is assumed when
W = 0 and agreement is considered when
W = 1. Results are depicted (Table
9), where no values for the variable
Government Effectiveness & Efficiency exist due to the structure of the AHP questionnaire that contained only one comparison for this entity. So, the Kendall's W test could not be applied.
To interpret the evaluation that the experts gave to each the entity, the authors decided to use the
mean rank: this method compares the ranking that each respondent gave to a component, with the rest of the values that this entity collected during the survey. Practically,
a low mean rank value denotes variables that are close to
1 (
equally important). As such and with the use of
Friedman's ANOVA process, figures like Figure
6 are extracted, which “classify” the entities. The extracted figures contain diagrams with two
axes (each one is a model's dimension) and
bubbles, the
size of which represents the
mean rank of the entities. For example, in Figure
6 the horizontal axis depicts the mean rank of each question in the
Smart City dimension (E1) and the vertical axis the mean rank of the same question in the
Open Government dimension (E2). As such, the yellow bubble that represents the comparison
Economic growth VS Key Enablers belongs to the quartile I, as it received values (25.15, 24.75). The size of this bubble corresponds to the mean rank of the question in the Digital (e-) Government (E3) which is equal to 21.54. Bubbles that are big and located in the
quartile I express a strong importance of the first variable compared to the second variable, and this importance holds for all the entities. Similarly, bubbles of small size in the
quartile III expressed same importance of the compared variables in all the entities.
This method generated Figure
7; Figure
8, which return the following outcomes:
citizen engagement appears to be the most important entity from the
digital government perspective, while
accountability and
government efficiency follow. The
Economic Growth and
Key-enablers entities on the other hand, appear to be the less important from the digital government perspective. Similarly,
citizen engagement is the most important entity from the
open government perspective, while
economic growth follows. On the other hand,
key enablers and
policy making are the less important entities from the
open government perspective. Finally, from the
smart city perspective, e
conomic growth is the most important entity with
citizen engagement to follow, while
key enablers and
interoperability appear to be the least important entities.
3.3.2 Round 2.
After the completion of the above analysis about the results from round one, a second structured online questionnaire
2 was given to the same experts. The questionnaire asked the experts to express their agreement (or not) with the use of the Likert scale (1: absolute disagreement to 5: total agreement) regarding the collected results from the first round. The online form contained 18 questions (six per each of the three model's dimensions:
E1-Smart City; E2: Open Government; E3: Digital Government). This was the final round of interviews and the results were consolidated and statistically processed (Table
10; Table
11).
The authors assumed that collected data follow a normal distribution in the evaluations (the Kolomogorov – Smirnov test was applied) and they checked the mean values that were equal to 4 (agreement on the results of the 1st Delphi round). The corresponding p-values showed that the null hypothesis (μ = 4) could not be rejected. Moreover, the outcomes showed that (i) for the Smart City (E1) dimension the mean value (3.64) expresses that most of the experts agree with this outcome, while 50% of them (Median) ranked with a value up to 4 this question; (ii) for the Open Government (E2) dimension the mean value (3.64) shows that most of the experts agree with the collected outcome and 50% of them ranked with a value up to 4; and (iii) for the Digital Government (E3) dimension the mean value (3.5) demonstrates again that most of the experts agree with the first round's outcome and 50% of them ranked with up to 3.5 this question.
Regarding the rest of the questions, Table
11 shows interesting results too:
Citizen Engagement was considered more important than the other entities, in all the dimensions. The
Economic Growth was also evaluated significantly important in all the dimensions, except from the
Digital Government one, where
Accountability was ranked more important. Moreover, the questions that referred to the less importance factors (for each dimension), were not supported by the participants. So, the question “
Key enablers and interoperability are the less important factors for the smart city dimension” was ranked with 2.86, since 50% of the respondents answered up to 3.00. The same observation stands for the question “
Key enablers and policy making are the less important factors for the open-government dimension”, where the experts ranked it with 2.79 and 50% ranked it with up to 2.50; and the question “
Economic growth and key-enablers are the less important factors for the digital government dimension” the average score was 3.14 and mode equals to 3.50. The previous statements were supported by the experts’ answer on the question “Figure
6: points that lie in the quartile III are of limited importance for all entities. Express your agreement with the findings”. They did not agree as they were evaluated with the average score 3.36 and with mode equal to 3.00. The authors concluded that this disagreement is related with the role of
Key Enablers, which was evaluated of low importance during the first round. The same finding was extracted from the corresponding question for the Figure
7, “Figure
7: points that lie in the quartile III are of limited importance for all entities. Express your agreement with the findings”. The experts evaluated the question with an average score 3.64 and on mode 3.50.