Government influence on e-government adoption by citizens in Colombia: Empirical evidence in a Latin American context.

This study aims to identify government influence in the adoption of e-government by citizens (AEC) through a case study analyzing actions in Antioquia, Colombia, to increase AEC in annual vehicle tax filing and payment services. We classified these actions employing institutional theory, institutional interventions, and legitimation strategies. An analysis correlating AEC actions (including the COVID-19 containment measures) with over 16 million transactions in these two services during 2015-2020 found a strong government influence on AEC. We established coercive pressure and conformance to the environment as important predictors of AEC, but the COVID-19 containment measures only influenced electronic tax payments. Service type was also an essential predictor for these services; however, mobilization was not. Increasing AEC should be considered a necessary objective for public administrations, especially in developing countries that face shortages of resources and facilities.

Introduction

The use of information and communication technologies (ICT) has transformed interactions with all government stakeholders [1], particularly with citizens [2]. The use of the Internet and other digital means to access government information and services has been defined as electronic government or e-government [3-5]. It emerged in the early 1990s and uses web-based IT as an important part of outreach to citizens [6, 7]. E-government has been a key catalyst in the transformation agenda of new public management for reinventing public administration and making it more efficient and effective [8, 9]. The value of e-government is recognized worldwide, and research has identified several benefits related to its implementation and use, such as transparency, efficiency, cost reduction, service delivery improvement, accountability, and reduction of corruption [8, 10–13]. The UN placed it at the center of the 17 Sustainable Development Goals for 2030 [14], and in recent years, e-government has demonstrated a positive global trend towards a higher level of development [15]. Nevertheless, the adoption of e-government by citizens (AEC) is crucial to achieving such benefits [11]. Indeed, e-government creates public value when citizens adopt it and also promotes private value acquisition for users, encouraging continuous use [16, 17].

According to Kumar et al. [18], AEC starts with citizens’ decision to use online services (for example, choosing this channel over visiting the government office). The next stage of AEC involves increasing the frequency of use. However, various scientific studies indicate that AEC levels remain low [19, 20]. Savoldelli, Codagnone, and Misuraca [21] have referred to this as the e-government paradox, implying that public money and effort are wasted if investments in technology for e-government implementation do not translate into appropriate adoption responses from citizens.

Consequently, increasing AEC should be considered a necessary objective for public administrations. Therefore, this subject has been broadly studied, and various perspectives have been proposed [22, 23]. One perspective focused on the inhibitors of AEC [24], viewing non-adoption as an option for some citizen groups [25]. Another studied the digital divide’s effects on the AEC process, acknowledging that not all citizens would benefit from such services as some may not have access to the Internet or do not know how to use such services [15, 26–28]. A third group examined channel choice and public service delivery, comparing e-government to traditional service delivery channels, including telephoning or visiting government offices [29].

Furthermore, Lamberti, Benedetti, and Chen [30] determined that citizens’ benefits were essential determinants in choosing AEC over offline channels. Moreover, some studies on AEC for various services found that adoption levels have varied according to service type [10, 31, 32]. Finally, strong support has been found in 80% of AEC studies [24, 33, 34] for technology adoption theories and models, such as the Technology Acceptance Model (TAM) [35], the Unified Theory of Acceptance and Use of Technology (UTAUT) [36], the Diffusion of Innovation Theory (DOI) [37], and the Information System Success Model (ISS) [38].

Technology adoption theories and models have contributed to e-government, presenting the bases for implementing electronic services and pursuing adoption [24, 33, 34]. Thus, technology characteristics are strong determinants of AEC. Different technology aspects (usefulness, usability, look and feel, reliability, data quality, content quality, information availability, performance, integrity, efficiency, accessibility, compatibility, security, and privacy) are recognized as essential for persuading more citizens to use e-government actively [39-42]. However, AEC goes beyond technology adoption and relates to institutional and political issues; the relationship between the government and the citizens is also influential [21]. Government entities, the legislature, and the legal system are considered political institutions [43, 44], providing ways to reduce uncertainty and increase cooperation in the political arena [45] through guidelines for human action or appropriate behavior in society [44].

Consequently, despite its responsibility on the offer side, the government also has the potential to influence the demand side and increase AEC. For example, the E-Government Development Index, published by the United Nations [15], highlights the key role of governments, not only in increasing the services offered (the Online Service Index) but in improving the enabling conditions for AEC (the Telecommunications Infrastructure Index and the Human Capital Index). Also, Savoldelli et al. [21] noted shifts in the literature over time from technical and operational to institutional and political issues and found institutional and political barriers to be among the main factors affecting AEC. The supply of high-quality, attractive e-government services is the necessary but insufficient condition of use. Assuming citizens will use e-government services automatically once they are available is a serious mistake [46]; hence, government institutions must create an environment conducive to increasing AEC.

This study used an institutional framework as the theoretical basis. Considering the various pressures, developmental goals, and the existing social norms and beliefs favoring legitimacy over efficiency that are relevant to the operationalization of e-government, an institutional perspective is appropriate for studying e-government initiatives and their adoption [47, 48]. Previous literature has similarly proposed institutional theory as a powerful lens to understand e-government [49, 50] and provided a framework to investigate the phenomenon in a wider context with multi-level analyses [48].

Institutional theory literature has identified three kinds of institutional pressures that can influence organizations and individuals: coercive, mimetic, and normative [51, 52]. King et al. [53] listed the institutions influencing IT innovation, highlighting the power of government entities in the adoption process (e.g., national government agencies, provinces, prefectures, states, and municipalities). Scott [52] recognized that states rely on institutional pressures to exert influence, and state actors are more likely to employ coercion. The legal framework, as an example of coercive pressure, is determinant in the context of AEC [54]. Several studies have confirmed that the government’s entities use this coercive pressure to encourage citizens and other groups to embrace e-government [13, 49, 55–58], even in mandatory use contexts, as Denmark and the UK have defined [59]. Consequently, we have defined that coercive pressure predicts AEC as our first hypothesis.

Adoption and legitimacy are connected: as innovation spreads, adoption provides legitimacy [60], and from an institutional perspective, adoption becomes a way of demonstrating legitimacy [61]. Suchman [62] presented conformance to the environment as the easiest legitimacy acquisition strategy where the organization modifies internally to respond or satisfy the constituent tastes. Some studies have explored legitimacy’s importance in e-government [47, 63–65]. For example, Trakhtenberg [66] proposed viewing AEC as the process of securing institutional legitimation for each of the government agencies whose services are offered by the e-government. Overall, we have defined that government conformance to the environment (citizens’ needs) predicts AEC as our second hypothesis.

Furthermore, King et al. [53] described the institutional interventions (knowledge building, knowledge deployment, subsidizing, mobilization, standard-setting, and innovation directives) employed by various institutions to influence IT innovations. They built a model of potential institutional actions with two dimensions: the influence and regulation that institutions might exert and the “supply-push” and “demand-pull” forces that provide a context for those actions. Some e-government studies have referenced this model [67-69]. Governments can effectively employ institutional interventions to obtain desired behavioral changes and increase AEC [70, 71]. For example, mobilization has been recognized as an important element in AEC. Designing robust marketing campaigns to promote the benefits of e-government engagement can influence channel choice and behavior, guiding citizens to choose cost-effective channels to access e-government [13, 46]. Governments should run advertising campaigns informing people about the services to help develop positive attitudes towards e-government among citizens and the intention to use and the actual use of e-government services [3, 72, 73]. Indeed, good marketing campaigns are considered directly influential in promoting the intention to use e-government [59]. Consequently, we have defined that mobilization predicts AEC as our third hypothesis. Also, we defined that knowledge deployment predicts AEC as our fourth hypothesis.

Finally, some studies focused on cross-sectional analysis and suggested longitudinal analysis in future research [2, 74]. For example, Hofmann, Räckers, and Becker [22] recommended longitudinal studies to eliminate short-term effects and understand the e-government domain development. Therefore, this research used a case study in Colombia, a developing Latin American country. We analyze the government actions to increase AEC in the annual vehicles’ tax filing and payment (offered by the same government entity using the same physical and electronic channels), correlating with the adoption behavior from 2015 to 2020. Additionally, we analyzed the impact of the measures implemented in 2020 to contain the COVID-19 pandemic on AEC. Our research question is: how does government influence AEC? We built our hypothesis using institutional theory [51], institutional interventions [53], and legitimation strategies [62]. Studying AEC in developing countries is essential for providing theoretical and practical contributions to the literature [23, 75].

The following section presents the methodology, followed by a section on the results, discussion, and conclusions.

Methodology

The case study

Colombia is a Latin American developing country located in South America’s northern region and has over 50 million inhabitants, a capital district, and 32 government departments. According to ICT Ministry of Colombia [76], 64% of the population has Internet access at home, and 74% has access to a smartphone. However, 97% of the population mainly uses the Internet to communicate and interact—90% use it daily, while 27% utilize it for transactions. Additionally, social media platforms are preferred among the population—88% use Facebook, 87% use WhatsApp, 48% use YouTube, 34% use Instagram, and 20% use Twitter.

Latin American governments are emulating the actions taken years ago by developed countries and need to move towards more locally-tailored technologies to enhance e-government [77]. For example, Acosta, Acosta-Vargas, and Lujan-Mora [78] found an absence of compliance with the Web Content Accessibility Guidelines (WCAG) 2.0 among websites of Latin American countries that offer e-government services, exacerbating digital divide consequences. Consequently, the Colombian government has applied different strategies to improve e-government. For example, on May 25, 2019, the government issued Law 1955, as part of the National Development Plan 2018–2020, which defined digital government in terms of institutional management and performance policy; alongside this, the Digital Government Direction was created to coordinate all government entities’ efforts [79]. Complementarily, in July 2021, the Colombian government enacted Law 2108, declaring the Internet an essential and universal public service [80].

Colombia was the Latin American leader in the UN e-government survey [81] in 2010 due to the unique strategies of its government. The Organization for Economic Co-operation and Development (OECD) placed Colombia third in its 2019 Digital Government Index [82]. Similarly, in the national Digital Government Performance Report for 2018, Colombia’s sixth-largest department, Antioquia (capital: Medellín), had the second-best performance [83]. The UN E-Government Survey for 2020 also accords a high e-government Development Index (EGDI) of 0.7164 to Colombia. Other countries in the region are also progressing rapidly; Argentina, Brazil, Chile, and Costa Rica significantly improved their EGDI values between 2018 and 2020 and transitioned to the “very high EGDI” group [15]. Among the three components of Colombia’s EGDI (human capital, online services, and telecommunication and infrastructure), the first two elements ranked very high, but telecommunications and infrastructure still need attention.

Our case study centers on the Department of Antioquia’s annual vehicle tax filing and payment services, which are available via a website (www.vehiculosantioquia.com), a mobile application, and physical offices. The department has total control for these services and can issue regulations ruling the process, the service design, technology selection, and the channels through which the services are offered. For 2020, the Department of Antioquia had 1.5 million vehicles, and approximately 1.2 million are subject to this tax (motorcycles with engines smaller than 125 cm3, public transport vehicles, and government vehicles are exempt). The annual payment rate is 79%, with the remaining 21% being defaulters. Almost 25% of the department’s annual income comprises overdue amounts, including penalties and interest.

This case study refers to the significant increase in AEC between 2016 and 2020 for these two services (see Table 1). The adoption of tax filing e-service rose from 52.07% to 78.94%, representing an increase of 51.60%. Additionally, the adoption of tax payment e-service went from 12.64% to 41.24%, representing an increase of 226.27%. The objective is to study AEC between 2015 and 2020 to identify the actions performed by the government to promote AEC and determine which of these actions could be used as AEC predictors. This case study is relevant because AEC is generally low in Latin America, and the adoption of these services stands out among regional adoption rates. Furthermore, studies revealed that approximately 7% of citizens have attempted to use e-government services, with only 4% conducting 100% online [84, 85].

Table 1

Annual adoption of e-filings and e-payments.

Year	Total filings	E-filings	Adoption	Total payments	E-payments	Adoption
2015	1,365,811	707,935	51.83%	678,674	90,484	13.33%
2016	1,350,360	703,167	52.07%	730,165	92,290	12.64%
2017	1,658,132	1,070,757	64.58%	805,175	140,307	17.43%
2018	1,778,395	1,203,783	67.69%	767,274	159,456	20.78%
2019	2,081,664	1,490,005	71.58%	949,449	268,346	28.26%
2020	3,001,148	2,369,015	78.94%	1,076,093	443,728	41.24%
TOTAL	11,235,510	7,544,662	TOTAL	5,006,830	1,194,611

Tax payment requires a previous tax filing, and e-payments require e-filings. Multiple filings for one payment are typical. Also, having filings without payment is usual. For example, a tax filing to pay with discounts should be updated if not paid before the deadline. Another example is presented when the vehicle’s information should be updated (e.g., the new owner or updated value). Thus, the payment is performed based on the last tax filing. It is impossible to pay through e-government channels once the tax is filed physically in an office, and this means a selection process affects e-payments defined by the filing presented by the electronic channels.

Additionally, it is vital to mention that the national government generates information every year to configure the basis for tax calculation. This configuration is performed in the first weeks of each year, during which the system is inaccessible, but some manual filing may still be generated. Additionally, the government currently closes and balances the previous year’s records.

Data and methods

This is an explanatory case study [86, 87] that used interviews and documentation analyses combined with quantitative analyses of the historical data from 2015 to 2020 (more than 11 million filing records and more than 5 million payments, as illustrated in Table 1) to identify how the government influences AEC (our dependent variable). We employed descriptive methods, including aggregated results for the use of e-government services in graphical and tabular output [88, 89]. Additionally, we conducted multiple regressions for the data correlation analysis. Similar approaches have been used in previous research related to AEC [90, 91].

We conducted interviews to gather the information, verify data quality, interpret the information, and discuss and confirm the findings [86]. This method provides an opportunity for interaction and dialogue between the interviewees and interviewer that can be used to clarify, explore, and raise new issues [92, 93]. We prepared a resume for every interview validated in the following interview. Additionally, we created a repository to consolidate all the information we gathered. In summary, we conducted 27 interviews from 2018 to 2021 with the general manager (GM), the operational manager (OM), and two data analysts (DA1, DA2) in charge of strategic and operational activities. We explained the study’s objective and focused on defining adoption in the initial five interviews. We concluded that our AEC measurement would be the number of services requested in the e-government channel divided by the total number of services requested (e-payments/total payments and e-filings/total filings). Other alternatives analyzed were based on the total amount paid by channel or the number of citizens accessing the services by channel (a citizen may request several services by different channels). However, we decided that number of services was the best option for this research to avoid dealing with personal data. Then, we worked on the data analysis (how to organize the data and connect actions with the history of adoption) in 16 sessions. We obtained authorization to access the data for this research for ethical considerations, and the data provided did not include sensitive or personal information.

Further, we collected data from physical records and documents, digital records, and relevant laws and regulations. We analyzed this information to identify all actions related to the services that might predict AEC. Each action performed was classified according to the institutional factor of our hypothesis (coercive pressure, conformance to the environment, mobilization, and knowledge deployment) to be used as independent variables (see next section and S1 Appendix). Next, we analyzed the available data from 2015 to 2020, and we found that a weekly consolidation was suitable for our study. We initially executed daily analysis, but we identified some problems because physical offices were closed on weekends and holidays. Thus, all services were accessed only by electronic channels, making adoption 100% with low transactions for those days. Consequently, we analyzed the weekly consolidated data, represented over 313 weeks, including service type (e-filings, e-payments), week (represented by the Sunday dates), and the number of services per channel. We studied the behavior of the weekly AEC rate and worked on defining how to relate the institutional factors identified in the previous step with the AEC rate behavior.

Finally, we used six sessions to discuss and confirm the results and prepare the report. All the data from different sources support the credibility of the findings as they allow triangulation and capture contextual complexity [94].

Government actions performed to increase AEC

The first action we identified was the alliance signed by the government in 2017 with a govtech company (IE University, 2020) [95], which focused on service improvement. As a result, an imported system (SAP) was integrated with an indigenous system developed by the govtech company. This type of action is recognized as knowledge deployment [53]. Additionally, we found other actions influencing the services presented in S1 Appendix.

First, the government establishes two payment due dates every year. The first date allows payment due at a 10% discount, and the second date is for the amount due without penalties, but interest accrues after this date. We also found some laws and regulations establishing special discounts in specific periods. Initial analyses showed that citizens that pay with discounts have higher AEC levels than citizens that pay with penalties. Also, we found that weeks with due dates have presented an increase in AEC, possibly since physical offices cannot receive the demand in these periods. Consequently, we classified these actions as coercive pressure [51], defining two variables: Payment and laws and regulations. We used the variable payment to establish the value of paying; 0.8 for the 20% discount period, 0.9 for the 10% discount period, 1.0 for the no-discount period, and 1.0 plus monthly interest rate for the weeks after the due date. For the variable laws and regulations, we used 1 if there was a deadline date in that week and 0 if it did not.

Second, we found several technical solution updates to the e-services since 2017. We analyzed the objectives of these updates, and they refer to software improvements, mainly for security, privacy, usability, information quality, and new functionalities. We classified this software update as conformance to the environment [62]. To define an independent variable, we used a cumulative variable that added one to the value for every software update in the corresponding week. The objective was to reflect the solution’s maturity accumulated over time.

Third, we identified different promotion campaigns starting in 2018. These campaigns were executed using SMS and social media advertising. We classified these actions as mobilization [53]. To define an independent variable, we consolidated the campaigns for every week, adding records promoting e-government and subtracting records promoting physical offices. We then normalized the values for this variable.

Fourth, we found that during the COVID-19 pandemic in 2020, the government (at national, regional, and local levels) undertook virus containment measures limiting physical offices services for many weeks. We classified these actions as coercive pressure [51] and defined the variable COVID-19. For this variable, we used 1 for the weeks the lockdown lasted and 0 for the other weeks.

Furthermore, we identified other actions to manage AEC. For example, we found evidence that permanent staff training sessions help improve service. Also, we found that monitoring of satisfaction, ease, and effectiveness of use and effective implementation of e-government services has been in place since November 2018. Additionally, after every transaction, the system requests the completion of a voluntary three-question survey—Tables 2 and 3 present the survey questions and the consolidated results, respectively.

Table 2

Survey for citizens’ feedback on e-government services from 1 January 2018 to 10 March 2021.

Question one and two.

Q1: Tell us the degree of satisfaction obtained when using our platform.
Year	Excellent	%	Good	%	Acceptable	%	Deficient	%	Total answers
2018	1.100	51.0	681	31.6	193	8.9	183	8.5	2.157
2019	35.122	63.5	14.519	26.2	3.404	6.1	2.307	4.2	55.352
2020	39.733	65.4	15530	25.6	3496	5.8	1994	3.3	60.753
2021	16.789	71.9	5279	22.6	861	3.7	416	1.8	23.345
Q2: Rate how easy it was to carry out your procedure when accessing our platform and using its tools.
Year	Excellent	%	Good	%	Acceptable	%	Deficient	%	Total answers
2018	1.061	49.2	849	39.4	167	7.7	80	3.7	2.157
2019	34.507	62.3	17479	31.6	2286	4.1	1080	2.0	55.352
2020	38.407	63.2	18392	30.3	873	1.4	3081	5.1	60.753
2021	17.041	73.0	5070	21.7	865	3.7	369	1.6	23.345

Table 3

Survey for citizens’ feedback on e-government services from 1 January 2018 to 10 March 2021.

Question three.

Q3: Did you manage to complete your procedure promptly?
Year	Yes	%	No	%	Total answers
2018	1.680	77.9	477	22.1	2.157
2019	48.240	87.2	7112	12.8	55.352
2020	55.073	90.7	5680	9.3	60.753
2021	22.282	95.4	1063	4.6	23.345

The above actions are important because service quality (e.g., technical or output quality, functional or process quality, and direct customer service from employees) has been recognized as a strong predictor of AEC, and the recurring use of e-government will increase confidence in AEC if citizens perceive better customer service [96-98]. Complementarily, technical quality plays an essential role, influencing citizens’ intentions to use e-government [2, 42, 99, 100]. Therefore, we classified these actions as conformance to the environment [62]. However, we did not identify sufficient action to implement an independent variable for the statistical analysis.

In summary, we observed actions for knowledge deployment, conformance to the environment (fulfilling demand-side expectations with software updates, quality of service, and quality of software), mobilization, coercive pressure (laws and regulations for due dates, discounts and penalties, and measures in 2020 related to COVID-19). After analyzing these actions and the available information, we defined the dependent variable and five independent variables to propose a model for the statistical analysis of AEC (Fig 1). For the independent variables, we added columns in the dataset according to the explanation given in this section.

Fig 1

Adoption model based on the findings and the information available for analysis.

Results

Exploratory analysis

Descriptive analysis

We analyzed tax filings and tax payments separately. Fig 2 presents the annual behavior of AEC using the data from Table 1. For both services, we found differences between the AEC rate in the no-management period (2015–2017) and the AEC in the managed period (2018–2020). Without AEC management, the adoption rate can decrease, as it did for e-payments during 2015–2016.

Fig 2

AEC behavior for e-payments and e-filings from 2015 to 2020.

Figs 3 and 4 illustrate the weekly volume of services requested (bottom of the graph) and the AEC rate (top of the graph). The X-axis represents the 52-week year with the year (first digit, 1 to 3) and the week (01 to 52). Initial analysis identified six peaks in volume, two per year, with a notable increase in volume during the weeks preceding the due dates. These peaks in volume also represent AEC’s increase, probably because the capacity of physical offices was overloaded, and citizens should use virtual channels. Furthermore, AEC for the first due date (payment with discount) was higher than the second due date from 2015 to 2019; for 2020, the second due date was during the pandemic contention period, and AEC was higher than the first. This might indicate that citizens that pay early, with discounts, tend to adopt more electronic channels.

Fig 3

E-payments: Comparing non-management (2015–2017) vs. adoption-management (2018–2020) periods—conformance to the environment (C), mobilization (M), and laws and regulations (L).

Fig 4

E-filings: Comparing non-management (2015–2017) vs. adoption-management (2018–2020) periods—conformance to the environment (C), mobilization (M), and laws and regulations (L).

Figs 3 and 4 also contain government actions that may have helped enhance AEC. Having observed adoption behavior for six years, we identified historic highs in adoption in 2020, especially during the pandemic-induced lockdown from late March (week 312) to July (week 330). Additionally, we found the service behavior shaped by the due dates that also seemed to predict AEC; the most significant adoption rate changes are associated with these dates. Fig 4 illustrates higher volumes of transactions observed in e-filings than in e-payments. For example, in week 30 of 2020 (third year of the second period, 330), more than 300,000 filings were performed, compared to less than 130,000 payments previously. Finally, we observed different changes in AEC related to mobilization and conformance to the environment activities since 2018. For example, we identified several mobilization campaigns, especially in 2019 and 2020, increasing the volume of services requested and AEC weeks before the due dates. However, it was not easy to find trends from this type of analysis.

Statistical analysis

We conducted different regression analyses considering the adoption model proposed in Fig 1 and the available data. First, we started with ordinary least squares (OLS) regression, as we will present the results in Tables 4 and 5.

Table 4

OLS regression results (dependent variable: adoption of e-payment).

Predictor	Coefficient	Std. error	z	P>|z|	[0.025	0.975]
Constant	21.8467	2.421	9.024	0.000	17.083	26.611
Mobilization	-1.2302	2.595	-0.474	0.636	-6.337	3.877
Conformance to the environment	1.5594	0.127	12.307	0.000	1.310	1.809
Payment	-10.4630	2.060	-5.078	0.000	-14.517	-6.409
Laws and regulations	6.7654	2.523	2.682	0.008	1.802	11.729
COVID-19	26.2523	3.173	8.274	0.000	20.009	32.495

N = 313; R2 = 0.722; Adj. R2 = 0.717; F = 159.2; p = 0.000.

Table 5

Regression results (dependent variable: adoption of e-filing).

Predictor	Coefficient	Std. error	z	P>|z|	[0.025	0.975]
Constant	49.3419	4.072	12.118	0.000	41.330	57.354
Mobilization	2.6855	4.365	0.615	0.539	-5.903	11.274
Conformance to the environment	1.9364	0.213	9.087	0.000	1.517	2.356
Payment	1.8944	3.465	0.547	0.585	-4.924	8.713
Laws and regulations	14.9517	4.242	3.524	0.000	6.604	23.300
COVID-19	-3.7885	5.336	-0.710	0.478	-14.288	6.711

N = 313; R2 = 0.345; Adj. R2 = 0.335; F = 32.4; p = 0.000.

We identified in Table 4 that, from the five variables, only mobilization did not predict AEC for e-payment with a p = 0.636. However, OLS regression has favorable properties if its assumptions are met but can give misleading results if those assumptions are not met. Thus, OLS is not robust to violations of its assumptions (normality and homoscedasticity). In this regression, we found that errors were not normally distributed across the data (Prob. Omnibus = 0.000) and heteroscedasticity in the variance of the errors across the dataset (Durbin-Watson: 0.639). This situation is typical of regressions applied to time series data, like our case.

Similarly, in the results presented in Table 5, we identified that mobilization, payment, and COVID-19 did not predict AEC for e-filing. However, we also found that errors were not normally distributed across the data (Prob. Omnibus = 0.000) and heteroscedasticity in the variance of the errors across the dataset (Durbin-Watson: 0.517). Consequently, we executed robust OLS for heteroscedasticity and autocorrelation consistency (HAC) regression. Robust regression methods are designed to be not overly affected by violations of the assumptions. Results will be presented in Tables 6 and 7.

Table 6

Robust regression results (dependent variable: adoption of e-payment).

Predictor	Coefficient	Std. error	z	P>|z|	[0.025	0.975]
Constant	21.8467	2.311	9.452	0.000	17.299	26.395
Mobilization	-1.2302	3.508	-0.351	0.726	-8.133	5.673
Conformance to the environment	1.5594	0.135	11.589	0.000	1.295	1.824
Payment	-10.4630	1.762	-5.939	0.000	-13.929	-6.996
Laws and regulations	6.7654	2.135	3.169	0.002	2.565	10.966
COVID-19	26.2523	6.877	3.817	0.000	12.720	39.785

N = 313; R2 = 0.722; Adj. R2 = 0.717; F = 64.07; p = 0.000; Covariance Type: HAC.

Table 7

Robust regression results (dependent variable: adoption of e-filing).

Predictor	Coefficient	Std. error	z	P>|z|	[0.025	0.975]
Constant	49.3419	6.218	7.935	0.000	37.106	61.578
Mobilization	2.6855	1.903	1.411	0.159	-1.059	6.430
Conformance to the environment	1.9364	0.231	8.378	0.000	1.482	2.391
Payment	1.8944	4.514	0.420	0.675	-6.988	10.777
Laws and regulations	14.9517	2.824	5.295	0.000	9.395	20.508
COVID-19	-3.7885	3.259	-1.162	0.246	-10.202	2.624

N = 313; R2 = 0.345; Adj. R2 = 0.335; F = 51.44; p = 0.000; Covariance Type: HAC.

The results presented in Table 6 identified that, from the five variables, only mobilization did not predict AEC for e-payment with a p = 0.726. Similarly, in the results presented in Table 7, we identified that mobilization, payment, and COVID-19 did not predict AEC for e-filing.

Complementarily, we expanded the analysis with a different regression model to verify previous results. Thus, we conducted a generalized linear regression (GLS): a Tweedie family distribution configured as a Poisson and Gamma distribution compound. In the GLS, errors can follow any distribution of the exponential family, and homoscedasticity is not essential for the distribution of the errors. Consequently, Table 8 presents the GLS regression results for AEC e-payment—conformance to the environment, payment, and laws and regulations measures predicted AEC for e-payment, but mobilization did not. This regression identified a variation with the COVID-19 variable; its p = 0.066 is above but close to the 5% limit. However, considering previous results, we claim that the COVID-19 variable should continue as a relevant predictor of AEC for e-payment.

Table 8

GLS regression results (dependent variable: adoption of e-payment).

Predictor	Coefficient	Std. error	z	P>|z|	[0.025	0.975]
Constant	3.1506	0.129	24.437	0.000	2.898	3.403
Mobilization	-0.0093	0.126	-0.074	0.941	-0.256	0.237
Conformance to the environment	0.0923	0.006	14.369	0.000	0.080	0.105
Payment	-0.7490	0.110	-6.786	0.000	-0.965	-0.533
Laws and regulations	0.4064	0.126	3.222	0.001	0.159	0.654
COVID-19	0.2815	0.153	1.837	0.066	-0.019	0.582

Notes: Model: GLM; Model Family: Tweedie (var_power = 1.8, meaning for a compound of Poisson and Gamma); Link Function: log; Method: IRLS; No. Iterations: 12; Covariance Type: nonrobust; No. Observations: 313; Df Residuals: 307; Df Model: 5; Scale: 0.33690; Log-Likelihood: nan; Deviance: 250.85; Pearson chi2: 103.

Table 9 shows the GLS regression results for e-filing. From the five variables, only laws and regulations and conformance to the environment were shown to predict AEC, while mobilization, payment, and COVID-19 actions did not predict this variable.

Table 9

GLS regression results (dependent variable: adoption of e-filing).

Predictor	Coefficient	Std. error	z	P>|z|	[0.025	0.975]
Constant	3.8750	0.076	51.192	0.000	3.727	4.023
Mobilization	0.0407	0.079	0.515	0.607	-0.114	0.196
Conformance to the environment	0.0327	0.004	8.366	0.000	0.025	0.040
Payment	0.0582	0.064	0.905	0.365	-0.068	0.184
Laws and regulations	0.2445	0.077	3.174	0.002	0.094	0.396
COVID-19	-0.0967	0.096	-1.003	0.316	-0.286	0.092

Notes: Model: GLM; Model Family: Tweedie (var_power = 1.8, meaning for a compound of Poisson and Gamma); Link Function: log; Method: IRLS; No. Iterations: 12; Covariance Type: nonrobust; No. Observations: 313; Df Residuals: 307; Df Model: 5; Scale: 0.15754; Log-Likelihood: nan; Deviance: 288.36; Pearson chi2: 48.4.

In summary, considering the results presented in this section, we found that the laws and regulations and conformance to the environment are important predictors of AEC in the two services. We did not find predictive power for mobilization activities, but we found that payment and COVID-19 actions were essential AEC predictors for e-payments but not e-filings. Thus, factors predicting AEC also relate to the type of service. It is important to mention that all regression models had equivalent results identifying which factors predicted AEC, despite the coefficients and the standard errors being different. We highlight that the COVID-19 variable was present only in 2020 but became a relevant predictor for e-payment.

Discussion

Considering the results presented in the last section, we consolidated the status of every hypothesis for the two analyzed services in Table 10. As a result, only H2 was confirmed for both services, H1 was partially confirmed, H3 was not confirmed, and we could not get enough information for statistical analysis for H4.

Table 10

Hypothesis summary.

Hypothesis	e-payments	e-filings
H1. Coercive pressure predicts AEC	Confirmed	Partially confirmed
H2. Conformance to the environment (citizens’ needs) predicts AEC	Confirmed	Confirmed
H3. Mobilization predicts AEC	Not confirmed	Not confirmed
H4. Knowledge deployment predicts AEC	NA	NA

The predictive power of coercive pressure in AEC indicates that government entities possess an important tool for AEC management. Coercive pressure is represented in the laws and regulations establishing due dates (laws and regulations), discounts and penalties (payment), and access restrictions to physical offices during the COVID-19 lockdown in 2020, making e-government use mandatory [51]. Only laws and regulations predicted e-filings and e-payments; neither payment nor COVID-19 variables predicted e-filings. After analyzing the behavior and adoption levels of e-filings, we consider that this service has reached a high level of adoption, close to 80% for 2020, which may be hard to increase, and the predictors seem to be different. Future research could study the maximum level of adoption of different e-government services and the predictors of services with high levels of AEC. For example, Becker et al. [28] studied the different levels of AEC compared to the total population, users of the Internet, and users of e-commerce.

These results align with those in the literature. Coercive pressures can be enablers or constraints in the AEC process [55, 58]. For example, some activities constrained AEC, including those related to due dates in the administrative collection available only in physical offices. The legal framework is important in shaping government services [54, 101]. Mandatory use was presented by Ghareeb et al. [59] as the strategy implemented for Denmark and the UK to force citizens to explore e-services and their benefits. This occurred during the COVID-19 lockdown; this variable was present only in 2020, when only the electronic channels were available, forcing new citizens to use the e-service, becoming an important predictor for e-payment. However, it was only effective in technically developed regions. Consequently, caution is required because the digital divide is an important issue in the Latin American region [15, 27]. Latin American governments could explore the gradual implementation of mandatory use, for example, starting with some services (e.g., accessing information), geographical areas (e.g., IT developed areas), or groups (e.g., public servants, students, teachers, and professors).

Indeed, Latin American governments have an essential role in combating the factors that widen or retain the digital divide, especially in rural areas and developing countries [102]. Increasing access to computers and the Internet (as the Colombian government promotes) is not a complete solution but a good start nonetheless [103]. The Internet can promote citizenship and citizen participation in Latin America [104]. Also, improving websites’ accessibility to avoid discrimination, even imposing sanctions for non-compliance of standards as another example of coercive pressure, could be another strategy followed by developing countries [78].

We also found conformance to the environment to be an important predictor of AEC for both services. Improving technical solutions and monitoring satisfaction, ease of use, and effectiveness can help increase conformance with citizens’ expectations of the technical solution quality and service quality [62]. We identified that the maturity of electronic services, represented in conformance to the environment variable in the model, supports the annual increase of AEC we observed in the analysis. We found that the literature highlights the important influence on AEC of quality of service and technical quality [96-98], but this is the first study to classify these as part of a legitimation strategy. We concluded that no law, regulation, or mobilization could influence AEC increase without improving the quality of technical solutions and services. Additionally, governments should take action to fulfill the needs of AEC in developing countries in a broader form. For example, creating the legal framework that supports AEC and removes legal barriers, investing in creating reliable databases, enhancing interoperability among government agencies, investing in education and training, combating the digital divide, and improving technical aspects of the systems to make them easy to use for all audiences [15, 103, 105].

We found that mobilization was not significant for any service. We identified results similar to those of Henriksen and Damsgaard [70], where the demand-pull-based approach of the Danish government was successful, as was changing the strategy for imperatives and regulations. However, many studies have highlighted mobilization’s importance in AEC [3, 72, 73]. We consider this result related to the nature of the services studied. This study looked at yearly, mandatory services, and most users would be recurrent users that know these services well. In this case, mobilization might impact other variables as early payments and increasing volume of transactions but not for AEC. Promotion campaigns are considered essential in the early stages of adoption, creating awareness among citizens and influencing the intention to use [59]; for example, in developing countries, promotion campaigns should leverage the social characteristics of the society [106, 107]. Additionally, the mobilization of services studied was mainly through electronic channels (SMS, social media advertising, web portal messages, and mobile applications), and non-adopters scarcely use these electronic channels. Therefore, conducting mobilization by combining traditional channels and electronic channels would be valuable for engaging non-adopters [11].

Although we did not have enough information for statistical analysis, we found evidence of the importance of knowledge deployment in AEC in three actions performed by the government: delegating services operation to experts, training public servants to improve service, and creating training videos to enhance citizen use [53]. We found similar results in previous literature mentioning the long-term contributions of this type of intervention in generating a critical mass of ICT (social infrastructure) users who can integrate technologies into their activities [67, 68]. Training public servants, citizens, and key stakeholders is also important for AEC [68, 71]. Montealegre [69] presented how IT can be successfully adopted even in less developed countries with contextual imperfections and scarcities. However, local contexts and traditions are essential for understanding how Colombia develops e-government [108].

Notably, we found the type of services to be a strong predictor in AEC. Thus, we found different behavior in AEC for each service; e-filings had higher levels of adoption. Also, the factors predicting each service were different. This result is consistent with Vrček and Klačmer [10], who demonstrated that 82% of the people in Croatia did not oppose less sophisticated electronic services (e.g., information-gathering), and 54% were willing to use e-government services with medium complexity (e.g., citizens sending information to the government). However, only 32% were disposed towards using e-government services for payment activities.

Overall, we found that government has a strong influence on AEC. While governments are responsible for the supply-side of e-government [11, 46], they must also play an essential role in the demand-side. Particularly, they can intervene to shape AEC’s potential [11, 109]. We found that the government has important tools to manage AEC, specifically the power to establish laws and regulations that shape services and define AEC guidelines. Furthermore, governments can configure internal aspects to conform to citizens’ expectations of e-government. Additionally, governments can deploy (directly or indirectly) the knowledge required to improve e-services and increase the number of citizens using ICT. Finally, they can promote e-government services, building awareness of their existence and benefits.

Additionally, we consider that AEC should be managed mainly by government entities, as they are responsible for simplifying service processes, integrating relevant entities, offering quality e-services, creating an adequate environment for AEC (e.g., accessibility, norms and regulations, knowledge, removing barriers for adoption), promoting awareness of e-government and the multiple benefits related to its use, monitoring e-government performance and citizen satisfaction, and continuously improving services by incorporating new technologies and good practices.

Conclusions

This research expands the knowledge of the adoption of e-government by citizens (AEC) using an institutional framework to study the government actions to increase AEC and answer our research question: how does government influence AEC? Consequently, we conducted a case study that analyzed the actions of the Antioquia Government in Colombia to increase AEC for annual vehicle tax filing and payment services and determine which of these actions could be used as AEC predictors. We employed institutional theory, institutional interventions, and legitimation strategies to classify these actions presenting four hypotheses to identify if coercive pressure, conformance to the environment, mobilization, and knowledge deployment predict AEC. Thus, we analyzed the correlation of these actions with AEC for the two services from 2015 to 2020.

We found that governments have a strong influence on AEC. First, the government controls coercive pressure, an essential factor in predicting e-filings and e-payment services. For e-payments, different elements of coercive pressure predicted AEC: laws and regulations establishing the deadlines, discounts, penalties, and access restrictions during the COVID-19 lockdown in 2020 that forced new citizens to use the e-services. For e-filings, only the laws and regulations (due dates) had predictive power. Second, governments can configure their internal components to conform to citizens’ expectations for the quality of technical solutions and services regarding conformance to the environment. We identified conformance to the environment variable (representing software updates) predicting both services. Thus, improving technical solutions and monitoring satisfaction, ease of use, and effectiveness can help increase conformance with citizens’ expectations of the technical solution quality and service quality. We concluded that no law, regulation, or mobilization could influence AEC increase without improving the quality of technical solutions and services.

Third, we found evidence of the importance of knowledge deployment in AEC in three government actions: delegating experts for operating services, training public servants to improve services, and creating training videos to enhance their usefulness for citizens. Although we could not test this statistically, based on the evidence gathered in the different interviews, we argue that this type of intervention would, in the long term, contribute to generating a critical mass of ICT (social infrastructure) users with the ability to integrate the technologies into their activities. Fourth, we found that mobilization was not significant for any service. We determined that this result was related to the nature of the studied services (yearly mandatory services, and most users are recurrent users). We consider promotion campaigns important in the early stages of adoption, creating awareness among citizens and influencing the intention to use. Finally, we found the type of service to be a crucial determinant of AEC; the levels of adoption and the factors predicting AEC were different.

This study makes significant theoretical contributions to the literature by proposing a new perspective to understand the development of AEC and by exploring the influence of institutional aspects on individuals. Analyzing the effects of the COVID-19 measures in AEC and comparing AEC for two different services offered on the same platform are important contributions. Developing insights in a Latin American context is another significant contribution. Finally, this study makes practical contributions identifying primary factors predicting AEC in Latin America that governments from developing countries can find helpful for policy development and prioritization.

This study has some limitations. First, we selected variables for the proposed model from the available information. Future research should explore additional institutional factors to complement our findings. Moreover, this work was limited to the Department of Antioquia and a specific population (car owners); therefore, caution should be exercised in generalizing the findings. Future research should explore different services using the institutional framework in other regions. Finally, in this work, we focused on identifying the predictor of AEC more than defining a model. Future researches could include institutional variables in existing or new models to study AEC.

Government actions related to the services.

Note. Promoting e-government for the current term (PE-CT); Promoting e-government for past due debts (PE-PDD); Promoting physical office for past due debts (PPO-PDD); Conform to the environment (C), mobilization (M), and laws and regulations (L).

(DOCX)

Click here for additional data file.

(XLSX)

Click here for additional data file.

(DOCX)

Click here for additional data file.

25 Oct 2021

Submitted filename: Comments_and_responses.docx

Click here for additional data file.

15 Nov 2021

A rebuttal letter that responds to each point raised by the academic editor and reviewer(s). You should upload this letter as a separate file labeled 'Response to Reviewers'.

A marked-up copy of your manuscript that highlights changes made to the original version. You should upload this as a separate file labeled 'Revised Manuscript with Track Changes'.

An unmarked version of your revised paper without tracked changes. You should upload this as a separate file labeled 'Manuscript'.

If you would like to make changes to your financial disclosure, please include your updated statement in your cover letter. Guidelines for resubmitting your figure files are available below the reviewer comments at the end of this letter.

If applicable, we recommend that you deposit your laboratory protocols in protocols.io to enhance the reproducibility of your results. Protocols.io assigns your protocol its own identifier (DOI) so that it can be cited independently in the future. For instructions see: https://journals.plos.org/plosone/s/submission-guidelines#loc-laboratory-protocols. Additionally, PLOS ONE offers an option for publishing peer-reviewed Lab Protocol articles, which describe protocols hosted on protocols.io. Read more information on sharing protocols at https://plos.org/protocols?utm_medium=editorial-email&utm_source=authorletters&utm_campaign=protocols.

We look forward to receiving your revised manuscript.

Kind regards,

Rogis Baker, Ph.D

Academic Editor

PLOS ONE

Journal Requirements:

When submitting your revision, we need you to address these additional requirements.

1. Please ensure that your manuscript meets PLOS ONE's style requirements, including those for file naming. The PLOS ONE style templates can be found at

https://journals.plos.org/plosone/s/file?id=wjVg/PLOSOne_formatting_sample_main_body.pdf and

https://journals.plos.org/plosone/s/file?id=ba62/PLOSOne_formatting_sample_title_authors_affiliations.pdf

[Note: HTML markup is below. Please do not edit.]

Reviewers' comments:

Reviewer's Responses to Questions

Comments to the Author

1. Is the manuscript technically sound, and do the data support the conclusions?

The manuscript must describe a technically sound piece of scientific research with data that supports the conclusions. Experiments must have been conducted rigorously, with appropriate controls, replication, and sample sizes. The conclusions must be drawn appropriately based on the data presented.

Reviewer #1: Yes

Reviewer #2: Partly

Reviewer #3: Partly

Reviewer #4: Yes

**********

2. Has the statistical analysis been performed appropriately and rigorously?

Reviewer #1: No

Reviewer #2: No

Reviewer #3: No

Reviewer #4: Yes

**********

3. Have the authors made all data underlying the findings in their manuscript fully available?

The PLOS Data policy requires authors to make all data underlying the findings described in their manuscript fully available without restriction, with rare exception (please refer to the Data Availability Statement in the manuscript PDF file). The data should be provided as part of the manuscript or its supporting information, or deposited to a public repository. For example, in addition to summary statistics, the data points behind means, medians and variance measures should be available. If there are restrictions on publicly sharing data—e.g. participant privacy or use of data from a third party—those must be specified.

Reviewer #1: Yes

Reviewer #2: Yes

Reviewer #3: Yes

Reviewer #4: Yes

**********

4. Is the manuscript presented in an intelligible fashion and written in standard English?

PLOS ONE does not copyedit accepted manuscripts, so the language in submitted articles must be clear, correct, and unambiguous. Any typographical or grammatical errors should be corrected at revision, so please note any specific errors here.

Reviewer #1: Yes

Reviewer #2: Yes

Reviewer #3: Yes

Reviewer #4: Yes

**********

5. Review Comments to the Author

Please use the space provided to explain your answers to the questions above. You may also include additional comments for the author, including concerns about dual publication, research ethics, or publication ethics. (Please upload your review as an attachment if it exceeds 20,000 characters)

Reviewer #1: The manuscript is well conceptualized in the manner that it highlights the importance of the E-Government adoption by citizens. However, the organization and structure of the paper does not provided an appetite for a reader to continue reading. the introduction of the paper is very long and has carried so many information that could be summarized. we suggest the author to use IMRAD approach to re-organize this work. literature part can be omitted or merged with the introduction.

The methodology of the study is not explanatory enough to know exactly what was done specifically in quality control, number of interviews, data analysis.

I have found that data analysis have been presented as a part of the result, i thought this could be seen in the methodological part.

Discussion should be done basing on the key areas of the study. i suggest that the discussion being guided by the central question/gap of the study so that we establish if the study answers what was motivated a researcher to conduct a study.

Conclusion should be aligned with the discussion and findings of the study which all are informed by the objective of the study

Reviewer #2: The paper discusses an interesting topic and explores a problematic that is increasingly relevant due to the global pandemic. The current version of the paper shows an important improvement compared to the previous one. New sections provide a more accurate theoretical context to the research. However, there are several problems with the paper listed below, in no particular order:

1. The paper is well written, but the structure of the empirical analysis is not clear.

2. Authors suggest that “Multiple filings for one payment are typical” which brings up the question of the number of e-fillings written by unique users or institutional users. At least some information and discussion on this topic is needed.

3. In addition, the second dependent variable (e-payment) seems to be associated with the main one. The authors claim that “Tax payment requires a previous tax filing, and e-payment requires e-filing”, so, technically there is a selection process that affects the estimation method or the population of analysis at least in the case of the e-payment.

4. Regarding the creation of the set of independent variables: Authors conducted several interviews with the service operation manager and staff, however, results are neither provided nor explained. The explanation and analysis of these interviews are crucial to understand the construction and accuracy of the independent variables.

5. The operationalization of the independent variables needs more explanation or at least some descriptive statistics. For instance, authors use two measures of environment conformance at the same time (see lines 379 and 385) without discussing the consequences of this decision and therefore making the role of each variable unclear. Another example is the use of the mobilization variable, which is the net between policies regardless of the strength of the policy.

6. The econometric analysis is limited due to the time series structure of the data. Simple regressions might provide biased results. Authors should provide more information regarding the selection of the econometric model and their implications regarding the characteristics of the data.

7. Discussion focused on comparisons with some developed countries, which is useful, but it needs to be complemented with other cases of developing countries with similar characteristics.

Reviewer #3: - In this current form the paper is really underdeveloped on the analysis part. The balance of the article leans heavily towards the literature review and institutional literature. The methods section should be more robust as all we learn about the actual analysis carried out is that: "multiple linear regression for data correlation analysis" (line 342 and 422)

- The data used for the regression is time series data and using pooled OLS as an estimator will yield biased results. The figures of the data clearly show seasonality and trend which needs to be addressed. The standard error correction should also reflect this fact (e.g.: Newey-West estimator, report unit root tests)

- The interpretation of the estimated coefficients are also problematic. In case of the year, as it is a numerical variable, it represents a time trend (and not that individual years are strong predictors).

- The method and analysis part should see major revisions, including a more detailed explanation of the regression approach used (and the merits of it given the data); robustness checks are also missing (different models, different estimators should point towards similar outcomes for the results to be considered robust and not just accidental), the authors should make efforts showing that all the relevant control variables are included.

- In particular, the construction of the laws and regulations variable is not transparent, the reader has little to work with as to how laws were coded as influental or not. Who measured it, how is influence defined, and how can a law have weekly influence? These questions are all the more important as this variable is statistically significant (altough with the flawed pooled OLS estimation).

- Based on the actual analysis carried out, I'm not sure that the lenghty literature review is neccesary (same stands for the Table 1). If that part is vital for the analysis then it should reflect on the enumerated institutional factors.

Reviewer #4: Thank you for the opportunity to read this paper. My comments are below:

- Scope of the article seems a bit niche, but generally the framework is tightly presented and the data collected is impressive and speaks to the questions asked.

- Table 3 is very difficult to parse; please re-format at the least to have Q3 take up equal space as previous cells.

- In Table 4 and 5, please include sample sizes; also I would rather see the significant p-values highlighted / boldened, than the non-significant ones.

- In the absence of any discussion of a causal inference strategy, I would be careful to avoid causal language. For example, on page 25, you use "determines"; in the abstract you also say "influencing". Please change these to "predicts/predicting" or similar language; or provide some rationale as to why the multiple regressions analysis can be interpreted causally.

**********

6. PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #1: No

Reviewer #2: No

Reviewer #3: No

Reviewer #4: No

[NOTE: If reviewer comments were submitted as an attachment file, they will be attached to this email and accessible via the submission site. Please log into your account, locate the manuscript record, and check for the action link "View Attachments". If this link does not appear, there are no attachment files.]

While revising your submission, please upload your figure files to the Preflight Analysis and Conversion Engine (PACE) digital diagnostic tool, https://pacev2.apexcovantage.com/. PACE helps ensure that figures meet PLOS requirements. To use PACE, you must first register as a user. Registration is free. Then, login and navigate to the UPLOAD tab, where you will find detailed instructions on how to use the tool. If you encounter any issues or have any questions when using PACE, please email PLOS at figures@plos.org. Please note that Supporting Information files do not need this step.

7 Jan 2022

29 December 2021

Prof. Rogis Baker,

Academic Editor

PLOS ONE

Dear Editor:

We are submitting a revised version of the manuscript with number PONE-D-21-33977 entitled “Government influence on e-government adoption by citizens in Colombia: Empirical evidence in a Latin American context” for its acceptance for publication. We appreciate and thank the reviewers for providing their comments that have considerably improved the quality and presentation of our work.

A point-by-point answer to the referees’ comments is presented below:

Reviewer #1:

The manuscript is well conceptualized in the manner that it highlights the importance of the E-Government adoption by citizens. However, the organization and structure of the paper does not provided an appetite for a reader to continue reading. the introduction of the paper is very long and has carried so many information that could be summarized. we suggest the author to use IMRAD approach to re-organize this work. literature part can be omitted or merged with the introduction.

Answer 1: Thank you very much for your time and valuable comments that helped us improve this new version. According to this recommendation, we have updated this new version to summarize and merge the literature review with the introduction. We have also suppressed Table 1 that lists all the institutional factors identified. Instead, we only mentioned the theoretical referents that we used to define our hypothesis in the introduction (page 3, line 69).

The methodology of the study is not explanatory enough to know exactly what was done specifically in quality control, number of interviews, data analysis.

Answer 2: We have updated de data and methods subsection (page 11, line 260) of the methodology section detailing the process we followed for the methodology.

For example, we indicated in the first paragraph the following (page 11, line 261):

“This is an explanatory case study [86,87] that used interviews and documentation analyses combined with quantitative analyses of the historical data from 2015 to 2020 (more than 11 million filing records and more than 5 million payments, as illustrated in Table 1) to identify how the government influences AEC (our dependent variable). We employed descriptive methods, including aggregated results for the use of e-government services in graphical and tabular output [88,89]. Additionally, we conducted multiple regressions for the data correlation analysis. Similar approaches have been used in previous research related to AEC [90,91].”

We mentioned that we had 27 interviews, and what we did in the different interviews. For example (page 13, line 301): “Finally, we used six sessions to discuss and confirm the results and prepare the report. All the data from different sources support the credibility of the findings as they allow triangulation and capture contextual complexity [94].”

I have found that data analysis have been presented as a part of the result, i thought this could be seen in the methodological part.

Answer 3: We moved the data analysis to identify the actions performed by the government to increase AEC to the methodology section. Additionally, we included the reference to Appendix A that was unintendedly omitted in the last version, despite the file being uploaded.

Discussion should be done basing on the key areas of the study. i suggest that the discussion being guided by the central question/gap of the study so that we establish if the study answers what was motivated a researcher to conduct a study.

Answer 4: This research expands the knowledge of AEC using an institutional framework to study the government actions to increase AEC and answer our research question (how does government influence AEC?). We also presented four hypotheses to identify if coercive pressure, conformance to the environment, mobilization, and knowledge deployment predict AEC.

At the end of the discussion section, we answered the research question in the following paragraph (page 25, line 568): “Overall, we found that government has a strong influence on AEC. While governments are responsible for the supply-side of e-government [11,46], they must also play an essential role in the demand-side. Particularly, they can intervene to shape AEC’s potential [11,109]. We found that the government has important tools to manage AEC, specifically the power to establish laws and regulations that shape services and define AEC guidelines. Furthermore, governments can configure internal aspects to conform to citizens’ expectations of e-government. Additionally, governments can deploy (directly or indirectly) the knowledge required to improve e-services and increase the number of citizens using ICT. Finally, they can promote e-government services, building awareness of their existence and benefits.”

We also presented the result for the four hypotheses for both services in Table 6 at the beginning of the discussion section (page 22, line 483).

Conclusion should be aligned with the discussion and findings of the study which all are informed by the objective of the study

Answer 5: We adjusted the conclusion according to your comment. We removed the third paragraph from this section, referring to how AEC should be managed. We also expanded the initial paragraph of this section as follows (page 26, line 587): “This research expands the knowledge of the adoption of e-government by citizens (AEC) using an institutional framework to study the government actions to increase AEC and answer our research question: how does government influence AEC? Consequently, we conducted a case study that analyzed the actions of the Antioquia Government in Colombia to increase AEC for annual vehicle tax filing and payment services and determine which of these actions could be used as AEC predictors. We employed institutional theory, institutional interventions, and legitimation strategies to classify these actions presenting four hypotheses to identify if coercive pressure, conformance to the environment, mobilization, and knowledge deployment predict AEC. Thus, we analyzed the correlation of these actions with AEC for the two services from 2015 to 2020.”

Reviewer #2:

The paper discusses an interesting topic and explores a problematic that is increasingly relevant due to the global pandemic. The current version of the paper shows an important improvement compared to the previous one. New sections provide a more accurate theoretical context to the research. However, there are several problems with the paper listed below, in no particular order:

1. The paper is well written, but the structure of the empirical analysis is not clear.

Answer 6: Thank you very much for your time and valuable comments that helped us improve this new version. As we mentioned in answer 2, we have updated the methodology section to clarify the empirical analysis.

2. Authors suggest that “Multiple filings for one payment are typical” which brings up the question of the number of e-fillings written by unique users or institutional users. At least some information and discussion on this topic is needed.

Answer 7: At the end of the case study sub-section, we included some examples to clarify this situation, expanding the discussion and giving examples. We have the following paragraph (page 11, line 247): “Tax payment requires a previous tax filing, and e-payments require e-filings. Multiple filings for one payment are typical. Also, having filings without payment is usual. For example, a tax filing to pay with discounts should be updated if not paid before the deadline. Another example is presented when the vehicle’s information should be updated (e.g., the new owner or updated value). Thus, the payment is performed based on the last tax filing. It is impossible to pay through e-government channels once the tax is filed physically in an office, and this means a selection process affects e-payments defined by the filing presented by the electronic channels.”

3. In addition, the second dependent variable (e-payment) seems to be associated with the main one. The authors claim that “Tax payment requires a previous tax filing, and e-payment requires e-filing”, so, technically there is a selection process that affects the estimation method or the population of analysis at least in the case of the e-payment.

Answer 8: In complementing answer 7, we recognized this in the same sub-section. In addition, we discussed this situation, and considering that e-filing has high levels of adoption, we concluded that the change in the size of the population for e-payment will not be significant for our analysis. In addition, we have defined how to measure AEC for each service (page 12, line 277): “We concluded that our AEC measurement would be the number of services requested in the e-government channel divided by the total number of services requested (e-payments/total payments and e-filings/total filings).”

4. Regarding the creation of the set of independent variables: Authors conducted several interviews with the service operation manager and staff, however, results are neither provided nor explained. The explanation and analysis of these interviews are crucial to understand the construction and accuracy of the independent variables.

Answer 9: In the data and methods subsection (page 11) and in the government actions to increase AEC sub-sections (page 13), we included the explanation and analysis of these interviews and how we defined the variables. We also included the reference to Appendix A that was unintendedly omitted in the last version, despite the file being uploaded.

For example, to define the variable conformance to the environment, we included the following paragraph (page 14, line 322): “Second, we found several technical solution updates to the e-services since 2017. We analyzed the objectives of these updates, and they refer to software improvements, mainly for security, privacy, usability, information quality, and new functionalities. We classified this software update as conformance to the environment [62]. To define an independent variable, we used a cumulative variable that added one to the value for every software update in the corresponding week. The objective was to reflect the solution’s maturity accumulated over time.”

5. The operationalization of the independent variables needs more explanation or at least some descriptive statistics. For instance, authors use two measures of environment conformance at the same time (see lines 379 and 385) without discussing the consequences of this decision and therefore making the role of each variable unclear. Another example is the use of the mobilization variable, which is the net between policies regardless of the strength of the policy.

Answer 10: As we mention in answer 9, we have expanded the operationalization process of the independent variables.

6. The econometric analysis is limited due to the time series structure of the data. Simple regressions might provide biased results. Authors should provide more information regarding the selection of the econometric model and their implications regarding the characteristics of the data.

Answer 11: We expanded the statistical analysis sub-section (page 18). This version presented the different models used for the analysis and explained the selection process. In addition, in this section, we explained the implications regarding the characteristics of the data. For example, we have included the following paragraph (page 18, line 419): “We identified in Table 4 that, from the five variables, only mobilization did not predict AEC for e-payment with a p=0.636. However, OLS regression has favorable properties if its assumptions are met but can give misleading results if those assumptions are not met. Thus, OLS is not robust to violations of its assumptions (normality and homoscedasticity). In this regression, we found that errors were not normally distributed across the data (Prob. Omnibus = 0.000) and heteroscedasticity in the variance of the errors across the dataset (Durbin-Watson: 0.639). This situation is typical of regressions applied to time series data, like our case.”

On page 19 (line 433) we have the following text: “…Consequently, we executed robust OLS for heteroscedasticity and autocorrelation consistency (HAC) regression. Robust regression methods are designed to be not overly affected by violations of the assumptions. Results will be presented in Tables 6 and 7.”

Finally, on page 20, we have: “Complementarily, we expanded the analysis with a different regression model to verify previous results. Thus, we conducted a generalized linear regression (GLS): a Tweedie family distribution configured as a Poisson and Gamma distribution compound. In the GLS, errors can follow any distribution of the exponential family, and homoscedasticity is not essential for the distribution of the errors. Consequently, Table 8 presents the GLS regression results for AEC e-payment…”

7. Discussion focused on comparisons with some developed countries, which is useful, but it needs to be complemented with other cases of developing countries with similar characteristics.

Answer 12: We have included discussions on Latin American and Asian developing countries, and including references to Colombia.

For example, on page 23 (line 511), we included the following paragraph: “Indeed, Latin American governments have an essential role in combating the factors that widen or retain the digital divide, especially in rural areas and developing countries [102]. Increasing access to computers and the Internet (as the Colombian government promotes) is not a complete solution but a good start nonetheless [103]. The Internet can promote citizenship and citizen participation in Latin America [104]. Also, improving websites’ accessibility to avoid discrimination, even imposing sanctions for non-compliance of standards as another example of coercive pressure, could be another strategy followed by developing countries [78].”

Another example is presented on page 24: “… Training public servants, citizens, and key stakeholders is also important for AEC [68,71]. Montealegre [69] presented how IT can be successfully adopted even in less developed countries with contextual imperfections and scarcities. However, local contexts and traditions are essential for understanding how Colombia develops e-government [108].”

Reviewer #3:

- In this current form the paper is really underdeveloped on the analysis part. The balance of the article leans heavily towards the literature review and institutional literature. The methods section should be more robust as all we learn about the actual analysis carried out is that: "multiple linear regression for data correlation analysis" (line 342 and 422).

Answer 13: Thank you very much for your time and valuable comments that helped us improve this new version. We have updated the article’s structure (please see answer 1) for a better balance. Also, we have updated our methodology section to make it more robust (please see answers 2 and 9).

- The data used for the regression is time series data and using pooled OLS as an estimator will yield biased results. The figures of the data clearly show seasonality and trend which needs to be addressed. The standard error correction should also reflect this fact (e.g.: Newey-West estimator, report unit root tests)

Answer 14: You are right. Consequently, we have updated the statistical analysis sub-section, as we mentioned in answer 11.

- The interpretation of the estimated coefficients are also problematic. In case of the year, as it is a numerical variable, it represents a time trend (and not that individual years are strong predictors).

Answer 15: You are right. Initially, we have identified this variable as an indicator of the system's maturity through time. However, according to your comment, we reviewed the model and decided to remove the year as a predictor of AEC. For the maturity of the system, we kept the conformance to the environment variable.

- The method and analysis part should see major revisions, including a more detailed explanation of the regression approach used (and the merits of it given the data); robustness checks are also missing (different models, different estimators should point towards similar outcomes for the results to be considered robust and not just accidental), the authors should make efforts showing that all the relevant control variables are included.

Answer 16: As we mentioned in Answers 11 and 14, we expanded our explanation about the statistical analysis indicating what analyses we conducted and why we decided to use that type of regression. Additionally, we mentioned that we executed three separate regressions with consistent results. Also, we identified the type of service as a relevant control variable in the available data, and we decided to run separated models for each service.

- In particular, the construction of the laws and regulations variable is not transparent, the reader has little to work with as to how laws were coded as influental or not. Who measured it, how is influence defined, and how can a law have weekly influence? These questions are all the more important as this variable is statistically significant (altough with the flawed pooled OLS estimation).

Answer 17: Expanding on Answer 9, we included details on how we defined this variable. In the sub-section actions performed to increase AEC (page 13, line 310), we included the following paragraph: “First, the government establishes two payment due dates every year. The first date allows payment due at a 10% discount, and the second date is for the amount due without penalties, but interest accrues after this date. We also found some laws and regulations establishing special discounts in specific periods. Initial analyses showed that citizens that pay with discounts have higher AEC levels than citizens that pay with penalties. Also, we found that weeks with due dates have presented an increase in AEC, possibly since physical offices cannot receive the demand in these periods. Consequently, we classified these actions as coercive pressure [51], defining two variables: Payment and laws and regulations. We used the variable payment to establish the value of paying; 0.8 for the 20% discount period, 0.9 for the 10% discount period, 1.0 for the no-discount period, and 1.0 plus monthly interest rate for the weeks after the due date. For the variable laws and regulations, we used 1 if there was a deadline date in that week and 0 if it did not.”

- Based on the actual analysis carried out, I'm not sure that the lenghty literature review is neccesary (same stands for the Table 1). If that part is vital for the analysis then it should reflect on the enumerated institutional factors.

Answer 18: You are right. Please see Answer 13.

Reviewer #4:

Thank you for the opportunity to read this paper. My comments are below:

- Scope of the article seems a bit niche, but generally the framework is tightly presented and the data collected is impressive and speaks to the questions asked.

- Table 3 is very difficult to parse; please re-format at the least to have Q3 take up equal space as previous cells.

Answer 19: Thank you very much for your time and valuable comments that helped us improve this new version. We have decided to create a new table (Table 3) for Q3, as you can see on page 15 (line 348). Q1 and Q2 remain in Table 2 (line 345).

- In Table 4 and 5, please include sample sizes; also I would rather see the significant p-values highlighted / boldened, than the non-significant ones.

Answer 20: We have included the sample sizes, and we have boldened the significant p-values instead of the non-significant ones.

- In the absence of any discussion of a causal inference strategy, I would be careful to avoid causal language. For example, on page 25, you use "determines"; in the abstract you also say "influencing". Please change these to "predicts/predicting" or similar language; or provide some rationale as to why the multiple regressions analysis can be interpreted causally.

Answer 21: We have updated our manuscript, avoiding causal language, and consequently defined our hypothesis. For example, in the introduction, we have the following paragraph (page 6): “The legal framework, as an example of coercive pressure, is determinant in the context of AEC [54]. Several studies have confirmed that the government’s entities use this coercive pressure to encourage citizens and other groups to embrace e-government [13,49,55-58], even in mandatory use contexts, as Denmark and the UK have defined [59]. Consequently, we have defined that coercive pressure predicts AEC as our first hypothesis.”

Additionally, we included a table in the discussion section consolidating the hypothesis results (page 22, line 483). There you can confirm that we implemented your recommendation.

Submitted filename: PLOS_ONE_Response to Reviewers.docx

Click here for additional data file.

14 Feb 2022

Government influence on e-government adoption by citizens in Colombia: Empirical evidence in a Latin American context

PONE-D-21-33977R1

Dear Dr. Juan Pablo RAMIREZ-MADRID,

We’re pleased to inform you that your manuscript has been judged scientifically suitable for publication and will be formally accepted for publication once it meets all outstanding technical requirements.

Within one week, you’ll receive an e-mail detailing the required amendments. When these have been addressed, you’ll receive a formal acceptance letter and your manuscript will be scheduled for publication.

An invoice for payment will follow shortly after the formal acceptance. To ensure an efficient process, please log into Editorial Manager at http://www.editorialmanager.com/pone/, click the 'Update My Information' link at the top of the page, and double check that your user information is up-to-date. If you have any billing related questions, please contact our Author Billing department directly at authorbilling@plos.org.

If your institution or institutions have a press office, please notify them about your upcoming paper to help maximize its impact. If they’ll be preparing press materials, please inform our press team as soon as possible -- no later than 48 hours after receiving the formal acceptance. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information, please contact onepress@plos.org.

Kind regards,

Rogis Baker, Ph.D

Academic Editor

PLOS ONE

Additional Editor Comments (optional):

Reviewers' comments:

16 Feb 2022

PONE-D-21-33977R1

Government influence on e-government adoption by citizens in Colombia: Empirical evidence in a Latin American context

Dear Dr. Ramirez-Madrid:

I'm pleased to inform you that your manuscript has been deemed suitable for publication in PLOS ONE. Congratulations! Your manuscript is now with our production department.

If your institution or institutions have a press office, please let them know about your upcoming paper now to help maximize its impact. If they'll be preparing press materials, please inform our press team within the next 48 hours. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information please contact onepress@plos.org.

If we can help with anything else, please email us at plosone@plos.org.

Thank you for submitting your work to PLOS ONE and supporting open access.

Kind regards,

PLOS ONE Editorial Office Staff

on behalf of

Dr. Rogis Baker

Academic Editor

PLOS ONE