Content area
Purpose
This paper aims to measure the potential for electronic participation of Brazilian citizens via Facebook as social media, identifying latent factors that provide a favorable environment for such participation by Brazilian municipalities with more than 100,000 inhabitants. Based on these factors, the Potential Index for Electronic Participation (PIEP) of municipalities is calculated.
Design/methodology/approachThe theoretical discussion is based on the literature on electronic government and citizen participation. In the methodology, exploratory factor analysis and cluster analysis have been used to identify latent factors and to classify PIEP according to the clusters.
FindingsThe results of the research point to serious regional discrepancies regarding the level of participation in social media, highlighting an urgent need for national e-government policies to be rethought from a regionalized point of view.
Originality/valueThe research enhances understanding of the relation between sociodemographic indicators such as income, education, employment and those concerning the access to and effective use of social media technologies by citizens and local governments.
1. Introduction
The phenomenon of social media in government has produced various discussions as to the benefits and limitations of the application of these tools in interactions between government and society (Alenezi et al., 2015; Khanra and Joseph, 2019). However, there is an empirical gap regarding studies that describe and analyze these interactions in developing countries (Abdelsalam et al., 2013; Seo and Hasan, 2015), with emphasis on those located in South America, where 51% of local governments in that region have an active Facebook profile (Gálvez-Rodríguez et al., 2018).
In Brazil, despite the increasing use by local governments (Brazilian Internet Steering Committee - CGI, 2018), there are still few studies on social media in the government. For example, Silva et al. (2015), studying municipalities in northeastern Brazil, observed that 70% of municipalities used Facebook as a way to communicate their actions and dialogue with citizens. Therefore, we discuss the use of social media in Brazilian municipal governments, because of the important challenges and consequences of social media use for more qualified citizen participation (Mergel, 2017).
The factors with influence capacity on citizen participation have been widely discussed over the years (Piñedo-Nebot, 2004; Sæbø et al., 2008; Vicente and Novo, 2014; Sabioni et al., 2016), but when this participation becomes eminently electronic, it brings new aspects that must be carefully studied. This paper reflects on some of these aspects, based on the theory that human action is intertwined with the given social structures, although it can change them over time (Giddens, 2009). From this perspective, this research aims to measure the potential of electronic participation via social media, identifying the latent factors that make up the favorable environment for this participation in social media in Brazilian municipalities with more than 100,000 inhabitants and, based on these factors, calculating the Potential Index for Electronic Participation (PIEP) of the municipalities.
Thus, the research contributes to a better understanding of the relation between sociodemographic indicators such as income, education, employment and those concerning the access to and effective use of these technologies by citizens and local governments. To do this, the paper is based on the literature about citizen participation determinants to develop an indicator that makes it possible to measure the potential of electronic participation via social media, based on what has already been happening in the use of these technologies.
2. Literature review
2.1 Contextual factors of the human and social developmental structure of municipalities
Electronic participation arises in the context of governance enlargement, referring to the participation of different social individuals, coming from different groups and communities, in a structure that allows changes in the relations between state and society (Braga and Gomes, 2016; Gil-de-Zuñiga et al., 2018; Ohme, 2018; Silva et al., 2019a, 2019b). Prior studies affirm that there are a number of determinants that must precede the citizen’s choice of participation in public topics, such as socioeconomic characteristics, digital skills, social networks, online development of public institutions, institutional context and social mobilization (Pinho, 2011; Vicente and Novo, 2014; Sabioni et al., 2016; Siyam et al., 2020).
In addition to these aspects, the research by Kavanaugh et al. (2014), carried out in different North American cities, brings some problematic issues to the discussion. First, the link between educational level and participation for civic purposes, when comparing people from different social groups and classes located in an environment of urbanization, presented with weak correlation. In the same way, extroverted behavior, which is typical of leaders, had no statistical relation with participation in the online environment.
Similarly, research by Vicente and Novo (2014) in Spain, about two types of electronic participation (expression of opinions and participation in petitions and online consultations), observed that once people are online, socioeconomic differences tend to disappear. On the other hand, unemployed people tended to be more participative. However, citizens with more digital skills – that is, those who had familiarity with and knowledge of the use of information and communication technologies (ICTs) – were more participative. Therefore, from the point of view of the citizen, digital skills are predictors of electronic participation.
In Latin America and the Caribbean, since the 2010s there has been an increase in the levels of education and internet use, and a greater ingress of mobile telephony. These aspects have been identified as drivers of the population’s expectations of public services, creating pressure for more participatory decision-making processes (Ricart and Ubaldi, 2016).
In a study carried out in Brazil, Sabioni et al. (2016) pointed out the importance of the context for citizens’ engagement in participation projects, where the human and social developmental factors of the municipality would be decisive for maximizing participation. Variables such as per capita income, people with higher education, employed persons, urbanization and fixed internet lines were organized like only one factor-denominated municipal structure. These are variables that interfere in citizens’ decision to inform themselves and consider electronic participation.
Given this context, it is important to analyze the phenomenon of electronic participation via social media beyond the use of technology, taking into account the issues of the human and social developmental structure of municipalities that take policies of virtualization of participation for themselves. In this sense, the routines of use of social media by citizens can be an important indicator of the possibilities of electronic participation. In the following section, this use is discussed.
2.2 Use of social media: metrics for governments and citizens
The theoretical revisions of Ngai et al. (2015) and Medaglia and Zheng (2017) take as their assumption the definition of Kaplan and Haenlein (2010), in which social media is an “internet-based technology that enables users to easily create, edit, evaluate and link content or link to other content creators as well.” When the term “technology” is used to encompass the idea of tools, services, applications and websites, it has its etymological root in the Greek word tekhnología, from the junction of tekhno-radical (tekhnē, “art, crafts, industry, science”) and the suffix-logy (from logos or “language, proposition”; Google Dictionary and online information). Thus, technology is conceptualized as a general theory or systematic study of the techniques, processes, methods, means and instruments of one or more trades or domains of human activity. This concept agrees with the one presented by Leonardi (2013, p. 60) as “a practical utility and an application of knowledge to a particular domain.”
In the current case, we can understand social media as a technology, on the assumption of its basis on the internet and with a particular domain of knowledge, focused on the interaction between humans and/or nonhumans. This considers the interference of the man–machine relation, which is especially interesting in the context of social media, mainly because of the manipulation of tendencies provided by the specialized algorithms of the software that manages and embraces social media (Tufekci, 2017; Ruediger, 2017; Lugosi and Quinton, 2018).
Hence, the use of social media can enable the government to access innovation and knowledge, through broadening interactions with different stakeholders, which can generate greater efficiency and effectiveness for government actions. However, not responding to the cravings and questions raised from the decision to use social media may indicate misuse of the tool. In other words, interaction and the internet are fundamental partners in the present days (Mergel, 2013).
The Brazilian Internet Governmental Committee (Brazilian Internet Steering Committee - CGI, 2018) has shown that a large number of public organizations in the country are present on social media, frequently updating their profiles and accounts. This technology has become routine in the actions of public communication, citizen services and institutional marketing. Taking the interaction from citizen to government as a precondition of possibilities for effective electronic participation (Abdelsalam et al., 2013; Bonsón et al., 2012; Capone et al., 2017; Silva et al., 2015), the research of these metrics becomes relevant from the managerial and academic point of view, especially at local governments (Mergel, 2017; Kagarise and Zavattaro, 2017; Bellström et al., 2016).
But despite the constant investments by governments in maintaining social media actions, characterized as staff expenses and developed technology, there is now a growing problem regarding the return on this investment, as the logic of cost vs benefits loses its strength when applied to the configuration of the public sector, whose ideal focus should be on the citizen’s engagement in quality public information, rather than on the increase of profit (Kagarise and Zavattaro, 2017).
Therefore, there is a current literature concerning the need for evidence of the effectiveness of the use of social media for these local government entities, which are pointed out in terms of both:
range or organizational reach (Mergel, 2017; Kagarise and Zavattaro, 2017), which shows that the organization has achieved the necessary relational reach to maintain the number of followers, number of page views and number of mentions (government citations on Facebook page); and
depth of relation, in which the organization has already reached its followers, but strives to capture interactivity, dialogue and knowledge sharing, also relating to the organizational ability to disseminate information, expanding the users’ engagement and feelings of approval from the generated content.
In this latter group, there are measures relating to the number of comments, reactions (the sum of different types of likes with gradations of positive and negative feelings) and sharing to the organization, as well as the number of organizational responses (Mergel, 2017; Kagarise and Zavattaro, 2017).
There are many studies about social media and electronic participation at municipalities (Bonsón et al., 2015; Silva et al., 2015, 2019a, 2019b), but, despite this, there are no indexes of electronic participation of municipal governments that are related to social media and tested with a large number of cases. The amplitude will be an index developed to measure the effectiveness of work performed online by public organizations. The depth measurements are able to bring indicators of the quality of the relationship maintained in a network, revealing the importance of the municipal governmental actors in the networks that they themselves elicit, from their online participation.
3. Research methodology
This study is a descriptive research because it aims to analyze the existing latent factors in the structure of Brazilian municipalities and the use of social media technology. It is developed with a quantitative approach, with a cross-sectional data design in 2010, according to the Brazilian Institute of Geography and Statistics (IBGE), in view of the absence of more up-to-date data from all municipalities in the study focus, and in 2017, through Netvizz application, available on Facebook site.
Facebook is an American online social media and social networking service (Facebook, 2020). Facebook is used for both communication and interaction between people and people and organizations too; the site shows tools for market, streams and other aspects to interaction. Around the world, this site has a large number of users, corresponding to the majority of online adults in countries such as the USA (Pew Research Center, 2019) and Brazil (Brazilian Internet Steering Committee - CGI, 2018).
Based on Moon’s (2002) assumption that large cities are more inclined to adopt ICTs, 276 Brazilian cities with more than 100,000 inhabitants were considered as large cities (both urban and rural municipalities), representing 89.32% of the total number of Brazilian cities classified in this population stratum (Brazilian Institute of Geography and Statistics – IBGE, 2010).
These cities were chosen because their municipal government has an active profile on Facebook, the social media platform that includes the largest number of users among Brazilian municipalities (Brazilian Internet Steering Committee - CGI, 2018). The concept of an active profile in this work was based on CGI’s (2016) e-government survey that considered the frequency of monthly posts, which is why a minimum of 100 postings was taken based on municipalities in a period of 9 months of interaction, which means more than about 10 monthly posts.
3.1 Sample and variables
The social media platform chosen was Facebook, as this represented the largest number of users among Brazilian municipalities (Brazilian Internet Steering Committee - CGI, 2018), and the collection period was nine months for each local government, between January and September 2017. The municipalities’ socioeconomic data were extracted from downloads on the IBGE and PNUD websites, and refer to the past census conducted in Brazil in 2010.
The Facebook data were accessed with Netvizz application and we followed an approach proposed by Rieder (2013). For this purpose, we first assessed the time period covered by the available data. As in 2016 there was a change of management in the Brazilian local governments, we proceeded to collect data starting from 2017, once the new mayors had taken office. Data were collected for each of the 276 municipalities studied, and were obtained from the municipalities’ respective Facebook pages.
The collection was carried out for a period of three months at a time; this decision was made because the number of posts between municipalities was divergent – for example, some cities had thousands of comments on each post, while others had only a very small number of posts per month. Thus, the responses considered for each post always lasted more than a month, but never more than three months. In addition, Facebook’s data access policy was modified after three collection periods (nine months), which prevented further access. Nevertheless, the amount of data obtained during the collection period was significant in statistical terms.
After download, all data related to the main variables studied were organized using an Excel spreadsheet, and treated and analyzed using the SPSS software. The study variables were identified from the theoretical framework (Sabioni et al., 2016; Mergel, 2017; Kagarise and Zavattaro, 2017) and their sources and definitions are presented in Table 1.
The conciliation of variables based on indicators of the existing structure in the municipality and the level of real response obtained by these governments, as well as the participation of their residents and progresses toward the consolidation of an indicator that can evaluate the potential of electronic participation in each municipality. This is especially because of the fact that the social media platform studied is already of common use in Brazil; that is, it has diminished its barriers of usability and cost of learning for the population (Brazilian Internet Steering Committee - CGI, 2018).
The sample database of this research points to about 11 million followers, 141,075 posts and 4,155,916 comments, which is why it is assumed that the study of social media in the government of Brazilian municipalities is a vast empirical field for the discussion of how citizen – government relations happen, presumed for an environment of civic and social participation.
3.2 Potential Index for Electronic Participation
The rapid pace of growth of social media and the use of this information technology by governments have contributed to raising the level of transparency, providing citizens with access to public services, and reducing information asymmetry regarding government acts (Guillamón et al., 2016). However, the degree of citizen participation fluctuates according to a combination of demographic, economic, social and technological factors that ultimately define the state of electronic participation (Ingrams et al., 2018).
To measure this degree of participation, the PIEP was developed, a synthetic indicator calculated from the latent factors resulting from exploratory factor analysis (EFA), using the principal components technique to reduce the number of variables and reach the latent constructs that express the largest portion of the variance that is explained by the data set (Hair et al., 2009).
The PIEP was developed like an indicator that attributes the level of electronic participation of residents to the municipality’s social media on Facebook in each municipality, ranging from 0 to 1, with the values close to 0 showing less potential for electronic participation in the municipality and values close to 1 indicating more participation. This indicator is similar to the E-Government Development Index (EGDI), calculated by the United Nations, to compare the level of development of e-government for 193 countries, in which EGDI is a weighted average with equal people for telecommunications, infrastructure and capital dimensions human.
The adequacy of the data for the application of EFA was evaluated using the Kaiser-Mayer-Olkin (KMO) test value, which if it is close to 1.0 indicates perfect suitability and values around 0.5 indicate no application of the EFA. The hypothesis that the correlation matrix of the variables is equal to the identity matrix was evaluated using the Bartlett sphericity test at a statistical significance of 5.0%. Components with an eigenvalue less than 1.0 were excluded from the set of constructs used in composing the PIEP. The measure of commonality was used as a criterion for the exclusion of variables with a value below 0.400, which represents the proportion of the variance that was shared with all other basic variables of the study (Hair et al., 2009).
The PIEP of the municipalities was calculated according to the model used by Santana (2007), Gomes et al. (2016) and Sabione et al. (2016) for index construction. This index aims to present in a comparative measure, the Brazilian municipalities above 100,000 inhabitants that have a greater or lesser potential for the adoption of electronic participation via social media. The final result of the indicator [equation (1)] reflects the linear combination of the main components for each of the dimensions considered by the portion of the total variance explained by each component, whose equation was defined by:
In which:
PIEPm = potential electronic participation index of municipality m;
θj = percentage of the variance explained by the factor j;
p = number of chosen factors; and
FPji = the factorial score standardized by the range method.
The results of the PIEP were used in the development of the analysis of clusters or groups, aiming to develop groups of cities with a greater potential for electronic participation. The number of distinct groups was defined by the application of the K average technique, observing the dendogram (Hair et al., 2009).
4. Results analysis
The interpretation of the statistical measures summarized in Table 2 was divided into two parts. The first part analyzes the results for variables X2, X3,…, X7, all representative of the level of electronic participation of citizens, and the second, the relative socioeconomic characteristics of the surveyed municipalities.
Table 2 shows that, on average, 511 posts were published by the municipalities, with 20,621 shares and 32,106 likes. On average, 1 post has been shared 40 times and liked 63 times. The average number of reactions, positive or negative, made by citizens was 36,865, with 72 reactions per post. The metrics obtained from the relationship of interaction measures with the number of posts were proposed from the study of Svidroňová et al. (2018). The municipality with the highest number of posts was Franco da Rocha, in the state of São Paulo, at 1,086, which were shared 20,474 times and liked 52,475 times, with an average of 19 shares and 48 likes. The municipality with the lowest number of posts in the period was Cambé, in the state of Paraná, at 66 posts, with 2,049 shares and 244 likes, on average 31 shares and 4 likes. The identified municipalities are part of the Southeast and South regions, which together comprise 56.5% of the total population of the country. The municipality of Florianópolis, in Santa Catarina in the Southern region, had the highest number of reactions to posts, 73,338, and the municipality with the lowest number of reactions was Bacabal, in the state of Maranhão located in the Northeast of Brazil, with 112 reactions. The number of reactions per post was 1,667 for Florianópolis and 0.62 for Bacabal, a municipality located in one of the poorest states in Brazil.
The second part of the variables used in this study are the socioeconomic variables, which, in the research developed by Ingrams et al. (2018), play an important role in the implementation and development stages of electronic government, known in the literature as success factors, because, depending on the level at which these variables are found, the effect on e-government can be either positive or negative.
The data in Table 2 show that the average resident population in the sample municipalities was 382,700 inhabitants, and in 218 of them (79.0%), the population was below average. The high level of dispersion shows that the municipalities are heterogeneous regarding this variable. The other variables show the homogeneity of the municipalities for each one of them because of their low level of dispersion.
4.1 Potential Index for Electronic Participation
Factor analysis (FA) requires some initial assumptions, such as the normality and linearity of the data that were achieved by logarithmic transformation. The sample size of 276 municipalities was greater than 100 observations and followed the recommendation of at least 10 observations for each variable. The KMO test result was 0.769 and shows the suitability of the data to the application of the FA technique. The value of the Bartlett sphericity test statistic was 3747.342, with a high significance to reject the hypothesis that the correlation matrix is equal to the identity matrix. The sample adequacy measure was above 0.682, which reflected the good suitability for factor extraction and factor score calculation. Thus, the application of EFA contributed to group the original variables X1, X2,…, X13 into subsets creating new mutually unrelated variables called factors. The percentage variables of the population in households with electricity and activity rate of persons over 18 years old were excluded from the data set because they presented a commonality value below 0.400, in addition to having weak correlation. The variable posting by the municipal government, despite the low commonality, was maintained because of its importance on moderate correlation with the other variables of the group.
Table 3 summarizes the factorial load of the three extracted factors, using main component analysis. They together represent 81.18% of the total variance explained by the factors after the varimax rotation, which aims to maximize the variance between the weights of each main component and with eigenvalues above the unit. The commonality is presented in the last column of Table 3 and expresses the variant portion of each variable explained by the three components.
The first factor is strongly correlated with the social and economic variables of the municipalities. It explains 29.7% of the total variance and was called human development. It refers to aspects of the municipality over which citizens do not necessarily have direct influence, but where the resources necessary to promote their participation in deciding public policies of interest to municipalities, such as income, education and access to technologies, are derived (Sabioni et al., 2016). The study by Ingrams et al. (2018) reinforces the importance of these variables in the implementation and development stages of electronic services in the municipalities.
The second factor is related to social media metrics. It explains 29.4% of the total variance and was called reach of social media, because the factor has a strong correlation with variables such as number of followers, reactions, likes and their relation to the estimated population of the municipality. This factor indicates how municipalities can approach citizens and encourage them to participate in government actions and not just access information on those actions (Kagarise and Zavattaro, 2017; Mergel, 2017). The intensity with which AMS occurs strongly depends on the gross domestic product (GDP) and physical structure of information technology available in the municipality (Ingrams et al., 2018).
The third factor explains 22.0% of the total variance. The variables with the highest level of correlation with this factor were comments, participations and posts by the government. The intensity with which they occur reflects, on the one hand, the degree of citizen engagement aspects of municipal management (Sabioni et al., 2016) and, on the other hand, the current state of the information technology infrastructure in relation to internet access and quality (Guillamón et al., 2016). This factor was called virtual relations deepening between municipal governments and citizens, which reflects the public organization’s capacity for dialogue, interactivity and knowledge sharing, and citizen engagement with the content generated by the municipal government.
4.2 Groups of municipalities by Potential Index for Electronic Participation
The PIEP general data showed the high level of heterogeneity among the municipalities, with an average of 0.528 and a coefficient of variation of 202.2%. Of the municipalities in the sample, 52.2% had below-average PIEP. The municipality of Codó in the state of Maranhão, in the Northeast region of Brazil, was the one with the lowest index (0.204). Maranhão is the Brazilian state with the second worst HDI for 2010, according to UNDP. The municipality with the highest PIEP was Curitiba (0.874), in the Southern region, the city with the fifth highest HDI in the country. In sum, the results are consistent with previous findings in the literature, where the potential for electronic participation is higher in municipalities where socioeconomic development is higher (Sabioni et al., 2016; Ingrams et al., 2018).
The general PIEP data show high heterogeneity, but they present a normal distribution according to the results of the Kolmogorov–Smirnov test. Given the high amplitude of the PIEP, the municipalities were classified into three strata obtained from the application of cluster analysis: low potential, medium potential and high potential. After creating the groups, the means were compared among them using the variance analysis technique, which rejected the hypothesis of equality of the PIEP means per group, at a level of 1.0% of significance. Table A1 shows the results of applying the Bonferroni test performed a posteriori, in which the difference between the mean of the groups taken two by two is significantly different from zero.
After the formation of the groups, the one-way analysis of variance (ANOVA) technique was used to assess whether there is a significant difference in the mean between the groups in each of the study variables. The results are summarized in Table 4 and, as the value of the F-statistic is high, the hypothesis of equality of the group means was rejected with a significance of 5.0%. The results suggest that there are three potential levels for electronic participation in the municipalities of the sample, in which 24.3% of them are part of the low-potential group; 54.0% form the group with medium potential; and 21.7% are part of the group with high potential. The results show that the averages per group are increasing in each of the variables, a characteristic that was also observed in Sabioni et al.’s (2016) study, for the variables per capita income and proportion of occupied people.
The interpretation of the results in Table 4 was achieved by aggregating the variables that characterize socioeconomic or structural aspects and the characteristic variables of citizen engagement with the municipality website in each municipality (comments and shares) with the average number of posts made by the government (Svidroňová et al., 2018).
Group 1 congregates the municipalities of the sample with low potential for electronic participation, in which the municipal HDI (0.684) is classified as average considering the developmental classification ranges, although five municipalities have an HDI classified as low. The indicators of per capita income, percentage of households with internet access and the number of people at least 25 years old with higher education in the municipalities of the low-potential group represent less than 50.0% of the value registered for the program. This is a group with greater electronic participation. In the case of higher education, this representation is 5.0%. These results are in line with previous studies that show the positive relationship between the socioeconomic development stage of municipalities and the electronic participation of citizens (Guillamón et al., 2016; Sabioni et al., 2016; Manoharan and Holzer, 2012; Ingrams et al., 2018).
Considering the set of indicators of electronic participation of citizens living in the municipalities in the sample, it is observed that all indicators have increasing average values considering the level of electronic participation of each group. For the variable average number of posts, there is a small gap between Group 1 and Group 2; however, when compared to Group 3, the value almost doubles. Using the metrics developed by Svidroňová et al. (2018), each post in Group 1 resulted in 19 likes, 23 posts, 29 shares, 24 comments and 23 followers. In Group 2, each post led to 41 likes, 47 reactions and 45 followers, but fewer comments (19) and shares (28). The indicators for Group 3 are significantly higher than for the other groups, where each post made by the municipality belonging to this group resulted in 143 likes, 165 reactions, 203 followers, 56 comments and 76 shares.
Municipalities with low participation potential have a small number of citizens who interact with their posts, where few people have a higher education background, per capita income is low and access to the internet is incipient. These are characteristics present in regions of low socioeconomic development, where the interaction with the electronic services offered boils down to the search for general information (Sabioni et al., 2016), which results in incipient levels of citizen interaction with the government. In Group 2, the interaction is higher than in Group 1 to a large extent, to the structural variables that correlate with citizens’ interaction variables with the municipal governments’ website, which ends up providing a moderate dynamic of citizen engagement (Guillamón et al., 2016; Svidroňová et al., 2018). The municipalities in Group 3 have a strong positive correlation between socioeconomic variables and measures of interaction with e-government, in particular citizens with higher education and the number of likes, comments and reactions, which leads to a strong dynamic of interaction (Ingrams et al., 2018).
4.3 Regional characteristics of Potential Index for Electronic Participation
The differences in PIEP between the groups largely reflect the existing socioeconomic inequality between Brazilian municipalities, resulting from a development policy centered on the industrialization of the South and Southeast regions of the country, while the North, Northeast and Midwest regions developed based on exports of natural and agricultural extractive products (Brazilian Institute of Geography and Statistics – IBGE, 2019). Table A1 summarizes the descriptive statistics of PIEP considering the large regions of Brazil, where the one-way ANOVA result rejects the null hypothesis of equality of averages, considering that at least one of the regions presents a different mean from the others, at a 5.0% significance level. The results of the Bonferroni test (Table A1) show that there is a difference in the means for combinations of the North and Northeast with the South, Southeast and Midwest. Among the municipalities of the sample, 63.4% are located in the South and Southeast, with the highest PIEP averages; in these regions, the productive activity contributed 70% to the national GDP in 2017 (Brazilian Institute of Geography and Statistics – IBGE, 2019) and they concentrate 56.0% of the resident population of the country. The North and Northeast regions comprise 30.0% of the municipalities in the sample, with an average PIEP of 0.488 and 0.456, respectively, and correspond to 5.6% and 14.5% contribution to GDP.
Table 5 illustrates the discrepancy between the regions. The municipalities with the highest potential for electronic participation are located mainly in the South and Southeast of the country, except in the municipalities of Salvador and Fortaleza in the Northeast and Goiânia in the Midwest. The city of Curitiba was the one with the second highest number of posts among the respondents, each of which resulted in 429 likes, 524 reactions, 812 followers, 92 comments and 116 shares. This shows that the citizens interact more with local government, because there is a great exchange of information and feedback from the citizen. This municipality had the seventh largest HDI in the country in 2010, which accentuates the positive relationship of socioeconomic development indicators and measures of citizen interaction with municipal governments, as presented in the articles by Kagarise and Zavattaro (2017), Mergel (2017) and Ingrams et al. (2018).
On the other hand, the municipalities with the lowest potential for electronic participation are located in the North and Northeast, except in the cities of Cambé, Vespasian and Itumbiara, in the South, Southeast and Midwest, respectively. Although Codó in the state of Maranhão has the lowest PIEP; the municipal government made more posts (110) than the managers of the municipality of Altamira (70 posts) in the state of Pará, where the Belo Monte hydroelectric dam was built, holding back the waters of the state. The Xingu River is to produce energy to be largely consumed by the Southeast. Citizen interaction with the media posted by the Codó municipality management resulted in 31 followers, 3 likes, 3 reactions, 25 comments and 27 shares. In the case of the posts made by Altamira’s managers, the result was 143 followers, 3 likes, 16 reactions, 28 comments and 30 shares. The places with incipient electronic interaction of citizens with municipal governments can be explained by the number of Facebook posts by governments as a way of creating citizen – government interaction (Abdelsalam et al., 2013) and by the positive relationship of participation metrics (posts, likes, comments and shares). In these cases, the interaction from the citizen did not occur or occurred to a lesser extent.
5. Discussion
This article contributes new results to the field of electronic government. It does so by analyzing the level of electronic participation of citizens residing in Brazilian municipalities with more than 100,000 inhabitants, and the use of Facebook for communication by municipal governments. Specifically, this study contributes to the field by filling a gap related to electronic participation in municipalities of developing countries, especially those located in South America (Gálvez-Rodríguez et al., 2018).
The literature contains numerous studies related to citizens’ electronic participation in municipal governments. Extant work has emphasized electronic participation’s presence, forms of use, dialogues and effectiveness in communication (Abdelsalam et al., 2013; Bellström et al., 2016; Lovari and Parisi, 2015; Marino and Presti, 2018; Svidroňová et al., 2018) and the statistical analysis of engagement metrics (Bonsón et al., 2015; Agostino and Arnaboldi, 2015). It has also produced synthetic indicators representing the dissemination of information and dialogue with citizens of municipalities, and identified determinants of government–citizen relationships, using multiple regression models (Guillamón et al., 2016; Gálvez-Rodríguez et al., 2018) with socioeconomic factors as independent variables (Ingrams et al., 2018; Manoharan and Ingrams, 2018) and applying structural equation models (Khanra and Joseph, 2019). However, research in developing countries, such as Brazil, is still incipient.
This article calculated the PIEP of Brazilian municipalities, using EFA to identify three latent factors or variables considering the correlation between socioeconomic variables and government–citizen relationship metrics. The results of the PIEP enable the identification of municipalities with less or greater potential for electronic participation, after ranking the indicator results for each municipality. This methodological procedure used represents a second contribution to the field of empirical research in electronic government and the participation of citizens. It also differs from the EGDI developed by the United Nations for 193 countries, on all continents, in that each factor should be weighted by its share of contribution in forming the total variance within the set of correlations.
In Brazil, the process of redemocratization was resumed from the mid-1980s. In the 1990s, the country implemented the reform of the Brazilian state, in which public management began to be decentralized, using information technologies, the internet and its platforms as a way to provide interaction between citizens and public administrators at the federal, state and municipal levels (Silva, 2015).
The first aspect studied in this work confirms research by Sabioni et al. (2016) on the importance of municipal structure in the context of participation. The depth factor combined the number of posts in the city with the number of comments and shares, applying a measure of citizen engagement that occurs in response to Facebook posts by the municipal government. This is in line with an approach developed by Vicente and Novo (2014), which associates greater involvement of citizens in the political issues of the municipality with the greatest use of electronic services.
The results of the PIEP show that there is a high level of heterogeneity among the Brazilian municipalities; after applying the cluster analysis, it was found that there are groups of municipalities that generate different levels of electronic participation. When comparing the groups, significant differences were observed between the means for the groups with low vs high PIEP. Variables such as population size, level of income per capita and percentage of people with complete higher education play an important role in municipalities’ development and maintenance of their information technology and digital communication structures, as discussed by Gálvez-Rodríguez et al. (2018) and Ingrams et al. (2018). The growing value of the metrics of reach and engagement of social media in Brazilian municipalities is linked to the level of support provided to users, the municipality’s socioeconomic structure, the level of digital openness, the strategies used in dialogues, the experience of citizens with social media (specifically Facebook) and the level of democracy, as noted by Gálvez-Rodríguez et al. (2018), Manoharan (2012) and Svidroňová et al. (2018).
This study revealed that municipal governments in Brazil do use social media, but there are cases in which the relationship such use generates is very basic, and can be expressed as a unidirectional government–citizen relationship (Abdelsalam et al., 2013). At the opposite extreme, some municipalities generate a high level of interaction with citizens, as their social media use provides services to citizens in addition to posts related to cultural aspects, legal instructions and public consultations (Marino and Presti, 2018).
As Brazil is a country of large expanse of territory, with 5,565 municipalities and a high level of socioeconomic inequality, the results show that municipalities with the highest level of PIEP are in the South and Southeast regions, where the level of socioeconomic development is highest, while those with lower levels of PIEP are in the North and Northeast regions, where socioeconomic indicators are low. This inequality in economic indicators and the scope and engagement from citizens related to electronic interaction with Brazilian municipalities indicates differences across the country by region and municipality, as also pointed out in studies by Ingrams et al. (2018), Manoharan (2012) and Svidroňová et al. (2018). These differences reflect, to a large extent, the degree of socioeconomic development and infrastructure of Brazilian municipalities (Brazilian Institute of Geography and Statistics – IBGE, 2018). Similar results applied to other countries, for example, the empirical results summarized by the EGDI show that the lowest indices pertain to poor African countries (Somalia – 0.0566 and Nigeria – 0.1095), while the highest are from rich European countries (Sweden – 0.8815 and Denmark – 0.9150).
6. Conclusion
The results of this study contribute to the theoretical and empirical knowledge on citizens’ electronic participation in social media, especially Facebook, with municipal governments. Use of such social media provides governments with a way to improve their transparency and accountability, and to offer citizens information that will aid their decision-making on the adoption of public policies and the improvement of participation mechanisms (Guillamón et al., 2016).
Initially, the PIEP was created based on socioeconomic data and reach and engagement at the municipal level; values ranged from 0 to 1, with those closest to 0 indicating the municipality’s lower potential for participation and those closest to 1 indicating higher potential for participation. Three groups of municipalities, namely, low, medium and high, were then created using cluster analysis techniques. The survey was conducted in 276 Brazilian municipalities with more than 100,000 inhabitants, in which all regions of the country were represented.
The PIEP results show that the average was 0.527 and the standard deviation was 0.107, while the lowest value was 0.204 and the highest was 0.875. The groups created show that the average values for each socioeconomic indicator and metric of electronic participation (number of posts, likes, reactions, comments and shares) are increasing, indicating structural and engagement difference between the groups. This result corroborates with those of Ingrams et al. (2018), who found that there are stages to the level of electronic participation that are initially related to levels of socioeconomic development, the adoption and development of electronic governments and the institutional level of governments.
The North and Northeast regions of Brazil are those with the lowest potential for citizens’ participation in social media. Based on the profile of these regions in relation to other Brazilian regions, it is assumed that these inequalities occur because of three aspects: higher level of education, higher level of income and greater acceptance of electronic media by citizens. Education contributes to raising participation levels because better-educated citizens have a greater understanding of their role in relation to governments and a greater affinity for electronic media; at the income level, higher income allows citizens to purchase electronic equipment by which they can access social media; at the level of electronic media used by governments, the initiative of electronic media use by governments opens an important communication channel with citizens and facilitates access to digital technologies.
A limitation of the research lies in the number of variables observed; some elements pointed out in the literature in the area of electronic participation, such as gender (Vicente and Novo, 2014), or even those related to the number of companies in the city (Sabioni et al., 2016), could not be cataloged for calculating the indicator either because data are not available for all 276 cities studied. The coincidental change in the institutional policy of offering data on Facebook during the last collection period also brought a limitation to the study. The addition of these and other elements could enhance the indicator created and provide deeper insights. Furthermore, this study lacked content analysis of citizens’ dialogues in response to posts; considering the positive or negative nature of posts could also generate more information about the relationship between the citizens and governments studied.
We propose that future research evaluate the effects of participation reach metrics on citizens’ engagement with electronic governments, taking the socioeconomic variables as a control, which will help to better qualify engagement considering aspects of the socioeconomic structure. Research could also subject the base data obtained here to content analysis, refining proposals made by Bellström et al. (2016) and considering the government post typology suggested by DePaula et al. (2018). Such content analysis could include statistical methods to expand understanding of citizen – government relationships.
Variables, descriptors, units of measure and source of data
| Variable | Descriptor | Unit of measure | Source |
|---|---|---|---|
| Estimated population in 2017 | IBGE estimated number of people living in the city (urban and rural areas) in a given year | Number | IBGE/Brazil |
| Activity rate of persons aged over 18 years | Ratio of persons over 18 years old who were economically active, that is, who were employed or unemployed in the census reference week, and the total number of people at this age multiplied by 100. A person who, not being employed in the reference week, had sought work in the month before this survey | Percentage | PNUD/Brazil |
| Percentage of population living in private households with electricity | Ratio of the population living in permanent private households with electricity and the total population living in permanent private households multiplied by 100. Illumination is considered as coming from a general network, with or without an energy meter | Percentage | PNUD/Brazil |
| MHDI | Municipal Human Development Index. Geometric mean of the indexes of income, education and longevity dimensions, with equal weights | Index | PNUD/Brazil |
| Per capita income | Ratio of the sum of the income of all individuals living in permanent private households and the total number of these individuals. Amounts in Brazilian reais at August 1, 2010 | Amount in reais | PNUD/Brazil |
| Percentage of microcomputers with internet access in the house | Ratio of the population living in households with microcomputers with internet access and the total population living in permanent private households multiplied by 100 | Percentage | IBGE/Brazil |
| Number of people who had finished their higher education at the age of 25 years or older | Number of people who had finished their higher education at the age of 25 years or older | Percentage | IBGE/Brazil |
| Followers | Number of profiles that automatically accompany updates of information provided by the municipal government page | Number | Local government pages on Facebook |
| Number of posts | Number of posts from each municipal government on its home page | Number | Local government pages on Facebook |
| Likes | Number of supporting symbols postings by citizens to municipal governments | Number | Local government pages on Facebook |
| Reactions | Synthesized number of different symbols to the municipal government posts, divided into six types: support, love, fun, surprise, sadness and anger | Number | Local government pages on Facebook |
| Comments | Number of written interactions by different citizens | Number | Local government pages on Facebook |
| Shares | Number of times the posted content was replicated on the platform | Number | Local government pages on Facebook |
Source: PNUD/BRAZIL, IBGE and Facebook
Exploratory analysis of the sample
| Variable | Minimum | Maximum | Average | SD |
|---|---|---|---|---|
| Estimated population in 2017 – X1 | 101,237.0 | 12,106,920.0 | 382,792.967 | 888,029.7425 |
| Followers – X2 | 2,161.0 | 858,146.0 | 41,125.746 | 74,515.9351 |
| Posts by the municipal government – X3 | 66.0 | 1,086.0 | 511.141 | 162.3479 |
| Likes – X4 | 182.0 | 671,509.0 | 32,105.873 | 62,089.7094 |
| Reactions – X5 | 112.0 | 733,386.0 | 36,864.917 | 70,887.7707 |
| Comments – X6 | 163.0 | 188,111.0 | 15,057.667 | 21,394.1348 |
| Shares – X7 | 556.0 | 209,707.0 | 20,620.859 | 24,973.7605 |
| Activity rate – 18 years or older in 2010 – X8 | 53.24 | 79.85 | 68.2326 | 4.45960 |
| % of the population in private households with electricity in 2010 – X9 | 83.19 | 100.00 | 99.4253 | 1.86463 |
| MHDI in 2010 – X10 | 0.524 | 0.862 | 0.73912 | 0.053975 |
| Per capita income in 2010 – X11 | 181.54 | 2,043.74 | 802.1842 | 321.09824 |
| % of microcomputers with internet access in private households – X12 | 3.25 | 74.07 | 36.4848 | 13.70138 |
| People with higher education – 25 years or older – X13 | 632,986 | 1,461,291.214 | 31,714.14494 | 108,944.141383 |
Source: Research results (2018)
Eigenvalues and the number of factors extracted by the main components method and the variance explained by eigenvalues
| Variable | Factors | Commonality | ||
|---|---|---|---|---|
| Human development | Reach of social media | Virtual relations deepening | ||
| MHDI in 2010 | 0.950 | 0.124 | 0.031 | 0.918 |
| Per capita income in 2010 | 0.948 | 0.164 | 0.053 | 0.929 |
| % of microcomputers with internet access in private households | 0.900 | 0.216 | −0.048 | 0.859 |
| People with higher education – 25 years old or more | 0.653 | 0.571 | 0.131 | 0.770 |
| Likes | 0.117 | 0.907 | 0.146 | 0.857 |
| Reactions | 0.122 | 0.869 | 0.297 | 0.859 |
| Followers | 0.337 | 0.747 | 0.396 | 0.827 |
| Estimated population in 2017 | 0.283 | 0.695 | 0.200 | 0.604 |
| Comments | −0.027 | 0.115 | 0.957 | 0.931 |
| Shares | 0.001 | 0.283 | 0.922 | 0.929 |
| Posts by the municipal government | 0.086 | 0.334 | 0.573 | 0.447 |
| Square rotation of the sum of the uploads | 3.268 | 3.239 | 2.423 | 8.931 |
| Percentage of the measure of the components | 29.708 | 29.449 | 22.030 | 81.187 |
Features of the clusters of the PIEP
| Variables | Group 1 – low potential | Group 2 – medium potential | Group 3 – high potential | F-statistics |
|---|---|---|---|---|
| PIEP interval | 0.204-0.462 | 0.463-0.599 | 0.603-0.875 | 24.2* |
| Number of municipalities | 67 | 149 | 60 | 26.2* |
| Average number of posts | 416.4 | 518.6 | 598.3 | 23.5* |
| Average number of likes | 8,037.9 | 21,476.5 | 85,377.8 | 36.9* |
| Average number of reactions | 9,420.5 | 24,304.2 | 98,703.7 | 38.4* |
| Average number of comments | 9,983.4 | 9,938.5 | 33,436.5 | 35.3* |
| Average number of shares | 12,001.5 | 14,370.6 | 45,767.4 | 54.2* |
| Population average | 152,253 | 231,601 | 1,015,689 | 22.7* |
| MHDI in 2010 | 0.684 | 0.746 | 0.782 | 91.7* |
| Average activity rate (%) | 65.6 | 68.8 | 69.7 | 17.5* |
| Average per capita income | 531.8 | 786.5 | 1143.3 | 99.2* |
| Average % of computers with internet | 22.5 | 37.9 | 48.4 | 99.6* |
| Average number of people with completed higher education | 5,800.7 | 12,982.8 | 107,166.7 | 21.2* |
| Average number of followers | 9,531.7 | 23,101.3 | 121,166.6 | 66.3* |
Source: Research results (2018). * The average difference is significant at the 0.05 level
Municipalities with the highest PIEP contrasted with municipalities with the lowest
| 10 municipalities with the highest potential | 10 municipalities with the lowest potential | ||||
|---|---|---|---|---|---|
| Region | Municipality | Index | Region | Municipality | Index |
| South | Curitiba | 0.8747 | Northeast | Codó | 0.204 |
| Southeast | São Paulo | 0.8604 | North | Cametá | 0.2308 |
| Southeast | Rio de Janeiro | 0.8415 | Northeast | Bacabal | 0.2424 |
| Southeast | Belo Horizonte | 0.8150 | North | Altamira | 0.2627 |
| South | Florianópolis | 0.7983 | North | Abaetetuba | 0.2802 |
| Northeast | Salvador | 0.7924 | North | São Félix do Xingu | 0.2938 |
| Southeast | Santos | 0.7785 | Northeast | Açailândia | 0.3020 |
| Southeast | Niterói | 0.7625 | South | Cambé | 0.3273 |
| Central west | Goiânia | 0.7348 | Southeast | Vespasiano | 0.33 |
| Northeast | Fortaleza | 0.7342 | Central West | Itumbiara | 0.3343 |
Source: Research results (2018)
Bonferroni a posteriori test for clusters
| Test | (I) Cluster case number** | (J) Cluster case number** | Average difference (I – J) | Standard error | Sig. | 95% confidence interval | |
|---|---|---|---|---|---|---|---|
| Lower limit | Upper limit | ||||||
| Bonferroni | Low | Medium | −0.2614* | 0.0086 | 0.000 | −0.2821 | −0.2406 |
| High | −0.1199* | 0.0069 | 0.000 | −0.1368 | −0.1031 | ||
| Medium | Low | 0.2614* | 0.0086 | 0.000 | 0.2406 | 0.2821 | |
| High | 0.1414* | 0.00818 | 0.000 | 0.1218 | 0.1610 | ||
| High | Low | 0.1199* | 0.0069 | 0.000 | 0.1031 | 0.1368 | |
| Medium | −0.1414* | 0.0081 | 0.000 | −0.1610 | −0.1218 | ||
Notes:Multiple comparisons with dependent variable: index; *Average difference is significant at the 0.05 level; **I group formed by municipalities with a category and J group formed by municipalities of a different category than I
Source: Research results (2018)
Confidence interval for average in each region
| Region | No. of municipalities | Average | SD | Standard error | 95% confidence interval | Minimum | Maximum | |
|---|---|---|---|---|---|---|---|---|
| Lower limit | Upper limit | |||||||
| Central west | 18 | 0.501 | 0.100 | 0.023 | 0.451 | 0.550 | 0.334 | 0.734 |
| Northeast | 57 | 0.488 | 0.118 | 0.015 | 0.456 | 0.519 | 0.204 | 0.792 |
| North | 26 | 0.455 | 0.112 | 0.022 | 0.410 | 0.501 | 0.230 | 0.650 |
| Southeast | 128 | 0.555 | 0.090 | 0.008 | 0.539 | 0.571 | 0.330 | 0.860 |
| South | 47 | 0.552 | 0.099 | 0.014 | 0.522 | 0.581 | 0.327 | 0.874 |
| Total | 276 | 0.527 | 0.106 | 0.006 | 0.515 | 0.540 | 0.204 | 0.874 |
Source: Research results (2018)
Bonferroni sample test
| (I) Region | Average difference (I – J) | Standard error | Sig. | Confidence interval 95% | |
|---|---|---|---|---|---|
| Lower limit | Upper limit | ||||
| Central west | |||||
| Northeast | 0.0127257 | 0.0273297 | 1.000 | −0.033064 | 0.058515 |
| North | 0.0454823 | 0.0309943 | 1.000 | −0.006447 | 0.097412 |
| Southeast | −0.0543019* | 0.0254456 | 0.337 | −0.096935 | −0.011669 |
| South | −0.0511177* | 0.0280188 | 0.692 | −0.098062 | −0.004174 |
| Northeast | |||||
| Central west | −0.0127257 | 0.0273297 | 1.000 | −0.058515 | 0.033064 |
| North | 0.0327566 | 0.0239217 | 1.000 | −0.007323 | 0.072836 |
| Southeast | −0.0670276* | 0.0160961 | 0.000 | −0.093996 | −0.040059 |
| South | −0.0638435* | 0.0199163 | 0.015 | −0.097212 | −0.030475 |
| North | |||||
| Central west | −0.0454823 | 0.0309943 | 1.000 | −0.097412 | 0.006447 |
| Northeast | −0.0327566 | 0.0239217 | 1.000 | −0.072836 | 0.007323 |
| Southeast | −0.0997842* | 0.0217444 | 0.000 | −0.136216 | −0.063353 |
| South | −0.0966001* | 0.0247061 | 0.001 | −0.137994 | −0.055206 |
| Southeast | |||||
| Central west | 0.0543019* | 0.0254456 | 0.337 | 0.011669 | 0.096935 |
| Northeast | 0.0670276* | 0.0160961 | 0.000 | 0.040059 | 0.093996 |
| North | 0.0997842* | 0.0217444 | 0.000 | 0.063353 | 0.136216 |
| South | 0.0031841 | 0.0172402 | 1.000 | −0.025701 | 0.032069 |
| South | |||||
| Central west | 0.0511177* | 0.0280188 | 0.692 | 0.004174 | 0.098062 |
| Northeast | 0.0638435* | 0.0199163 | 0.015 | 0.030475 | 0.097212 |
| North | 0.0966001* | 0.0247061 | 0.001 | 0.055206 | 0.137994 |
| Southeast | −0.0031841 | 0.0172402 | 1.000 | −0.032069 | 0.025701 |
Note:**Average difference is significant at the 0.05 level
Source: Research results (2018)
© Emerald Publishing Limited 2020