Streaming city councils' meetings on social media before, during and after the pandemic

The main objectives of this article are to understand the extent to which the COVID-19 pandemic affected the streaming of city council meetings on social media networks and whether, having this effect existed, it has persisted over time. A complementary objective is to find if there are particular characteristics of municipalities that favor the streaming of city council meetings on social media networks. To the best of the author's knowledge, this is the first study to address these objectives. To achieve them, data were collected and analyzed from all 308 Portuguese municipalities. These included the registration of 4905 streams published on Facebook and YouTube, corresponding to 2639 city council meetings from 74 municipalities, held between January 1, 2019, and December 31, 2022. The main findings are: the number of Portuguese municipalities that streamed their city councils in each month more than doubled during the acute phase of the pandemic; this effect started to fade immediately after the easing of the pandemic; and it started to grow again immediately after the September 2022 municipal elections. The main theoretical implication is that disruptive events such as sanitarian crises and local elections can trigger the adoption of innovative forms of the use of social media networks by municipalities, namely those related to transparency, such as streaming city councils. The study also contributes to confirm population size, voter turnout, and margin of victory as relevant determinants of online transparency at the local level.


INTRODUCTION
The first cases of COVID-19 were reported in Wuhan, China, in December 2019.On March 11 th of the following year, COVID-19 spread to more than 100 countries and the World Health Organization (WHO) declared it as a pandemic.Lockdowns were one of the key measures implemented by governments around the world to try to control the spread of the pandemic.These lockdowns had a tremendous impact on day-to-day life of individuals, businesses, and private and public institutions [1].
This was also the case of the political process, as public gatherings have been limited or prohibited in many parts of the world.Naturally, this also affected executive and legislative bodies, at national, regional, and local levels.City council meetings have also been impacted.In response, many have shifted to holding meetings remotely, often via videoconferencing platforms.As some city council meetings used to be public, some municipalities chose to start to stream them online.As many of these municipalities were already actively using social media networks to reach their citizens [2,3], these channels became an obvious option to start streaming the city council meetings.Considering this, the first research question of this study is: • Has the pandemic had an impact on the number of municipalities that stream city council meetings via social media networks?
As vaccination rates have increased and infection rates have declined in some parts of the world, city councils have gradually begun to return to in-person meetings.Naturally, this return to normality may have affected the number of municipalities that stream their meetings through social media networks.Considering this, the second research question of our study is: • Assuming there was an impact of the pandemic on the number of municipalities that stream city council meetings via social media networks, has it faded quickly after the end of the pandemic, or has it lasted over time?
It may have been the case that some municipalities did not start streaming their meetings on social media networks because of some intrinsic limitations, such as the lack of technical and human resources available for this purpose.Considering this possibility, the third and final research question of our study is: • Are there particular characteristics of municipalities that disfavor the streaming of city council meetings on social media networks?
The cases of the use of Facebook and YouTube to stream city council meetings by Portuguese municipalities were selected to answer these three research questions.Portugal was chosen because it has a large number of municipalities, but not so large as to make quantitative studies unfeasible [4].Facebook was selected because it is the social media network most used by Portuguese municipalities [5] and it allows live streaming of videos.YouTube was selected because it is the reference platform for live video streaming and because it is also widely used by Portuguese municipalities [6].
The remainder of this paper is organized as follows: in the next section we build the hypothesis of the study by resorting the theory and related work; in Section 3 we present the methods used to collect the data and validate the hypotheses of the study; in Section 4 we present the results; in Section 5 we discuss those results; and, finally, in Section 6 we answer the research questions and address implications and the future work.

HYPOTHESES BUILDING
As has been documented for several countries in the world [7,8,9,10,11], Portuguese municipalities are very active in social media networks [12,13].Thus, as previously addressed, they may have taken that opportunity to start streaming their council meetings on those platforms, as a response to COVID-19 lockdowns.In consequence, it can be expected that the number of Portuguese municipalities that streamed city council meetings on social media networks has increased during the height of the pandemic.In Portugal, the height of the pandemic can be considered as the period during which successive states of emergency were in force, which imposed total or partial confinements (from March 18, 2020, to March 30, 2021).Considering this, the first hypothesis of this study states as follows: • The number of municipalities that have streamed city council meetings in each month has raised during the duration of the states of emergency.
In opposite direction, it can be expected that the number of Portuguese municipalities that streamed city council meetings via social media networks has decreased after the height of the pandemic, because city councils began to return to in-person meetings.To test if this was actually the case, the second hypothesis of this study states as follows: • The number of municipalities that have streamed city council meetings in each month has lowered after the end of the last state of emergency.
In a recent study, Dias et al. [6] have demonstrated that local elections had a very visible impact in the adoption of social media network by Portuguese municipalities.This effect is characterized by "peaks in adoption that are clearly associated with inauguration dates of newly elected governments".Thus, it is plausible that the same effect may have occurred with regard to new ways of using the same technologies, as is the case with streaming city council meetings.It is relevant to consider this possibility in this study because less than six months elapsed between the end of the states of emergency (on March 30, 2021) and the municipal elections (on September 26, 2021).Therefore, it is possible that, if they existed, the two effects may have overlapped in time.Considering this, the third hypothesis of this study states as follows: • The number of municipalities that have streamed city council meetings in each month has raised after the 2021 municipal elections.
Typically, Portuguese city councils hold four meetings per month, one of which is public (and allows public intervention), and the rest are private.As this frequency is high, it is to be expected that the pandemic has not led to an increase in the number of meetings held and streamed on social media networks.Anyway, to control this possibility, the fourth hypothesis of this study states the following: • The average number of city council meetings streamed by municipalities in each month has remained stable during the period of analysis.
Several studies have established that more populous municipalities are more successful in implementing digital government initiatives [14].The same has also been observed in the case of Portugal [15].This has previously been explained as resulting from more populous municipalities being in more dynamic regions and having more technical, managerial, and financial capacity [16].Thus, it can also be expected that more populous municipalities are better able to meet the necessary conditions to stream their city council meetings on social media networks.Moreover, it has also been observed that municipalities located in more developed regions in Portugal have been more successful in encouraging the participation of citizens through their social media sites [13].Assuming that this might also happen for streaming city council meetings, the fifth hypothesis of this study states as follows: • More populous municipalities are more likely to broadcast their council meetings on social media networks than less populous municipalities.
It has previously been established that the willingness of citizen to participate in the political process at the local level contributes to online transparency [17,18,19].Thus, it could be expected that voter turnout -as a possible measure of citizens' willingness to participate -could be associated with a greater propensity of municipalities for streaming their city council meetings through social media networks.However, it has also been found that, at least in the case of Portugal, "lower levels of turnout increase local government's activity on social media" [5].Additionally, it has been demonstrated that "municipalities with voter turnouts below the average are more likely [to use social media networks] to publish" contents that "encourage the participation of citizens" [13].Assuming this might also be the case for streaming city council meetings, the sixth hypothesis of the study states as follows: • Municipalities with lower voter turnout are more likely to stream their meetings on social media networks than municipalities with higher voter turnout.
It is a well-known fact that the local political environment and the electoral competition are associated with higher levels of online transparency and online participation [20,21,22,23].In Portugal, it has been demonstrated that "higher electoral competition in local elections is associated with greater local government activity on social media" [5].Thus, it could be expected that electoral competition may also be associated with the propensity of municipalities to stream their city councils on social media networks.Considering this, the seventh hypothesis of this study states as follows: • City councils with lower margin of victory are more likely to stream their meetings on social media networks than city councils with higher margin of victory.Finally, the promotion of participation by local governments through digital means has also been associated with the political orientation of the local government and its alignment with the central government [24,25,26,27,28].In Portugal, at the time of the study, the Socialist Party ruled most of the Portuguese municipalities, had an absolute majority in the national assembly, and held the central government.Considering this, the eighth and final hypothesis of this study states as follows: • Left-wing city councils are more likely to stream their meetings on social media networks than others.

DATA AND METHODS
The methods used in the study are described in the three following subsections, including: the methods used to collect primary data (Subsection 3.1); the secondary data retrieved from official sources (Subsection 3.2); and the analyses performed to test the eight hypotheses of the study (Subsection 3.3).

Primary data
The primary data used in the study relate to the broadcasting of city council meetings on Facebook and on YouTube by Portuguese municipalities from January 1, 1999, to December 31, 2022.These data were collected specifically for the purpose of this study.The identification of the official accounts on Facebook and YouTube was performed by referring to the corresponding links in the official websites of the 308 Portuguese municipalities, as suggested by Dias, Tupia, and Bruzza [11].A total of 305 Facebook and 268 YouTube official accounts were identified by using this procedure.Once official accounts were identified, all video streams published between 2019 and 2022 on those accounts were inspected to identify those corresponding to streams of city council meetings.A total of such 4905 streams were identified, corresponding to 2639 city council meetings from 74 municipalities.For each stream, the date of publication and the corresponding link were registered.This was done in two phases, with the help of 31 undergraduate students: in the first phase, held during April 2022, the streams between January 1, 1999, and December 31, 2021, were registered; in the second phase, held during March 2023, the streams between January 1 and December 31, 2022, were also registered.The quality of the students' observations was assessed by the authors by using a sampling technique.After relevant corrections were made, the data were merged to a single spreadsheet and subsequently imported to a SQL database to allow aggregation and more elaborated analyses.

Secondary data
Table 1 presents the secondary data used in the study.These data are used to evaluate hypotheses H5 to H6 by means of statistical test, as explained in the next subsection.Four variables are used: • Population -the resident population in Portuguese municipalities as collected by the National Institute of Statistics in the scope of the last population census (2021).The variable is logarithmized to acquire an approximately normal distribution, which is relevant to the subsequent statistical analysis.It originally exhibits an exponential distribution since there are many sparsely populated municipalities and few highly populated municipalities in Portugal.The logarithmic transformation reduces the skewness of data while keeping the original monotony of variables [15].• Voter turnout -percentage of voters in the universe of registered voters for the municipal elections prior to the pandemic (2017).The variable is calculated using data available from the National Election Commission.It has originally been used by Silva et al. [5] for studying social media usage by local governments in Portugal.• Margin of victory -difference between the percentage of votes of the winning party or coalition and the percentage of votes of the second-most-voted party or coalition, in the same municipal elections (2017).The variable is also calculated using data available from the National Election Commission.It has also been previously used by Silva et al. [5].• Left political wing -dichotomous variable equal to 1 if the winning party or coalition belongs to the left wing of the political spectrum and 0 if it belongs to the write-wing or cannot be associated with one of the two wings.The variable is calculated by aggregating data from the National Election Commission on the winning parties and coalitions in the 2017 municipal elections.

Data analyses
Hypotheses H1 to H4 are evaluated by visual inspection of histograms.Two histograms were built with this purpose: an histogram showing the evolution of the number of municipalities that streamed at least a city council meeting in each month, for all months between January 2019 and December 2022, to assess H1, H2, and H3; and an histogram showing the average number of city council meetings streamed by municipalities in each month, in the same period, to assess H4.Hypotheses H5 to H8 were evaluated using statistical tests that are suited to test significant statistical differences between two independent groups of municipalities, for each of the secondary variables defined (see Table 1).For the Population variable (used to evaluate H5) the municipalities were divided into the group of municipalities that streamed at least one meeting during the entire period of analysis (from January 1, 2019, to December 31, 2022) and the group of those that did not.For the variables Voter turnout, Margin of victory, and Left political wing (used to evaluate H6, H7, and H8, respectively) the same type of grouping was used, but adapting the period of analysis: from January 1, 2019, to September 31, 2021, since a new election, with different electoral results, took place in September 2021.
The statistical tests used depended on the type of secondary variable in question: a T-test was used for the variable Population (algorithmized), because it exhibits an approximately normal distribution within groups; Mann-Whitney U tests were used for variables Voter turnout and Margin of victory, because normality within groups cannot be assumed for these variables; and a Chisquared test was used for variable Left political wing, because it is a dichotomous variable.Normality of continuous variables was assessed using the Kolmogorov-Smirnov normality test.All tests were made using SPSS 28.

RESULTS
In the following subsections we present the results of the study by analyzing the evolution of the number of municipalities that streamed city council meetings, the evolution of the average number of meetings streamed by each municipality, and the determinants of adoption of streaming of those meetings on social media networks.

Evolution of the number of municipalities
Figure 1 shows the evolution of the number of Portuguese municipalities that streamed at least one city council meeting in each month, between January 2019 and December 2022, on Facebook or YouTube.The period in which Portugal had states of emergency (lockdown period) is shaded in the Figure .Because a new municipal mandate started in October 2021, after the September 26 elections, the subsequent period is also shaded.
From Figure 1 it is immediately visible that the number of municipalities that streamed their city council meetings grew steadily during the states of emergency, from 14 municipalities in March 2020 to 40 municipalities in April 2021 (an increase of 186% during that period).It is also visible that this number began to decrease immediately after the end of the states of emergency and until the 2021 municipal elections, from 40 to 31 municipalities (a decrease of 23% during this period).After the municipal elections, the number of municipalities that streamed their city council meetings started to grow again and stabilized at 52 municipalities per month from September 2022 onwards (an increase of 68% since the municipal elections).Overall, the number of municipalities that have streamed their city councils in each month has grown 271% since just before the states of emergency (March 2019), and 420% since the beginning of the analysis (January 2019).Thus, in summary, it can be concluded that: • The number of municipalities that have streamed city council meetings in each month has raised during the duration of the states of emergency (H1 is supported).• The number of municipalities that have streamed city council meetings in each month has lowered after the end of the last state of emergency (H2 is supported).• The number of municipalities that have streamed city council meetings in each month has raised after the 2021 municipal elections (H3 is supported).

Evolution of the average number of meetings
Figure 2 shows the evolution of the average number city council meetings streamed by municipalities on Facebook or YouTube, in each month, between January 2019 and December 2022.As can be observed, the average number of meetings remained reasonably  Left political wing Chi-squared 0.627 0.429 constant throughout the period (varying between 1,10 and 1,38) and without any well-defined trend.In the graph, only the municipalities that streamed at least a meeting in a given month are included for that month.Although the highest number of meetings that a municipality streamed in one month is 11, the most frequent value (mode) is two (two meetings per month).Based on the Figure, it can be concluded that: • The average number of city council meetings streamed by municipalities in each month has remained stable during the period of analysis (H4 is supported).

Overall determinants of adoption
Table 2 presents the results of the statistical tests performed to assess relevant determinants (hypotheses H5 to H8).Statistically significant differences between groups have been found for the variables Population (logarithmized), Voter turnout, and Margin of victory, but nor for Left political wing.Based on these results, it can be concluded that: • More populous municipalities are more likely to broadcast their council meetings on social media networks than less populous municipalities (H5 is supported).• Municipalities with lower voter turnout are more likely to stream their meetings on social media networks than municipalities with higher voter turnout (H6 is supported).• City councils with lower margin of victory are more likely to stream their meetings on social media networks than city councils with higher margin of victory (H7 is supported).
• No significant association was observed between the political wing of the municipalities and the streaming of their city council meeting on social media networks (H8 is rejected).

DISCUSSION
The results presented in the previous section are clear to what the impacts of the COVID-19 lockdowns were: the number of municipalities that streamed their city council on social media networks grew significantly and steadily during the states of emergency; and this number started to diminish immediately after their end.What started afterwards, coinciding with the municipal elections, deserves further discussion.As previously addressed, it has been observed that adoption of social media networks by municipalities is highly impacted by local elections [6].In this study, we observed that the same is true when new ways of using these media networks are at stake.This is a solid conclusion.However, the concrete reasons that lead to this happening and the influence that the pandemic may have had on that process remain to be understood.Additional studies will be needed to fully explain these issues.
To what concerns the relevance of the population size, voter turnout, and the margin of victory as determinants of city council streaming on social media networks, our results are in line with the extant literature.Indeed, it has previously been observed that those are relevant determinants for social media networks adoption and use by municipalities [5,13,17,18], and of municipalities online transparency in general [19,20,21,22,23].Our results confirm these previous observations.To what concerns the irrelevance of the political wing for the results, our case contradicts some of the previous studies [24,25,26,27,28].However, it is not the first time that this irrelevance has been observed [13].Thus, our study contributes to assert the political wing as a non-universal determinant: it may or may not be relevant, depending on the concrete object of study and the concrete country case.
The main limitation of this study is that it is based in a single country case.Despite this, we believe the results are relevant because this is the first study to assess the impact of COVID-19 pandemic in streaming city councils on social media networks.Indeed, there are several examples in extant literature of studies that draw conclusions upon single cases but have latter be confirmed for other countries, namely to what concerns social media networks adoption and use by municipalities.Our expectation is that the same could happen in this case.However, it will be essential to conduct similar studies in other countries and regions of the world to confirm this.

CONCLUSIONS, IMPLICATIONS, AND FUTURE WORK
This study addressed the impact of the COVID-19 pandemic in the streaming of city councils meeting on Facebook and YouTube by Portuguese municipalities.Findings are that: (i) during the states of emerging related to the pandemic, the number of municipalities that streamed their city councils in each month increased by 186%; (ii) six months after the end of the last state of emergency, this number decreased by 23%; and after the subsequent municipal elections, the number raised again by 68%.The three effects combined, the number of municipalities that streamed their city councils in each month grew 271%.The average number of city council meetings streamed by each municipality monthly remained stable during the period of analysis.Conclusions are that the COVID-19 pandemic and the associated lockdowns had a huge impact on the number of Portuguese municipalities that streamed their city councils monthly, and that this effect started to fade immediately after the easing of the most restrictive measures.Six months later, this effect was contradicted by the occurrence of municipal elections, which triggered a new growth.Altogether, the COVID-19 pandemic and the subsequent municipal elections contributed to more than triple the number of municipalities that streamed their city council meetings in each month.
This study also contributes to confirm population size, voter turnout, and the margin of victory as relevant determinants of municipalities online transparency.Indeed, it has been proved that municipalities that have more population, had lower turnout in local elections, and had lower margins of victory in local elections are more likely to stream their city councils on social media networks than others.The same did not happen with the political wing of the ruling party, which was not confirmed as a relevant determinant.
Besides confirming determinants of municipalities' online transparency, the conclusions have another very relevant implication for theory: it was demonstrated that the disruptive effect of local elections in the adoption of innovative uses of social media networks among municipalities [6] can also happen in face of emergencies such as the COVID-19 pandemic, at least in the Portuguese case.
It remains to be demonstrated whether these results can be generalized to other countries or to the adoption of other technologies.These two questions constitute two lines for future work.A third line for future work is related to understanding the reasons behind the results obtained in this study.The first two research lines should be pursued by resorting to quantitative studies, eventually by replicating this study in other contexts.The third research line should be pursued using qualitative studies, seeking to understand the role and motivation of the main actors involved.

Figure 1 :
Figure 1: Evolution of the number of municipalities that streamed at least one city council meeting on Facebook or YouTube, monthly.

Figure 2 :
Figure 2: Evolution of the average number city council meetings streamed by municipalities on Facebook or YouTube, monthly.

Table 1 :
Secondary data used in the study.

Table 2 :
Results of the statistical tests.