In the Web of the Parties: Local Politicians on Facebook in Hungary

The study examines the Facebook use of elected local politicians over two years in Hungary. To gain insights into the role of local politicians in social‐media‐based local publics in Hungary, a large‐scale data collection has been conducted to capture the Facebook activity of all elected local representatives (mayors and councilors; N = 19,503) from the 3,152 Hungarian municipalities. Our research uncovers the level (adoption, activity) and direct audience (number of followers) of their Facebook activity and shows how these patterns are conditioned by political (party, electoral competitiveness, bandwagon effect) and contextual (size, average income of the population, development level of the local Facebook sphere) factors. We show that local politicians are mostly active in larger municipalities, while a larger proportion of the population can be reached directly in smaller communities. The activity of local politicians is largely driven by political considerations, while demand‐side factors are less important


Introduction
While the emergence of social media is usually associated with expanded global connectedness, it also helps to create densely connected local communities, thereby providing space for evolving local public spheres of actors, ideas, and actions. These digital local publics can be incredibly important to maintain local community life in the age of the "news desert" when news media consumption is increasingly focused on national media outlets (Martin & McCrain, 2019). Local politicians can be key actors in and promoters of these local digital publics, whose pages can function as central hubs of local public life (Thorson et al., 2020). In this way, their activities play an important role in the political mobilization of the local population by enabling citizens to engage with locally relevant content, helping residents to connect, and thereby developing a well-functioning local digital public sphere (Ellison & Hardey, 2013).
However, our knowledge of local politicians' social media activity is limited. People's personal social mediabased experiences with local politics differ according to their local politicians' social media activities. To understand these differences, our central goal is to uncover the patterns of local politicians' Facebook usage and the structural factors that shape them. In this study, we investigate local politicians' Facebook use in Hungary. We map the extent to which mayors and councilors use Facebook and are followed by residents and test various theoretical explanations of how political incentives and contextual factors influence their activities on this platform. For this purpose, we draw upon an exceptionally broad dataset that covers all elected local politicians from every Hungarian municipality and their Facebook activity over two years. We focus on Facebook because previous research has shown that it is the central platform of the local digital public since other social media sites tend to be more interest-or entertainment-driven with much weaker local-level activities (Thorson et al., 2020), and also because it is the only social media platform in Hungary that is extensively used politically (Bene & Farkas, 2022). Hungary is a system that is "democratic in form but authoritarian in substance" (Körösényi et al., 2020, p. 22). Since 2010, the right-wing party Fidesz has won four elections with two-thirds of the seats in parliament. These supermajorities have been used to implement centralizing reforms that have fundamentally changed the entire political system. This also applies to the field of political communication sphere: A highly centralized, extensive network of media outlets was created, including the entire public service media conglomerate, which transmits governmental actors' messages uncritically to the audience (Merkovity & Stumpf, 2021). In this context, social media has become an increasingly important mobilization channel for the opposition (Bene et al., in press). Although Fidesz's dominance has been virtually unchallenged over the past 12 years, stronger cooperation between opposition parties has led to electoral victories in several urban areas, including the capital. Thus, while the national level is dominated by a single party, the local level shows much greater diversity. Further, as the relative success of the opposition was attributed to its social media activity (Bene et al., in press), since the local election, Fidesz has made significant efforts to strengthen its position on these platforms, especially on Facebook (Bene & Farkas, 2022). Nonetheless, it is unclear to what extent local digital publics are affected by the fierce political competition between competing blocs. While the institutional autonomy of local governments has shrunk over the last decade (Dobos, 2022), the more demand-driven (contextual factors) and less politically motivated patterns of Facebook use by local elected officials suggest some kind of substantial autonomy at the communication level. In this highly polarized political context, it is therefore important to question whether local politicians' Facebook activity is driven by their political motives or the contextual characteristics of the respective municipality.

Costs and Benefits of Using Social Media in Local Politics
The literature on social media and politics demonstrated that there are benefits but also costs and even risks associated with active social media presence (for an overview, see Jacobs & Spierings, 2016). Most of these can also be relevant to activities at the local level, albeit with slightly different meanings (Mabillard et al., 2021). Benefits include politicians' opportunity to inform their constituents about locally relevant issues and their political work, to manage and perform their public image, and to interact directly with their voters. These are particularly important aspects in the local political context, where it is difficult to communicate publicly important information to the respective population, which mostly consumes national or regional news content (Martin & McCrain, 2019).
However, the costs, risks, and other demotivating factors associated with using social media are also significant. While creating a page is costless, maintaining an effective, active, and sophisticated social media presence requires resources, time, and expertise (Jacobs & Spierings, 2016). These are in short supply at the local level, especially in lower-ranked positions and smaller communities (Jacobs & Spierings, 2016). Moreover, interactivity does not have only positive sides. As at the national level, local politicians also fear losing control over their image and are concerned about reputational risk due to negative, challenging, or even aggressive interactions (Mabillard et al., 2021;Thorson et al., 2020). In a more densely linked local community, such interactions may cause more damage to politicians, as commenters are more likely to have personal relationships with the recipients of the message than in a national context. On the other hand, the lack of interactivity is also a deterring force. Local politicians are often followed less and many of their posts receive little or no response (Silva et al., 2019). This perceived indifference can demotivate political actors to invest more effort, time, and resources in their social media presence, but it can also be seen as something that threatens their reputation.
At the national level, we have extended knowledge that benefits outweigh costs and other demotivating factors, and politicians actively use social media platforms (Williams & Gulati, 2013). However, it seems that benefits and costs have somewhat different meanings in local contexts. Most importantly, fewer resources are available at the local level to address challenges and build an effective social media presence, while the "danger of interactivity" is more pronounced in both terms (i.e., challenging comments and limited engagement).
There has been little research on the social media activities of politicians at the local level. The few available studies were conducted in the early phase of social media platforms (Djerf-Pierre & Pierre, 2016;Metag & Marcinkowski, 2012) and often focused only on larger municipalities (Szmigiel-Rawska et al., 2018). This study aims to fill this gap by covering all elected local politicians from all Hungarian municipalities on Facebook. We focus on three frequently investigated dimensions of Facebook use: adoption (opening a page), activity (number of posts), and size of the directly available audience (number of followers). Given the lack of knowledge on this topic, our first research question is a descriptive one about the presence of local politicians on Facebook: RQ1: To what extent do elected local politicians, such as mayors and members of city councils, (a) use Facebook in terms of adoption and activity and (b) are followed by users in different types of municipalities?

Structural Explanations Behind Local Politicians' Social Media Use
Our second research question goes beyond the descriptive approach and investigates the structural factors that explain the activities of local politicians. Several studies have been conducted to examine the factors shaping politicians' social media performance at the national level (e.g., Williams & Gulati, 2013); however, as Larsson and Skogerbø (2018) pointed out, local politicians have markedly different social media preferences.
We argue that while the benefits and demotivating factors of using Facebook for local politicians are more or less identical, certain political and contextual factors create additional incentives for politicians to overcome the aforementioned challenges and establish an active presence on this platform. Undoubtedly, structural factors are not the only factors shaping these decisions, as personal characteristics and perceptions (Metag & Marcinkowski, 2012) may also be important. However, there are no theoretical reasons to expect that personal factors such as age, gender, education level, or attitudes toward social media operate differently in local politics than at the national level, where they have been extensively studied in the literature. In addition, a practical limitation of including personal factors in our analysis is that this information is hardly accessible to politicians in lower-ranked positions. For these reasons, we focus exclusively on structural factors that are thought to be specific to local-level political activity. By identifying these factors, we can understand why local Facebook publics function differently across municipalities. We argue that politicians' Facebook usage can be shaped by both political and context-specific considerations. Beyond uncovering the role of different political and contextual factors in these individual decisions, we are particularly interested in whether political or municipality-specific factors are more decisive in local politicians' social media usage behavior: RQ2: What political and contextual factors explain elected local politicians' (a) adoption, (b) activity on Facebook, and (c) size of the follower base?

Political Factors
Mayors and local representatives are political actors whose actions are likely to be shaped by political considerations. These political considerations may be determined both by the political background of the politicians and by the specific political context of their municipalities. Starting with the former, broadly speaking, there are two types of politicians at the local level. Some elected politicians are official representatives of national political parties, while others are independent or supported by local organizations. Party politicians arguably have the extra motivation to be active on Facebook. They are local promoters of their national parties as they represent them at the level closest to people's everyday experiences. Research has demonstrated that internal organizational pressure is a crucial factor in politicians' use of social media (Graham & Avery, 2013) and that it is a typical party strategy to reach and persuade voters at the local level (Karlsen & Skogerbø, 2015;Schäfer, 2021). Moreover, as important local information hubs, local politicians may have the benefit of being followed by non-partisan and politically less engaged voters, who are the most difficult to reach by party campaigns. Therefore, parties may pressure local politicians to engage in effective social media activities. In addition, party-affiliated local politicians may be able to draw on resources from their parties; parties can provide their local representatives with instructions, training, consultancy, and shareable materials (Klinger & Russmann, 2017). For these reasons, we expect that local politicians affiliated with national parties are more likely to use Facebook (H1a), post more actively (H1b), and gather more followers (H1c) than non-partisan politicians.
In the Hungarian context, the differences between national political blocs are also important, as the political sphere is highly unbalanced in terms of access to resources and media visibility (Körösényi et al., 2020). However, we can argue for both scenarios. On the one hand, the ruling Fidesz-KDNP party alliance is an extremely centralized and disciplined political organization (Metz & Várnagy, 2021), which makes it easier to coordinate and instruct local representatives in line with the party's national strategy. Moreover, Fidesz-KNDP is the most well-resourced actor in the political arena, and the party's local branches can also benefit from this situation. Opposition parties, on the other hand, may place greater emphasis on social media, as their access to both national and local media is very limited. In contrast, Fidesz-KNDP can rely on an extensive and locally embedded partisan media network and a more resourceintensive offline billboard and door-to-door campaigning (Bátorfy & Urbán, 2020). As a result, they may see Facebook as a complementary tool rather than a central communication tool. Given these contradictory arguments, an open research question is formulated to investigate which party bloc is more active in terms of adoption and activity and more followed on Facebook (RQ2a).
However, local politicians may be motivated to be active on Facebook not only by their parties but also by the electoral context. Even if they believe that an active social media presence provides limited electoral benefits (Mabillard et al., 2021), this small contribution (Elvestad & Johannessen, 2017) is still important in a highly competitive electoral context. Most local politicians want to maintain their positions, and if these are more engendered, the perceived benefits may outweigh the costs and risks associated with using social media. Studies have shown that heightened political competition measured by the electoral margin of the last election, significantly increases the activity of national political actors (Williams & Gulati, 2013), as well as local political actors, such as municipalities (Faber et al., 2020;Silva et al., 2019), local parties (Whitesell et al., 2022), and mayors of large cities (Szmigiel-Rawska et al., 2018). However, there is no evidence regarding the extent to which ordinary local politicians are affected by the political context. In line with the theoretical argument and previous research, we expect that local politicians in municipalities where the last election was more competitive, as measured by electoral victory, are more likely to use Facebook (H2a), be active (H2b), and be followed (H2c).
Local politicians can also be politically motivated by the examples of their peers; if elected representatives see that their fellow politicians are members of the local Facebook public, they are also incited to be actively present there. The bandwagon effect is a well-known and widely demonstrated explanation for politicians' social media activities (Klinger & Russmann, 2017). This effect was investigated at different political levels in an early study by Metag and Marcinkowski (2012), who found that it exists only at the level of local politicians regarding their attitudes toward personal websites. In line with this theory, we expect that politicians are more likely to use Facebook (H3a), be active on the site (H3b), and have more followers (H3c) in municipalities where other politicians are also present on the platform.

Contextual Factors
It is not only the political composition in which municipalities differ. The social media strategies of political actors may be shaped by the non-political contextual characteristics of their municipalities.
In smaller communities, people can be reached effectively in other ways, while mediatized communication is a greater need with a larger population (Jacobs & Spierings, 2016). In addition, the relative cost of using social media is higher when targeting a smaller population, as an active social media presence requires the same amount of time and effort regardless of audience size. In addition, perceived indifference (i.e., low levels of engagement) may be a more common experience in smaller municipalities, where a smaller number of people are targeted by communications from local politicians. The association between population size and social media use has been confirmed in studies investigating municipalities' Facebook activity (Faber et al., 2020;Guillamón et al., 2016;Lev-On & Steinfeld, 2015;Silva et al., 2019), as well as in an early study focusing on local government officials (Djerf-Pierre & Pierre, 2016). Therefore, we hypothesize that local politicians in larger municipalities are more likely to adopt Facebook (H4a), post more frequently (H4b), and be more followed (H4c).
The supply side can also be shaped by the demand side. A more social media-savvy population can be an important incentive to use social media. The general economic status of residents may play a role in this regard, as it can be seen as a proxy for the sociocultural status of the population. More privileged people tend to be more educated, more interested in politics, and more likely to use social media for news consumption; therefore, they have a greater demand for locally relevant political information (Haro-de-Rosario et al., 2018). Research on municipalities' social media use produced mixed findings: Positive (Lev-On & Steinfeld, 2015; for local parties, see Whitesell et al., 2022), negative (Guillamón et al., 2016), and insignificant relationships (Silva et al., 2019) can be found in the literature. Nevertheless, in line with our theoretical argument, we assume that local politicians in municipalities with wealthier populations are more likely to use Facebook (H5a), post (H5b), and be followed more (H5c).
Nonetheless, local politicians do not exist in a vacuum but are embedded in the local political public sphere. This is an important contextual factor that, despite its theoretical relevance, has hardly been studied in the local political context. Arguably, the role of Facebook in the local public sphere varies from municipality to municipality: In some places, it is a central part of the local community's life, while in others it plays a rather marginal role in the discussion of local issues. Previous research has shown that the presence of hyperlocal digital media outlets can stimulate online activity among local community members, including political actors (Hujanen et al., 2021). However, the role of the local Facebook public in a more general sense has not yet been studied. Accordingly, we expect local politicians' adoption rates (H6a), activities (H6b), and follower rates (H6c) to be associated with the development level of the local Facebook public because the more important Facebook's role is in a community, the more motivated local politicians are to actively use it.

Data and Methods
The local level in Hungary has a strong mayor-form horizontal power-sharing structure (Dobos, 2022). Municipalities are governed by a body of elected councilors and a directly elected mayor. While formally the councilors are responsible for decision-making and the mayor oversees executive functions, in practice the mayor can effectively control the work of the council. As we aimed to map Facebook activity across the full spectrum of Hungarian local politics, we included every elected politician, both mayors and councilors, of Hungary's 3,152 municipalities, with the exception of the capital, Budapest, which can be considered a mezzo-level unit (see Dobos, 2022). Between March and May 2021, we collected data on nearly 20,000 local politicians (3,152 mayors and 16,351 councilors) with the help of 20 coders. To ensure quality, we implemented double coding with a supervisor in charge of comparing coding results. This procedure results in highly valid observations, as only cases confirmed by two observers (at least one coder and the supervisor) are considered acceptable. The coders checked whether politicians had Facebook pages, recorded the number of their followers, and identified the local Facebook groups and official pages of the municipalities. Using this process, we captured the Facebook pages of 330 mayors and 901 councilors. To measure the activity of local politicians, we downloaded every post on their pages in the two years following the 2019 local elections (14/10/2019-12/10/2021) via CrowdTangle. The fact that we covered the entire Hungarian local public over an extended period of time means that our observations can be automatically generalized to the population. Nonetheless, to avoid treating meaningless associations as proof of our hypotheses, we draw upon common significance tests.
An important limitation of our data collection method is that, although we covered two years, the data were collected at two points in time: first, when we collected the pages, and second, when we downloaded the posts. This means that local politicians who had pages at any point during this period, but not at the time of data collection, were treated as if they had not had pages during the entire period. In addition, posts that were posted during this period but later deleted could not be included in our data collection. While it is impossible to determine the extent to which this circumstance biases our data, it is not plausible that elected officials delete pages during the first half of their term. Although personal reasons may lead a few politicians to make this decision, a small number of such cases would not bias our results given the size of our dataset. This is also true for deleted posts; it is possible that some posts were deleted, but they are probably not on a level in our aggregated dataset that could significantly distort our observations. The political context is described using two variables. Local political competitiveness is measured by the electoral margin, that is, the vote difference between the first and second strongest candidates in the 2019 mayoral election, based on data from the National Election Office. Furthermore, we categorized the partisan leaning of the local actors by identifying whether they were politicians of the governing party Fidesz, the opposition coalition, any other minor national party, a local party, or whether they were independent. The categorization was based on previously published works (Dobos, 2022;Kovarek & Littvay, 2022).
We also completed our database by using municipallevel contextual variables. Regarding population size, we used the data of the Hungarian Central Statistical Office for 2021, while we downloaded information on the per capita payroll taxes of the citizens (to measure the average level of income of the population) from the National Regional Development and Spatial Planning Information System.
To measure the intensity of the local Facebook public, we used two proxies: the number of followers of official municipal pages and the number of members of the largest local Facebook group. Clearly, the local Facebook public is a complex phenomenon with several important actors and activities (e.g., hyperlocal media outlets, local NGOs, celebrities, etc.), but in large-scale data collection, it would be difficult to obtain more fine-grained data from each municipality in a valid manner. For this reason, we opt for these two proxies, which are relatively easy to define and identify but say something relevant about overall Facebook activity at the local level. To obtain comparable results, we divide these numbers by the population size of each municipality.
Regarding the modeling strategy, separate models were fitted to mayors and councilors. Since our three dependent variables have different properties, we need to apply different regression models. Adoption was measured as a dummy variable (1 = has a Facebook page); therefore, it was explained by the logit model. The number of posts is an overdispersed count variable that was analyzed with a negative binomial regression. The number of followers was measured as a proportion of the population size of a given municipality (number of followers/population size). We fitted a linear regression model to the logarithmically transformed version of this variable. Although there is only one mayor in each municipality, there are multiple council members; therefore, in the case of councilors, our data are nested. For this reason, a multilevel modeling strategy is used in models drawing upon councilor data with random intercepts at the municipality level, while the models based on the mayors' data are simple.

Descriptive Results
In line with RQ1, we first descriptively assess the Facebook presence, activity, and follower base of local politicians. Overall, there are not any public pages for elected politicians in 85% of municipalities, but this high percentage is mainly due to the large overrepresentation of small villages (85%) in our sample; moreover, 92% of small villages, 71% of larger villages, and 38% of ordinary cities are municipalities without a public Facebook presence of elected politicians. Our results suggest that there is a connection between municipality size and adoption rate (Figure 1). Politicians are less active in smaller communities, both in terms of the existence of Facebook pages and the number of posts. The political elite of small villages is rarely present on Facebook. Above the 5,000-inhabitant threshold, the usage rate increases significantly, and nine out of 10 mayors of large cities have Facebook pages. The proportion of mayors using Facebook is much higher in each population category, and they also post more frequently than councilors. The only exception is the category of the smallest   villages, but, in this group, there are only 25 councilors who use Facebook. Facebook presence seems to be important for local councilors only in larger cities. In general, politicians' activities are not very high in either municipality type. Since the research period is 730 days, this means that even mayors of large cities publish on average less than one post per day, while mayors of smaller municipalities are characterized by an activity of one post per week. Councilors have even lower activity: They post on Facebook every three days, even in the largest municipalities.
Our results suggest that it is easier to reach citizens in smaller municipalities because a higher proportion of residents follow local politicians in these municipalities. The average mayor of a small village is followed by two-thirds of citizens, while this rate drops to one-third in larger villages and to one-fifth in towns and cities. These numbers are smaller for councilors, as they reach only 1-5% of the population in municipalities with more than 5,000 inhabitants. Table 1 shows the results of the regression models. As the (pseudo) R 2 values show, most of the models have substantial explanatory power, but leave a large part of the variance unexplained. This variance is likely associated with personal factors. Analysis of political factors shows that party affiliation influences Facebook-related behavior among local politicians ( Figure 2). All else being equal, the probability of having a Facebook page is significantly higher for Fidesz-affiliated mayors (19%) than for independent mayors (11%). While this probability is about the same for mayors from other national parties as it is for Fidesz mayors, the differences between these party-affiliated and nonparty-affiliated mayors are not significant, as the confidence intervals are quite large due to the small number of non-Fidesz-affiliated partisan mayors. In terms of activity, the predicted count of Facebook posts for Fidesz-affiliated mayors is 538, while for independent mayors it is 320, which is not significantly different from the post count of other partisan mayors.

Explanatory Analysis
However, the results for councilors paint a different picture. The probability of having a Facebook page is higher for opposition politicians (7%) than for Fideszaffiliated (2.7%) or independent (2.7%) representatives. Meanwhile, Fidesz and opposition councilors are much more active than independents. In terms of followers, there are less clear differences. Although the mayors of the two dominant political blocs (Fidesz and the opposition coalition) have more followers, there are no significant differences in the follower rates of the partisan councilors.
In conclusion, H1 can only be partially confirmed: Politicians' activities are influenced by partisan leaning, Notes: * p < 0.05, ** p < 0.01, *** p < 0.001; 1 logit-model; 2 multilevel logit-model with a random intercept on the level of municipalities; 3 negative binomial model; 4 multilevel negative binomial model with a random intercept on the level of municipalities; 5 OLS model; 6 multilevel linear model with a random intercept on the level of municipalities; for odds and incidence rate ratios values below 1 indicate a negative relationship; for OLS estimates negative relationships are indicated by a negative sign. but these effects vary for different positions and parties. Adoption and follower rates are higher for Fideszaffiliated mayors than for independents, and both ruling party mayors and councilors are more active. For opposition party mayors, the small number of cases prevents us from having robust results, but their councilors have a Facebook page with a higher probability, are more active, and have more followers than their independent colleagues. For the two largest party blocs, significant differences can only be found in the rate of adaptation of councilors, as opposition representatives are more likely to open a page.
Our results show that the intensity of the electoral contest clearly influences the adoption rate: The probability of having a Facebook page is higher in the case of local politicians in municipalities with a lower electoral margin in the 2019 election (i.e., H2a is supported). The electoral margin does not affect other factors: The politicians of more competitive local political arenas are neither more active on Facebook nor do they have more followers (H2b and H2c are rejected).
The analysis revealed some evidence of the bandwagon effect: If there is at least one politician in the local government who has a Facebook page, the mayor of that municipality will create a Facebook page with a higher probability (with no Facebook activity: 12%; with a Facebook page of a fellow politician: 21%), and they will have more followers. Similar effects are not observed for councilors; moreover, the activity of councilors even decreases with the emergence of another politician's Facebook page. The reason for this could be that this "other politician" is usually the mayor, who is more in the spotlight and likely to have precedence in sharing information (H3a and H3c are supported for mayors but not for councilors; H3b is rejected).
Regarding contextual factors, our results show that population size has a massive effect on local social media communication. The probabilities of adaptation and activity are higher for mayors and councilors of larger municipalities (H4a and H4b are supported). Interestingly, there is a negative relationship between population size and follower rate, meaning politicians reach a higher proportion of citizens in smaller communities (H4 is rejected). An increase in citizens' income has no effect on politicians' social media activity (H7a and H7b are rejected). Surprisingly, the wealthier the citizens are, the less likely they are to follow the mayor's Facebook page (H7c is rejected).
Finally, the analysis shows that the intensity of the local Facebook public has only a minor effect on politicians' activities. In municipalities where the local Facebook group has more followers relative to the population size, the councilors have a higher adoption rate; however, this does not lead to higher activity or more followers, and the follower rate of the local government's official page has no effect at all (H8a is partially supported for councilors, but not for mayors; H8b and H8c are rejected).

Conclusions
While social media is an established tool for actors in national politics, we have little knowledge of social media usage at the local level. However, it is important to increase our knowledge of local politicians' social media presence, as their activities can play an important role in shaping the amount of local content on social media platforms, and thereby, the local information supply for citizens' information consumption (Ellison & Hardey, 2013). Our research is the first attempt to explore the entire local political Facebook sphere of a single county, with the goal of mapping politicians' social media usage, not limited to larger cities.
We argued that the costs and benefits associated with politicians' social media use are different at the local level than at the national level, as fewer available resources (Jacobs & Spierings, 2016) are paired with heightened "danger of interactivity" (Mabillard et al., 2021;Thorson et al., 2020). Indeed, our findings show that social media usage at the local level is not as universal as in the national political scene, while Facebook has been generally and actively used by national politicians in Hungary for years (Bene & Farkas, 2022). Most local politicians, especially in smaller municipalities, do not open public pages. Our results also confirmed that Facebook activity is related to the type of political position: Politicians in larger communities and in a more prominent position (mayor vs. councilor) are more likely to use Facebook actively. In large cities, mayors have a similar adoption rate to politicians at the national level, while politicians in municipalities with fewer than 5,000 inhabitants rarely communicate via Facebook. Even in municipalities with populations of 5,000-10,000, the adoption rate does not reach the 50% threshold, despite the fact that popular demand for social media activity by politicians in smaller municipalities appears to be high. This is indicated by the large proportion of citizens in small communities who follow existing Facebook pages, although in smaller communities personal relationships may also motivate the following activity. Nonetheless, this is an important insight because it is a widely shared claim that social media is generally used by political actors. However, our findings suggest that this view is based on "methodological nationalism" (Chernilo, 2006) and that the use of social media in subnational political contexts is not so self-evident, even in a country with extremely high nationwide adoption and activity rate (Bene & Farkas, 2022). This also means that many people, especially those who do not live in large cities, can only connect with their national-level representatives on Facebook, which can lead them to focus on national rather than local political processes.
In light of these patterns, we aimed to uncover the structural factors that shape the level of Facebook use by local-level elected politicians. The research concludes that different types of Facebook-related outcomes are explained by different factors; moreover, these factors also differ by political position. According to our findings, local politicians' Facebook activity is primarily influenced by political factors, but less so by contextual factors, with the exception of population size. Political incentives are clearly important: Belonging to a national party motivates local politicians to be more active, as they represent and popularize their political group at the local level (Karlsen & Skogerbø, 2015). This motivation seems to be particularly important for Fidesz-affiliated mayors, while in the case of opposition parties, councilors are the more active actors. This can be explained by the specificities of the Hungarian context. The opposition is a wide coalition of several parties and councilors usually represent different parties, while opposition mayors have to represent the whole coalition and not just individual parties. Therefore, opposition councilors may be motivated to actively represent their parties in the local public sphere. These findings suggest that national parties use and support their local representatives to reach voters at the level they live (Schäfer, 2021). However, future research is needed at the content level to uncover the extent to which partisan mayors and councilors use their pages to explicitly promote their parties.
The intensity of the electoral contest primarily affects entry costs; that is, politicians of the more contested arenas tend to create Facebook pages but are not motivated to be active after the election. The electoral contest may only be relevant in the context of campaigns.
We observed the first part of the electoral cycle, when elected officials may not be very concerned about their positions. Nonetheless, although local politicians clearly respond to the local electoral contest, we can exclude the possibility that a permanent Facebook campaign is motivated by strong competitors. However, in an electoral situation, these politicians can easily activate their existing social media infrastructure and conduct more active Facebook campaigns. The presence of other politicians on Facebook stimulates mayors' Facebook presence, which also shows that mayors are sensitive to the local political context. This is a sign of a local "bandwagon" effect (Metag & Marcinkowski, 2012) but it is limited to mayors. Councilors are not affected by the Facebook presence of other politicians, perhaps because the "other" politician is usually the mayor, who is perceived as standing at a different level, where Facebook presence is more of a requirement than at the level of councilors.
As noted earlier, the most important influencing contextual factor is population size. However, politicians in smaller municipalities have the comparative benefit of being able to reach a larger portion of the population than elected representatives in larger communities. Mayors and councilors in larger municipalities face an uncomfortable situation: While they need to use social media platforms to communicate with constituents, it is not enough to rely solely on their central page, as it still has a limited reach. City mayors and councilors need to draw on other social media-based (e.g., Facebook groups, ads) or offline channels to impact their voters. This aspect has yet to be explored, as existing studies only emphasize the positive effect of population size on social media activity (Faber et al., 2020;Guillamón et al., 2016;Lev-On & Steinfeld, 2015;Silva et al., 2019) and ignore the audience dimension.
The local Facebook context does not seem to be an important factor, and politicians' strategies are less related to the state of the local Facebook public. This is surprising given the few reports showing that welldeveloped local Facebook publics are integrated, densely connected, and interactive entities that play an important role in local political processes (e.g., Thorson et al., 2020). It seems that while politicians are receptive to the local political context, they are less sensitive to the demands of the local public; their activity in this context is not demand-driven. Nevertheless, it is important to note that the extension or development of the local Facebook public is measured by two proxy variables that are raw indicators of the concept. For instance, hyperlocal media (Hujanen et al., 2021) and local NGOs (Thorson et al., 2020) play a crucial role in local public spheres; however, methodological limitations prevented us from including them in this study. Future studies may help to understand how political actors' social media activities and the local public sphere as a whole relate to each other.
The main limitation of our study is the generalizability of our findings to a different context. Our research employed a uniquely large dataset; however, it is limited to the Hungarian political context. We assume that political incentives and population size may be important factors in other countries as well, but future studies should confirm this assumption. Nonetheless, this is the first study that covers politicians from the entire field of local politics and is not limited to large cities. Moreover, our empirical investigation was limited to a few key dependent and independent variables; however, other factors could be at play. We have already mentioned other potential actors in the local public sphere, but beyond structural aspects, personal factors may also be relevant. An important but unavoidable limitation of our approach is that only public Facebook pages are considered. Mayors and councilors may use their private profiles for political communication, especially in smaller municipalities. However, these private profiles cannot be included in large-scale data collection because it is difficult to define a "threshold" from which a private profile can be considered "official," and also because Facebook does not allow data scraping from private pages. Nonetheless, because of this fact, we might underestimate the extent of Facebook activity at the local level if we do not take into account these "semi-public" forms of Facebook communication. Finally, our study focused only on raw indicators of Facebook activity but did not address actual communication by mayors and councilors on these platforms. Future studies should employ content analysis to determine how local politicians use social media platforms.
If we agree with the proposition that the presence of local politicians on Facebook plays a key role in the development of local Facebook public and the provision of locally relevant information to users (Ellison & Hardey, 2013;Thorson et al., 2020), it is important to note that, based on our results, this effect may be most prevalent in larger municipalities, where national parties are important actors in local political life. This conclusion suggests that local public activity is largely driven by national political interests, raising questions about the substantive autonomy of local political spheres. This is a particularly important question in the Hungarian context, where the institutional autonomy of local politics is shrinking and strong centralizing tendencies prevail (Dobos, 2022). Independent, locally embedded elected representatives can counter these trends by contributing to the functioning of strong local publics, but it seems that these actors are reluctant to rely more on social media. The lack of partisan incentives for these actors is not counterbalanced by the demands of the local population. Another important lesson that can be drawn from the analysis of the Hungarian case is that the local public does not create itself; the key actors of the local public sphere do not become active because there is a demand from citizens for local public life on social media. It seems that politicians do not react to local conditions, but rather are motivated by external factors such as the interests of their nominating parties and the intensity of the electoral contest. Therefore, it can be assumed that the local public is an area for the mobilization efforts of national parties and that local politicians serve as local promoters of their national parties (Karlsen & Skogerbø, 2015). However, it is a task for future research to assess the role that national political interests play in the local digital public.