Local engagement online: Municipal Facebook pages ...

2 downloads 0 Views 1MB Size Report
Apr 22, 2013 - engagement score of only 13.37 (Mdn = 1, SD = 372.05). Significant differences were found between the engagement levels generated by ...
Government Information Quarterly 32 (2015) 299–307

Contents lists available at ScienceDirect

Government Information Quarterly journal homepage: www.elsevier.com/locate/govinf

Local engagement online: Municipal Facebook pages as hubs of interaction Azi Lev-On ⁎, Nili Steinfeld School of Communication, Ariel University, Israel

a r t i c l e

i n f o

Article history: Received 15 September 2014 Received in revised form 14 April 2015 Accepted 28 May 2015 Available online 27 June 2015 Keywords: Social Media Local Government e-Government Big Data Engagement e-Participation

a b s t r a c t The article presents the results of a comprehensive study of the scope and character of the activities of the Israeli municipalities on Facebook.1 The study maps the Facebook presence of Israeli municipalities and checks for correlations between Facebook presence, municipality size, and the geographic and socio-demographic characteristics of the municipalities. We then measure several engagement indices of municipal activities on Facebook (number of fans, average likes, comments, and shares of posts) and check for correlations of these engagement indices with municipality size and geographic and social demographic characteristics of the municipalities. The article concludes with a review of several indices of municipality Facebook page activity, including scope of content uploaded by the page and by fans, content formats (statuses, images, and videos) and their reception. Published by Elsevier Inc.

1. Literature review Do social media facilitate communications between local governments and their constituencies? What characterizes the content and dynamics of the online discussions between citizens and local governments that take place on social media platforms? While many studies analyze online citizen engagement in politics at the national level and on websites of government agencies, MPs, and political parties, very few studies have explored online interactions at the local level between citizens and municipalities (Medaglia, 2012). Still, nowhere does the promise of interactivity and responsiveness online seem more viable than in the municipal arena. Arguably, studying the interaction between residents and local governments on Facebook at the municipal level is important for the following reasons (Bertot, Jaeger, & Grimes, 2012; Deakin, 2010; Shkabatur, 2010): First, decisions made at the municipal level have a direct impact on the daily lives of residents. Issues such as kindergarten and school enrolment, public transportation, welfare, and zoning regulations are decided at the local rather than national level, and municipal ⁎ Corresponding author. E-mail addresses: [email protected] (A. Lev-On), [email protected] (N. Steinfeld). 1 This study was conducted with the assistance of the institute for the study of New Media, Politics and Society at Ariel University. The authors express their gratitude to Bar Ram, Hadas Schlussel, Valerie Aronov, Varonika Makogon, Avital Mandel and Yakira Azulay for their assistance in collecting and processing the data and preparing the manuscript.

http://dx.doi.org/10.1016/j.giq.2015.05.007 0740-624X/Published by Elsevier Inc.

Facebook pages, whose content remains accessible to all may function as attractive arenas for residents to ask and respond to questions and discuss these issues. Second, not only is the number of relevant citizens significantly smaller at the local than the national level, but municipal elections and the division of seats in municipal councils are frequently decided by a small number of votes. Hence, a post that captures residents' attention (especially prior to elections) can potentially promote significant change. Thus, municipal Facebook pages have the potential to encapsulate incentives and opportunities to generate discussions and mobilize residents, and support deep and vibrant engagement at the local level. Third, studies demonstrate that municipal website usage is correlated with trust in local authorities (Tolbert & Mossberger, 2006) and with the scope of interaction between citizens and local governments (Feeney, Welch, & Haller, 2011; Garrett & Jensen, 2011). The more people use municipal websites, the more they communicate with local authorities and trust them, even more so than at the national level. Thus, municipal Facebook activities can generate beneficial results for the municipality. Still, decision makers may have reservations about establishing a municipal Facebook page since negative content, critical comments, and improper language that might appear on the page are seen by large publics. Indeed, studies demonstrate that decision makers often

300

A. Lev-On, N. Steinfeld / Government Information Quarterly 32 (2015) 299–307

view the use of interactive platforms in general and Facebook in particular as problematic and they tend to limit their use of interactive platforms in order to avoid creating platforms for problematic issues and opposition (Feeney et al., 2011; Lev-On, 2011; Moon, 2002; Mossberger, 2013; Norris & Reddick, 2013; Stromer-Galley, 2000). To summarize, municipal Facebook pages represent significant potential to improve communications between citizens and authorities, yet introduce obstacles that may prevent the authorities from making optimal usage of these social media tools. This is why the study of Facebook usage by municipalities, and mapping the landscape of municipal Facebook usage are of significant value for students of communication and public administration. Few studies have explored the characteristics of municipal egovernment and even fewer studies have studied the relatively recent phenomenon of municipal Facebook usage. Below are some important insights from these studies.

The most important predictor of diffusion of e-government (and specifically Facebook usage) has been found to be municipality size, which influences not only media adoption but also the scope of usage. Municipal websites and Facebook pages of large cities attract significantly more activity than those of small cities (Ahn, 2011; Borge, Colombo, & Welp, 2009; Garrett & Jensen, 2011; Haug, 2008; Holden, Norris, & Fletcher, 2003; Moon, 2002; Norris & Reddick, 2013; Wohlers, 2009). Additional variables that predict usage are peripherality (central/ peripheral municipalities; Haug, 2008; Norris & Reddick, 2013; Wohlers, 2009), structural attributes such as the functions in charge of maintaining the municipality's online presence (Carrizales, 2008; Norris & Reddick, 2013; Reddick & Norris, 2013; Wohlers, 2009), and population income and education levels (Reddick & Norris, 2013).

questions posted by residents on municipal Facebook walls remain without public response (Hand & Ching, 2011; Strecker, 2011), as municipalities prefer to respond through other channels. This is the first study in Israel and one of a few worldwide to focus on municipal-level Facebook usage and practices. The few studies exploring Internet adoption by municipalities in Israel have focused on municipality websites, and have shown that while Internet use by local governments in Israel is growing constantly, large and small municipalities differ in adoption scope and practices in several respects. The studies also suggest that local governments may lag in Internet adoption due to the profile of their residents (for example, UltraOrthodox or Arab populations) (Purian-Lukach, 2011; Ravitsch, 2005; Rotem, 2007). This is likely due to the conservative character of these populations, which is manifest on lower Internet penetration rates as well as more conservative usage patterns (Lev-On, 2013; Lissitsa & Lev-On, 2014). This study is also the first to use computer-based tools to analyze interactions on municipal Facebook pages. Such analysis may offer insights on the character of the interactions involving citizens and municipal authorities focusing on, for example, agendas and framing, content that becomes popular, patterns of association and responsiveness, and more (Steinfeld & Lev-On, 2015). Using such automatic tools to study “big data” enables us to dramatically expand the scope of research while simultaneously significantly reducing the time required for data analysis (Lazer et al., 2009). Facebook appears to be the most promising novel arena for econtact between municipalities and their residents. Unlike municipal websites, whose access requires special effort by users, Facebook municipality pages are much more noticeable and accessible directly from Facebook, especially where Facebook usage is high. Israel has been the world leader in Facebook usage in terms of percentage of the entire population who uses Facebook (53% in 2011, right before data for this study was collected). and average amount of time these users spend on Facebook (ComScore, 2011), and ranked 15th in the World Economic Forum's Networked Readiness Index of 2013 (Bilbao-Osorio, Dutta, & Lanvin, 2013). Hence, Israel is a promising arena for our efforts to take a fresh look at recent patterns of interactions between municipalities and residents. In Israel, state law requires every municipality to maintain a website, and close to 100% of all cities indeed have a website. The law does not require authorities to maintain a Facebook page: Facebook presence or absence and the activities that take place on the page are subject to municipality discretion. This fact adds another element to the current analysis, as it allows us to compare municipalities that maintain Facebook pages with those that do not.

1.3. Character of activities

2. Research questions and hypotheses

A recurrent finding in e-government studies is that municipality websites place greater weight on static contents, such as tenders and information on municipal activities, with much less emphasis on interactive contents (Haug, 2008; Mossberger, 2013; Musso, Weare, & Hale, 2000; Norris & Reddick, 2013; Scott, 2006; Torres, Pina, & Acerete, 2006). Municipalities tend to disregard the transactional potential of social media, choosing instead to post informational materials that also appear on other, more traditional, media (Graham & Avery, 2013; Lovari & Parisi, 2012; Oliveira & Welch, 2013; Perlman, 2012). The static character of municipality website communications is apparently reproduced in municipality Facebook pages, despite the inherently interactive character of Facebook. Several studies illustrate that municipalities rather than citizens are the dominant actors in uploading content to municipal websites (Graham & Avery, 2013; Hofmann, Beverungen, Räckers, & Becker, 2013; Magnusson, Bellström, & Magnusson, 2012). Moreover, municipal Facebook pages may create an impression of organizational indifference to the problems raised by constituencies, since many of the

The study analyzes the scope and character of activities on municipal Facebook pages. In view of the paucity of studies on this topic, several of the research questions are presented without providing corresponding hypotheses.

1.1. Scope of diffusion and usage The scope and complexity of municipal Facebook usage shows steady, incremental improvement (Deakin, 2010; Mossberger, 2013; Oliveira & Welch, 2013) and Facebook is becoming an essential medium for communication between municipalities and citizens in Europe and the US (Bonsón, Torres, Royo, & Flores, 2012; Norris & Reddick, 2013). Still, the scope of diffusion of e-government and specifically social media arenas initiated by local authorities is not uniform: Local authorities have adopted e-government and social media at different times and have been using them on different scales and levels of sophistication (Bonsón et al., 2012; Norris & Reddick, 2013). 1.2. Correlates with e-government adoption of Facebook

RQ1 What is the scope of Facebook adoption by municipalities, and what is the relation between Facebook adoption and various municipalities' characteristics? In view of the significant use of Facebook by the population, we hypothesize that a significant percentage of authorities maintain a Facebook page. Additionally, we hypothesize that: H1. Facebook page adoption will be significantly lower in municipalities populated by ultra-Orthodox or Arab populations. This hypothesis follows previous studies conducted in Israel (Purian-Lukach, 2011; Ravitsch, 2005; Rotem, 2007) that found significant differences in use between the municipal websites of these compared to authorities without significant ultra-Orthodox or Arab populations. Such differences are

A. Lev-On, N. Steinfeld / Government Information Quarterly 32 (2015) 299–307

likely due to the conservative character of these populations, which is manifest on lower Internet penetration rates as well as more conservative usage patterns (Lev-On, 2013; Lissitsa & Lev-On, 2014). H2. We also expect to find correlations between municipal-level Facebook adoption and population size, location (central/peripheral), as well as age, income, and education of the population, in line with these previous studies. RQ2 What is the scope of engagement (i.e., likes, comments, shares) with municipal Facebook pages, and what is the relation between engagement and various municipalities' characteristics? To develop an understanding of the practices on municipality Facebook pages, we collected data on several engagement metrics (number of fans, likes, comments, shares, and comment-likes). H3. The scope of engagement on Facebook will be correlated with municipality size and the socio-economic status of the population. RQ3 Who is the primary producer of the content posted on Facebook — municipalities or residents? What is the distribution of the different information formats posted — video, images, and texts? RQ4 Which content stimulates more citizen engagement — content uploaded by the municipality or by the residents? RQ5 Which content stimulates more citizen engagement — video content, images, or texts?

301

running it to extract data from pages or groups. The output includes all texts of posts and comments published by users or by the pages in a given period of time or most recent posts, subject to the user's preferences, and the Facebook engagement measures for each post (i.e., number of likes, comments, shares, comment-likes, and a total sum of all the engagement measures). In addition, the tool outputs a network file that summarizes and visualize links between users (in the case of groups) or between users and posts (in the case of official pages). A link between a user and a post is created when the user likes, shares or comments on a post. This study focuses on the content posted on the municipal pages, and therefore does not include an analysis of the networks generated by the tool. The application was used to gather all posts and comments published on official municipality pages, by either the pages or other users. We also manually documented the number of friends on each municipality Facebook page for each month since June 2013. We used Netvizz to track and summarize the activity on the pages over a period of 6 months, between October 22, 2012 and April 22, 2013, when the data was extracted. For the pages that were created during this period, we analyzed all of the posts from page launch date to April 22, 2013. No municipal elections were held or other significant events in this period that might bias the results of the study. In general, we extracted 23,768 posts and 71,338 comments from the 43 municipal Facebook pages in the study. 4. Findings

3. Methodology

4.1. General findings

To analyze the activities of local authorities on Facebook, we first generated a list of all 75 cities in Israel based on data from the Israeli Central Bureau of Statistics website (Central Bureau of Statistics 2011). To locate the official Facebook page of each city, we searched in Google's search engine and in Facebook's internal search function for the name of each city on the list in Hebrew, Arabic, and English, Israel's three official languages. The search in Arabic returned no relevant results. Three pages were found using an English search, and the remainder was found through the Hebrew search. In addition, we scanned the municipalities' websites for possible links to their Facebook pages. One municipality uses a personal profile instead of an official page and was therefore excluded from the study, due to Facebook's policy to disallow the use of personal profiles by organizations, which sometimes results in these profiles being shut down by Facebook. In a small number of cases, we found pages that had the names of the cities and concerns local affairs, yet had been opened and were administered by residents, and the cities' websites had not linked to them officially, so they were not officially associated with the municipality. An example is the Facebook page “Kiryat Gat”, presented as “the official page of Kiryat Gat”, which displays the city logo and content created by the municipality. Nevertheless, the page is run by local residents and is not officially affiliated with the municipality. Such pages were not included in the study, which focused exclusively on official municipal pages. A total of 43 Facebook pages were found and included in the study. For all 75 cities, socio-demographic data was collected from the Central Bureau of Statistics website (Central Bureau, 2011). Data included the city's socio-economic classification (cluster), population size, age median, percentage of matriculation certificate holders among all twelfth grade students in the city in the year 2010–2011 (as an indicator of average educational attainment, the most recent metric for the entire country), and geographic location (periphery cluster). These measures were chosen due to their role in predicting Internet usage in previous research. After mapping the cities' Facebook pages, we used Netvizz, a Facebook application developed by Digital Methods Initiative labs.2 The application is executed from within Facebook, and allows the user

Out of 75 cities in Israel, 43 cities (58.1%) maintained an official Facebook page when the data for the study was collected (October 2012–April 2013). The average number of posts published on each page during the six months of data collection is 552.74 (Mdn = 374, SD = 625.34), and the average number of comments is 1659.02 (Mdn = 703, SD = 2865.7). The average number of likes per post for all pages combined is 23.69 (Mdn = 2, SD = 650.26), the average number of comments per post is 3.09 (Mdn = 0, SD = 20.11), and the average number of shares per post is 2.1 (Mdn = 0, SD = 40.2). The average number of commentlikes (i.e., likes for the comments and not for the posts themselves) is 2.66 (Mdn = 0, SD = 27.03). Average engagement per post (which is a sum all of these engagement measures) is 31.55 (Mdn = 4, SD = 692.24). Figs. 1 and 2 present the distribution of posts and comments in the various municipal pages, and Fig. 3 presents the distribution of overall likes to posts made to the pages, per page. Of the 23,768 posts included in the sample, 9930 were posted by the municipal pages (page administrators) themselves (on average 248 posts per page), and 13,838 posts were posted by fans (an average of 419 user posts per page). The most prevalent post type is an image: 43% of the posts (10,117) are photos. 38% (9071) are text-only statuses, 14% (3341) of the posts include only links, and 5% (1216) are videos. In June 2013 (shortly after the conclusion of the data collection phase of this study), we also counted the number of fans on each page. The average number of fans per page is 7653.62 (Mdn = 2958, SD = 13,555.97), with results ranging from 67,413 fans (for Haifa) to a mere 45 fans only (for Or Yehuda) (Fig. 4). Examining the number of fans per page while controlling for municipality size indicates that the number of likes per page does not necessarily correspond to municipality size: Several small- and mediumsize cities lead in the number of fans divided by population size (Fig. 5).

2

We thank the Digital Methods Initiative and Bernhard Rieder for the use of the tool.

4.2. Correlations of Facebook presence with the background variables Some immediate distinctions are clearly indicated from this primary mapping. First, no Arab city or ultra-orthodox Jewish city maintains presence on Facebook. These data confirm the hypothesis H1.

302

A. Lev-On, N. Steinfeld / Government Information Quarterly 32 (2015) 299–307

Fig. 1. Posts, by municipality Facebook page.

Fig. 2. Comments, by municipality Facebook page.

A. Lev-On, N. Steinfeld / Government Information Quarterly 32 (2015) 299–307

303

Fig. 3. Likes to posts, per municipality Facebook page.

To test hypothesis H2, municipality Facebook presence as a function of city population size was tested using chi-square tests, where cities were classified into one of four population-size categories (each contains a quartile of the cities — lowest quartile with smallest populations, highest quartile with the biggest populations. The division into quartiles was done due to an abnormal distribution of population size between the cities). The relation between population size and Facebook adoption was significant (Cramer's V = .36; p b .05), where among the largest

cities — 84.2% maintain a Facebook page, while among the smallest cities only 35.3% maintain a Facebook page. When comparing cities with and without Facebook presence, we found that these two groups differed significantly by sociodemographics characteristics, which were calculated by a chi-square test: • A significant association was found between a city's sociodemographic cluster and Facebook presence, such that cities who

Fig. 4. Number of fans per page.

304

A. Lev-On, N. Steinfeld / Government Information Quarterly 32 (2015) 299–307

Fig. 5. Number of fans by municipality. The Y axis represents number of fans per thousand residents.

belong to higher socio-economic clusters were more likely to have a Facebook page (Cramer's V = .51, p b .001). Here, we grouped cities into three socio-demographic clusters, based on the classification from the Central Bureau of Statistics which uses a ten-level scale: Clusters 1–3 were classified as “low”, clusters 4–7 were classified as “medium” and clusters 8–10 were classified as the “high” sociodemographic cluster. The test found that while among the cities which belong to the high socio-demographic clusters, 73.3% maintain a Facebook page, only 11.8% of the cities which belong to the low socio-demographic clusters maintain a Facebook page. • Median population age was also found to be significantly related to Facebook presence (Cramer's V = .49, p b .001). For this test, the cities were classified into quartiles — from the “youngest” to the “oldest” population quartile. The test found that among the “oldest” cities — 80% maintain a Facebook page, compared to only 16.7% of the “youngest” cities. • Education level, measured by the percentage of matriculation certificate holders of all twelfth grade students in the city in the year 2010–2011, is also significantly related to Facebook presence (Cramer's V = .36, p b .05). Here, too, cities were classified into quartiles. The test found that cities with a higher percentage of matriculation certificate holders are more likely to maintain a Facebook page (for example, among the highest ranked cities 78.9% maintain a Facebook page, while among the lowest ranked cities only 33.3% maintain a Facebook page). • Peripheral cluster was not significantly related to Facebook presence. At the time of the research, several geographically peripheral cities attracted prominent Facebook activity.

4.3. Correlations of overall posts and comments with the background variables Municipal pages vary in the number of posts and in their engagement measures. The most inactive page belongs to the small peripheral town of Kiryat Malachi, where only one new post was published in the

6-month data collection period. The city of Nazareth-Illit came second with a total of 9 posts in 6 months, with no comments. Unsurprisingly, the two most active pages are Tel Aviv and Haifa, two of the largest cities in Israel. Tel Aviv's page had 2995 posts in the 6 months examined and 11,451 comments, and Haifa had 2189 posts with 10,458 comments. On a series of Pearson tests, significant relationships were found between various socio-economic indicators and the number of posts and comments found on municipal pages. Population size is significantly and strongly related to the number of posts on a municipal page (r = 0.59, p b .001) and the number of comments (r = 0.55, p b .001), such that the larger the city's population, the more posts and comments appear on the page (published by citizens or the municipality). The socio-economic cluster is also significantly associated with the number of posts (r = 0.29, p b .05) and the number of comments (r = 0.29, p b .05), such that the higher a city's socio-economic status, the more posts and comments appear on the municipality page. These findings confirm H3, with regard to the correlation between municipality size and the socio-economic status of the population, and the scope of activity on the page in the sense of number of posts and comments. Whereas no significant correlation was found between population education and median age and the number of posts and comments on the page, a correlation was found between a city's peripheral cluster and the number of posts and comments on the city's Facebook page: the closer the city is to the country's central area, the greater the number of posts and comments (r = 0.31 and r = 0.32, respectively, p b .05). 4.4. Distributions of engagement metrics per post Kfar Saba's municipality Facebook page has the highest mean of likes per post (between October 22, 2012 and April 22, 2013): 94.78. This is an unusual number, considering that the city with the next-highest mean number of likes per post is Ashkelon, with only 60.23 likes per post. This anomaly was caused by an unusual post on the city page, which alone attracted an outstanding 93,676 likes. This anomaly is reduced when considering the median number of likes per post in these two

A. Lev-On, N. Steinfeld / Government Information Quarterly 32 (2015) 299–307

cities: The median likes per post on the Kfar Saba Municipality page is 1, that is to say, half of the posts attract one like or fewer. On Ashkelon Municipality's Facebook page, the median is 33. Activity on this page is significantly more lively compared to the Kfar Saba municipality page. Haifa and Tel Aviv are in the fourth and eighth places, respectively, in the mean number of likes per post. Ashkelon leads the list of highest mean number of comments per post (5.73), and Rishon Lezion has the highest mean of shares per post (5.21). Note that different municipal pages lead in different engagement indices (likes, comments, shares). 4.4.1. Mean post engagement per page The overall engagement index sums up all the engagement measures: likes, comments, shares of posts, and comment-likes. This index is mainly affected by the number of likes since this is the most common type of engagement on Facebook pages, and therefore the distributions of overall engagement and of likes are quite similar. Kfar Saba leads in overall engagement per post, with a mean overall engagement score of 105.16 (the median, however, is only 4), followed by Ashkelon (mean of 72.38, median of 40), and Rishon Lezion (mean 55.99, median 3). Fig. 6 shows the distribution of overall engagement (summary of all engagement measures) per post on the various municipalities' pages. 4.5. Correlations of engagement metrics per post with the background variables A Pearson test was used to measure the relationship between population size and socio-demographic variables, and the various Facebook page engagement measures per post. The only variable that was significantly correlated with any engagement measure was the size of the population. There is a significant and strong correlation between population

305

size and mean post shares (r = 0.65, p b .001), a significant moderate correlation between population size and mean average overall engagement (r = .32, p b .05), mean comments per post (r = 0.4, p b .01), and mean comment-likes (r = 0.46, p b .001), and a weak correlation between population size and mean likes per post (r = 0.26, p b .05). For all other socio-demographic indicators — educational level, peripheral cluster, socio-economic cluster, and median age — no significant correlation was found with any engagement measures. These findings add to RQ2 and partly confirm H3, with regard to the correlation between population size and engagement with posts on the municipality's page.

4.6. Correlations between post sources, formats, and engagement Posts uploaded by the municipalities generated significantly greater engagement than posts posted by citizens. Generally, Facebook page managers are relatively more active than users when it comes to uploading content: Of the 23,768 posts collected during the study period, 9930 posts were posted by the page managers (representing the city), and 13,838 posts were posted by users. Given that managers represent one entity — the city, and users are many (dozens to hundreds of thousands in the case of large cities) it is clear that cities are more active than users on the city's page. The results of t-tests for independent samples demonstrate that the differences between engagement with page-initiated posts and user-initiated posts are significant. Thus, the number of likes per post uploaded by page managers (M = 44.99, SD = 946.98) was significantly higher than the number of likes per post by users (M = 8.41, SD = 286.82) (t(11,243.06) = 3.73, p b .001). In addition, users almost never share posts created by other users on municipal pages, compared to posts uploaded by page managers: The number of shares per post posted by municipality page managers

Fig. 6. Mean overall engagement per post by municipality.

306

A. Lev-On, N. Steinfeld / Government Information Quarterly 32 (2015) 299–307

(M = 3.73, SD = 19.09) is significantly higher than the number of shares per post posted by users (M = 0.94, SD = 50.11) (t(18,922.5) = 5.96, p b .001). These findings are consistent with the average of comments per post, while the number of comments per post posted by page managers (M = 4.48, SD = 21.86) is significantly higher than the number of comments per post posted by users (M = 2.1, SD = 18.7) (t(19,280.78) = 8.78, p b .001). In general, page manager-initiated posts have a mean overall engagement score (shares, comments, likes and comment-likes) of 56.89 (Mdn = 13, SD = 976.26), and posts of users have a mean overall engagement score of only 13.37 (Mdn = 1, SD = 372.05). Significant differences were found between the engagement levels generated by different post formats. Different post formats generated significant differences in overall engagement per post (f(4, 23,752) = 5.46, p b .001). In a follow-up Least Significance Different (LSD) test, significant differences were found in the engagement levels generated by images (M = 55.63, SD = 1058.03) compared to statuses (M = 14.97 SD = 65.49) and links (M = 7.10, SD = 34.74). In general, the posts that receive the highest engagement scores are photos: Posts that include photos received more likes, comments, and shares than any other post format. Photos received an average of 43.95 (Mdn = 6, SD = 995.12) likes (53.23 likes when posted by the page, and 24.42 likes when posted by users). Photos generated a mean of 4.10 comments (Mdn = 0, SD = 29.51) (4.74. when posted by the page, 2.77 when posted by users), and a mean of 4.13 shares (Mdn = 0, SD = 60.71) (4.21 for photos posted by pages, 3.97 for photos posted by users). Questions (Facebook surveys) were the post format that attracted the next-highest engagement scores. Only 13 posts in our sample are survey-type posts. All survey posts were uploaded by the page managers, and they attracted a mean of 22.92 likes (Mdn = 3, SD = 47.72) and 12 comments (Mdn = 1, SD = 18.17). Surveys were not shared as much as videos, which averaged 2.23 shares (Mdn = 0, SD = 13.57) (in comparison to only 0.15 average shares per survey) but only 16.62 likes on average (Mdn = 2, SD = 73.60). Fig. 7 presents engagement indicators by post format and source.

5. Discussion and conclusion The article examined the scope and features of official Facebook presence of municipalities in Israel. Previous research that examined the communications of the local governments and citizens on the Internet were conducted before the significant penetration of social networks, and focused on the websites of local governments (Purian-Lukach, 2011; Ravitsch, 2005; Rotem, 2007). This article represents an effort to fill the gap in research and map the presence of municipalities on Facebook. We found that municipalities' Facebook adoption is significant. More than one-half of the municipalities in the country (58.1%) maintained an official Facebook page at the time of data collection, with an average of 7565 fans per page. The municipalities that did not maintain Facebook presence include prominent proportion of Arab and ultra-Orthodox cities, likely due to the conservative character of their populations. The policy of these municipalities to avoid using Facebook to communicate with their residents may exacerbate even further the gaps between disadvantaged and affluent populations (Lev-On, 2013; Lissitsa & Lev-On, 2014). In line with many previous studies, the present study found that population size is the main factor that was found to influence municipality presence on Facebook. The larger the city's population, the greater the probability that the municipality will adopt and use Facebook pages (Ahn, 2011; Borge et al., 2009; Garrett & Jensen, 2011; Haug, 2008; Holden et al., 2003; Moon, 2002; Norris Reddick, 2013; Wohlers, 2009). In addition, it was found that population-related variables (socioeconomic status, education, and the median age) were significantly correlated with municipality Facebook presence. Population size not only had a significant effect on Facebook presence, but also significantly affected the scope of engagement on Facebook: Facebook pages of municipalities with larger populations attracted greater numbers of post and comments. A municipality's socio-economical cluster and location (center/periphery) also affected the volume of activity on Facebook. In general, we can conclude that urban populations are generally more likely to use Facebook excessively for engaging with their local authorities, and that predictors of internet

Fig. 7. Engagement by media format and by uploader (where page is the page manager, and user is any Facebook user).

A. Lev-On, N. Steinfeld / Government Information Quarterly 32 (2015) 299–307

usage found in previous studies (such as education level, centrality and population size) also predict activity on municipal Facebook pages. However, different authorities lead in the various engagement measures (likes, comments, shares), and in the ratio of fans to total population, and it is difficult at this point to draw decisive conclusions on the characteristics or predictors of the cities whose Facebook pages are more engaging. The findings also indicate that content posted by the municipality generated higher levels of engagement than content published by the citizens on the same page. Municipality-initiated posts attracted more likes, shares and comments, and sparked more discussions. It is important to note that the structural features of Facebook may have affected these patterns: At the time this study was conducted, Facebook page fans (users who follow an official page by clicking Like) automatically receive updates made by the page to their News Feed, but view other users' contents only when they visit the page itself or if other users are their Friends. For this reason, it is not surprising that users engage more strongly with contents uploaded by municipality pages to which they are passively exposed and which do not require users to actively access the municipality page. We also found that posts containing photos accounted the majority of all posts, and generated the highest engagement levels compared to other post formats (videos, text) on municipality pages. Thus, the most effective posts for a municipality that wishes to promote involvement of its residents, are photos, which are more eye catching than text and convey a story or message more quickly than video. As far as we know, this is the first study to use data mining and automated analysis tools to analyze Big Data drawn from municipal Facebook pages, page engagement indicators, post formats, discourse, and the content itself. While this study demonstrates the capabilities of these tools (Lazer et al., 2009), automated tools do not enable us to answer every type of question, and when it comes to perceptions of decision makers or cataloging of content, more traditional methods such as quantitative content analysis, surveys and interviews are still needed to produce the insights required for research. Further research on content types and cataloging is recommended, including efforts to identify the most popular contents that trigger citizen engagement. Such research efforts could explore differences between the texts posted by users and posts posted by the page, or between the photos users upload and the photos the page uploads. Such studies may also seek to support the findings using surveys and interviews as to gain a better understanding of residents' satisfaction, and the operating strategy of municipalities that manage their presence on social media platforms. References Ahn, M. J. (2011). Adoption of E-communication applications in US municipalities: The role of political environment, bureaucratic structure, and the nature of applications. The American Review of Public Administration, 41(4), 428–452. Bertot, J. C., Jaeger, P. T., & Grimes, J. M. (2012). Promoting transparency and accountability through ICTs, social media, and collaborative e-government. Transforming Government: People, Process and Policy, 6(1), 78–91. Bilbao-Osorio, B., Dutta, S., & Lanvin, B. (Eds.). (2013). The global information technology report 2013 growth and jobs in a hyperconnected world (http://www3.weforum.org/ docs/WEF_GITR_Report_2013.pdf). Bonsón, E., Torres, L., Royo, S., & Flores, F. (2012). Local e-government 2.0: Social media and corporate transparency in municipalities. Government Information Quarterly, 29(2), 123–132. Borge, R., Colombo, C., & Welp, Y. (2009). Online and offline participation at the local level: A quantitative analysis of the Catalan municipalities. Information, Communication & Society, 12(6), 899–928. Carrizales, T. (2008). Functions of e-government: A study of municipal practices. State & Local Government Review, 40(1), 12–26. ComScore (2011). It's a social world: Top 10 need-to-knows about social networking and where it's headed. Retrieved from www.comscore.com/it_is_a_social_world Deakin, M. (2010). A review of city portals: The transformation of service provision under the democratization of the fourth phase. In C. Reddick (Ed.), Politics, democracy and e-government: Participation and service delivery (pp. 215–224). Hershey, PA: Information Science Reference.

307

Feeney, M., Welch, E. W., & Haller, M. (2011). Transparency, civic engagement, and technology use in local government agencies: Findings from a national survey. Institute for Policy and Civic Engagement, University of Illinois at Chicago. Garrett, R. K., & Jensen, M. J. (2011). E-democracy writ small: The impact of the Internet on citizen access to local elected officials. Information, Communication & Society, 14(2), 177–197. Graham, M., & Avery, E. (2013). Government public relations and social media: An analysis of the perceptions and trends of social media use at the local government level. Public Relations Journal, 7(4), 1–21. Hand, L. C., & Ching, B. D. (2011). You have one friend request. Administrative Theory & Praxis, 33(3), 362–382. Haug, A. V. (2008). Local democracy online: driven by crisis, legitimacy, resources, or communication gaps? Journal of Information Technology & Politics, 4(2), 79–99. Hofmann, S., Beverungen, D., Räckers, M., & Becker, J. (2013). What makes local governments' online communications successful? Insights from a multi-method analysis of Facebook. Government Information Quarterly, 30(4), 387–396. Holden, S. H., Norris, D. F., & Fletcher, P. D. (2003). Electronic government at the local level: Progress to date and future issues. Public Productivity and Management Review, 26(3), 1–20. Lazer, D., Pentland, A., Adamic, L., Aral, S., Barabási, A., Brewer, D., et al. (2009). Life in the network: The coming age of computational social science. Science, 323(5915), 721–723. Lev-On, A. (2011). Campaigning online: Use of the internet by parties, candidates and voters in national and local election campaigns in Israel. Policy and Internet, 3(1). Lev-On, A. (2013). Another flew over the digital divide: Internet usage in the ArabPalestinian sector in Israel during the municipal election campaigns, 2008. Israel Affairs, 19(1), 154–169. Lissitsa, S., & Lev-On, A. (2014). Gaps close, gaps open: A repeated cross-sectional study of the scope and determinants of the ethnic digital divide. International Journal of Electronic Governance, 7(1), 56–71. Lovari, A., & Parisi, L. (2012). Public administrations and citizens 2.0: Exploring digital public communication strategies and civic interaction within Italian municipality pages on Facebook. In F. Comunello (Ed.), Networked sociability and individualism: Technology for personal and professional relationships (pp. 238–263). Hershey, PA: Information Science Reference. Magnusson, M., Bellström, P., & Magnusson, M. (2012). Facebook usage in government: A case study of information content. Proceedings of the 18th Americas conference on information systems 2012 (pp. 353–362). Red Hook, NY: Curran Associates. Medaglia, R. (2012). eParticipation research: Moving characterization forward (2006–2011). Government Information Quarterly, 29(3), 346–360. Moon, M. J. (2002). The evolution of e-government among municipalities: Rhetoric or reality? Public Administration Review, 62(4), 424–433. Mossberger, K. (2013). Connecting citizens and local governments? Social media and interactivity in major US cities. Government Information Quarterly, 30(4), 351–358. Musso, J., Weare, C., & Hale, M. (2000). Designing web technologies for local governance reform: Good management or good democracy? Political Communication, 17(1), 1–19. Norris, D. F., & Reddick, C. G. (2013). Local e‐government in the United States: Transformation or incremental change? Public Administration Review, 73(1), 165–175. Oliveira, G. H. M., & Welch, E. W. (2013). Social media use in local government: Linkage of technology, task, and organizational context. Government Information Quarterly, 30(4), 397–405. Perlman, B. J. (2012). Social media sites at the state and local levels operational success and governance failure. State and Local Government Review, 44(1), 67–75. Purian-Lukach, R. (2011). Municipal e-government: Comparative index, success factors and models for management and evaluation. In E. Cohen, & A. Lev-On (Eds.), Connected: politics, Technology and society in Israel. Tel Aviv: Israeli Political Science Association Press (Hebrew). Ravitsch, P. (2005). Difference between Israeli municipal websites, 2005: To what extent the inventory of online services fulfills the potentials of the medium and meets public expectations? (MA thesis) Bar-Ilan University, Department of Political Science (Hebrew). Reddick, C. G., & Norris, D. F. (2013). Social media adoption at the American grass roots: Web 2.0 or 1.5? Government Information Quarterly, 30(4), 498–507. Rotem, E. (2007). Municipal e-government in Israel: A comparative perspective. Tel Aviv: Hartog School for Governance and Policy (Hebrew). Scott, J. K. (2006). “E” the people: Do US municipal government web sites support public involvement? Public Administration Review, 66(3), 341–353. Shkabatur, J. (2010). Cities at crossroads: Digital technology and local democracy in America. Brooklyn Law Review, 76, 1413–1485. Steinfeld, N., & Lev-On, A. (2015). “Well done, Mr. mayor!”: Linguistic analysis of municipal Facebook pages. International Journal of E-Planning Research, 4(2), 26–38. Strecker, A. (2011). Flocking to Facebook: How local governments can build citizen engagement. Doctoral dissertation University of North Carolina. Stromer-Galley, J. (2000). On-line interaction and why candidates avoid it. Journal of Communication, 50(4), 111–132. Tolbert, C. J., & Mossberger, K. (2006). The effects of e-government on trust and confidence in government. Public Administration Review, 66(3), 354–369. Torres, L., Pina, V., & Acerete, B. (2006). E-governance developments in EU cities: Reshaping government's relationship with citizens. Governance, 12(9), 277–302. Wohlers, T. E. (2009). The digital world of local government: A comparative analysis of the United States and Germany. Journal of Information Technology & Politics, 6(2), 111–126.