Municipal Campaigns on Facebook: What influences the scope of engagement and does it win votes?

Full text

Municipal Campaigns on Facebook: What influences the scope
of engagement and does it win votes?
Azi Lev-On
School of Communications, Ariel University, Ariel, Israel
Nili Steinfeld
School of Communications, Ariel University, Ariel, Israel
ABSTRACT
The paper examines which variables predict engagement on Face-
book pages of candidates running in municipal elections, and
whether candidate Facebook activity predicts vote share as much
as institutional and population-level variables. Using the Facebook
activity of 387 candidates running in municipal elections in Israel
as a case study, it was found that institutional variables, primarily
number of candidates, incumbency status and size of constituency,
signicantly correlate with candidate vote share. Number of likes
candidates receive for their Facebook posts mediates a weak re-
lationship between number of posts the candidate uploads to the
platform, number of page fans and population size, to candidate
vote shares. Number of candidates and incumbency status are the
strongest indicators of candidate vote share. The ndings suggest
that Facebook activity is a signicant, positive, but weak indicator
of municipal election results.
CCS CONCEPTS
•Human-centered computing →
Web-based interaction;
•So-
cial and professional topics;
KEYWORDS
Social Media, Political Communication, e-Campaigning, Municipal
Government, Facebook, Engagement
ACM Reference Format:
Azi Lev-On and Nili Steinfeld. 2021. Municipal Campaigns on Facebook:
What inuences the scope of engagement and does it win votes?. In
DG.O2021: The 22nd Annual International Conference on Digital Govern-
ment Research (DG.O’21), June 09–11, 2021, Omaha, NE, USA. ACM, New
York, NY, USA, 9 pages. https://doi.org/10.1145/3463677.3463689
1 INTRODUCTION
Facebook has emerged as a central communicative arena for politi-
cal representatives and constituencies. Scholars have long pointed
to the political potential of the Internet, not merely for information
distribution or image building, but also as a tool to generate and
strengthen the interaction between political representatives and the
public [
6
,
7
,
55
] Facebook makes the possibilities of such interaction
all the more viable.
Permission to make digital or hard copies of all or part of this work for personal or
classroom use is granted without fee provided that copies are not made or distributed
for prot or commercial advantage and that copies bear this notice and the full citation
on the rst page. Copyrights for components of this work owned by others than ACM
must be honored. Abstracting with credit is permitted. To copy otherwise, or republish,
to post on servers or to redistribute to lists, requires prior specic permission and/or a
fee. Request permissions from permissions@acm.org.
DG.O’21, June 09–11, 2021, Omaha, NE, USA
©2021 Association for Computing Machinery.
ACM ISBN 978-1-4503-8492-6/21/06. . . $15.00
https://doi.org/10.1145/3463677.3463689
Studying Facebook usage in political campaigns attracts schol-
ars because of the signicant usage of Facebook by national-level
candidates [
11
,
13
,
22
,
31
,
39
,
42
,
54
], and due to the public interest
these campaigns generate.
Israel is a promising arena to study online campaigning. Accord-
ing to Internet World Stats, in December 2013 (immediately after
data for this study were collected), the rate of Internet penetration
in Israel was 70.8%, compared to the global average of 34.3%, and
has reached 87% by end of 2019 [
57
]. Israel is also a world leader in
Facebook usage: No less than 81% of all Internet users in Israel today
use Facebook regularly, and it is the second most common social
network in Israel after WhatsApp [
2
]. Members of the Parliament
(MPs) also have a signicant presence on Facebook. As of 2017, 108
of the 120 MPs had active Facebook pages [
59
] which are extremely
active and used by the MPs, and even more so - by users [
40
]. There
is evidence of a substantial penetration of social media at the mu-
nicipal level as well: In 2013, 43 of Israel’s 75 municipalities had
active Facebook pages [29].
Still, the study of the political usage of Facebook is lacking in
many respects. First, there are only a handful of studies about the
scope of Facebook usage during municipal election campaigns (for
exceptions see [
25
,
38
,
50
]). Second, existing studies usually do not
reference the various aspects of candidate activities on Facebook,
such as number of fans, number of posts per page, and audience
reactions. Addressing this lacuna is the signicant and unique con-
tribution of the current study. Furthermore, this study uses auto-
mated computerized tools to analyze municipal-level issues usually
studied using manual content analysis methods. The use of comput-
erized tools can supply preliminary insights on a massive volume
of data signifying interactions between citizens and municipal can-
didates, “from a bird’s eye view”, which manual analysis methods
will not be able to achieve [41].
2 THEORETICAL BACKGROUND
Scholars have long praised the relative advantage of Internet interac-
tivity over other media in political campaigns [
34
,
45
,
46
]. However,
studies that analyzed the political usage of the Internet found that
interactive features, such as forums and chatrooms, were rarely
used by candidates. Studies also demonstrated that (with excep-
tions) o-line and online campaign experiences were similar, and
the signicant majority of campaign websites primarily included
unidirectional messages from politicians to their followers, with
very few interactive functions. Websites tended to be active exclu-
sively during the campaign season, and, even then, the majority of
parties were risk averse in their online activities, focusing mainly
on the website as a tool to disseminate information and to make
secondary use of content that was created for the oine campaign,
while interactive tools were used rarely and were closely monitored
[10, 31].

DG.O’21, June 09–11, 2021, Omaha, NE, USA Azi Lev-On and Nili Steinfeld
Studies in Israel also consistently found that the interactive use
of the Internet has been almost absent in a variety of political
contexts, including the websites of MPs [
14
], municipalities [
36
],
and municipal web campaigning in 2008, municipal elections that
preceded the campaigns covered in this paper [27].
In a classic study, Stromer-Galley [
43
] found that politicians tend
to be cautious of interactive usage of the Internet for three main
reasons. First, the ability to maintain contact with the public online
can make the campaign more costly, as additional personnel are
required to operate the digital channels. Second, direct contact with
the public may invite negative messages and criticism that expose
problematic issues on the candidate’s own page. Third, to maintain
an eective dialogue with the public, politicians must provide clear,
well-formulated answers to questions posed by the public, whereas
politicians generally prefer to address wider audiences through
general and ambiguous messages.
Still, the rapid, massive penetration of Facebook may have
changed this state of aairs. Facebook is free, user-friendly, and
widely used by the public; Facebook may be more attractive as a
channel of political communications to the public, since Facebook
campaign pages may be easily reached while browsing the social
network, in contrast to candidate websites that must be intention-
ally accessed.
Facebook is, however, problematic due to the lack of options it
aords to control content, design, and community management,
which explains the cautiousness of political players with respect
to online interaction with the public. Bürger and Ross [
1
] found
that Facebook was used by candidates as a supplemental channel to
convey their message in the same way they used traditional media,
without trying to create genuine engagement with their constituen-
cies. Other studies demonstrate that while many candidates spoke
about creating a direct channel to citizens, most have used social
media to send unidirectional messages, and their main goal, by
updating statuses, is to show that they are present on Facebook, not
to start public debates that might become heated and spiral out of
control [
20
]. For example, Vergeer et al. [
49
], who studied electoral
twitter usage by the European Parliament in 2009, observed that
tweeting was performed "reluctantly”, and Twitter was adopted by
only a fraction of all candidates (only 36 out of the 288 candidates
had Twitter accounts).
Change has nonetheless taken place in recent years, as Facebook
pages have become a signicant tool for interactions between rep-
resentatives, contenders for public oces, and the public itself, and
are being used by a growing number of politicians during their
term in oce and while campaigning. A signicant milestone in
the growing awareness of potential Facebook functions in political
campaigning is the success of Obama’s campaign for the US presi-
dency in 2008, in which social media, and particularly Facebook,
were used extensively to reach potential voters, expose the candi-
date, listen to and provide an outlet for voter opinions, and also to
mobilize voters to the polls [1, 21, 48].
Consequently, there is evidence of an increase in the political
use of Facebook in recent years to reach dierent audiences and
to recruit volunteers, party members and resources. Many politi-
cians open pages on Facebook in response to pressure by party
colleagues, other candidates, and even voters themselves [
8
,
26
].
Many politicians maintain a high prole of activity throughout the
year and not only during campaigns [
18
,
31
]. For example, In the
US, in 2012, 90% of representatives and senators were registered
with Facebook [12].
2.1 Political uses of Facebook at the municipal
level
The focus of this municipal-level research is justied for several
reasons. First, addressing the municipal elections as the unit of
analysis allows us to analyze Facebook campaigning at a higher
resolution compared to analyses of national-level parliamentary
elections. Municipal elections include numerous candidates who
run simultaneously in a large number of municipalities, which
makes it possible to construct a broader and more ne-grained
picture of the distribution of Facebook usage for political purposes,
allowing for a study of correlations between candidate Facebook
usage and consequent vote shares in the polls.
Second, in contrast to parliamentary elections, municipal elec-
tions have a signicantly smaller number of voters, and outcomes
may be decided by a small number of votes. In cases of close compe-
tition, Facebook may be the instrument that distinguishes victory
from defeat at the polls. Therefore, once a signicant percentage
of the constituency become active Facebook users, politicians are
likely to pay close attention to the Facebook arena.
Third, partisan identication is weaker in municipal than in na-
tional parliamentary elections, and the strength of partisan identi-
cation is often replaced with assessments of contender competence
or incumbent performance. Since voting behavior appears to be
better predicted by variables such as political ideology and party
aliation than by media messages [
19
] a study of Facebook’s po-
litical impact might be better carried out at the municipal rather
than national level. Reduced partisan identication at the municipal
level leads to greater openness to persuasion through Facebook or
other communication channels.
Furthermore, the relatively low cost of Facebook campaigning is
particularly important at the municipal level, where campaign bud-
gets are signicantly lower than in the campaigns for nationwide
elections.
Which variables inuence the scope of Facebook activity of mu-
nicipal election candidates? In view of the absence of literature
that directly addresses this question, hypotheses can be derived
from current knowledge of the use of websites in municipal cam-
paigns (excluding Facebook) and of Facebook usage of municipali-
ties (municipal e-government). Studies illustrate that the adoption
of e-government and especially social media platforms by munici-
palities is not uniform; not all municipalities adopted e-government
tools at the same time, or at the same scope or level of sophistica-
tion. According to studies, the most important factor that inuences
municipal adoption of e-government, in general, and Facebook, in
particular, is the size of the constituency. The larger the munici-
pality is, the more dicult it becomes to provide information and
respond to the needs of diverse population groups, and the more
advantages social media channels oer in the provision of services
compared to traditional channels. The size of the constituency is
directly linked not only to the rst stage of platform adoption but
also to the regular maintenance of a municipality’s online presence:
larger municipalities are more active on their Facebook pages and
websites [15, 33, 35, 41, 56].

Municipal Campaigns on Facebook: What influences the scope of engagement and does it win votes? DG.O’21, June 09–11, 2021, Omaha, NE, USA
Apart from size of constituency, additional variables predict the
usage of municipal websites, such as the location of the municipal-
ity (center vs. periphery) [
15
,
35
,
56
,
58
]; the level of income and
education of the local population [
36
,
41
]; and the median age of
the population [41].
Existing literature on websites usage in municipal campaigns
posits that the most important variables determining usage are
institutional-level variables related to the characteristics of the
election campaign. The most signicant variable is incumbency,
where websites are used more intensely by new candidates, who
are under-exposed to the public compared to incumbent candi-
dates, regarding whose activities the public already receives regular
information. Hence, newcomer candidates nd digital channels
to be relatively inexpensive and useful for gaining needed public
exposure to constituencies [16, 17, 25, 26, 51, 58].
On the other hand, focusing on US congressional elections,
Williams and Gulatti [
54
:65] demonstrate that “incumbents have
more capacity to generate the content that constitutes or creates
usage,” a suggestion that seems at least partially valid in other con-
texts as well. For example, in the municipal Facebook context it
is likely that incumbents have access to more information so that
they can post more about regular aairs of the city as compared to
challengers who lack access to such information (e.g. [42]).
In a study of website usage in the 2008 municipal elections in
Israel, Lev-On [
26
] found that exactly half of the contenders had
websites and there were signicant dierences between candidates
with and without websites in two respects: the size of municipality,
and the intensity of the competition, measured using a proxy of
the dierence between the voting shares of the winner and the
runner-up (see also [
51
,
52
]). Additional signicant - although weak
- dierences between candidates with and without websites were
found in population-level variables such as degree of peripherality,
population income, and education levels.
In a study that complements the current research, interviews
were conducted with 67 municipal election candidates in the munic-
ipal elections of 2013, exploring candidates’ general media strategy
and specically Internet strategy, their presence in social networks,
the aims of their Facebook activities, Facebook uses, opinions re-
garding the most eective political activity on social networks,
and their impact [
28
]. The interviews indicate that Facebook is no
longer considered an esoteric tool: Candidates treat it seriously
and professionally by allocating budgets and even hiring profes-
sionals to manage their Facebook activities. Although challengers
perceived that Facebook constitutes an arena that oers both op-
portunity and risk, they consider Facebook presence a necessity.
Even if the activity on Facebook is not a game changer, absence
from this arena may be severely damaging. Still, large dierences in
the perceptions of Facebook as an important factor in the elections
were found between candidates in large and small municipalities.
In smaller municipalities, contenders preferred to invest their time
and resources in parlor meetings and other arenas that oer more
direct contact with voters. On the other hand, in larger municipal-
ities direct contact with a signicant portion of the population is
unfeasible, and candidates attributed greater signicance to their
Facebook presence.
2.2 Does Facebook engagement inuence
election results?
Previous studies indicate that the use of the Internet, in general,
and social media, in particular, in election campaigns generally
has a positive, but small, impact on election results. Earlier studies
looked for correlations between website usage and election out-
comes. In the Irish general elections of 2007, website usage was
correlated with vote share [
44
]. In the 2006 US midterm elections,
Democrat web presence (measured using the Google PageRank of
the campaign webpage of each candidate in the sample) signi-
cantly correlated with their support levels [
51
]. In the 2011 Polish
parliamentary elections, website existence correlated with votes
gained [
23
]. In Australia, use of websites and social media in a
series of general election campaigns between 2001 and 2010 also
correlated with election results [11].
In line with the growing penetration and usage of social media
as election campaign tools, initial studies analyzing their impact
on election results have consistently found similar patterns. In the
US 2006 midterm elections, vote share correlated with the number
of Facebook supporters [
52
]; in the US 2008 presidential primaries,
with the relative Facebook strength of candidates in each state
[
53
]; and in the Netherlands 2010 parliamentary elections, with the
number of Hyves (a social networking site in the Netherlands) and
Twitter followers [
39
]. In the same 2010 general elections in the
Netherlands, Kruikemeier [
24
], who studied the scope and charac-
ter of activity and number of followers, found a signicant positive
correlation with election results, and that using Twitter in an inter-
active way (i.e., retweeting) had an additional positive correlation
with vote share. Votes received also correlated with the number of
Facebook fans and Twitter followers in the 2011 general elections
in New Zealand [
5
]. In Finland in 2015, Facebook post likes and
candidate vote share were found to have a signicant positive re-
lationship [
48
]. A study of candidate Twitter use during UK 2015
and 2017 elections shows that Twitter campaigning has a weak but
signicant contribution to vote share [
3
]. Concentrating on the
2018 Brazilian Presidential Election, a study found strong correla-
tions between user engagement on politicians social media pages,
especially Instagram, and vote shares [4]
In contrast to the above studies that analyzed the correlation
between social media usage and vote share at the national level,
few studies have looked at these correlations at the municipal level.
In Canada, Wagner [
50
] used a nationwide survey of 307 candidates
for various municipal oces between 2010 and 2012, and found
a positive correlation between Internet presence and vote share.
Sobaci et al. [
28
] found a positive correlation between presence
on Twitter and vote share in the 2014 Turkish Local Elections. In
a study of Facebook usage in the Greek 2014 municipal elections,
Lappas et al. [
25
] found that while only 30% of the candidates had
a Facebook page, Facebook usage as well as the number of fans
correlated with candidate vote shares. Facebook presence correlated
with incumbency and location, such that newcomer candidates
running in large municipalities who had never been elected were
more likely to use Facebook.

DG.O’21, June 09–11, 2021, Omaha, NE, USA Azi Lev-On and Nili Steinfeld
3 RESEARCH QUESTIONS
On the basis of previous research reviewed above, the hypotheses
of the current study were formulated concerning the predictors of
the scope of Facebook usage in municipal elections, and indicators
of candidate vote share.
3.1 Institutional-level variables and Facebook
activity.
H1.1 Challengers will generate more engagement on Facebook than
incumbents [16, 17, 26, 51].
H1.2 Greater Facebook engagement will be generated in more
competitive elections, where candidates use every possible tool to
increase voter turnout [16, 17, 26, 51, 52].
H1.3 The larger the number of eligible voters, the more activ-
ity will be observed on contender Facebook pages, based on the
assumption that candidates will nd it dicult to use conventional
methods to reach greater numbers of potential voters [25, 26, 50].
3.2 Population-level variables and Facebook
activity:
H2. Activity on the Facebook pages of contenders will be greater
in municipalities where the population is younger, more educated,
and socio-economic status is higher, and in central vs. peripheral
cities, as these variables are related to various aspects of the digital
divide in Israel [
27
] and have also emerged in previous studies
of Internet usage in campaigns in reference to website activities
[
17
,
27
] as well as in reference to municipal Facebook usage in
general [13, 28, 41, 56, 58].
3.3 Predictors of candidate vote share:
Based on the extensive scope of research at the national level, and
several municipal-level studies, social media - particularly Facebook
- seems to indicate, to some degree, the level of success or failure
of campaigns. Additionally, and presumably to a larger degree, the
characteristics of the race itself, incumbency status, competitiveness
and other institutional variables are indicators and predictors of
election results. Hence:
H3. Candidate Facebook activity and user engagement with can-
didates on Facebook will indicate elections results, however institu-
tional variables, candidate and race characteristics will be stronger
indicators of candidate vote share.
4 METHODOLOGY
The study used automatic scraping tools based on access to Face-
book’s public API. Due to changes in Facebook’s policy, which
resulted in a general restriction of the company’s API in April 2018
[
9
], access to broad and inclusive data from public pages for re-
search purposes became practically nonviable. The last municipal
elections in Israel were held on October 30, 2018, 6 months after
Facebook’s API policy changes, therefore the research data in based
on the prior election campaigns which were held on October 22,
2013.
To the best of our knowledge, this is the rst study in Israel,
and one of the rst worldwide, to use automatic and computerized
tools to investigate questions pertaining to municipal level election
campaigns and candidate success.
4.1 Data Collection: Institutional- and
Population-Level Variables
On October 22, 2013, municipal elections were held in 111 local
authorities throughout Israel, in which a total of 387 contenders ran
for mayoral oce. The names of contenders were obtained from the
oce of the election comptroller in the Israeli Ministry of Interior
Aairs about one month prior to elections, after all candidates
ocially led their paperwork and were approved as candidates by
the comptroller. The number of eligible voters in each municipality
was obtained from the website of the election comptroller. After the
elections, the number of candidates in each municipality, as well as
the number of votes and vote shares received by each candidate,
were calculated. From the website of the Administration of Local
Authorities, information on the candidates who were incumbent
mayors was obtained. As a proxy for election competitiveness, two
indices were created: the number of contenders in each municipality
and the dierence between the percentages of votes received by
the winner and by the runner up.
To analyze the impact of population variables on the scope of
Facebook activity and candidate vote shares, data were collected
on the relevant variables from the website of the Central Bureau of
Statistics (CBS): The authority’s social economic cluster (reecting
residents’ socio-economic status), the median age and education
of the municipality’s residents, and peripherality data
1
. As a proxy
for educational attainment, the percentage of residents between
the ages of 17 and 18 who received their matriculation certicate
from the Ministry of Education was used.
4.2 Candidate Website and Facebook Activity
To locate candidate personal websites, after obtaining the names
of the contenders from the oce of the election comptroller, we
searched the Internet for candidate websites among the 500 leading
results in the Google search of each candidate three times - two
months prior to the elections, a month before the elections, and
two days prior to the elections. We also searched for information
about the candidates in local forums on the Internet where some
published personal information, including links to their websites
and Facebook pages. To locate and map candidate Facebook ac-
tivity, ve research assistants concurrently searched on Google,
Facebook, YouTube, and local websites for contender Facebook
pages. After locating the pages, all the posts on these pages were
scraped for an entire month prior to the election using Netvizz, an
application that extracts data from Facebook’s API
2
. We archived all
posts, engagement measures per post (likes, comments, shares), and
cumulative engagement (cumulative number of likes, comments,
comment likes, and shares) in the month prior to the elections. The
number of fans was recorded manually on Election Day.
1
The Peripherality Index is constructed by the Israeli Central Bureau of Statistics (CBS)
and takes into account the distance of a municipality from the Tel-Aviv District, which
is the economic and business center of Israel; and the Potential Accessibility Index – a
function of the distance between a municipality and other municipalities, as well as the
size of population – which functions as a proxy for a variety of economic parameters.
2
The authors express their gratitude for the developer, Bernhard Rieder, for enabling a
free and public use of the tool.

Municipal Campaigns on Facebook: What influences the scope of engagement and does it win votes? DG.O’21, June 09–11, 2021, Omaha, NE, USA
Table 1: Matrix of Correlation Coecients between the Study’s Variables
Fans# Posts Likes Comments Shares Engagement
Institutional
Variables
1 Newcomer facing Incumbent? .13*
2 #Candidates -.12* -.16** -.14**
3 Population size .60** .52** .62** .65** .64** .63**
Population
Variables
4 SE Status -.11* -.11* -.12* -.11*
5 Peripheriality .15** .18** .16** .22** .20** .18**
6 Age .14*
7 Education -.14** -.16** -.17** -.17**
FB Variables 8 Page fans# .75** .92** .88** .80** .91**
9 Posts 0.75** .82** .82** .78** .82**
10 Likes 0.92** 0.82** .96** .88** .99**
11 Comments .88** .98**
12 Shares .89**
13 Engagement 0.91** 0.82** 0.99** 0.98** 0.89**
5 FINDINGS
The elections were held in 111 local authorities in Israel. Of the
387 contenders, 92 were incumbents, 221 were challengers who
competed against incumbents, and 74 competed in authorities in
which the incumbent did not run for reelection. Out of the 92 in-
cumbents, 67 were reelected and 25 were ousted. The mean number
of candidates per municipal campaign was 3.5.
While only 78 of the 387 candidates (20.2%) maintained a web-
site, Facebook pages were extensively used by 316 (81.6%) con-
tenders. Facebook pages were used by candidates in dierent re-
gions and throughout the country, with one prominent exception:
no candidates running for mayorship in local governments with
predominantly ultra-orthodox populations had a Facebook page.
This nding reects the conservative character of the Haredi (ultra-
Orthodox) population, which is manifest on lower Internet penetra-
tion rates as well as more conservative usage patterns. A previous
study also found that no ultra-Orthodox municipality has main-
tained a formal Facebook pages during the year of 2013 as well
[30].
In the month prior to the elections, contender Facebook pages
included on average 73 posts which attracted on average 2,968 likes,
351 comments, and 266 shares. The average cumulative engagement
(including likes, comments, comment likes, and shares) was 4,113.
The average number of fans per page was 2,598.
Interestingly, website-based campaigning signicantly dropped
compared to the previous round of municipal elections. In 2008, 50%
of candidates maintained a website, compared to slightly more than
20% in the current elections. Clearly, in 2013 the center of gravity
of online activity shifted from websites to Facebook pages. For
example, contenders used Facebook pages compared to websites
at a ratio of 4:1, with only eight cases of candidates who had an
ocial website but no Facebook page. Many candidates, however,
had a Facebook presence but no website.
5.1 Correlations between study variables
Table 1 summarizes signicant correlations between institutional,
population and Facebook activity variables. Strong and signicant
correlations were found between several Facebook activity indices:
number of fans, number of posts, and engagement measures (likes,
comments, shares and cumulative engagement). Election competi-
tiveness (specically the number of candidates) signicantly corre-
lated with several Facebook activity variables. Correlations were
found between (a) municipality population size, Facebook activity,
and most of the population variables; (b) between peripherality
and Facebook activity; and (c) between education and most of the
Facebook activity variables.
5.2 Predictors of engagement on candidate
Facebook pages
For predicting Facebook engagement, a hierarchical linear regres-
sion analysis was performed (see Table 2). In the rst block, in-
stitutional variables were entered (whether the candidate was an
incumbent, whether the candidate ran against an incumbent, the
number of eligible voters in the municipality, the number of can-
didates and the dierence in vote shares between the winner and
the runner-up as two signiers for level of competitiveness of the
election).
In the second block population-level variables were entered:
social economic cluster and peripherality, education, and age. The
third block included variables related to Facebook activity: number
of fans and posts. Due to multicollinearity between the variables
and the depended variable of the analysis, the results of the third
block were not further analyzed, but are presented for illustrative
purposes.
From the rst block of the regression, it emerges that incum-
bency, competitiveness (reected in number of candidates), and
municipality size predict Facebook engagement. Facebook pages
of incumbents show greater activity than Facebook pages of non-
incumbents (when the incumbent is running for reelection). The
number of eligible voters positively predict Facebook engagement,
whereas number of candidates negatively predict Facebook engage-
ment. The variables in the rst block of the model explain 43% of
the variance of Facebook engagement.
After introducing the second block of the regression, which in-
cludes the population variables, incumbency and municipality size
continue to signicantly and positively predict engagement. Median

DG.O’21, June 09–11, 2021, Omaha, NE, USA Azi Lev-On and Nili Steinfeld
Table 2: Findings of the Linear Regression for Predicting Facebook Engagement
Model 1 Model 2 Model 3
B Beta B Beta B Beta
(Constant) 6307.79 15878.70 6823.12
Newcomer facing
Incumbent?
-3337.41** -0.15 -3411.52** -0.15 558.30 0.03
Newcomer- Incumbent not
running?
-1947.30 -0.07 -2651.01 -0.09 28.93 0.00
#Residents 0.06** 0.63 0.07* 0.65 0.01** 0.10
#Candidates -752.43* -0.11 -583.78 -0.09 -289.69* -0.04
Gap #1- #2 -26.91 -0.04 -15.21 -0.02 -28.21* -0.04
SE Status 672.57 0.11 473.71* 0.08
Peripheriality -69.44 -0.01 193.68 0.03
Age -251.65* -0.10 -204.52 -0.08
Education -99.53 -0.13 -81.20** -0.10
Fans 1.03** 0.60
Posts 38.34** 0.32
R Square 0.43 0.44 0.90
R Square change 0.43** 0.01 0.44**
Table 3: Goodness-of-t indices for the 2 SEM models
Model χ2df ∆χ2NFI CFI TLI RMSEA SRMR
1-with
population
variables
239,083 38 .91 .92 .83 .12 Not dened
due to
missing data
2- without
population
variables
31,573 14 207,510*** ,
df=24
.98 .99 .98 .06 .05
population age negatively predicts Facebook page activity. Compet-
itiveness, socio-economic cluster, peripherality and education do
not signicantly predict Facebook engagement. The contribution
to the explained variance over and above the contribution of the
institutional-level variables in the rst block of the regression was
insignicant.
These ndings support H1. Incumbency and size of the con-
stituency positively predict engagement levels on contender Face-
book pages. However, H2 was not supported: Population-level vari-
ables had no impact on engagement level on contender Facebook
pages, except for a small and negative impact of age.
5.3 Predictors of candidate vote share
To analyze how the dierent variables may indicate election results,
we performed a Structural Equation Modeling (SEM) using AMOS
software. Based on the correlations and regression results, we hy-
pothesized that institutional variables and characteristics of the
race, namely incumbency status (is the candidate a challenger fac-
ing incumbent? Is the candidate a new candidate in a race where the
incumbent is not running?), number of candidates in the race (i.e.
competitiveness) and size of constituency (number of citizens) will
predict candidate vote share. We also hypothesized that popularity
of candidate posts (i.e. the average number of likes candidate posts
receive), which if aected by size of constituency, number of page
fans and scope of candidate Facebook page activity (i.e. number of
posts), will also indicate to some degree the candidate vote share.
We were also interested in inquiring whether population variables
(SE status, peripherality, population age and education level) predict
either the popularity of candidate posts or vote shares. Comparing
between the two models, one which includes population variables
and one which does not, made clear that removing the population
variables, which had no signicant contribution to either of the
model dependent variables, improved the model signicantly. Table
3 summarizes goodness-of-t indices of the two models. For model
2, all goodness-of-t indices are very good, and the chi-square dif-
ference between the models is signicant, therefore model no. 2 is
the preferred model.
The nal model is presented in gure 1, and indicates that vote
share is best predicted by institutional variables, mainly incum-
bency status, number of candidates in the race and to a much lesser
degree- size of constituency. Number of likes candidates receive for
their Facebook posts mediates a weak relationship between number
of posts the candidate uploads to the platform, number of page fans
and population size, to candidate vote shares.

Municipal Campaigns on Facebook: What influences the scope of engagement and does it win votes? DG.O’21, June 09–11, 2021, Omaha, NE, USA
Figure 1: SEM model for predicting municipal election candidate vote share
6 DISCUSSION AND CONCLUSIONS
This study examined Facebook activity in political campaigns in
Israel. Municipal elections held in many dierent constituencies
across the country enabled the analysis of political uses of the
Internet at a higher resolution than possible in national elections
(which in Israel are conducted in a single district, the entire country).
Municipal election data enable the study of several institutional-
level and population-level variables and their impact on Facebook
usage as well as on candidates’ vote shares.
Facebook is an important campaigning tool, and is considered
obligatory for candidates. While only a small fraction of the con-
tenders used websites (20.2%), Facebook pages were very prevalent
(81.6%). Instead of developing websites that require considerable
maintenance and costs, contenders ock to Facebook, where pres-
ence is easier to establish and maintain, and is more accessible
to voters as well. In the 2013 municipal elections in Israel, Face-
book was widely used both by candidates and by voters throughout
the country, with the exception of candidates running for mayor-
ship in ultra-Orthodox municipalities. The scope of engagement
was predicted by constituency size and incumbency, but not by
population-level variables. Although incumbents have more con-
tent to present on Facebook, a result of their activities in oce,
challengers are more active and attract more engagement during
the campaigns.
Nevertheless, Facebook activity had only minor impact on the
vote shares received by candidate. Again, incumbency, municipal-
ity size, and competitiveness of the election campaign were the
strongest predictors of vote share. Population-level variables did
not predict vote share at all. The SEM analysis reveals that incum-
bency status and competitiveness are strong indicators of election
results and candidate vote share. Number of Facebook post likes,
inuenced by number of fans, number of candidate posts and size
of constituency, is a weak yet signicant indicator of candidate vote
share.
This study explores several Facebook variables that represent
activity initiated by campaigners (number of posts) and by users
(number of fans and engagement). In contrast to past studies, the
ndings indicate that campaign-initiated Facebook activity is a
small indicator of election results. Recruiting fans and facilitating
engagement, so often discussed as the core of the political activity
on Facebook, had very little impact, if any, on candidate vote share.
Thus, in contrast to the widespread belief about the potential of
Facebook to pave the way to success in the polls, or to serve as an
indicator of candidate popularity, it seems that Facebook has a lim-
ited role that is eclipsed by institutional-level variables, specically
incumbency state and size of constituency. Nonetheless, Facebook
usage may still be a game changer or an important indicator in very
close races.

DG.O’21, June 09–11, 2021, Omaha, NE, USA Azi Lev-On and Nili Steinfeld
In the years following the time of data collection for this study,
social media has established its status and role as a central commu-
nication channel for citizens to communicate with public represen-
tatives at the national and municipal level, and for candidates in
national as well as municipal level campaigns to reach voters and
disseminate messages. As the results of this study show, Facebook
activity and engagement is an important factor and can serve to
increase candidate popularity, or to some degree- indicate candidate
popularity, but it is far from being a leading factor in candidate
success.
Future studies can further analyze the impact of Facebook on
election results in other countries with varying degrees of Facebook
penetration. It is also interesting to comparatively study Facebook
adoption in municipal campaigns over time, including the scope of
use and impact of Facebook usage on these election results. Such
studies will also enable development of a broader theory to explain
the scope, character, and impact of social media usage on election
results in various circumstances.
REFERENCES
[1]
Tobias Bürger and Karen Ross. 2014. Face to Face (book): Social Media, Political
Campaigning and the Unbearable Lightness of Being There. Political Science,
66,1,46-62.
[2]
Bezek. Digital life: Bezek Internet report 2019-2020. Retrieved January 10, 2021
from https://media.bezeq.co.il/pdf/internetreport_2019.pdf.
[3]
Jonathan Bright, Scott Hale, Bharath Ganesh, Andrew Bulovsky, Helen Margetts
and Phil Howard. 2020. Does Campaigning on Social Media Make a Dierence?
Evidence from candidate use of Twitter during the 2015 and 2017 UK Elections.
Communication Research, 47,7, 988-1009.
[4]
Kellyton Brito, Paula Natalia, Fernandes Manoel and Meira Silvio. 2019. Social
media and presidential campaigns–preliminary results of the 2018 Brazilian
presidential election. In Proceedings of the 20th Annual International Conference
on Digital Government Research, 332-341.
[5]
Michael P. Cameron, Patrick Barrett, and Bob Stewardson. 2016. Can Social
Media Predict Election Results? Evidence from New Zealand. Journal of Political
Marketing, 15, 4, 416-432.
[6]
Stephen Coleman. 2004. Connecting Parliament to the Public via the Internet:
Two Case Studies of Online Consultations. Information, Communication, and
Society, 7,1,1-22.
[7]
Stephen Coleman and Jay G. Blumler. 2009. The Internet and Democratic Citi-
zenship: Theory, Practice and Policy. Cambridge: Cambridge University Press.
[8]
Gunn Sara Enli and Eli Skogerbø. 2013. Personalized Campaigns in Party-
Centered Politics: Twitter and Facebook as Arenas for Political Communication.
Information, Communication & Society, 16,5,757-774.
[9]
Facebook. 2018. Restricting Data Access and Protecting People’s
Information on Facebook. Retrieved January 1, 2019 from https:
//www.facebook.com/business/news/restricting-data-access-and-protecting-
peoples-information- on-facebook
[10]
Rachel Gibson and Stephen Ward. 2009. Parties in the Digital Age: A review
Article. Representation, 45,1,87-100.
[11]
Rachel Gibson and Lan Mcallister. 2015. Normalising or Equalising Party Com-
petition? Assessing the Impact of the Web on Election Campaigning. Political
Studies, 63,3,529-547.
[12]
Matthew E. Glassman, Jacob R. Straus, and Colleen J. Shogan. 2013. Social net-
working and constituent communications: Members’ use of Twitter and Facebook
during a two-month period in the 112th Congress. Congressional Research Ser-
vice, CRS Report for Congress. Washington DC, USA
[13]
Girish J. Gulati and Christine B. Williams. 2013. Social Media and Campaign
2012: Developments and Trends for Facebook Adoption. Social Science Computer
Review, 31,5,577-588.
[14]
Sharon Haleva-Amir. 2011. Online Israeli Politics: The Current State of the Art.
Israel Aairs, 17,3,467-85.
[15]
Are Vegard Haug. 2008. Local Democracy Online: Driven by Crisis, Legitimacy,
Resources, or Communication Gaps? Journal of Information Technology and
Politics, 4,2,79-99.
[16]
Paul S. Herrnson. 2004. Congressional Elections: Campaigning at Home and in
Washington. Washington, DC: CQ Press.
[17]
Paul S. Herrnson, Atiya Kai Stokes-Brown, and Matthew Hindman. 2007. Cam-
paign Politics and the Digital Divide: Constituency Characteristics, Strategic
Considerations, and Candidate Internet Use in State Legislative Elections. Politi-
cal Research Quarterly, 60,1,31-42.
[18]
Nigel Jackson and Darren Lilleker. 2011. Microblogging, Constituency Service and
Impression Management: UK MPs and the Use of Twitter. Journal of Legislative
Studies, 17,1,86-105.
[19]
Thomas J. Johnson, Weiwu Zhang, Shannon L. Bichard, and T Seltzer. 2010.
United We Stand? Online Social Network Sites and Civic Engagement. In The
Networked Self: Identity, Community, and Culture on Social Network Sites, ed.
Z Papacharissi. New York, NY: Routledge.
[20]
Timo Wandhoefer, Mark Thamn, and Somya Joshi. 2011. Politician 2.0: Infor-
mation behavior and dissemination on social networking sites: Gaps and best
practices. eJournal of eDemocracy and Open Government, 3,2,207-214.
[21]
Rune Karlsen. 2013. Obama’s Online Success and European Party Organizations:
Adoption and Adaptation of US Online Practices in the Norwegian Labor Party.
Journal of Information Technology & Politics, 10, 158-170.
[22]
Karolina Koc-Michalska, Darren G. Lilleker, Pawel Surowiec, and Pawel Bara-
nowski. 2014. Poland’s Online Election Campaign: New Tools, New Professional-
ism, New Ways to Win Votes. Journal of Information Technology and Politics,
11,2, 186-205.
[23]
Karolina Koc-Michalska, Rachel Gibson, and Thierry Vedel. 2014. Online Cam-
paigning in France, 2007–2012: Political Actors and Citizens in the Aftermath
of the Web. 2.0 Evolution. Journal of Information Technology and Politics, 11,2,
220-244.
[24]
Sanne Kruikemeier. 2014. How Political Candidates use Twitter and the Impact
on Votes. Computers in Human Behavior, 34,131-139.
[25] Georgios Lappas, Amalia Triantallidou, Prodromos Yannas, Anastasia Kavada,
Alexandros Kleftodimos, and Olga Vasileiadou .2015. The Role of Facebook in the
2014 Greek Municipal Elections. In Multidisciplinary Social Networks Research.
Springer Berlin Heidelberg.
[26]
Azi Lev-On. 2011. Campaigning online: Use of the Internet by parties, candidates
and voters in national and local election campaigns in Israel. Policy & Internet,
3,1, 1-28.
[27]
Azi Lev-On. 2013. Another ew over the digital divide: internet usage in the Arab-
Palestinian sector in Israel during municipal election campaigns, 2008. Israel
Aairs, 19,1, 154-169.
[28]
Azi Lev-On. 2018. Perceptions, uses, visual aspects, and consequences of social
media campaigning: Lessons from municipal Facebook campaigning, Israel 2013.
In Sub-national democracy and politics through social media .149-168. Springer,
Cham.
[29]
Azi Lev-On and Nili Steinfeld. 2016. Social media and the city: Analyzing conver-
sations in Municipal Facebook pages. In Social Media and Local Governments
,243-261, Springer, Cham.
[30]
Azi Lev-On and Nili Steinfeld. 2015. Local engagement online: Municipal Face-
book pages as hubs of interaction. Government information quarterly, 32,3,
299-307.
[31]
Darren Lilleker and Thierry Vedel. 2013. The Internet in Campaigns and Elec-
tions. In Oxford Handbook of Internet Studies, ed. WH Dutton. Oxford: Oxford
University Press.
[32]
Julia Metag and Frank Marcinkowski. 2012. Strategic, Structural, and Individual
Determinants of Online Campaigning in German Elections. Policy and Internet,
4,3-4, 136-158.
[33]
Jae M. Moon. 2002. The Evolution of E-Government among Municipalities:
Rhetoric or Reality? Public Administration Review, 62, 4, 424-433.
[34]
Merrill Morris and Christine Ogan. 1996. The Internet as Mass Medium. Journal
of Communication, 46, 1, 39-50.
[35] Donald F. Norris and Christopher G. Reddick. 2013. Local E-Government in the
United States: Transformation or Incremental Change? Public Administration
Review, 73, 1, 165-175.
[36]
R. Purian-Lukach. 2011. Municipal E-Government: Comparative Index, Success
Factors and Models for Management and Evaluation. In Connected: Politics,
Technology and Society in Israel, ed. A. Lev-On and E Cohen. Tel-Aviv: The
Israeli Political Science Association Press. (Hebrew).
[37]
Christopher G. Reddick and Donald F. Norris. 2013. Social media adoption at the
American grass roots: Web 2.0 or 1.5? Government Information Quarterly, 30,4,
498-507.
[38]
Mehmet Zahid Sobaci, Kadir Yasin Eryiğit, and Ibrahim Hatipoğlu. 2016. The Net
Eect of Social Media on Election Results: The Case of Twitter in 2014 Turkish
Local Elections. In Social Media and Local Governments. Springer International
Publishing.
[39]
Niels Spierings and Kristof Jacobs. 2014. Getting Personal? The Impact of Social
Media on Preferential Voting. Political Behavior, 36,1, 215-234.
[40]
Nili Steinfeld and Azi Lev-On. 2020. MPs on Facebook: Dierences between
Members of Coalition and Opposition. Digital Government: Research and Practice,
1,2, 1-14.
[41]
Nili Steinfeld and Azi Lev-On. 2015. Well-Done,Mr. Mayor!. Linguistic Analysis of
Municipal Facebook Pages. International Journal of E-Planning Research ,IJEPR,
4,2, 26-38.
[42]
Kim Strandberg. 2013. A Social Media Revolution or Just a Case of History
Repeating Itself? The Use of Social Media in the 2011 Finnish Parliamentary
Elections. New Media & Society, 15,8, 1329-1347.

Municipal Campaigns on Facebook: What influences the scope of engagement and does it win votes? DG.O’21, June 09–11, 2021, Omaha, NE, USA
[43]
Jennifer Stromer-Galley. 2000. Online Interaction and Why Candidates Avoid It.
Journal of Communication, 50, 111-132.
[44]
Matia Laura Suddulich and Matthew Wall. 2010. Cyber Campaigning in the
2007 Irish General Election. Journal of Information Technology and Politics, 7,
340–355.
[45]
Shyam S. Sundar, Sriram Kalyanaraman, and Justin Brown. 2003. Explicating
Web Site Interactivity: Impression Formation Eects in Political Campaign Sites.
Communication Research, 30, 1, 30-59.
[46]
Sonja Utz. 2009. The (Potential) Benets of Campaigning via Social Network
Sites. Journal of Computer-Mediated Communication, 14, 2, 221-243
[47]
Tapio Vepsäläinen, Li Hongxiu and Reima Suami. 2017. Facebook likes and
public opinion: Predicting the 2015 Finnish parliamentary elections. Government
Information Quarterly, 34 ,3, 524-532.
[48]
Maurice Vergeer. 2012. Politics, Elections and Online Campaigning: Past, Present
. . . and a Peek into the Future. New Media & Society, 15, 1, 9-17.
[49]
Maurice Vergeer, Liesbeth Hermans, and Steven Sams. 2013. Online Social Net-
works and Micro-Blogging in Political Campaigning: The Exploration of a New
Campaign Tool and a New Campaign Style. Party Politics, 19, 3, 477-501.
[50]
Angelia Wagner. 2015. Candidate orientation to ICTs in Canadian municipal
elections. In Citizen participation and political communication in a digital world,
Alex Frame and Gilles Brachotte eds. Routledge. New-York, USA. 81-94.
[51]
Kevin M. Wagner and Jason Gainous. 2009. Electronic Grassroots: Does Online
Campaigning Work? The Journal of Legislative Studies, 15, 4, 502-520.
[52]
Christine B. Williams and Girish J. Gulati. 2007. Social Networks in Political
Campaigns: Facebook and the 2006 Midterm Elections. Paper presented at the
annual meeting of the American Political Science Association, Illinois, Chicago,
August 30- September 30.
[53] Christine Williams and Girish J. Gulati. 2008. What is a Social Network Worth?
Facebook and Vote Share in the 2008 Presidential Primaries. Paper presented at
the 2008 Annual Meeting of the American Political Science Association. Boston,
MA, August 28-31.
[54]
Christine B. Williams and Girish J. Gulati. 2012. Social Networks in Political
Campaigns: Facebook and the Congressional Elections of 2006 and 2008. New
Media & Society, 15,1 ,52-71.
[55]
Andy Williamson. 2009. The Eect of Digital Media on MPs’ Communication
with Constituents. Parliamentary Aairs, 62,3, 514-527.
[56]
Tony E. Wohlers. 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.
[57]
World bank. Individuals using the Internet (% of population). Retrieved January
01, 2020 from https://data.worldbank.org/indicator/IT.NET.USER.ZS
[58]
Prodromos Yannas and Georgios Lappas. 2004. E-Campaign in Greek Elections:
2000-2004. WSEAS Transactions on Information, Science and Applications, 5,1,
1332-1338.
[59]
Shahaf Zamir and Gideon Rahat. 2019. Online Political Personalization. Israeli
Democracy Institute. Jerusalem, Israel.