Campaigning online: social media in the 2010 Niagara municipal elections.
Hagar, Douglas
Abstract
Although social media have become an increasing feature in Canadian
electoral politics and have been given increased attention in research
on Canadian political communication, the use of social media in
municipal elections has yet to be thoroughly studied. This paper employs
a quantitative analysis to assess the extent and effectiveness of social
media use by candidates during the 2010 Niagara municipal elections. The
types of content posted online, the level of interaction between
candidates and voters, and the impact on electoral success are examined.
An interaction scale developed for this study helps gauge the level of
interactivity between candidates and voters. Although it is determined
that social media were being used during the campaigns, the interaction
between candidates and voters was low. There are differences in
electoral success for incumbents and challengers and a positive
correlation is identified between the success of challenger candidates
and the number of Facebook "likes" and posts. Conversely, the
use of social media has a negative correlation with success for
incumbents. A number of directions for future research on social media
in municipal campaigns are suggested.
Keywords: elections, political campaigns, social media,
interactivity
Resume
Les medias sociaux occupent une presence accrue dans la politique
electorale canadienne et font de plus en plus l'objet de recherches
sur la communication politique canadienne. Toutefois, l'utilisation
des medias sociaux dans les campagnes electorales municipales est un
sujet encore peu etudie. Cette recherche a recours a une analyse
quantitative afin dexaminer lefficacite et l'ampleur de
l'utilisation des medias sociaux par les candidats a
l'election municipale de 2010 a Niagara. Elle examine le type de
contenu publie en ligne, le niveau d'interaction entre les
candidats et les electeurs ainsi que l'incidence sur le succes
electoral. Une echelle d'interaction a ete developpee afin de
mesurer le niveau d'interaction entre les candidats et les
electeurs. Meme si les medias sociaux ont ete utilises pendant la
campagne, l'interaction entre les candidats et les electeurs etait
faible. DifFerents niveaux de succes electoral ont ete identifies pour
les candidats sortants et les candidats aspirants. Une relation positive
entre le succes des candidats aspirants et le nombre de <<
j'aime >> et de commentaires sur Facebook a ete identifiee. A
l'oppose, une relation negative a ete trouvee entre
l'utilisation des medias sociaux et le succes des candidats
sortants. Des orientations pour la recherche future sur
l'utilisation des medias sociaux dans les campagnes electorales
municipales sont proposees.
Mots cles: elections municipales, medias sociaux
Introduction
Studies on the use of social media in Canadian federal politics, as
well as on the political discussion taking place on the popular #cdnpoli
Twitter hashtag, have proliferated in recent years (Small 2008b; Small
2011, Chen and Smith 2011; and Small 2008a). However, within this new
field of study, elections have been given minimal attention. Social
media may contribute to political success because of the opportunity
provided to candidates to have continued interaction with voters in a
scale and intensity not possible through more traditional campaign
methods (i.e. door-to-door campaigning, leafleting, print or television
coverage). In addition, there are no direct costs associated with using
Twitter, Facebook, and YouTube and the platforms are user-friendly for
individuals with minimal computer skills. However, within research on
social media use during elections (e.g. Goodman and Copeland 2010;
Raynauld and Greenberg 2014), the impact of social media on political
success has not been widely explored.
This article analyzes the use of Facebook, Twitter, YouTube and
campaign websites by a selection of municipal candidates in the Niagara
region, Ontario. The research addresses three primary questions: First,
to what extent do municipal candidates use websites and social media as
a component of their campaigns? Second, what is the level of online
interaction between voters and municipal candidates? Third, does the use
of social media contribute to the municipal candidates' electoral
success? It is hypothesized that more extensive use of social media (an
increased number of tweets and Facebook posts) positively impacts the
success of municipal candidates. It is also hypothesized that the impact
of social media is positive for both incumbents and challengers.
The article begins with a literature review highlighting the
existing findings on politician-citizen engagement on social media. This
is followed by a detailed description of the methodology employed in
this study, including the use of an interaction scale to determine the
extent of two-directional discussion on social media platforms between
citizens and municipal candidates. Finally, the results of the study are
discussed and avenues for future research are identified.
Candidates and Social Media
Social media use in politics has been increasing in most advanced
democratic countries, including Canada, In the 2011 federal election,
245 Canadian members of parliament (1) had Twitter accounts which were
commonly referenced by the news media. Additionally, when the Green
Party leader, Elizabeth May, was denied participation in the televised
debates during that election, she used the Twitter hashtag #emayin to
respond to debate questions (Francoli et al. 2011). As well, Twitter
jargon penetrated the television debates themselves, with Jack Layton
proclaiming that Stephen Harper's crime policies were "a
hashtag fail" (Paperny 2011).
Raynauld and Greenberg (2014, 8) argue that social media, including
Twitter, have "helped lower the threshold to political
participation, thus giving resource-poor political players previously on
the edges of the political system more opportunities to participate
actively without institutional restraints in public political
processes." Since using Twitter and other social media only require
an Internet connection and a computer, smartphone or tablet, political
actors can easily develop and share content with large audiences
(Vergeer, Hermans et al 2013). Research on the use of social media by
candidates contributing to electoral success has produced mixed
findings. Through an analysis of the use of social media in the 2008
Canadian federal election, Chen and Smith (2011) identify that
structural and financial factors are the main determinants of political
success. Although optimal use of social media has some impact on
individual candidate electoral success, they claim that this benefit can
"not be separated from the electoral advantages of established
political parties and incumbent candidates such as greater resources and
higher levels of access to 'free media' [exposure and
journalistic coverage]" (2011,400). Thus, this study suggests that
social media is of minimal benefit to electoral candidates.
In her study of Canadian and US politicians in their respective
2008 federal elections, Small (2008b) suggests that of those federal
candidates who did use social media, very few utilized it to its full
potential. Candidates' Facebook profiles provided minimal
information and did not engage voters to any substantial extent (2008b,
86). Additionally, Small (2010) notes that despite the increased number
of politicians on Twitter and other social media, Canadian politicians
have primarily used Twitter as a press-release sharing mechanism rather
than to interact with citizens. Scholarship on social media and
political communication more broadly, similarly reveals that social
media have not significantly increased political dialogue. For instance,
Sweetser and Weaver-Lariscy's (2008) study of US House and Senate
Facebook pages suggests that there is very little evidence of
discussion, as most public posts consisted of shallow supportive
messages and candidates rarely responded to comments. Similarly, in the
study of the 2010 UK general election, Graham et al (2013), identify
that Twitter is mainly being used for unidirectional, non-interactive
communication. In addition to a lack of politician-voter dialogue,
existing research on social media in politics also reveals minimal
voter-to-voter dialogue. For instance, through an analysis of the most
popular Canadian politics hashtag, #cdnpoli, Small (2011) suggests that
only 7.4 percent of the tweets using this hashtag could be categorized
as conversational (Small 2011, 883), while most are unidirectional,
where users state an opinion or share news articles and other existing
online content.
There are some differences between municipal and federal politics
that could suggest that social media is more effective when used in
municipal elections compared to federal. At the municipal level, expert
teams rarely control the communication with voters and political parties
or leaders do not shape the discussion. The small, local scale of
municipal campaigns may allow social media to be more effectively used
to engage voters, particularly youth. Xenos, Vromen and Loader (2014)
identify that there is a strong, positive relationship between social
media use and political engagement of young people. Park (2013) notes
that online campaigns engaging Twitter opinion-leaders are particularly
effective at motivating the political engagement of the general public.
At the municipal level, research on the factors that contribute to
electoral success have rarely included social media. Existing research
has identified multiple factors affecting candidate electoral success in
municipal campaigns. For example, Kushner, Siegel and Stanwick (1997)
identify that incumbency, campaign expenditures and the number of
candidates contesting seats are factors influencing electoral success in
Ontario municipal elections. Both incumbents and candidates with higher
levels of campaign expenditures are more likely to be elected. Gerber
and Green (2000) argue that face-to-face canvassing interaction is a
more important determinant in voter turnout in city elections than
direct mail and phone canvassing. Because social media emulates the
personal interactive elements of face-to-face campaigning, it may have a
similar positive impact on electoral success.
Although some newspapers have drawn attention to the use of social
media in municipal campaigns (such as the case of Naheed Nenshi in
Calgary (2)), there have been few systematic studies examining the use
of social media in municipal elections (e.g. Raynauld Sc Greenberg 2014;
Goodman & Copeland 2010) and existing research has not examined the
relationship between social media use and electoral success. By
examining the extent and nature of social media usage in municipal
political campaigns, including the level of interaction between voters
and candidates, to assess the relationship between social media use and
candidate success in municipal elections, this article attempts to bring
more insight into the potentials and shortfalls of social media in
political communication.
Methodology
This paper is guided by three research questions:
1. To what extent do municipal candidates use websites and social
media in their campaigns?
2. What is the level of interaction between voters and municipal
candidates online?
3. To what extent does the use of social media impact the candidate
electoral success?
To answer these questions, a content analysis of municipal
candidates' websites, Facebook pages, Twitter feeds, and YouTube
videos for candidates across the Niagara Region was conducted. A sample
of 105 candidates was selected for analysis during the last 30 days of
the municipal campaign period between September 25 and October 25, 2010.
The sample included 21 mayoral candidates from 11 lower-tier
municipalities, (3) as well as a stratified random sample of 84 from a
total of 450 municipal candidates from the twelve city/town council and
upper-tier elections. A stratified sample of candidates was conducted
that included 12 upper-tier candidates and 6 candidates from each of the
12 lower-tier municipalities in the Niagara Region. The stratified
sample allowed for a diverse selection of candidates from rural and
medium-sized urban municipalities (4) as well as upper-tier and
lower-tier candidates. To identify candidate social media accounts,
references were noted in campaign materials and newspaper ads in local
newspapers. Additionally, searches on Google, Facebook, and Twitter were
conducted for candidate names as well as the relevant city or position,
(5) examining a maximum of two pages of Google for each candidate and a
maximum of the first 25 results on Twitter, Facebook and YouTube. These
limits were chosen because it seemed unlikely that voters would search
beyond these parameters for candidate pages. Although there is no
benchmark for how far from the top search listings should be ranked,
past research by search engine optimization firms has indicated that 75
percent of users do not scroll past the first page
(Marketshare.hitslink.com 2010) and primarily follow the first three
search results (Slingshot SEO 2011; Unger 2011). Therefore, the search
maximum used in this study surpassed common consumer trends.
The Niagara Region (6) was selected as the sample because it
contains several small and medium municipalities as well as one large
municipality and it reasonably represents the average-sized municipal
area in Ontario. The Niagara Region is a two-tier municipal system with
31 regional councillors in the upper tier. Of the 31 councillors, 18 are
directly elected while the remaining 12 are mayors from the lower-tier
municipalities. The 12 lower-tier municipalities in the Niagara region
include St. Catharines, Thorold, Welland, Niagara Falls,
Niagara-on-the-Lake, Port Colborne, Fort Erie, Wainfleet, Lincoln, West
Lincoln, Fonthill, and Grimsby. The Niagara Region consists of one large
urban city, St. Catharines, as well as medium-sized urban cities
including Welland and Niagara Falls and several small urban and rural
municipalities.
All social media discussion from the sample of 105 candidates,
including their tweets, (7) Facebook posts (8) and YouTube videos, are
analyzed. A directed content analysis (FIsieh and Shannon 2005,1281) was
followed. Data was collected on the number of posts and tweets by
candidates, the frequency of updates, and the number of responses or
questions posed by candidates and voters. A number of other variables
were collected (see Table 1 for a complete list).
Measuring interaction solely on the number of tweets or posts
provides only a limited depiction of the depth of online discussion. For
example, a social media platform may have a large number of posts, but
these posts may not build upon each other or make reference to each
other, or they may contain only minimal discussion. To overcome this, an
interaction scale is used to measure the level of interactive
discussion. Interaction, defined as a two-way communication where the
roles of the sender and receiver are interchangeable with near real-time
speed of communication (Kiousis 2002; McMillan and Hwang 2002), is thus
an important dimension in analyzing social media in municipal election
campaigns and vital to understanding the nature and effectiveness of
social media use in this study.
The interaction scale includes a number of variables, such as the
number of tweets/ posts and replies as well as the frequency of
tweets/posts. Interaction on each social media platform is rated on a
scale of 1 to 5 (1 being the lowest and 5 being the highest level). The
variables are coded according to three different methods: System A,
System B and System C.
System A and B are used for variables that likely have different
frequencies of occurrence. Because variables such as voters attacking
candidate positions, candidates responding to those attacks, candidates
sharing supporter stories, and answers to submitted questions on
websites are less likely to occur with great frequency online, System B
is used to quantify the value of attacks, candidate responses to
attacks, and candidate responses to questions posed by voters on
websites in the calculation of the interaction scale.
System C is used to calculate the frequency of candidate posts and
tweets. It analyzes the last 5 candidate posts to determine the
frequency of posts. Finally, the mean for System A, B, and C variables
are calculated for each platform to produce an overall measure of
interactivity on a scale of 1 to 5.
The relationship between electoral success and social media use is
tested by using chi-square and Pearsons correlation coefficient for
incumbents, challengers and all candidates. Chi-square is used to test
the relationship between electoral success and the presence of a
candidate Twitter, Facebook or YouTube account or campaign website. For
the chi-square tests where the expected cell values are less than 5,
Fisher's exact test is used because it more accurately calculates
p-values for samples of small size compared to asymptotic p-value tests
(Mehta and Patel 2011, 12). Because the extent of usage may affect
electoral success, the relationship between electoral success and the
number of Twitter followers, tweets, Facebook posts and
"likes" are analyzed using Pearson's correlation
coefficient.
I. How Candidates Used Social Media
The most common online platforms used by candidates were campaign
websites, followed by Facebook, Twitter, and YouTube. A description of
how candidates used social media by each platform follows.
Websites
The most common content observed on candidate websites were photos
of the candidate, discussion of their background qualifications, and
their platform on campaign issues. A total of 69 percent discussed the
main elements of their campaign on their home page, and 43 percent had
created additional webpages for further discussion of their platforms.
The majority of candidates mentioned their political accomplishments and
civic experience multiple times, with 94 percent including a biography.
Additionally, 56 percent of candidates had lawn sign request forms on
their website, 88 percent encouraged donations, and 94 percent had a
request for volunteers.
Since the perceived user friendliness of a website may impact a
voter's ability to interact with a candidate, it is significant to
note that a large majority of candidate websites were generally well
designed and easy to navigate, having logical layouts, tidy
organization, and no textual errors. Only 8.5 percent of candidate
websites were difficult to navigate, with broken links and no table of
contents. Overall, candidates used websites to promote the main issues
of their platforms and gain support.
Facebook and Twitter
For the 26 percent of municipal candidates who used Facebook for
their campaigns, generating a large number of likes may be important
because it enables candidates to send frequent updates on their campaign
issues and to request support through votes, donations, volunteering, or
other means. The number of "likes" (9) that each candidate
received varied widely from a few to over two thousand, with 225 as the
average number of likes per candidate. Additionally, the average number
of Facebook posts during the 2010 Niagara municipal campaign was 64.
Of the candidates that used Facebook and Twitter, only about half
used either platform on a regular basis. This group of candidates
updated Facebook and Twitter at least once per day. The remaining half
did not use Facebook and Twitter very often, with a span of over seven
days between updates. It was also observed that candidates who had a
Twitter account were more likely to tweet regularly than Facebook users
posted new content. It is possible that this may be due to Twitter
attracting more technologically savvy candidates than Facebook.
YouTube
YouTube was the least utilized platform in the municipal campaigns.
The YouTube videos were typically embedded directly on the candidates
'websites, making YouTube an intermediary host for the
candidates' videos rather than a primary campaign tool. There were
very few candidates who had extensive YouTube pages: most simply listed
the videos that were embedded on websites and did not contain any other
text. Of the candidates who created YouTube videos, most of the video
content depicted candidates discussing their main campaign priorities.
Additionally, candidates posted videos of supporters promoting their
candidacy. The Welland mayoral candidates were the largest users of
YouTube, posting a large number of well-produced YouTube videos. The
videos provided voters insight into the candidates' communication
abilities that would otherwise only be gained in person or through
televised debates.
II. Interaction on Social Media
[FIGURE 3 OMITTED]
As illustrated in Figure 3, minimal interaction occurred between
candidates and voters on social media during the campaign. The average
level of interaction for all platforms was a value of 1.83 out of 5.
Facebook was the most interactive social media platform with a score of
2.21 out of 5, followed by Twitter with an interaction value of 2.05.
Websites and YouTube had lower scores (1.94 and 1.11 respectively).
Facebook
[FIGURE 4 OMITTED]
Most candidates with a Facebook page had at least 10 posts from
themselves and voters during the course of the campaign. There were a
relatively equal number of posts from candidates and voters. An
examination of the content of the posts revealed a lack of dialogue
among candidates and voters.
[FIGURE 5 OMITTED]
The majority of Facebook posts by voters offered electoral support,
using common statements such as "good luck on election day,"
"I'm voting for you! and I hope you win!" Posts from
voters rarely consisted of questions posed by voters to candidates
related to their political platforms and campaign issues, and very few
posts by voters providing their opinions on campaign issues. A total of
63 percent of candidates posed no questions at all to voters, and 30
percent posted 1 to 3 questions. Only 7 percent of candidates posted
more than 4 questions to voters, and 26 percent of candidates posted
responses to questions or comments posed by voters on Facebook.
Thirty-eight percent of candidates posted their platform elements
on Facebook, 37 percent posted biographical information, 4 percent
requested donations, 22 percent requested volunteers, or 37 percent
requested placement of lawn signs. Additionally, 11 percent of
candidates attacked competitors' positions or past records. In
contrast to campaign websites, Facebook was not used by candidates to
encourage campaign participation. On websites, 88% of candidates
requested donations, 94% requested volunteers, and 56% requested
placement of lawn signs. Given the more interactive and personalized
nature of Facebook compared to websites, it is noteworthy how little
social media was utilized to encourage campaign participation. Overall,
Facebook discussion exhibited a low level of interaction, with neither
candidates nor voters initiating discussion with each other.
Twitter
Similar to Facebook, the level of interaction on Twitter was low.
Among the candidates who used Twitter, most used it as a venue to
announce new Facebook posts or to direct users to news stories on their
websites. A small number of candidates offered daily updates on the
campaign happenings, informing followers about debates that they had
attended or experiences that they had while door-to-door campaigning in
the community.
[FIGURE 6 OMITTED]
Overall, there was minimal interaction between candidates and
voters on Twitter. The content of tweets was candidate-driven and
consisted mainly of links to news articles, websites, and daily campaign
updates. A total of 60 percent of candidates had more than 10 daily
campaign updates throughout the course of their campaign. These daily
updates included statements such as "out knocking on doors
today" or pertained to events and rallies that the candidate had
attended or had planned to attend. During the course of the campaign, 40
percent of candidates tweeted about supporter stories more than four
times. A total of 90 percent of Twitter-using candidates shared links to
news media on campaign issues or articles on themselves, and all shared
links to websites. Very few candidates had Twitter followers who tweeted
comments or opinions to the candidates, resulting in a low level of
interaction on Twitter between voters and municipal candidates.
Websites
Although websites were the most widely used platform among
candidates in the 2010 Niagara elections, websites also exhibited low
levels of interaction. Candidate websites primarily served as venues for
the candidates' campaign issues and biographical information. While
almost all websites encouraged voters to send questions or comments by
email to the candidates, very few allowed voters to pose questions or
comments to candidates on the sites themselves. A few candidates posted
questions that had been received by email from voters, as well as their
responses to the questions, but in these cases, the questions may have
been selected based on consistency with the candidate's campaign.
Moderated comments provided candidates control over content and hindered
interactivity.
Although there was significant use of social media by candidates
and voters during the election, the level of interaction was low.
Candidates posed very few questions to voters, and vice versa. This
relatively unsophisticated use of social media might be explained by the
nature of individual non-partisan campaigns at the municipal level.
Candidates are often running for part-time positions, and as a result,
may lack the time and resources to engage in multiple methods of
campaigning simultaneously. Candidates may also prefer more traditional
methods of campaigning such as door-to-door campaigning. However, if
candidates perceive an electoral benefit to engaging voters on social
media, these candidates may devote more energy to social media
campaigning in the future.
III. Impact of Social Media on Electoral Success
For all candidates, the use of Facebook, Twitter, YouTube, and
websites did not significantly impact electoral success. (11) The
results from the chi-square tests indicate that there is not a
statistically significant relationship between electoral success and the
presence of Facebook, Twitter, YouTube or website campaign tools. In
addition, the results from the Pearson's correlation coefficient
tests suggest that there is not a statistically significant relationship
between electoral success and the number of Facebook "likes,"
Facebook posts, Twitter followers, or tweets. However, a number of
significant differences between electoral success and the use of social
media emerged when candidates are differentiated by incumbents and
challengers.
For challengers, (12) the chi-square test results identify that
there is not a statistically significant relationship between electoral
success based on the presence of campaign websites, Twitter or YouTube
accounts. However, the Pearson's correlation coefficient results
suggest that the number of Facebook posts and "likes" have
statistically significant moderate positive correlations with electoral
success. Interestingly, simply having a Facebook page does not correlate
with electoral success; candidates need to generate interaction as
demonstrated by the number of "likes" and posts.
For incumbents, (13) the chi-square results also identify that
there is not a statistically significant relationship between electoral
success and the presence of a campaign website, Twitter or YouTube
account. In contrast to what was hypothesized, incumbents' use of
Facebook negatively impacted electoral success. Pearson's
correlation coefficient results indicate strong negative correlations
between electoral success and the number of Facebook posts and
"likes". The results on the number of Twitter followers and
tweets had to be excluded from the analysis due to the small sample size
(n=2). (14) The negative correlation between Facebook use and electoral
success for incumbents could be attributed to a number of factors.
First, incumbents who use social media may not have the resources for
other campaign strategies such as door-knocking. Second, they may decide
to utilize social media as a result of difficulties reaching the public
through traditional campaign strategies or may initiate social media
campaigning in an attempt to reinvigorate a stalling campaign. Third,
the use of social media may cause candidates
to redirect time and resources away from traditional campaign
methods that may have a greater impact on electoral success.
Conclusion
As voters become more willing to acquire local political
information on social media due to the low resources required to monitor
candidates and increasingly seize the opportunity to directly inform
candidates of issues of concern while, at the same time, candidates
become more competent with using the platforms for voter engagement, the
presence and prominence of social media in municipal campaigns is likely
to expand. In 2010, the use of social media platforms for political
engagement was still in its infancy. Municipal candidates rarely used
social media and candidate-voter interaction was minimal, but a deeper
analysis of social media use in the Niagara municipal elections provides
some preliminary insights on the potential value of social media,
particularly for incumbents. The following briefly summarizes the key
findings from this study and identifies promising directions for further
research.
In the 2010 Niagara municipal elections, it was determined that
both the extent of social media use and the interaction between
candidates and voters were low. Between 9% and 26% of candidates used at
least one social media platform during the campaign, with Facebook being
the most popular platform, followed by Twitter and YouTube. Very little
discussion of electoral issues emerged online and candidates rarely
asked voters questions on issues and vice versa. Most voters' posts
consisted of shallow messages of support, and most candidates'
posts were campaign updates. The interaction scale identified that the
average level of interaction for all social media platforms was 1.83 out
of 5. Discussions on Facebook, Twitter, and YouTube were not very
interactive, and websites lacked elements such as forums or polls. The
relatively recent emergence of social media as campaign tools may
explain the low level of interaction observed, as candidates have only
recently become familiar with these mediums.
Despite the limited use of social media during the municipal
elections, this study reveals significant correlations between the use
of social media and electoral success. The number of Facebook posts and
"likes" are positively correlated with increased electoral
success for challenger candidates. The unmediated and dynamic nature of
social media as well as the low resources required to create and share
information with large audiences may provide challenger candidates with
an edge in encouraging voter support. Conversely, the increased use of
social media by incumbent candidates is negatively associated with
electoral success, since incumbent candidates with an increased number
of Facebook posts and "likes" were less likely to be elected.
Although this study does not explore social media in conjunction with
other campaign methods, it is possible that faltering campaigns or
difficulty engaging voters through other methods may motivate incumbents
to engage in campaigning through social media to a greater extent.
Therefore, future research should compare social media use with other
factors affecting campaign success such as newspaper coverage, campaign
expenditures, door-to-door canvassing, the use of signs and incumbency
to determine social media's relative impact on campaign success.
Multiple regression analysis would allow for an assessment of the
significance of each of these variables. In addition, post-campaign
interviews with candidates could provide insight into online and offline
campaign strategies as well as lessons learned in social media voter
engagement.
Further studies should also refine the methods for assessing
interaction on social media. The interaction scale is an initial attempt
to incorporate a large number of elements to measure interaction, such
as the number of posts, replies, and questions and the frequency of
posts. Future research could incorporate other elements, such as the
language and tone of discussion as well as examine the voters who engage
candidates online. Small (2011) examines the proportion of #cdnpoli
discussion participants who are individuals, bloggers, media,
politicians, or associated with interest groups, think tanks or other
organizations. A similar analysis may provide valuable insight into the
representativeness of individuals interacting with candidates on social
media during elections.
Appendix 1: Use of Social Media and Electoral Success for All
Candidates
Use of Twitter and Electoral Success
Use of Twitter Total
in campaign
none used
twitter
Elected not Count 59 5 64
elected % within Use 62.1% 50.0% 61.0%
of Twitter
in campaign
elected Count 36 5 41
% within Use 37.9% 50.0% 39.0%
of Twitter
in campaign
Total Count 95 10 105
% within Use 100.0% 100.0% 100.0%
of Twitter
in campaign
Pearson Chi-Square value=.557 with 1 df, p=.455
Use of Campaign Website and Electoral Success
Use of Website Total
in campaign
none had
website
Elected not elected Count 39 25 64
% within Use of 56.5% 69.4% 61.0%
Website in
campaign
elected Count
% within Use of 43.5% 30.6% 39.0%
Website in
campaign
Total Count 69 36 105
% within Use of 100.0% 100.0% 100.0%
Website in
campaign
Pearson Chi-Square value=1.660 with 1 df, p=.198
Use of You Tube and Electoral Success
Use of You Tube Total
in campaign
none had
You Tube
videos
Elected not Count 58 6 64
elected % within Use of 60.4% 66.7% 61.0%
You Tube in
campaign
elected Count 38 3 41
% within Use of 39.6% 33.3% 39.0%
You Tube in
campaign
Total Count 96 9 105
% within Use of 100.0% 100.0% 100.0%
You Tube in
campaign
Pearson Chi-Square value=.135 with 1 df, p=.713.
Fisher's Exact 2-sided test value is 1.000
Use of Facebook and Electoral Success
Use of Facebook Total
page in campaign
none used
Facebook
Elected elected Count 47 17 64
% within Use of 61.0% 60.7% 61.0%
Facebook page
in
elected Count 30 11 41
% within Use of 39.0% 39.3% 39.0%
Facebook page
in campaign
Total Count 77 28 105
% within Use of 100.0% 100.0% 100.0%
Facebook page
in campaign
Pearson Chi-Square value=.001 with 1 df, p=.976.
Fisher's Exact 2-sided test value is .507
Candidate Use of Social Media and Electoral Success
Elected Number Number
of of
Facebook Facebook
likes posts
Elected Pearson Correlation 1 .257 .203
Sig. (2-tailed) .187 .300
N 105 28 28
Number Pearson Correlation .257 1 .620"
of Sig. (2-tailed) .187 .000
Facebook N 28 28 28
Number
likes
Number Pearson .203 .620 ** 1
of Correlation
Facebook Sig. (2-tailed) .300 .000
posts N 28 28 28
Number Pearson Correlation -.219 -.162 .175
of Twitter Sig. (2-tailed) .543 .678 .653
followers N 10 9 9
Number Pearson -.297 -.302 -.047
of Tweets Correlation
Sig. (2-tailed) .404 .429 .904
N 10 9 9
Number Pearson -.041 .432 * .565 **
of social Correlation
media Sig. (2-tailed) .679 .022 .002
platforms N 105 28 28
used by
candidate
Number Number Number
of Twitter of of social
followers Tweets media
platforms
used by
candidate
Elected Pearson Correlation -.219 -.297 -.041
Sig. (2-tailed) .543 .404 .679
N 10 10 105
Number Pearson Correlation -.162 -.302 .432 *
of Sig. (2-tailed) .678 .429 .022
Facebook N 9 9 28
Number
likes
Number Pearson .175 -.047 .565 **
of Correlation
Facebook Sig. (2-tailed) .653 .904 .002
posts N 9 9 28
Number Pearson Correlation 1 .892 ** -.128
of Twitter Sig. (2-tailed) .001 .724
followers N 10 10 10
Number Pearson .892 ** 1 -.046
of Tweets Correlation
Sig. (2-tailed) .001 .899
N 10 10 10
Number Pearson -.128 -.046 1
of social Correlation
media Sig. (2-tailed) .724 .899
platforms N 10 10 105
used by
candidate
. Correlation is significant at the U.U1 level (2-tailed;.
*. Correlation is significant at the 0.05 level (2-tailed).
Appendix 2: Challenger Use of Social Media and Electoral Success
Challenger Use of Twitter and Electoral Success
Use of Twitter Total
in campaign
none used
twitter
Elected not elected Count 58 4 62
% within Use of 77.3% 50.0% 74.7%
Twitter in
campaign
elected Count 17 4 21
% within Use of 22.7% 50.0% 25.3%
Twitter in
campaign
Total Count 75 8 83
% within Use of 100.0% 100.0% 100.0%
Twitter in
campaign
Pearson Chi-Square value=2.858 with 1 df, p=.091.
Fisher's Exact 2-sided test value is .107
Challenger Use of Campaign Website and Electoral Success
Use of Website Total
in campaign
none had
website
Elected not Count 38 24 62
elected % within Use of 73.1% 77.4% 74.7%
Website in
campaign
elected Count 14 7 21
% within Use of 26.9% 22.6% 25.3%
Website in
campaign
Total Count 52 31 83
% within Use of 100.0% 100.0% 100.0%
Website in
campaign
Pearson Chi-Square value=.194 with 1 df, p=.660
Challenger Use of YouTube and Electoral Success
Use of You Tube Total
in campaign
none had
You Tube
videos
Elected not Count 57 5 62
elected % within Use of 76.0% 62.5% 74.7%
You Tube in
campaign
elected Count 18 3 21
% within Use of 24.0% 37.5% 25.3%
You Tube in
campaign
Total Count 75 8 83
% within Use of 100.0% 100.0% 100.0%
You Tube in
campaign
Pearson Chi-Square value=.697 with 1 df, p=.404.
Fisher's Exact 2-sided test value is .411
Challenger Use of Facebook and Electoral Success
Use of Facebook Total
page in campaign
none used
Facebook
Elected not elected Count 46 16 62
% within Use of 74.2% 76.2% 74.7%
Facebook page
in campaign
elected Count 16 5 21
% within Use of 25.8% 23.8% 25.3%
Facebook page
in campaign
Total Count 62 21 83
% within Use of 100.0% 100.0% 100.0%
Facebook page
in campaign
Pearson Chi-Square value=.033 with 1 df, p=.856
Challenger Electoral Success and Social Media Use
Elected Number Number of
of Facebook
Facebook posts
likes
Elected Pearson Correlation 1 .455 * .442 *
Sig. (2-tailed) .038 .045
N 83 21 21
Number Pearson Correlation .455 * 1 .642 **
of Facebook Sig. (2-tailed) .038 .002
likes N 21 21 21
Number Pearson Correlation .442 * .642 ** 1
of Facebook Sig. (2-tailed) .045 .002
posts N 21 21 21
Number Pearson Correlation -.168 -.191 .133
of Twitter Sig. (2-tailed) .691 .682 .776
followers N 8 7 7
Number Pearson Correlation -.294 -.350 -.035
of Tweets Sig. (2-tailed) .480 .441 .940
N 8 7 7
Number Pearson Correlation .043 .422 .497
of social Sig. (2-tailed) 698 057 .022
media N 83 21 21
platforms
used by
candidate
Number Number Number of
of of social
Tweets Tweets media
followers used by
candidate
Elected Pearson Correlation -.168 -.294 .043
Sig. (2-tailed) .691 .480 .698
N 8 8 83
Number Pearson Correlation -.191 -.350 .422
of Facebook Sig. (2-tailed) .682 .441 .057
likes N 7 7 21
Number Pearson Correlation .133 -.035 .497
of Facebook Sig. (2-tailed) .776 .940 .022
posts N 7 7 21
Number Pearson Correlation 1 .911 ** -.201
of Twitter Sig. (2-tailed) .002 .633
followers N 8 8 8
Number Pearson Correlation .911 ** 1 -.062
of Tweets Sig. (2-tailed) .002 .885
N 8 8 8
Number Pearson Correlation -.201 -.062 1
of social Sig. (2-tailed) .633 .885
media N 8 8 83
platforms
used by
candidate
*. Correlation is significant at the 0.05 level (2-tailed).
**. Correlation is significant at the 0.01 level (2-tailed).
Appendix 3: Incumbent Use of Social Media and Electoral Success
Incumbent Use of Facebook and Electoral Success
Use of Facebook Total
page in campaign
none used
Facebook
Elected not elected Count 1 1 2
% within Use of 6.7% 14.3% 9.1%
Facebook page
in campaign
elected Count 14 6 20
% within Use of 93.3% 85.7% 90.9%
Facebook page
in campaign
Total Count 15 7 22
% within Use of 100.0% 100.0% 100.0%
Facebook page
in campaign
Pearson Chi-Square value is .335 with 1 df and p= .563.
Fishers Exact 2-sided test value is 1.000
Incumbent Use of Twitter and Electoral Success
Use of Twitter Total
in campaign
none used
twitter
Elected not elected Count 1 1 2
% within Use of 5.0% 50.0% 9.1%
Twitter in
campaign
elected Count 19 1 20
% within Use of 95.0% 50.0% 90.9%
Twitter in
campaign
Total Count 20 2 22
% within Use of 100.0% 100.0% 100.0%
Twitter in
campaign
Pearson Chi-Square value 4.455 with 1 df and p=.035.
Fisher's Exact 2-sided test value is .177
Incumbent Use of Campaign Website and Electoral Success
Use of Website Total
in campaign
none had
website
Elected not elected Count 1 1 2
% within Use of 5.9% 20.0% 9.1%
Website in
campaign
elected Count 16 4 20
% within Use of 94.1% 80.0% 90.9%
Website in
campaign
Total Count 17 5 22
% within Use of 100.0% 100.0% 100.0%
Website in
campaign
Pearson Chi-Square value is .932 with 1 df and p=.334.
Fishers Exact 2-sided test value is .411
Incumbent Use of YouTube and Electoral Success
Use of You Tube Total
in campaign
none had
You Tube
videos
Elected not elected Count 1 1 2
% within Use of 4.8% 100.0% 9.1%
You Tube in
campaign
elected Count 20 0 20
% within Use of 95.2% 0.0% 90.9%
You Tube in
campaign
Total Count 21 1 22
% within Use of 100.0% 100.0% 100.0%
YouTube in
campaign
Incumbent Use of Social Media and Electoral Success
Elected Number of
Facebook
likes
Elected Pearson Correlation 1 -.937 **
Sig. (2-tailed) .002
N 22 7
Number of Pearson Correlation -.937 ** 1
Facebook likes Sig. (2-tailed) .002
N 7 7
Number of Pearson Correlation -.805 * .714
Facebook posts Sig. (2-tailed) .029 .072
N 7 7
Number of Pearson Correlation -1.000 ** 1.000 **
Twitter Sig. (2-tailed) . .
followers N 2 2
Number of Pearson Correlation -1.000 ** 1.000 **
Tweets Sig. (2-tailed) . .
N 2 2
Number of Pearson Correlation -.364 .746
social media Sig. (2-tailed) .095 .054
platforms used N 22 7
by candidate
Number of Number
Facebook of Twitter
posts followers
Elected Pearson Correlation -.805 * -1.000 **
Sig. (2-tailed) .029 .
N 7 2
Number of Pearson Correlation .714 1.000 **
Facebook likes Sig. (2-tailed) .072 .
N 7 2
Number of Pearson Correlation 1 1.000 **
Facebook posts Sig. (2-tailed) .
N 7 2
Number of Pearson Correlation 1.000 ** 1
Twitter Sig. (2-tailed) .
followers N 2 2
Number of Pearson Correlation 1.000 ** 1.000 **
Tweets Sig. (2-tailed) . .
N 2 2
Number of Pearson Correlation .869 * 1.000 **
social media Sig. (2-tailed) .011 .
platforms used N 7 2
by candidate
Number Number
of of social
Tweets media
platforms
used by
candidate
Elected Pearson Correlation -1.000 ** -.364
Sig. (2-tailed) . .095
N 2 22
Number of Pearson Correlation 1.000 ** .746
Facebook likes Sig. (2-tailed) . .054
N 2 7
Number of Pearson Correlation 1.000 ** .869 *
Facebook posts Sig. (2-tailed) . .011
N 2 7
Number of Pearson Correlation 1.000 ** 1.000 **
Twitter Sig. (2-tailed) . .
followers N 2 2
Number of Pearson Correlation 1 1.000 **
Tweets Sig. (2-tailed) .
N 2 2
Number of Pearson Correlation 1.000 ** 1
social media Sig. (2-tailed) .
platforms used N 2 22
by candidate
**. Correlation is significant at the 0.01 level (2-tailed).
*. Correlation is significant at the 0.05 level (2-tailed).
Acknowledgements
Expansion of research conducted for the Niagara Community
Observatory in preparation of the policy brief "The Use of Social
Media in the 2010 Niagara Municipal Elections."
References
Chen, R, and P. Smith. 2011. Digital media in the 2008 Canadian
election. Journal of Information Technology & Politics 8 (4):
399-417.
Endres, D., and B. Warnick. 2004. Text-based interactivity in
candidate campaign websites: a case study from the 2002 elections.
Western Journal of Communication 68 (3): 322-42.
Francoli, M., J Greenberg, and C. Waddell. 2011. The campaign in
digital media. In The Canadian Federal Election of 2011, ed. J. Pammett
and C. Doman. Toronto: Dundurn Press, 219-246.
Gerber, A., and D. P. Green. 2000. The effects of canvassing,
telephone calls, and direct mail on voter turnout: a field experiment.
American Political Science Review 94 (3): 653-664.
Gibson, R., and S. Ward. 2009. Parties in the digital age: a review
article. Representation 45 (1): 87-100.
Goodman, N. and L. Copeland. 2010. Candidates do it, why can't
election agencies? The Town of Markham's use of Facebook in the
2010 municipal election. In communication presentee a la conference des
etudiants de deuxieme cycle de la Chaire Bell sur lapolitique
canadienne, Ottawa (Vol. 11).
Graham, T., M. Broersma, and K. Hazelhoff. 2013. Closing the gap?
Twitter as an instrument for connected representation. In R. Scullion,
R. Gerodimos, D. Jackson, D. Lilleker & (Eds.) The Media, Political
Participation and Empowerment. New York: Routledge, 71-88.
Hsieh, H-F. and S. Shannon. 2005.Three Approaches to Qualitative
Content Analysis. Qualitative Health Research 15 (9): 1277-1288.
Jenkins, H., and D. Thorburn, 2004. Democracy and new media.
Cambridge, MA: MIT Press.
Kiousis, S. 2002. Interactivity: A Concept Explication. New Media
and Society 4 (3): 355-383.
Krashinsky, M., and W.J. Milne. 1986. The effect of incumbency in
the 1984 federal and 1985 Ontario elections. Canadian Journal of
Political Science 19 (2): 337-343.
Kushner, J., D. Siegel, and H. Stanwick. 1997. Ontario Municipal
Elections: Voting Trends and Determinants of Electoral Success in a
Canadian Province. Canadian Journal of Political Science 30 (3):
539-553.
Kushner, J., and D. Siegel. 2006. Why do municipal electors not
vote? Canadian Journal of Urban Research 15 (2): 264-278.
Marketshare.hitslink.com. 2010. Market share reports. Hitslink.
October 2010.
Martin, D. 2009. Teens don't tweet: Twitter's growth not
fuelled by word-of-mouth. Nielsenwire: The Nielsen Company,
http://blog.nielsen.com/nielsenwire/online_
mobile/teens-dont-tweet-twitters-growth-not-fueled-by-youth/ (Accessed
20 April 2011).
McMillan, S. J., and J. S. Hwang. 2002. Measures of perceived
interactivity: an exploration of the role of direction of communication,
user control, and time in shaping perceptions of interactivity. Journal
of Advertising 31 (3): 29-42.
Mehta, C. and N. Patel. 2011. IBM SPSS Exact Tests. Armonk, NY: IBM
Corporation.
Nielson Company. 2009. Global Faces and Networked Places: A Nielsen
report on Social Networking's New Global Footprint. The Nielsen
Company. http://www.
nielsen.com/content/dam/corporate/us/en/newswire/uploads/2009/03/nielsen_ globalfaces_mar09.pdf (Accessed October 2010).
Oliviera, M. 2013. How do you compare? New report reveals stats
about social media usage in Canada. Media Technology Monitor, 29 April,
http://www.mtm-otm.ca/
files/Press/New%20report%20revels%20stats%20about%20social%20media%20
in%20Canada%20-%20CP%20%20Michael%2001iviera%20April%2029%20 2013.pdf
(Accessed 10 September 2013).
Paperny, A. 2011. Jack Layton's debatable #fail. Globe and
Mail, 13 April 2011. http://
www.theglobeandmail.com/news/politics/jack-laytons-debatable-hashtag-fail/ article576224/
Park, C. 2013. Does Twitter motivate involvement in politics?
Tweeting, opinion leadership, and political engagement. Computers in
Human Behavior. 29(4): 1641-8.
Slingshot SEO. 2011. A tale of two studies: Google vs Bing
click-through rate. Moz, 6 December,
http://moz.com/blog/a-tale-of-two-studies-google-vs-bingclickthrough-rate (Accessed 2 September 2013)
Raynauld, V. and J. Greenberg. 2014. Tweet, Click, Vote: Twitter
and the 2010 Ottawa Municipal Election. Journal of Information
Technology and Politics (July 2014).
Small, T. 2008a. Equal access, unequal success: major and minor
Canadian parties on the net. Party Politics 14: 51-70.
Small, T. 2008b. The Facebook effect? Online campaigning in the
2008 Canadian and US elections. Policy Options November: 85-87.
Small, T. 2010. Canadian Politics in 140 Characters: Party Politics
in the Twitterverse. Canadian Parliamentary Review Autumn: 39-45.
Small, T. 2011. What the hashtag? Information, Communication &
Society 14(6): 872-95.
Sweetser, K., and R. Weaver-Lariscy. 2008. Candidates make good
friends: an analysis of candidates' uses of Facebook. International
journal of Strategic Communication 2 (3): 175-198.
Trent, J., and R. Friedenberg. 2008. Political Campaign
Communication: Principles and Practices. 6th ed. Plymouth, UK: Rowman
& Littlefield Publishers.
Unger, T. 2011. The changing face of SERPs: organic click through
rate. Optify. http://
www.optify.net/search-marketing/organic-click-through-rate
Vergeer, M., L. Hermans and S. Sams. 2013. Online social networks
and microblogging in political campaigning: The exploration of a new
campaign tool and a
Walton, D. 2011. Naheed Nenshi apologizes for "off your
meds" Twitter rebuke. The Globe and Mail 7 November,
http://www.theglobeandmail.com/news/politics/
naheed-nenshi-apologizes-for-off-your-meds-twitter-rebuke/article4183121/
Xenos, M., A. Vromen and B. Loader. 2014. The great equalizer?
Patterns of social media use and youth political engagement in three
advanced democracies. Information, Communication & Society 17(2):
151-167.
Douglas Hagar
Department of Political Science
Carleton University
Notes
(1) Obtained from politwitter.ca
(2) For example, see Walton, D. 2011. "Naheed Nenshi
apologizes for 'off your meds' Twitter rebuke." The Globe
and Mail. November 7, 2011. http://www.theglobeandmail.
com/news/politics/naheed-nenshi-apologizes-for-off-your-meds-twitter-rebuke/ article4183121/
(3) The twelfth municipality was not included because the mayoral
candidate was acclaimed no contest.
(4) Small municipalities have populations of less than 30,000,
medium-sized municipalities have populations of between 30-100,000 and
large-municipalities have populations of over 100,000.
(5) For example "Bob Smith", "Bob Smith
Councillor", "Bob Smith Lincoln".
(6) The total population of the Niagara region is 427,421.
(7) A tweet is a post on Twitter. Tweets are a maximum of 140
characters long and can include links, photos and hashtags. Hashtags are
used to self-categorize users tweets according to topics. Hashtags are
prefaced with a hash symbol (#). A re-tweet is a reposting of another
user's tweet.
(8) Facebook posts are status updates that allow users to share
their thoughts and other information with their friends. When users
update their status, it is shared on a user's personal wall and on
their friends' news feeds.
(9) Facebook Tikes' allow users to show their support for
comments, wall posts, fan pages and groups. The Tikes' feature
replaced the "become a fan" button of Facebook pages in 2010.
After a user Tikes' a page, their friends are notified that they
have 'liked' the page and the user's news feed displays
the latest updates on the pages.
(10) Twitter users can broadcast tweets and/or follow other users
tweets. Following twitter users allows users to be informed of new
tweets on their livefeeds on their mobile devices or when logged into
Twitter in a web browser.
(11) See appendix 1 for the results of the chi-square and
Pearson's correlation coefficient tables.
(12) See appendix 2 for the results.
(13) See appendix 3 for the results.
(14) The number of Twitter followers and number of tweets for
incumbents is based on a very small sample. This variable requires
further exploration in future research. new campaign style. Party
Politics 19(3): 477-501.
Table 1: Interaction Scale Variables
Facebook
--Number of posts by
Candidate+
--Number of posts by Voters+
--Daily updates on campaign+
--Frequency of posts **
--Questions posed to voters+
--Responses to questions posed
to voters+
--Voters' Questions+
--Responses to Voters'
Questions+
--Voter attacks +candidate's
position *
-Candidate responds to attack *
Twitter
--Number of tweets by
candidate+
--Number of tweets by others+
--Daily updates on campaigns-
--Frequency of posts **
--Questions posed to voters+
--Responses to questions posed
to voters+
--Voters' Questions+
--Responses to Voters'
Questions+
--Voter attacks candidate's
position *
--Candidate responds to attack *
Websites
--Daily updates+
--Link to blog or
presence of blog *
--Shares supporter's
stories *
--Requests to ask
candidate a question *
--Answers to
submitted questions *
--Introduces an
application or other
method to facilitate
interaction *
YouTube
-Number of
comments
on videos.+
+ System A * System B; ** System C
Table 2: Variable Coding Systems
System A
Value Content
1 Zero posts, tweets or
updates
2 1-3 posts, tweets or
updates
3 4-6 posts, tweets or
updates
4 7-9 posts, tweets or
updates
5 10+ posts tweets or
updates
System B
Value Content
1 Zero posts, tweets or
updates
2 1 post, tweet or update
3 2 posts, tweets or updates;
4 3 posts, tweets or updates
5 4+ posts tweets or updates
System C
Value Content
1 Average of more
than 7 days between
the last S posts,
tweets or updates
2 Average of 5-7 days
between the last
5 posts, tweets or
updates
3 Average of 3-4 days
between the last
5 posts, tweets or
updates
4 Average of 1-2 days
between the last
5 posts, tweets or
updates
5 Average of less than
one day between the
last 5 posts, tweets
or updates
Table 3: Average Facebook and Twitter Usage
Facebook Facebook Twitter Twitter
"Likes" Posts Followers (10) "Likes"
Average Number 255 64 46 98
Standard Deviation 401 67 29 148
Table 4: Relationship between Electoral
Success and Social Media or Website Use
Incumbents Challengers All Candidates
(n=22) (n=83) (n=105)
Facebook Use * Not significant Not significant Not significant
Twitter Use * Not significant Not significant Not significant
Website Use * Not significant Not significant Not significant
YouTube Use * Not significant Not significant Not significant
Number of Statistically Statistically Not significant
Facebook significant significant
"likes" negative positive
correlation correlation
Number of Statistically Statistically Not significant
Facebook posts significant significant
** negative positive
correlation correlation
Number of N/A (1) Not significant Not significant
Twitter
followers **
Number of N/A (1) Not significant Not significant
"tweets" **
Number of Not significant Not significant Not significant
platforms used
by candidate **
(1) value withheld due to small sample size
* tested using chi-square
** tested using Pearsons correlation coefficient
Figure 1: Percentage of candidates that
used social media by platform
Website 35%
Facebook 26%
Twitter 10%
YouTube 9%
Note: Table made from bar graph.
Figure 2: Frequency of Updates by Percentage
of Candidates
Facebook Twiter
Less than
1 day 29% 40%
1-2 days 7% 10%
3-4 days 25% 0%
5-7 days 10% 20%
over 7 days 29% 30%
Note: Table made from bar graph.
