2. Background and Objectives
This report builds on earlier studies examining social media activity involving the principal
candidates in the 2010 Toronto Mayoral election: Rob Ford, Joe Pantalone and George
Smitherman.
It reviews conversations taking place on social networks and reports on propensity of each
candidate to be mentioned online as well as the accompanying sentiment from September 20
to October 21
This report offers a final effort to understand which candidate is winning the “word of mouth”
war online, and will serve as the final reference for the relationship between online chatter
and actual voting results come election day October 25.
3. About Social Media
Social media is conversation or interaction between people online. Unlike traditional news
media, which tries to maintain objectivity and fairness in reporting the news, social media is a
domain of unfettered subjectivity and bias. It’s an environment where innuendo assumes
equal prominence alongside fact, and where special interests are forever seeking to leverage
the channel’s potential to influence opinion and alter outcomes in their favour. It truly is a new
and wild frontier.
Traditional polling asks voters who they plan to vote for. Social media provides a broader
perspective. The volume and tone of discussion provides a real-time indication of voter
engagement and intentions, possibly foretelling the fortune of the candidates based on the
response and potential influence of the medium.
4. Trended Mention Activity – Final 3 Candidates
In adding together all of the mentions of the 3 candidates during the last 30 days, 65,543
posts emerged
3807
2444
2499
2093
1019
714
3410
3300
2203
1240
1055
881
1374
1163
2133
2361
1384
2031
982 956
713
1921
2682
2171
1740
913
1912
2706
4188
3453
2837
3258
20/09/2010
21/09/2010
22/09/2010
23/09/2010
24/09/2010
25/09/2010
26/09/2010
27/09/2010
28/09/2010
29/09/2010
30/09/2010
01/10/2010
02/10/2010
03/10/2010
04/10/2010
05/10/2010
06/10/2010
07/10/2010
08/10/2010
09/10/2010
10/10/2010
11/10/2010
12/10/2010
13/10/2010
14/10/2010
15/10/2010
16/10/2010
17/10/2010
18/10/2010
19/10/2010
20/10/2010
21/10/2010
Daily Average:
2009
5. Trended Mention Activity: Ford
There’s been a downward trend in Ford mentions, with peaks in activity attributable to
televised debates, controversial policy announcements and polls challenging Ford’s front-
runner status
2894
1587
1482
1393
645
472
2200
1848
1219
612
549
696
868
700
1170
1018
730
1350
724
568
393
1155
1439
921
810
525
875
1235
1811
1493
1070
1245
20/09/2010
21/09/2010
22/09/2010
23/09/2010
24/09/2010
25/09/2010
26/09/2010
27/09/2010
28/09/2010
29/09/2010
30/09/2010
01/10/2010
02/10/2010
03/10/2010
04/10/2010
05/10/2010
06/10/2010
07/10/2010
08/10/2010
09/10/2010
10/10/2010
11/10/2010
12/10/2010
13/10/2010
14/10/2010
15/10/2010
16/10/2010
17/10/2010
18/10/2010
19/10/2010
20/10/2010
21/10/2010
Poll compels
Rossi to exit
race
Ford unveils
fiscal plan
Marathon ban,
“Two Horse” race
poll released
Rossi
attacks Ford
at televised
debate
Final
CP24
Debate
Daily Average:
1058
N= 35033
8. Trended Mention Activity – All Candidates
For the first time in the past 30 days, on October 20 Smitherman overtook Ford in terms of
absolute mentions
n=65543
0
500
1000
1500
2000
2500
3000
3500
20/9/1021/9/1022/9/1023/9/1024/9/1025/9/1026/9/1027/9/1028/9/1029/9/1030/9/101/10/102/10/103/10/104/10/105/10/106/10/107/10/108/10/109/10/1010/10/1011/10/1012/10/1013/10/1014/10/1015/10/1016/10/1017/10/1018/10/1019/10/1020/10/1021/10/10
Ford Smitherman Pantalone
9. Candidate Share of Voice
Ford
54%
Smitherman
28%
Pantalone
17%
Share of Voice is a proxy for online presence. It’s a relative measure of proportion of
instances where a candidate is mentioned during the specified time frame.
For the period measured, Rob Ford has a larger share of mentions than the other candidates
combined.
n=65543
10. Trended Share of Voice
In the last week of the campaign, Smitherman’s share of voice has eclipsed Ford and pulled
away significantly from Pantalone
0
10
20
30
40
50
60
70
80
90
20/9/1021/9/1022/9/1023/9/1024/9/1025/9/1026/9/1027/9/1028/9/1029/9/1030/9/101/10/102/10/103/10/104/10/105/10/106/10/107/10/108/10/109/10/1010/10/1011/10/1012/10/1013/10/1014/10/1015/10/1016/10/1017/10/1018/10/1019/10/1020/10/1021/10/10
Ford Smitherman Pantalone
%
n=65543
11. Sentiment – All Candidates
Sentiment analysis compares the percentage of opinionated (i.e. non-neutral) posts
Ford has the highest proportion of negative comments and Pantalone has the highest
proportion of positive remarks
68% 74% 68% 62%
32% 26% 32% 38%
Overall Ford Smitherman Pantalone
Positive
Negative
N= 8582 N=4695 N=2538 N=1349
12. Trended Sentiment: Ford
The week following October 10 had the highest proportion of positive comments, two days
after he released his fiscal plan and the week Rossi dropped out of the race.
75 78 75 69 74 74
25 22 25 31 26 26
Sept 20 to 25 Sept 26 to Oct 2 Oct 3 to 9 Oct 10 to 16 Oct 16 to 22 Total
Positive
Negative
N=1355 N=515 N=902 n=927 N=996 N=4695
13. Trended Sentiment: Smitherman
Smitherman received a boost in positive support following Thomson’s exit from the race,
which has softened slightly heading into the last week of the campaign. .
74 70 65 62
70 68
26 30 35 38
30 32
Sept 20 to 25 Sept 26 to Oct 2 Oct 3 to 9 Oct 10 to 16 Oct 16 to 22 Total
Positive
Negative
N=320 N=228 N=517 N=570 N=903 N=2538
14. Trended Sentiment: Pantalone
Pantalone has consistently enjoyed the highest proportion of positive comments online, albeit
with significantly lower volumes than the other candidates.
67
57 59 59 66 62
33
43 41 41 34 38
Sept 20 to 25 Sept 26 to Oct 2 Oct 3 to 9 Oct 10 to 16 Oct 16 to 22 Total
Positive
Negative
N=212 N=138 N=291 N=280 N=428 N=1349
15. Media Source
All candidates share a similar mix of comments on the various social media types, with
Twitter being the dominant medium of conversation.
Twitter, 73%
Blogs, 18%
Social Networks,
4%
Message Boards,
4%Mainstream Media,
3%
Forums, 1%
n=65543
16. Social Network Profiles
Pantalone gained the largest number of Facebook fans during the past month, while
Smitherman attracted the most followers. Ford actually lost followers towards the end of the
campaign.
1913
2087
2487
1934
2166
2598
1896
2416
3273
Ford
Smitherman
Pantalone 1617
2087
1931
2594
2869
3309
1494 1621
2081
Facebook Fans Twitter Followers
Sep-24 Oct-01 Oct-20
Sep-24 Oct-01 Oct-20
17. Summary
Elections are often characterized by momentum. As this particular campaign comes to a close, it would
appear as if George Smitherman has momentum on his side, at least in terms of the being the candidate
that people are talking about. For the first time in the past month, and in any of the social media studies
done on this election, Smitherman has surpassed Ford as the most discussed candidate.
Sentiment towards each of the candidates has remained relatively constant. Ford has always been a
controversial, polarizing figure which is supported by the fact the he’s consistently fostered the most
negativity. Then again, one might expect that the campaign front-runner would be subject to the most
criticism, particularly in social media where his opponents strive to knock him off of his pedestal.
Smitherman has maintained his positive sentiment scores, even during the last crucial week when his
volume of mentions have increased.
There are a couple of bright spots with the Pantalone campaign. While hasn’t really engendered as
much chatter as his opponents, he’s consistently achieved the most positive sentiment scores, and has
done remarkably well on the social networks, having more Facebook friends than both of his opponents
and more Twitter followers than Ford.
19. Search Terms Used
Rob Ford
Rob Ford
Robford
robfordteam
“Ford” and “voteto”
or “mayor”
Slobford
Exclusionary terms
Fiesta, mustang,
crown, lincoln,
focus, motor, co
George Smitherman
George Smitherman
G_Smitherman
“Smitherman” and
“voteto” or “mayor” and
not “carole” or
“birmingham”
“George” and “voteto”
“slitherman”
Joe Pantalone
Pantalone
jpantalone
Joey Pants and
not “movie” or
“festival” or
“pantoliano”
“Joe” and “voteto”
20. Methodology
We used specialized social media software to aggregate English language mentions of the
three remaining candidates in the race from September 20 through October 21th
. The
software collected information from social media sources like blogs, discussion forums,
Twitter, mainstream media articles and comments. Data from image sites like YouTube and
Flickr was not included. Data from Facebook was limited to publicly available content only,
meaning any private conversations between friends that are protected by passwords and
user privacy settings were not collected. Essentially, anything that is considered part of the
public domain was considered for the analysis.
The system looks for mentions based on a query consisting of key words. The complete list
of key words and exclusions is contained in the appendix. In general we used a combination
of the candidates formal names (e.g George Smitherman) in addition to naming conventions
commonly used as part of the online vernacular (e.g. Joey Pants).
Sentiment was calculated using an automated sentiment engine, which essentially looks for
overtly positive or negative words that modify the subject, i.e. the candidates names
21. About Patrick Gladney (@pgladney)
Patrick Gladney was one of the first marketers in Canada to specialize in the area of social
media research. He pioneered Social Currency™, a social media practice for Northstar
Research Partners. His approach utilizes traditional research practices and principles to
derive true insight from social media, a process that differs considerably from basic brand
monitoring that’s become commonplace today. His experience working with clients such as
BMW, Fairmont Hotels, RBC and The Toronto Star affords him a keen understanding of both
how and why consumers engage in social media and how to accurately measure and
understand consumer opinion. And because he has also worked for over 15 years in the
advertising and communications industry, Patrick can advise on online and offline marketing
solutions that respect the challenges confronting businesses in this new age of
communications.