A report into the online penetration of five leading political parties during the Singapore General Election 2011. Conducted by Grayling, the world's second largest independent PR consultancy, the report assesses share of voice, sentiment, conversation peaks, popular channels and key influencers.
2. Why are you reading this?
In the Singapore General Election 2011, the national political parties were allowed to campaign
using social media for the first time ever. With approximately 74% of internet users in Singapore
on facebook, close to 1 million tweeting and as many as 7 out of 10 writing blogs, the potential
impact of digital politics is immense. The engagement levels enjoyed by certain politcians are
well documented, but at Grayling we wanted to drill deeper to measure not just the volume of
noise, but also who was saying what, where, how and when. Using our unique online monitoring
tools we have compiled the following report that assesses the online penetration of five leading
political parties.
A word on methodology…
This report is a comparative study of conversations taking place outside official party channels,
in blogs, forums, microblogs, comments and news sites. No party facebook pages, twitter
accounts or websites are included in the study, though some individual candidate blogs
are. In order to return accurate results, all comparisons exclude other parties or events – for
instance, the percentage of votes polled (page 3) is the percentage from total votes polled by
the assessed parties, not the percentage of votes polled in the election as a whole. Similarly, in
order to assess metrics such as sentiment in a like-for-like manner, figures are normalised, rather
than evaluated as raw figures, as some parties generated significantly more buzz than others.
Posts have been analysed by man and machine to ensure their relevancy and to avoid skewed
results. The parties assessed are: the People’s Action Party (PAP), the Workers’ Party (WP),
the National Solidarity Party (NSP), the Singapore Democratic Party (SDP) and the Singapore
People’s Party (SPP). The study was conducted for four weeks, from 10 April to 7 May.
The interesting stuff…
Page
Executive summary 2
Share of voice 3
Sentiment 4
Conversation spikes 5
Channels 6
Influencers 7
1
3. EXECUTIVE SUMMARY
Results of the study show several interesting findings for digital communicators, including how other aspects
of an integrated campaign can influence online conversation.
Lots of noise doesn’t mean everyone is listening
Firstly, a share of voice does not translate into direct results. All opposition parties enjoyed a
higher share of voice than the percentage of either the votes polled or seats awarded. This can
partly be explained by the fact that share of voice includes neutral and negative mentions, as
well as positive ones, so not every online comment would translate into an advocate or vote. In
addition, the share of voice is not geographically defined. It is important to drive conversation
in target markets or areas. For instance, based on election results, we can assume that though
the NSP had a greater share of voice than the WP (16% v 15%) those people mentioning them
(NSP) were spread across Singapore rather than being concentrated in key areas.
It is people with a problem who speak up
Being an established entity tends to drive mentions, with PAP generating nearly three times as
many mentions as its closest competitor, the WP However, this also meant the PAP had the
.
largest proportion of negative comments – in fact it had more negative mentions than any party
(PAP included) had positive mentions. People were more likely to engage with the election in
order to criticise rather than support. With regards to sentiment, the percentage of positive
mentions more accurately reflects voting patterns. Only the PAP and WP achieved significant
proportions of positive mentions.
Offline events mean online engagement
Spikes in mentions and search do not directly correlate, indicating that an increase in awareness
and interest (search) does not automatically result in an increase in engagement (mentions).
Spikes appeared on both levels in the days of, or immediately following, the election and major
rallies. Engagement was driven by real-life events which provided users with direct reference
points and opportunities to post firsthand content.
Advocates get involved at crucial times
The channel that generated most mentions was microblogs (namely Twitter). Users were more
likely to make brief comments or share links than they were to produce lengthy content such
as blogs or forum posts. This was particularly relevant during election week and especially as
results were announced. Microblog sentiment in election week revealed more positive mentions,
indicating that advocates are more likely to post online in the immediate lead-up to an actual
event.
News outlets start onversations, individuals spread them
In terms of which users made most mentions, the online portals of traditional news outlets
posted much more frequently than any individual blogger or microblogger. The proportion of
total mentions made via news sites, however, was significantly smaller. News sites serve to
provide regular and reliable information and start the conversation – it’s social users who take
the content and popularise it online. There is a large number of social users who made frequent
mentions, many of whom have high followings and influence, which highlights the importance
of engaging both traditional and social media.
2
4. SHARE OF VOICE
PAP WP NSP SDP SPP
Seats won in Votes received in Online share of voice
Singapore GE 2011 (%) Singapore GE 2011 Singapore GE 2011
(% of total votes cast for (% of total mentions of
five focus parties) five focus parties)
93 65 5 49 13
7 14 2 15 7
13 16
Following the General Election 2011, the PAP continues to dominate the Singapore parliament, though the
party received a lower percentage of votes than in previous years. The PAP also enjoyed the largest online
share of voice, though as a percentage of total buzz it was significantly smaller than both the percentage
of seats and votes won. Although the NSP failed to win a seat in parliament, it enjoyed the second largest
online share of voice.
Online share of voice is important, but it doesn’t account for spread over electoral constituencies, thus
explaining why buzz didn’t necessarily translate into votes and seats. In addition the total share of voice
does not evaluate the manner in which parties were mentioned.
3
5. SENTIMENT
Data is normalised with 100 representing the highest level of
mentions within the study. Other volumes of mention are divided
by the highest level to produce all other points on the chart.
Overall
Positive Neutral Negative In the four weeks leading up
100 to the election the PAP was
the party mentioned most
frequently online, but it wasn’t
80 always good news. While more
supportive comments were
made about the PAP than any
60 other party, the incumbents
were actually mentioned
negatively more frequently
than positively. Netizens felt
40
more strongly about voicing
their dissatisfaction with the
PAP than sharing their support
20
or admiration for an opposition
party. Conversely, opposition
parties were barely referred to
0 negatively at all.
PAP WP NSP SDP SPP
Weekly Week 1 Week 2
The opposition’s share of
voice peaked in weeks two
and three, fuelled first by
a rapid increase in NSP
mentions and then a surge in
SDP related conversations.
Significantly, mentions were
primarily neutral and did not
translate into mass positive
conversation. After week
one the WP maintained a
consistent share of voice
though this gradually Week 3 Week 4
became more positive, a
trend reflected by the PAP .
Other opposition parties,
on the other hand, were
mentioned less frequently
and less positively in
election week than earlier in
their campaigns.
4
6. CONVERSATION SPIKES
Unsurprisingly, most WP SPP SDP PAP NSP
mentions of each party Buzz
came during the election
weekend. The buzz
dwarved other online
conversation peaks.
Search volume, on the
other hand, saw parties
achieve more consistent
peaks throughout,
while all opposition
parties other than WP
actually peaked before Search
election day. This would
suggest that offline
events like speeches
and news frequently
influence search but
only particpatory events
and gatherings have
a strong influence on
online conversation.
Party by party Buzz Search Peaks
Aside from election weekend, the time at which PAP
each party saw the closest correlation between
online conversation and search volume was within
a day of a major rally. In addition to generating
4 May
news headlines and awareness, major rallies
resulted in firsthand content from attendees in the
form of blogs, comments, photos and videos.
WP NSP
29 April 30 April
SDP SPP
4 May 2 May
5
7. CHANNELS
Throughout the campaign conversation was generated through blogs, news sites, comments, forums and
social media, but in election week itself the overwhelming majority of online mentions of the contesting
parties were made through microblogging, or more specifically Twitter. With the exception of the SDP all
parties were mentioned at least three times more frequently in microblogs than regular blog posts. The easy
and succinct nature of platforms like Twitter was the choice of an engaged electorate who wished to voice
their opinions instantly. This was particularly relevant on election day as netizens awaited and then shared
results from different constituencies.
Week 4 conversations by channel (%)
1. Microblogs 2. News 3. Blogs 4. Comments
1 74 3 16 1 70 3 12
2 5 4 5 2 6 4 12
Microblog sentiment (normalised)
1 59 3 24 1 64 3 18
2 16 4 1 2 18 4 0
Microblog sentiment in election week saw a big upsurge
in positive mentions of parties, when compared to total
76 7 online sentiment at any other stage of the campaign. The
1 3
SPP was the exception and was mentioned negatively
2 17 4 0 more frequently than positively, a trend reflected in a
drop in votes year-on-year, while the WP was the only
party mentioned positively more often than neutrally.
Negative mentions of the PAP were only slightly fewer
than positive mentions. Microblogging sentiment in
election week bears the closest reflection to actual
voting patterns, with proportion of positive mentions of
opposition parties comparing to actual votes.
6
8. INFLUENCERS
The frequency with which a site mentioned a party is
demonstrated by the size of it’s name. Sizes are dettermined in
relevance to the most frequent poster, or largest name.
Sites which mention PAP most frequently In all cases, parties were
mentioned most frequently
by the online portals of
traditional media outlets.
Only the two parties
who ultimately entered
parliament received similar
Sites which mention WP most frequently levels of conversation
outside of news sites,
accurately demonstrating
their popularity with the
masses.
The overall majority of
Sites which mention NSP most frequently mentions did not come
from traditional media
sites, rather from regular
bloggers and numerous
microbloggers. Thus it is
illustrated that established
news sites are instrumental
Sites which mention SDP most frequently in providing frequent
and reliable information
and ultimately in starting
conversations.
Non-news site
influencers
Sites which mention SPP most frequently Excluding news sites, there
are many blogs, forums
and microbloggers which
consistently discussed the
2011GE, highlighting the
importance of engaging
with voters online in future
Non-news sites which most frequently mentioned political parties elections.
7
9. Peter McFeely
Senior digital consultant
+65 6325 4606
peter.mcfeely@grayling.com
Grayling Singapore
6 Shenton Way #12-08A
DBS Building Tower Two
Singapore 068809