Politweet.Org

Observing Malaysian Social Media

Evaluating Voter Support in Kuala Kangsar and Sungai Besar Using General Election Results and Twitter

1. Background

Prior to the 13th General Election (GE13) we came up with a methodology of predicting election results based on voting patterns in previous elections.

Our method relied on mapping polling lane results to individual voters. This process assigned probability values (chance of turnout; chance of voting for each coalition) to the voter that was not affected if they migrated to another constituency. This is important because between GE12 and GE13 527,849 voters migrated to different constituencies.

The impact of voter migration cannot be measured for a single seat by comparing the results of GE12 and GE13 for that seat. An analysis of the whole country needs to be performed. New voter registrations, voters passing away and voters no longer eligible to vote are other factors that require deep analysis.

After GE13 we were able to apply the same estimation method to voters based on GE13 results. By comparing the shift in probabilities we are able to calculate the swing in support for each coalition. Because we base our calculations on individual voters, we are able to calculate shifts in support based on combinations of the following dimensions:

  • By Age
  • By Race
  • By Gender
  • By Urban Development Category (rural / semi-urban / urban)
  • By Parliament/State Assembly Seat
  • By Polling District
  • By Locality
  • By Seats Won by Specific Parties

Any voter whose level of support cannot be determined is assigned a probability of 50% and categorised as a fence-sitter. The most reliable metric is age because voters are separated into polling lanes based on age. Additionally we have also categorised the 222 Parliament constituencies as rural, semi-urban or urban based on satellite imagery. The descriptions of each category are:

Rural = villages (kampungs) / small towns / farmland distributed within the seat. Rural seats tend to be physically large with a low population.

Semi-urban = larger towns and/or numerous small towns, may include villages as well

Urban = cities where a majority of the seat is covered by some form of urban development

For this report we will focus on how Pakatan Rakyat (PR) and Barisan Nasional (BN) performed with regular voters (pengundi biasa) in P67.Kuala Kangsar and P93.Sungai Besar. This will give a sense of what to expect during the by-elections to be held on June 18th 2016.

In addition to this we will also briefly examine political interest from Twitter users based in these constituencies. This may identify patterns that can be linked to urban youth in these areas.

Postal and early voters are not part of this analysis. They need to be analysed separately due to their different voting process and difficulties in campaigning to both groups.

Please remember that unless otherwise stated, all statistics in this analysis refer to regular voters only. We do not have access to the electoral roll being used for these by-elections and will be relying on estimated figures from the electoral roll for 2015 Q4 (4th quarter).

2. Seat Demographics

Demographics for Sungai Besar and Kuala Kangsar are listed below.

Detail / Seat P93. Sungai Besar P67. Kuala Kangsar
State

 

Selangor

 

Perak

 

Voters (GE13)

 

 

42,923

(2.09% of Selangor voters)

 

 

33,607

(2.38% of Perak voters)

 

Urban Development Category

 

Rural

 

Semi-urban

 

Majority Race

 

Malay

 

Malay

 

Contesting Parties (GE13)

 

UMNO, PAS

 

UMNO, PAS, Independent

 

Winner (GE13)

 

UMNO

 

UMNO

 

 

Twitter Users

 

 

 

1,049

(0.66% of Selangor users)

89% primarily use Bahasa Malaysia

 

660

(2.39% of Perak users)

81% primarily use Bahasa Malaysia

 

The following charts show the estimated ethnic divide among voters in both seats based on our estimated electoral roll for 2015 Q4. This covers all voters (postal, early and regular).

sgbesar_ethnicpie

kkangsar_ethnicpie

Changes in Sungai Besar since GE13 (up to 2015 Q4):

  • Malay voters increased by 0.33 percentage points
  • Chinese voters decreased by 0.4 percentage points
  • Indian voters increased by 0.06 percentage points
  • 1,260 voters removed
  • 1,394 new voters
  • 171 voters migrated in from other constituencies

 

Changes in Kuala Kangsar since GE13 (up to 2015 Q4):

  • Malay voters increased by 0.46 percentage points
  • Chinese voters decreased by 0.45 percentage points
  • Indian voters decreased by 0.015 percentage points
  • 1,153 voters removed
  • 1,079 new voters
  • 185 voters migrated in from other constituencies

Both seats have had an increase in the percentage of Malay voters, and a decrease in the percentage of Chinese voters.

3. Measuring Regular Voter Support by Age

During the election voters are grouped into polling lanes based on age. The results of each lane are reflective of support by the age group. By mapping registered voters to polling lane results we can estimate the support by race and age.

Sungai Besar Support in Polling Lanes

The bubble chart below shows the average probability of Malay voters in a specific polling lane voting for PR in Sungai Besar:

saluran_scatter_malay_sgbesar

Each point in the graph represents one polling lane. The horizontal scale shows the percentage of Malay voters in the lane. The vertical scale shows the percentage probability of Malay voters in that lane of voting for PR. Fence-sitters are indicated along the 50% horizontal line. The size of the circle is based on the number of Malay voters registered to vote in that lane.

Any point above the 50% probability line is good for PR. From this chart we can tell that PR did well in 11 polling lanes where the percentage of Malay voters is between 47% – 99%. PR had lower than 50% support in 45 lanes where the percentage of Malay voters is between 90% – 100%.

For comparison, here is the bubble chart for Chinese voters:

saluran_scatter_cina_sgbesar

There is a trend of increasing support for PR as the percentage of Chinese voters increase.

Kuala Kangsar Support in Polling Lanes

The bubble chart below shows the average probability of Malay voters in a specific polling lane voting for PR in Kuala Kangsar:

saluran_scatter_malay_kkangsar

Like Sungai Besar, polling lanes with a high percentage of Malay voters have low support. However even mixed polling lanes where >50% of the voters are Malay also have low support for PR. This means that PR (PAS) performed worse with Malay voters in Kuala Kangsar compared to Sungai Besar.

Here is the bubble chart for Chinese voters:

saluran_scatter_cina_kkangsar

There are less polling lanes with a high percentage of Chinese voters compared to Sungai Besar. There is still a trend of increasing support for PR as the percentage of Chinese voters increase.

Sungai Besar Swing in Support

The following graph shows how average support has shifted by age and ethnic group in Sungai Besar. This was calculated by taking the average of individual support values from voters across all seats for each age group. Because they represent the majority of the electorate, only Malay and Chinese ethnic groups are shown.

swing_age_race_sgbesar

Some key points from this graph are:

  • There was a positive swing to PR from voters of all ages
  • Chinese voters had a much bigger swing in support, but the swing was smaller from Chinese voters aged 47 – 61 years.
  • Malay voters had a positive shift mainly from ages 36 and below, up to 6.8%. This is indicative of a Malay youth swing

This graph shows the average level of support from Malay and Chinese voters in Sungai Besar after the swing. Anything close to or above 50% is good for PR.

support_age_race_sgbesar

Key points from this graph are:

  • Support from Chinese voters is close to or above 60%, with support reaching 70% for voters below 42 years old
  • Support from Malay voters is below 50%, with the highest support from voters aged 34 years and below.

Kuala Kangsar Swing in Support

The following graph shows how average support has shifted by age and ethnic group in Kuala Kangsar.

swing_age_race_kkangsar

Some key points from this graph are:

  • There was a positive swing to PR from Chinese voters of all ages
  • Positive swing was much higher from Chinese voters aged 68 years and above. The higher swing is an indicator of lower support in GE12 from older Chinese voters.
  • There was a negative swing away from PR from Malay voters above 26 years old, with the biggest drop from those aged 44 – 57 years old
  • Positive swing from Malay voters ranged from 0.04% – 0.09%. Malay youth voting patterns in Kuala Kangsar were in-line with their community, unlike Sungai Besar

This graph shows the average level of support from Malay and Chinese voters in Kuala Kangsar after the swing. Anything close to or above 50% is good for PR.

support_age_race_kkangsar

Key points from this graph are:

  • Support from Chinese voters is close to or above 60%
  • Support from Malay voters is below 50%, with a downward trend starting at 47.5% from 22 year-olds and dropping to 34% and below for voters aged 62 years and above.

 

To summarise key findings from this section:

  • Chinese voter support and swing to PR (PAS) was good in both seats in GE13. However support from Malay voters was below 50% in both seats, particularly Kuala Kangsar.
  • The negative swing in Kuala Kangsar suggests that Malay voters will be harder for PAS to win over in the by-election
  • There is some evidence to suggest increasing support from Malay youth in Sungai Besar. This may serve to PAS’ advantage in the by-election.

 

This section describes voter support by age, taken as an average of support across voters within each age group. These graphs do not show the number of voters – a large negative swing from a small group of voters can be easily offset by a small positive swing from a large group of voters.

These graphs only serve as an indicator of where average support levels are at and what direction they are heading in.

Because averages of positive and negative swing values can results in a net positive or negative, we need to look at individual voter support levels to get a more accurate reading.

4. Measuring Overall Regular Voter Support

Because individual voters have their own support and swing statistics, we can calculate the proportion of the electorate that is leaning towards PR (>50% probability of voting PR); leaning towards BN (<50% probability of voting PR); and on the fence (50% probability of voting PR).

Voters who are leaning BN might still vote for PR, so this metric is only an indicator of how good the odds are for PR/BN to win support from the electorate.

The two charts below show the predicted and estimated support from regular voters in Sungai Besar. By comparing the two we can see how support has shifted.

sgbesar_ge12pie

 

sgbesar_ge13pie

Prior to GE13, BN had a clear lead with 68% of the voters leaning towards BN. After GE13 PR increased their share of support by 17% while BN lost 17%. The proportion of fence-sitters remains the same at 14%.

The two charts below show the predicted and estimated support from regular voters in Kuala Kangsar.

kkangsar_ge12pie

 

KKangsar_ge13pie

Prior to GE13, BN had a slim lead with 52% of the voters leaning towards BN. After GE13 BN increased their share of support by 6% while PR gained 2%. The proportion of fence-sitters has reduced by 8%.

As far as regular voter support goes, odds look good for PAS in Sungai Besar. Reducing BN’s support from 68% to 51% was an achievement for PAS in GE13. Kuala Kangsar does not look good for PAS as BN increased their share by 6%.

In the previous section we have seen the disparity between Chinese and Malay voter support, so to get a more accurate reading we need to examine both ethnic groups separately.

5. Measuring Malay Voter Support

The charts below show voter support from the Malay electorate in Sungai Besar:

sgbesar_ge12pie_mly

 

sgbesar_ge13pie_mly

After GE13 PR gained 8% while BN lost 10% share of support. The proportion of fence-sitters grew by 2%. Despite the loss, BN still maintains a large share of support with 75% of the voters leaning towards BN. Malay voters in Sungai Besar were largely pro-BN before and after GE13.

The charts below show voter support from the Malay electorate in Kuala Kangsar:

kkangsar_ge12pie_mly

 

kkangsar_ge13pie_mly

After GE13 PR lost 16% while BN gained 23% share of support. The proportion of fence-sitters reduced by 7%. This was a big improvement for BN, and the drop in fence-sitters also means that it is much harder for PR to increase their share of support.

6. Measuring Chinese Voter Support

The charts below show voter support from the Chinese electorate in Sungai Besar:

sgbesar_ge12pie_cina

 

sgbesar_ge13pie_cina

After GE13 PR increased their share of support from 43% to 81% of the voters. Support for BN dropped from 36% of the voters to only 2%. The proportion of fence-sitters reduced by4%.

The charts below show voter support from the Chinese electorate in Kuala Kangsar:

kkangsar_ge12pie_cina

 

kkangsar_ge13pie_cina

After GE13 PR increased their share of support from 17% to 70% of the voters. Support for BN dropped from 45% of the voters to only 2%. The proportion of fence-sitters reduced by10%. This was a dramatic shift in support from the Chinese electorate for PAS.

7. Predicted Support from Voters

Based on estimated figures from the 2015 Q4 roll, this is the predicted support from regular voters in both seats. The number of postal voters is too small to significantly affect the expected support levels.

sgbesar_2016pie

 

kkangsar_2016pie

BN is expected to win in both seats because:

  • Multi-corner contests will divide the vote
  • The Chinese electorate was the main factor behind PR’s performance in GE13 and it is not clear whether they will support PAS or AMANAH.
  • Malay voters showed a small positive swing to PR in Sungai Besar, but moved away from PR in Kuala Kangsar

8. Measuring Interest from Twitter Users

Since October 2014 we have been profiling users from Malaysia based on their tweets. Over time we compiled a database of over 630,000 users in Malaysia. This enabled us to do opinion-based analyses on popular issues using tweets from users, weighted by the state user populations.

Based on frequency of visits to a constituency, we can identify users who are possibly based in those areas. Our method is not 100% accurate as it is based on profiling, but in practice it has served as accurate enough for understanding user opinion at the state-level.

Because Twitter is a form of social media primarily used by urban youth, our expectation is that the number of users in both constituencies will be small. They may not be representative of the local youth. However it will give an idea of what users on Twitter in these areas like to talk about.

8.1 Location of users

This map shows the concentration of Twitter users in Sungai Besar (outlined in red). Users are represented as shaded areas ranging from green (few) to red (many). Each user shown is a unique user in the country.

SungaiBesar_Tweeple

There are 1,049 users in this constituency. Users can be found mainly in Sekinchan, Sungai Besar town, and residential areas along the coast line. 89% of users use Bahasa Malaysia as a primary language.

This map shows the concentration of Twitter users in Kuala Kangsar (outlined in red).

KualaKangsar_Tweeple

There are 660 users in this constituency, concentrated almost entirely in Kuala Kangsar town. 81% of users use Bahasa Malaysia as a primary language.

8.2 Measuring Political Interest

We have Twitter data going back to 2010, but for the purposes of this analysis we will focus on measuring interest for specific topics from 2015 – April 2016. By calculating the number of tweets each user has tweeted about a topic, we can measure their level of interest. The topics of discussion are listed below:

Topic Period Description
#BantahGST 2015

 

2015

 

A hashtag campaign opposing the GST

 

#BantahTPPA Jan-Feb 2016

 

Jan – Feb 2016

 

A hashtag campaign opposing the TPPA

 

#Bersih4 2015

 

 

2015

 

 

A hashtag campaign promoting the Bersih rally and the cause, one of the objectives being for PM Najib to step down

 

#DearNajib 2015

 

2015

 

A hashtag used to mock the PM, not specifically a goal-driven campaign

 

#KitaLawan 2015

 

2015

 

A hashtag campaign to protest Anwar Ibrahim’s jail sentence

 

#MansuhAktaHasutan 2015

 

2015

 

A hashtag campaign to abolish the Sedition Act

 

#NajibLetakJawatan 2015

 

2015

 

A hashtag campaign calling for PM Najib’s resignation

 

#Nothing2Hide 2015

 

2015

 

A hashtag campaign to promote a debate between PM Najib and Tun Mahathir, which subsequently turned into a hashtag used to mock the PM for not showing up

 

#RakyatHakimNegara 2015

 

2015

 

A hashtag campaign calling for Anwar Ibrahim’s guilty verdict to be overturned

 

#SaveMalaysia 2016

 

 

Mac 2 – May 7 2016

 

 

A hashtag campaign used to promote the ‘Save Malaysia’ movement and the ‘Citizen’s Declaration’. One of the campaign goals is for PM Najib to step down

 

#TangkapNajib 2015

 

2015

 

A hashtag campaign calling for PM Najib to be arrested

 

#UndurNajib 2015

 

2015

 

A hashtag campaign calling for PM Najib to step down

 

1MDB 2015

 

2015

 

Mentions of 1MDB

 

1MDB Jan-Apr 2016

 

Jan – Apr 2016

 

Mentions of 1MDB

 

BN YB 2015

 

2015

 

Mentions of BN politicians’ @usernames on Twitter

 

DAP YB 2015

 

2015

 

Mentions of DAP politicians’ @usernames on Twitter

 

Hudud 2015

 

2015

 

Mentions of Hudud (the Islamic penal code)

 

Hudud Jan-Apr 2016

 

Jan – Apr 2016

 

Mentions of Hudud (the Islamic penal code)

 

Islamic Phrases

(Socio-Political) 2015

 

 

 

 

 

 

2015

 

 

 

 

 

 

 

Mentions of words/phrases related to Islam, e.g. ‘Bismillah’; ‘Alhamdulillah’; ‘Allah’; ‘aurat’;’Quran’;’Al-Fatihah’. This is a general topic defined by a limited set of keywords drawn from political and social issue tweets. This does not include Hudud.

 

This topic is useful for comparison to see if users in one group are more/less interested in Islamic phrases within a political/social context compared to the national interest levels.

 

Mahathir 2015

 

2015

 

Mentions of Tun Dr Mahathir

 

Mahathir Jan-Apr 2016

 

Jan – Apr 2016

 

Mentions of Tun Dr Mahathir

 

Najib 2015

 

2015

 

Mentions of PM Najib Razak (including @NajibRazak)

 

Najib Jan-Apr 2016

 

Jan – Apr 2016

 

Mentions of PM Najib Razak (including @NajibRazak)

 

Opposition YB 2015

 

2015

 

Mentions of Opposition politicians’ @usernames on Twitter. This includes DAP, PKR, PAS, AMANAH and PSM

 

PKR YB 2015

 

2015

 

Mentions of DAP politicians’ @usernames on Twitter

 

UMNO 2015

 

2015

 

Mentions of UMNO on Twitter

 

Zakir Naik Apr 2016

 

Apr 10 – 25 2016

 

Mentions of Zakir Naik on Twitter, during the period when he was in Malaysia and the week after

 

Anti-Najib Campaigns 2015

 

 

 

 

 

2015

 

 

 

 

 

Mentions of hashtag campaigns considered to be critical of PM Najib Razak:

·         #UndurNajib 2015

·         #NajibLetakJawatan 2015

·         #TangkapNajib 2015

·         #DearNajib 2015

·         #Nothing2Hide 2015

 

 

For each topic, we are going to list the number of users in each constituency and compare it with the state-level and Peninsular Malaysia-level of interest.

By comparing the percentage of users in a constituency interested in a topic with the percentage of users in the state, we can see if they have higher or lower interest levels.

The table below lists the topics ranked by the percentage of users in Peninsular Malaysia expressing interest in each topic by tweeting / re-tweeting.

Rank Topic (% of Peninsular Msia)
1 Najib 2015 34.57
2 BN YB 2015 31.42
3 Najib Jan-Apr 2016 24.53
4 Islam Phrases (Socio-Political) 2015 21.27
5 UMNO 2015 17.56
6 1MDB 2015 15.94
7 Mahathir 2015 14.08
8 Anti-Najib Campaigns 2015 12.93
9 Opposition YB 2015 12.25
10 #Bersih4 2015 12.15
11 Hudud 2015 11.89
12 PKR YB 2015 10.44
13 Mahathir Jan-Apr 2016 9.87
14 Zakir Naik Apr 2016 9.30
15 #KitaLawan 2015 6.27
16 #TangkapNajib 2015 5.63
17 #Nothing2Hide 2015 5.50
18 1MDB Jan-Apr 2016 5.46
19 #SaveMalaysia 2016 4.99
20 #DearNajib 2015 4.12
21 #BantahGST 2015 3.17
22 DAP YB 2015 3.11
23 #UndurNajib 2015 2.79
24 #NajibLetakJawatan 2015 1.76
25 #RakyatHakimNegara 2015 1.65
26 #BantahTPPA Jan-Feb 2016 1.24
27 Hudud Jan-Apr 2016 0.89
28 #MansuhAktaHasutan 2015 0.43

 

The table below lists the topics ranked by the percentage of users in Selangor expressing interest in each topic by tweeting / re-tweeting.

Selangor

Rank

Peninsular Rank Topic (% of Selangor)
1 1 Najib 2015 36.02
2 2 BN YB 2015 33.12
3 3 Najib Jan-Apr 2016 24.78
4 4 Islam Phrases (Socio-Political) 2015 20.89
5 5 UMNO 2015 17.86
6 6 1MDB 2015 16.65
7 7 Mahathir 2015 14.21
8 10 #Bersih4 2015 14.11
9 9 Opposition YB 2015 13.87
10 8 Anti-Najib Campaigns 2015 13.62
11 12 PKR YB 2015 11.92
12 11 Hudud 2015 11.64
13 13 Mahathir Jan-Apr 2016 9.72
14 14 Zakir Naik Apr 2016 8.59
15 15 #KitaLawan 2015 6.60
16 17 #Nothing2Hide 2015 5.82
17 16 #TangkapNajib 2015 5.76
18 18 1MDB Jan-Apr 2016 5.58
19 19 #SaveMalaysia 2016 5.24
20 20 #DearNajib 2015 4.42
21 22 DAP YB 2015 3.73
22 21 #BantahGST 2015 3.31
23 23 #UndurNajib 2015 2.99
24 25 #RakyatHakimNegara 2015 1.82
25 24 #NajibLetakJawatan 2015 1.79
26 26 #BantahTPPA Jan-Feb 2016 1.30
27 27 Hudud Jan-Apr 2016 0.84
28 28 #MansuhAktaHasutan 2015 0.48

 

The table below lists the topics ranked by the percentage of users in Perak expressing interest in each topic by tweeting / re-tweeting.

Perak

Rank

Peninsular Rank Topic (% of Perak)
1 1 Najib 2015 39.31
2 2 BN YB 2015 34.74
3 3 Najib Jan-Apr 2016 29.31
4 4 Islam Phrases (Socio-Political) 2015 26.17
5 5 UMNO 2015 20.98
6 6 1MDB 2015 18.24
7 7 Mahathir 2015 16.36
8 8 Anti-Najib Campaigns 2015 15.05
9 11 Hudud 2015 14.50
10 9 Opposition YB 2015 13.10
11 13 Mahathir Jan-Apr 2016 12.53
12 10 #Bersih4 2015 12.38
13 14 Zakir Naik Apr 2016 11.97
14 12 PKR YB 2015 11.24
15 15 #KitaLawan 2015 7.19
16 16 #TangkapNajib 2015 6.83
17 18 1MDB Jan-Apr 2016 6.41
18 17 #Nothing2Hide 2015 6.22
19 19 #SaveMalaysia 2016 5.88
20 20 #DearNajib 2015 4.71
21 21 #BantahGST 2015 3.80
22 23 #UndurNajib 2015 3.14
23 22 DAP YB 2015 2.83
24 24 #NajibLetakJawatan 2015 2.23
25 25 #RakyatHakimNegara 2015 1.89
26 26 #BantahTPPA Jan-Feb 2016 1.38
27 27 Hudud Jan-Apr 2016 1.23
28 28 #MansuhAktaHasutan 2015 0.46

 

Compared to Peninsular Malaysia, Perak users show greater interest in the topics listed. The only exception is mentions of DAP politicians where 2.83% of Perak users showed interest compared to 3.11% of Peninsular Malaysia users.

The topics where Perak users showed a significantly higher interest compared to the Peninsular Malaysia level are listed below:

  1. Anti-Najib Campaigns 2015 (+2.12 points)
  2. Mahathir 2015 (+2.28 points)
  3. 1MDB 2015 (+2.3 points)
  4. Hudud 2015 (+2.61 points)
  5. Mahathir Jan-Apr 2016 (+2.65 points)
  6. Zakir Naik Apr 2016 (+2.66 points)
  7. BN YB 2015 (+3.32 points)
  8. UMNO 2015 (+3.41 points)
  9. Najib Jan-Apr 2016 (+4.07 points)
  10. Najib 2015 (+4.74 points)
  11. Islamic Phrases (Socio-Political) 2015 (+4.9 points)

Perak users also show higher interest in all topics compared to Selangor, except the following:

  1. #Bersih4 2015 (-1.73 points)
  2. #MansuhAktaHasutan 2015 (-0.03 points)
  3. DAP YB 2015 (-0.90 points)
  4. Opposition YB 2015 (-0.77 points)
  5. PKR YB 2015 (-0.68 points)

This is indicative of the stronger appeal these topics have for users in Selangor.

Other observations from comparing interest levels by state with Peninsular Malaysia:

  • Hudud ranked higher in Perak compared to Selangor and Peninsular Malaysia.
  • Bersih 4 ranked lower in Perak
  • Bersih 4 ranked higher in Selangor
  • PKR and DAP politicians ranked higher in Selangor
  • PKR and DAP politicians ranked lower in Perak
  • Zakir Naik ranked higher in Perak
  • Anti-Najib Campaigns ranked lower in Selangor

The following topics showed a decline in interest when comparing 2015 levels to Jan – Apr 2016 levels within Peninsular Malaysia:

  1. Najib (-10.04 points)
  2. Mahathir (-4.21 points)
  3. 1MDB (-10.48 points)

A similar drop in interest can be observed in Selangor and Perak, which indicates that users are starting to lose interest in these 3 issues, despite media coverage each month. Hudud did not resurface as a current issue until May 2016, so comparing the drop in interest would be misleading.

Another interesting fact is that while Anti-Najib Campaigns ranked in the Top 10 of each list, the individual campaigns drew less than 10% of the population’s interest. They did not have dedicated interest over the long-term. The most recent anti-Najib campaign, #SaveMalaysia only saw the participation of 4.99% of Peninsular Malaysia users.

Most of the 12.15% of the population that showed interest in #Bersih4 (which included a goal of Najib stepping down) did not show interest in #SaveMalaysia in 2016. The same is true of the population that showed interest in Anti-Najib Campaigns in 2015.

It is important to note that interest in a topic does not indicate partisanship. It only means those users have demonstrated enough interest to tweet/re-tweet content related to the topic. In the case of hashtag campaigns this may include tweets that are for or against the cause.

8.3 Political Interest of Sungai Besar Users

Topic Users (%) Peninsular (%) Diff from Pen. Selangor (%) Diff from SGR
#BantahGST 2015 55 5.24 3.17 2.07 3.31 1.93
#BantahTPPA JanFeb 2016 19 1.81 1.24 0.57 1.30 0.51
#Bersih4 2015 133 12.68 12.15 0.53 14.11 -1.43
#DearNajib 2015 60 5.72 4.12 1.60 4.42 1.30
#KitaLawan 2015 82 7.82 6.27 1.55 6.60 1.22
#MansuhAktaHasutan 2015 6 0.57 0.43 0.15 0.48 0.09
#NajibLetakJawatan 2015 16 1.53 1.76 -0.23 1.79 -0.26
#Nothing2Hide 2015 74 7.05 5.50 1.55 5.82 1.24
#RakyatHakimNegara 2015 16 1.53 1.65 -0.13 1.82 -0.29
#SaveMalaysia 2016 57 5.43 4.99 0.45 5.24 0.20
#TangkapNajib 2015 75 7.15 5.63 1.52 5.76 1.39
#UndurNajib 2015 47 4.48 2.79 1.69 2.99 1.49
1MDB 2015 188 17.92 15.94 1.98 16.65 1.28
1MDB Jan Apr 2016 89 8.48 5.46 3.03 5.58 2.90
BN YB 2015 362 34.51 31.42 3.09 33.12 1.39
DAP YB 2015 28 2.67 3.11 -0.44 3.73 -1.06
Hudud 2015 184 17.54 11.89 5.65 11.64 5.90
Hudud Jan Apr 2016 16 1.53 0.89 0.63 0.84 0.69
Islamic Phrases (Socio-Political) 2015 305 29.08 21.27 7.80 20.89 8.18
Mahathir 2015 191 18.21 14.08 4.13 14.21 4.00
Mahathir Jan Apr 2016 114 10.87 9.87 0.99 9.72 1.14
Najib 2015 430 40.99 34.57 6.42 36.02 4.97
Najib Jan Apr 2016 315 30.03 24.53 5.50 24.78 5.25
Opposition YB 2015 167 15.92 12.25 3.67 13.87 2.05
PKR YB 2015 154 14.68 10.44 4.24 11.92 2.76
UMNO 2015 247 23.55 17.56 5.98 17.86 5.69
Zakir Naik Apr 2016 151 14.39 9.30 5.09 8.59 5.81
Anti-Najib Campaigns 2015 173 16.49 12.93 3.56 13.62 2.87

 

The ‘Diff from Pen.’ and ‘Diff from SGR’ columns list the difference in percentage points between Sungai Besar user interest and Peninsular / Selangor level user interest. Key differences are highlighted in bold.

Out of the total 1,049 users in Sungai Besar, 625 users (59.6%) show interest in the topics listed. Excluding Islamic Phrases, 598 users (57%) show interest in the topics listed.

Measured by the percentage of users, the most popular topics with users in Sungai Besar are:

  1. Najib 2015
  2. BN YB 2016
  3. Najib Jan Apr 2016
  4. Islamic Phrases (Socio-Political) 2015
  5. UMNO 2015
  6. Mahathir 2015
  7. 1MDB 2015
  8. Hudud 2015
  9. Anti-Najib Campaigns 2015
  10. Opposition YB 2015

Compared to Selangor’s state-level interest, Sungai Besar users showed lower interest in the following topics:

  1. #NajibLetakJawatan 2015 (-0.26 points)
  2. #RakyatHakimNegara 2015 (-0.29 points)
  3. DAP YB 2015 (-1.06 points)
  4. #Bersih4 2015 (-1.43 points)

Compared to Selangor’s state-level interest, Sungai Besar users showed higher interest in other topics. The most significant topics are listed below:

  1. Mahathir 2015 (+4 points)
  2. Najib 2015 (+4.97 points)
  3. Najib Jan-Apr 2016 (+5.25 points)
  4. UMNO 2015 (+5.69 points)
  5. Zakir Naik Apr 2016 (+5.81 points)
  6. Hudud 2015 (+5.9 points)
  7. Islamic Phrases (Socio-Political) 2015 (+8.18 points)

The higher interest in Zakir Naik, Hudud and use of Islamic phrases suggests that Islam-related issues would be more appealing to the youth in Sungai Besar compared to the average Selangor youth.

8.4 Political Interest from Kuala Kangsar Users

Topic Users (%) Peninsular (%) Diff from Pen. Perak (%) Diff from PRK
#BantahGST 2015 25 3.79 3.17 0.61 3.80 -0.01
#BantahTPPA Jan-Feb 2016 8 1.21 1.24 -0.03 1.38 -0.16
#Bersih4 2015 76 11.52 12.15 -0.64 12.38 -0.87
#DearNajib 2015 33 5.00 4.12 0.88 4.71 0.29
#KitaLawan 2015 49 7.42 6.27 1.15 7.19 0.23
#MansuhAktaHasutan 2015 1 0.15 0.43 -0.27 0.46 -0.30
#NajibLetakJawatan 2015 14 2.12 1.76 0.36 2.23 -0.11
#Nothing2Hide 2015 38 5.76 5.50 0.25 6.22 -0.46
#RakyatHakimNegara 2015 14 2.12 1.65 0.47 1.89 0.23
#SaveMalaysia 2016 41 6.21 4.99 1.23 5.88 0.34
#TangkapNajib 2015 38 5.76 5.63 0.13 6.83 -1.07
#UndurNajib 2015 12 1.82 2.79 -0.97 3.14 -1.32
1MDB 2015 109 16.52 15.94 0.58 18.24 -1.72
1MDB Jan-Apr 2016 31 4.70 5.46 -0.76 6.41 -1.71
BN YB 2015 229 34.70 31.42 3.28 34.74 -0.04
DAP YB 2015 17 2.58 3.11 -0.53 2.83 -0.26
Hudud 2015 108 16.36 11.89 4.47 14.50 1.86
Hudud Jan-Apr 2016 11 1.67 0.89 0.78 1.23 0.44
Islamic Phrases (Socio-Political) 2015 194 29.39 21.27 8.12 26.17 3.22
Mahathir 2015 107 16.21 14.08 2.13 16.36 -0.15
Mahathir Jan-Apr 2016 82 12.42 9.87 2.55 12.53 -0.10
Najib 2015 256 38.79 34.57 4.22 39.31 -0.52
Najib Jan-Apr 2016 180 27.27 24.53 2.74 29.31 -2.03
Opposition YB 2015 90 13.64 12.25 1.38 13.10 0.54
PKR YB 2015 80 12.12 10.44 1.69 11.24 0.88
UMNO 2015 147 22.27 17.56 4.71 20.98 1.30
Zakir Naik Apr 2016 92 13.94 9.30 4.64 11.97 1.97
Anti-Najib Campaigns 2015 90 13.64 12.93 0.70 15.05 -1.41

 

The ‘Diff from Pen.’ and ‘Diff from PRK’ columns list the difference in percentage points between Kuala Kangsar user interest and Peninsular / Perak level user interest. Key differences are highlighted in bold.

Out of the total 660 users in Kuala Kangsar, 390 users (59.1%) show interest in the topics listed. Excluding Islamic Phrases, 363 users (55%) show interest in the topics listed.

Measured by the percentage of users, the most popular topics with users in Kuala Kangsar are:

  1. Najib 2015
  2. BN YB 2016
  3. Islamic Phrases (Socio-Political) 2015
  4. Najib Jan Apr 2016
  5. UMNO 2015
  6. 1MDB 2015
  7. Hudud 2015
  8. Mahathir 2015
  9. Zakir Naik Apr 2016
  10. Opposition YB 2015

As we noted earlier, Perak users show more interest in all topics (except DAP YB 2015) compared to Peninsular Malaysia. A higher percentage of Perak users show interest in all the topics when compared to Selangor, apart from #Bersih4, #MansuhAktaHasutan and Opposition politicians.

Compared to Perak’s state-level interest, Kuala Kangsar users showed interest levels very close to the state-level within a margin of +/- 1%. Beyond that margin, Kuala Kangsar users showed lower interest in the following topics:

  1. #TangkapNajib 2015 (-1.07 points)
  2. #UndurNajib 2015 (-1.32 points)
  3. Anti-Najib Campaigns 2015 (-1.41 points)
  4. 1MDB Jan-Apr 2016 (-1.71 points)
  5. 1MDB 2015 (-1.72 points)
  6. Najib Jan-Apr 2016 (-2.03 points)

Compared to Perak’s state-level interest, Kuala Kangsar users showed higher interest levels in the following topics:

  1. UMNO 2015 (+1.30 points)
  2. Hudud 2015 (+1.86 points)
  3. Zakir Naik Apr 2016 (+1.97 points)
  4. Islamic Phrases (Socio-Political) 2015 (+3.22 points)

Like Sungai Besar, Kuala Kangsar users also showed higher interest in Zakir Naik, Hudud and the use of Islamic phrases compared to state-level interest. A similar proportion of users are also interested in UMNO. When compared to Peninsular Malaysia’s interest levels, the percentage of Kuala Kangsar users interested in Islamic Phrases was 8.12 points higher.

The lower interest in 1MDB, PM Najib and Anti-Najib Campaigns may indicate these issues are less popular with the youth in Kuala Kangsar.

8.5 Comparing the Degree of Interest

The network graphs below show how users in Sungai Besar are connected to each topic. Each user is represented as a blue node (circle) that is connected to larger nodes labelled with the topic name.

The strength of the connection is based on the number of tweets by the user – stronger connections will draw the user closer to the topic. Stronger connections are also visualised as thicker lines with larger arrows.

The size of the topic node (and its font) is relative to its popularity – larger nodes represent the most popular topics. ‘Islamic Phrases (Socio-Political) 2015’ is not shown in the graph – this was to make it easier to identify closely-related topics and focus more on political issues.

gephi_SgBesar_TopicInterest_1200px

This is a close-up view of the graph:

gephi_SgBesar_TopicInterest_1200px_crop

Users are clustered based on their connectivity to each topic. For example, users clustered between the BN YB 2015 and the Najib 2015 nodes have the strongest connections to both topics.

Users clustered within the center of this network have a strong interest in political topics, while those on the outer edges of the graph have weak interest in a few topics. Using network graphs we can identify the potential local campaigners for activists and political parties.

Topics that draw similar audiences are also clustered together. For example, ‘#bersih4 2015’, ‘1MDB 2015’ and ‘Mahathir 2015’ are closer because they appeal to a shared audience.

This is the network graph for users in Kuala Kangsar:

gephi_KKangsar_TopicInterest_1200px

There are fewer users in Kuala Kangsar, but the patterns are similar to Sungai Besar.

This is a close-up view of the graph:

gephi_KKangsar_TopicInterest_1200px_cropped

The key findings from this section are:

  • 55% of users in Kuala Kangsar and 57% of users in Sungai Besar have an interest in political issues and Islam-related issues.
  • Hudud, Zakir Naik, Islamic phrases and UMNO draw a greater percentage of users compared to their respective state levels.
  • Sungai Besar users have more interest in Najib, Mahathir, 1MDB and Anti-Najib Campaigns compare to Selangor’s state level
  • Kuala Kangsar users have less interest in Najib, 1MDB and Anti-Najib Campaigns compared to Perak’s state
  • Perak users have a greater interest in political issues compared to the Peninsular Malaysia interest level

 

9. Conclusion

The table below lists the winning margins for both seats in GE12 and GE13:

Seat/Election GE12 (2008) GE13 (2013)
Sungai Besar 5,009 399
Kuala Kangsar 1,458 1,082

 

Looking at these figures, it would appear that PAS succeeded in reducing BN’s share of the vote in Malay-majority seats. However our analysis indicated that Chinese voters in both constituencies swung to PAS by a high percentage, while Malay voters swung away from PAS in Kuala Kangsar. Positive swing from Malay voters in Sungai Besar was mainly concentrated in voters aged 36 and below.

PAS’ improved performance in GE13 was largely due to support from the Chinese community. Now that PAS is not part of Pakatan Harapan it is not clear if the Chinese community will continue to support it. AMANAH may gain the support of the Chinese community instead, but this can be counter-productive.

As both PAS and AMANAH are Islamist parties, they need to appeal to the Malay electorate. If the election results are racially-divided where AMANAH gains votes mainly from Chinese voters while PAS and UMNO gain votes mainly from Malay voters, this can hurt AMANAH’s reputation as it didn’t prove its appeal to the Malay community.

In Sarawak, AMANAH was unsuccessful at drawing a significant number of votes away from PAS. These by-elections are really a test to see how well PAS and AMANAH do competing against each other.

If both parties combined get 50% of the vote that enables BN to win the seats with less than 50% of the vote. The results of these by-elections will serve as a preview of what might happen in GE14 if PAS and AMANAH contest each other nation-wide.

Our analysis of Twitter users indicates that PM Najib Razak, UMNO, Hudud, and Tun Dr. Mahathir are discussion topics that draw the most interest in both constituencies. However interest in 1MDB dropped considerably in 2016, while interest in #SaveMalaysia was lower than Anti-Najib Campaigns in 2015 in Peninsular Malaysia, Selangor, Perak and both seats. Interest in Najib and Mahathir has also declined, more for Najib than Mahathir.

Using Twitter is not as reliable at gauging sentiment compared to a survey. However comparing the interest levels at both the seat-level and state-level has yielded some insights as to the degree of importance each issue has for the youth there.

Interest in Anti-Najib campaigns is higher in Sungai Besar and lower in Kuala Kangsar. Kuala Kangsar users also showed less interest in 1MDB. This implies that Anti-Najib campaigns may find less appeal in Kuala Kangsar.

The above-average interest in Islamic Phrases also indicates that Islamic issues and Hudud have a strong appeal to the Malay youth there, even stronger than the respective state interest levels.

In conclusion, based on the patterns observed:

  • BN should win both seats, with a bigger margin in Kuala Kangsar compared to Sungai Besar. This assumes that turnout rates remain the same as GE13
  • Support from the Chinese electorate is essential for PAS/AMANAH to win each seat
  • Malay voters below 40 may show increased support for PAS/AMANAH in Sungai Besar, based on GE13 trends
  • Malay youth may show increased support for PAS/AMANAH depending on Islamist credentials
  • A racially-divided election result can affect AMANAH’s reputation negatively
  • Declining interest in Najib, 1MDB and Mahathir along with weak interest in #SaveMalaysia indicate that allegations against Najib and 1MDB are not strong campaign issues for the youth

 

Advertisements

Written by politweet

June 11, 2016 at 9:25 am

One Response

Subscribe to comments with RSS.

  1. Appreciate if you would forward me updates. Thank you.

    Ronnie Klassen

    June 16, 2016 at 3:50 pm


Comments are closed.

%d bloggers like this: