On January 20th 2018 a detailed listing of the 56 state seats being contested by Pakatan Harapan (PH) component parties in Johor was published by Malaysiakini (https://www.malaysiakini.com/news/409412). Based on this listing, the seat division is as follows:
- 18 seats contested by PPBM
- 12 seats contested by PKR
- 14 seats contested by DAP
- 12 seats contested by AMANAH
This report provides an overview of the seats contested by component parties in PH and the winning odds for each seat. Detailed election forecast results are not available at the moment due to the cost involved in running simulations, however the winning odds will give a rough idea of the swing in support that PH needs to win control of the state.
Analysis was performed based on the 2017 1st Quarter (Q1) electoral roll, State and Federal seat results from the 13th General Election (GE13) and individual historical voting patterns from GE12 (2008) and GE13 (2013).
A PDF copy of this report can be downloaded at https://drive.google.com/file/d/1zYr2-a3KlYJ_msfEqaeze4FrKrqWOeSC/view?usp=sharing
2. List of Acronyms
The following table shows a list of acronyms used in this document.
|PR||Pakatan Rakyat (coalition of PKR, DAP and PAS)|
|PH||Pakatan Harapan (coalition of PKR, DAP and AMANAH)|
|UMNO||United Malays National Organisation|
|GERAKAN||Parti Gerakan Rakyat Malaysia (also known as PGRM)|
|MCA||Malaysian Chinese Association|
|MIC||Malaysian Indian Congress|
|PBB||Parti Pesaka Bumiputra Bersatu Sarawak|
|PKR||Parti Keadilan Rakyat|
|DAP||Democratic Action Party|
|AMANAH||Parti Amanah Negara|
|PAS||Parti Islam Se-Malaysia|
|PPBM||Parti Pribumi Bersatu Malaysia|
|EC||Election Commission of Malaysia|
|GE12||12th General Election (held in March 2008)|
|GE13||13th General Election (held in May 2013)|
Prior to the 13th General Election (GE13) we came up with a methodology of predicting election results based on voting patterns in previous elections (https://politweet.wordpress.com/2013/05/05/predictions-for-malaysias-general-election-ge13-pru13/).
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 between different constituencies.
The impact of voter migration cannot be measured for a single seat just by comparing results of GE12 and GE13 for that seat. An analysis of the whole country needs to be performed. New voter registrations, transferred voters, voters passing away and voters no longer eligible to vote are 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
The most reliable metric is age because voters are separated into polling lanes based on age. The majority of polling lanes contain between 350 – 700 voters.
How probabilities were assigned
- The ethnic composition of each polling lane was calculated. This was used to adjust the probability levels for individual voters.
- The polling lane result was used to generate a probability figure e.g. 60% of votes for BN and 40% of votes for PR means a base probability value of 60% pro-BN and 40% pro-PR is generated for all voters in that polling lane.
- The base probability was then adjusted based on the ethnic composition. This assumes voter turnout was evenly spread among all ethnic groups. For example:
- In a polling lane of 600 voters (assume 100% turnout), BN received 360 votes (60%)
- 420 voters (70%) of the voters are Malay and 180 voters (30%) are Chinese.
- That means BN received votes from at least 180 Malays and at most 360 Malays.
- 180/420 = 42.86% of Malays
- 360/420 = 85.71% of Malays
- BN received votes from 42.86% – 85.71% of Malay voters. This is the range of probability that Malay voters in that lane voted for BN. The average probability is a 64% chance of voting BN.
- For the 180 Chinese voters, BN did not need their vote. BN received votes from 0% – 100% of Chinese voters. This translates to an average probability of 50% chance of voting for BN. This also applies to the probability that they voted PR.
- For PR, they received 240 votes (40%). Based on the ethnic ratio, at least 60 Malays voted for PR.
- 60/420 = 14.29% of Malays
- 240/420 = 57.14% of Malays
- PR received votes from 14.29% – 57.14% of Malays. This translates to an average probability of 35.7% that Malay voters in that lane voted for PR.
- A non-racial method would assign a probability of 60% for BN and 40% for PR. By adjusting for race we identified the higher probability of Malays voting for BN and the higher probability of Chinese voting for PR (50% instead of 40%).
- Any voter belonging to a minority group small enough to not affect the polling lane result either way is also assigned a probability of 50%. For example, if 5% of a polling lane’s voters are of Race X and the rest are Race Y, and the result is 90% in favour of Party A, then voters of Race X are assigned a value of 50%.
- This race-based approach proved more accurate at identifying patterns and predicting results in elections.
- For individuals who previously voted in seats with 3-corner fights between PR component parties, all PR votes were treated as one party. If the result was 60% BN, 20% PAS and 20% DAP then the base probability for voters in that lane would be 60% BN and 20% PR.
- For seats where BN did not contest and it was a contest between PR and Independents, Independent results were treated as pro-BN results.
- Any voter whose level of support cannot be determined is assigned a probability of 50% on voting BN/Opposition and categorised as a fence-sitter.
- Voting preferences and turnout rates for new voters are calculated based on averages from their peers – individuals with similar characteristics (age, race, location) in the electoral roll.
By calculating support and swing levels at the individual level we can address the impact of voter migration between seats; voters who have passed away; and new voters registered to vote since GE13.
Because individual voters have their own support and swing statistics, we can calculate the proportion of the electorate that is leaning towards Opposition (>50% probability of voting Opposition); leaning towards BN (<50% probability of voting Opposition); and on the fence (50% probability of voting Opposition). Voters who are leaning BN might still vote Opposition, so this metric is only an indicator of how good the odds are for the Opposition/BN to win support from the electorate.
To see previous applications of this method please read our previous reports at the following links:
- Election Forecast for Pakatan Harapan in Peninsular Malaysia (https://politweet.wordpress.com/2018/01/16/election-forecast-for-pakatan-harapan-in-peninsular-malaysia-ge14/)
- The Impact of Redelineation on The Selangor State Elections (https://politweet.wordpress.com/2016/11/09/the-impact-of-redelineation-on-the-selangor-state-elections/)
- How Barisan Nasional and Pakatan Rakyat Performed with Voters in Sarawak (GE13) (https://politweet.wordpress.com/2016/04/16/how-barisan-nasional-and-pakatan-rakyat-performed-with-voters-in-sarawak-ge13/)
- Analysing Pakatan Rakyat’s Performance with Malay Voters in Peninsular Malaysia (GE13) (https://politweet.wordpress.com/2015/07/27/analysing-pakatan-rakyats-performance-with-malay-voters-in-peninsular-malaysia-ge13/)
- Urban Development Categorisation of Parliament Seats in Malaysia (https://politweet.wordpress.com/2013/05/30/urban-development-categorisation-of-parliament-seats-in-malaysia/)
How Our Method Applies to State Elections
Our probabilities are calculated based on General Election results at the Federal level. They are not accurate at the State level in seats that exhibited split voting (where voters chose different coalitions at Federal and State level). Split voting was not significant enough to have overturned the winning majorities at the Federal level.
4. Overview of Johor Election Statistics
Total seats: 56
Total voters: 1,774,055
Net increase in voters: 168,744
New voters (since GE13): 235,733
Removed voters (since GE13): 66,989
Transferred-in voters (who voted in other states in GE13): 17,913
The division of voters by ethnic group is shown in the chart below.
Division of Seats Won by Each Party in GE13
In GE13 PR won 89 seats while BN won 133 seats. In Peninsular Malaysia PR won 80 seats while BN won 85 seats.
The breakdown of seats won by urban development category is shown in the table below:
|Urban Development Category / Party||DAP||PKR||PAS||BN||Total|
The division of voters by estimated ethnic group is shown in the chart below.
5. Overview of GE14 Contested Seats
The number of seats contested by each component party of PH is listed below.
The breakdown of seats contested by ethnic majority is shown in the table below:
The breakdown of voters in contested seats by ethnic group is shown in the table below:
|Malay / Muslim Bumiputera||343,298||208,990||197,016||223,714|
The composition of voters in contested seats by ethnic group on a party basis is shown in the following charts.
6. Political Facebook Interest in Johor
The table below shows the number of Facebook users interested in major political parties in Johor, aged 21 years and above. This is calculated based on their profile information, Page likes, posts, shares and other activity on Facebook. The number of users is considered to be the ‘audience’ or user-base for each party/politician. For comparison we have included Tun Mahathir and PM Najib. One common pattern is that DAP, PKR and PAS all have a user-base that is >70% male.
|Party/Politician||Total||Male||Female||Male (%)||Female (%)|
The following charts show how interest in UMNO, PKR and PAS is divided among users aged 21 years and above residing in Johor. This allows us to measure the size of overlapping audiences between parties, for every combination. Exclusive interest in a party or combination of parties means those users do not show exclusive interest in the other parties and combinations of parties.
We calculated these figures to get a sense of how interest is divided between parties with the best appeal to the Malay electorate. From these charts we can observe that PAS has little influence in Johor on Facebook and UMNO dominates both in terms of conversation and Page likes. Women also have very little interest in PKR and PAS.
Facebook currently does not make statistics for PPBM or AMANAH available. This is usually an indicator that interest is too low. However we can test for Tun Mahathir’s reach for users in Johor.
If we combine interest in Tun Mahathir with PKR and measure the divide, UMNO still retains the largest exclusive share in Johor.
Compared to the previous chart, PAS’ exclusive share is 0% because all of those users were interested in Tun Mahathir.
Out of the 460 thousand users in Johor who are interested in UMNO, PKR on its own can reach 30.4%. PKR and Tun Mahathir combined can reach 45.7%. This still leaves 250 thousand potential Johor voters on Facebook who are only interested in UMNO.
Another issue that should concern both BN and Opposition parties is that only 25.45% of the total 2.2 million potential voters in Johor on Facebook have interest in UMNO, PKR, PAS or Tun Mahathir. Both sides will have issues campaigning on Facebook though UMNO will have the advantage of a larger base of users and a greater proportion of women in its audience.
7. Using Average Support Values to Estimate Winning Odds
For each voter we have calculated the odds of them voting for Opposition or BN as a percentage between 0% – 100%. To simulate an election, we would take all the individual voters for a given constituency and run them through a process to get detailed election results for each seat.
However in this case we are only looking at the basic odds of winning the seat in a straight fight. To do this we calculated the average the average support for BN and the Opposition. This is calculated as the sum of probabilities of individuals voting BN / Opposition for each seat divided by the number of voters. This gives us the ‘Average Odds of Voting Opposition’ that we can use to work out the number of winning seats.
From our experience running simulations at the Federal level:
- Any seat with an average support of 52% and above is a safe seat
- Seats with an average support value of 49% and above is winnable though the margin would be very slim if it is below 51%
- Seats with an average support value below 49% would require a minimum swing in support for PH to win:
- 47% to <49% would require a +2 point swing to PH
- 44% to <47% would require a +5 point swing to PH
- 40% to <44% would require a +10 point swing to PH
Between 2013 and 2017Q1, the seat demographics have changed due to new voter registrations, deleted voters and voters transferred in or out. This has shifted the odds towards or away from the Opposition.
The most significant changes in averaged support levels for the Opposition occurred in the following seats:
- Support has dropped by 2.97 points in N12.BENTAYAN (won by DAP in GE13). DAP is still expected to win in GE14 as the average support value of BENTAYAN is 63%
- Support has dropped by 2.84 points in N3.PEMANIS (won by BN in GE13). PKR will have a challenge winning the seat as the average support value is now 43%
- Support has increased by 1.11 points in N46.PENGKALAN RINTING (won by DAP in GE13). It is still not a safe seat as the support level is now 50.81%
- Support has increased by 1.74 points in N50. BUKIT PERMAI (won by BN in GE13). It remains a challenging seat for PPBM as the average support value is now 41.6%
- Support has increased by 2.67 points in N56. KUKUP (won by BN in GE13). It remains a very challenging seat for AMANAH as the average support value is now 31%
8. Evaluating Winning Scenarios for PH
We used the average support values for each seat to evaluate the odds of PH winning under the following conditions for 4 scenarios:
- Straight fights between BN and a united Opposition
- PAS was not included, any support for PAS is assumed to be ‘pro-Opposition’ support
- No redelineation
- Voter sentiment unchanged since GE13
- Average voter turnout rate based on GE13
- Because we are using averaged support levels, only the best-case scenario results were calculated. This includes seats won with slim majorities
The 4 scenarios tested were:
- Scenario 1: Voter sentiment unchanged since GE13
- Scenario 2: +2 point increase in support for the Opposition i.e. a voter with a 48% probability of voting Opposition would now have a 50% probability of voting Opposition
- Scenario 3: +5 point increase in support for the Opposition
- Scenario 4: +10 point increase in support for the Opposition
29 seats are needed to win control of the state government. The full results are listed in Appendix 2. A summary of the results is in the table below:
|Party||Scenario 1: GE13 Odds||Scenario 2: PH +2pts||Scenario 3: PH +5pts||Scenario 4: PH +10pts|
Division of seats won by party
The table below shows the range of seats won by each party for each scenario.
|Party||Contested Seats||Scenario 1: GE13 Odds||Scenario 2:
|Scenario 3: PH +5pts||Scenario 4: PH +10pts|
(seats won = maximum possible seats won)
In a straight fight against BN, PH can form the state government with a 10-point swing of support leading to a win of 36 seats.
3-corner fights between PH, PAS and BN would most likely benefit BN. The only way for PH to overcome this is for their parties to win over pro-BN supporters.
For example in a seat contested by PAS, PPBM and BN, if 10% of the anti-BN vote went to PAS, PPBM would need to counter that by getting >10% of the pro-BN vote. If PH is able to do this in 3-corner fights then the results of Scenario 4 (a 10-point swing to PH) can be achieved.
Getting Malay voters to switch from BN to PH will be challenging. From our observations on Twitter since 2015, young Malays (aged 21-30 years) in Johor are more likely to express support for UMNO compared to young Malays in other states.
- PPBM and PKR are contesting the most difficult seats to win a straight fight
- The impact of PAS contesting against PH in 3-corner fights needs to be overcome by PH winning votes from pro-BN voters
- Our estimates on the number of seats won are ‘best-case’ scenarios based on a simple formula instead of election simulations. Simulations would be more accurate and reliable.
- PH likely needs to target more than a 10-point swing to win Johor in addition to neutralising the effect of PAS by winning over pro-BN voters
- UMNO has a large share of exclusive interest on Facebook, making it more difficult for PH and Tun Mahathir to spread their message
- UMNO and PM Najib draw more interest from women compared to the Opposition parties and Tun Mahathir
A PDF copy of this report can be downloaded at https://drive.google.com/file/d/1zYr2-a3KlYJ_msfEqaeze4FrKrqWOeSC/view?usp=sharing
10. Appendix 1: Johor State Seat Details
The table below lists the statistics and classification for each state seat (DUN).
|Parliament Seat Code||Parliament||State Seat Code||Dun Name||Voters (2017Q1)||Majority Ethnic Group||Malay (%)||Chinese (%)||Indian (%)||GE13 Winner||GE14 PH Contesting Party||Avg (Odds of Voting for Opposition)|
|147||PARIT SULONG||18||SRI MEDAN||27935||MALAY||90.15||9.28||0.32||BN||PPBM||29.49|
|148||AYER HITAM||19||YONG PENG||24402||CHINESE||33.54||59.59||6.50||DAP||DAP||52.88|
|149||SRI GADING||21||PARIT YAANI||26899||MALAY||55.31||43.30||0.98||PAS||AMANAH||49.57|
|149||SRI GADING||22||PARIT RAJA||22986||MALAY||76.56||21.58||1.57||BN||PPBM||35.30|
|156||KOTA TINGGI||37||JOHOR LAMA||18498||MALAY||74.72||21.22||3.62||BN||PKR||26.92|
|159||PASIR GUDANG||42||JOHOR JAYA||59379||MIXED||43.68||46.68||7.13||DAP||DAP||51.53|
|160||JOHOR BAHRU||44||TANJONG PUTERI||53704||MALAY||63.78||30.17||5.27||BN||PPBM||37.28|
|162||GELANG PATAH||49||NUSA JAYA||65494||MALAY||51.68||35.19||11.90||BN||AMANAH||46.11|
|165||TANJONG PIAI||55||PEKAN NENAS||34122||CHINESE||42.68||55.74||1.15||DAP||DAP||51.75|
11. Appendix 2: Winning Scenario Results
The table below lists the seats won by each party based on the scenarios given. If it is possible for a seat to be won under a scenario, ‘YES’ will be stated in the corresponding column. Please bear in mind that these are best-case estimates, with no indication of the size of the winning majority.
|Parliament Seat Code||Parliament||State Seat Code||DUN||GE13 Winner||GE14 PH Contesting Party||Avg (Odds of Voting for Opposition||S1: GE13 Odds||S2: PR +2pt Swing||S3: PR +5pt Swing||S4: PR +10pt Swing|
|147||PARIT SULONG||18||SRI MEDAN||BN||PPBM||29.49|
|148||AYER HITAM||19||YONG PENG||DAP||DAP||52.88||YES||YES||YES||YES|
|149||SRI GADING||21||PARIT YAANI||PAS||AMANAH||49.57||YES||YES||YES||YES|
|149||SRI GADING||22||PARIT RAJA||BN||PPBM||35.30|
|156||KOTA TINGGI||37||JOHOR LAMA||BN||PKR||26.92|
|159||PASIR GUDANG||42||JOHOR JAYA||DAP||DAP||51.53||YES||YES||YES||YES|
|160||JOHOR BAHRU||44||TANJONG PUTERI||BN||PPBM||37.28|
|162||GELANG PATAH||49||NUSA JAYA||BN||AMANAH||46.11||YES||YES|
|165||TANJONG PIAI||55||PEKAN NENAS||DAP||DAP||51.75||YES||YES||YES||YES|