Teluk Intan #GE13 Analysis

During the 13th General Election (GE13) we came up with a methodology for estimating individual voter support based on polling lane results from the 12th General Election (GE12). The results of these calculations were used to design a program to predict election results based on ‘swing’ parameters fed in e.g. ethnic support; age group; increasing partisanship etc. You can read a summary of this method at the end of this post.

Using the polling lane results of the 13th General Election we compared our previous calculations and come up with an estimated swing. This can help inform analysts and campaigners for Barisan Nasional (BN) and Pakatan Rakyat (PR).

For the current by-election in P76.Teluk Intan we performed an analysis of the GE13 results and found that:

1. Support for PR has increased by an average of 3.33% since GE12, from 43.85% to 47.18%
2. Support increased from Chinese voters of all ages
3. Support increased from Malay voters below 40 years old
4. Support decreased from Malay voters aged 40 years and above
5. Malay and Indian women voters may be a weak point for PR, particularly Indian women
6. PR is likely to retain the seat this by-election provided that GE13 voting patterns remain the same. The choice of candidates by BN and PR may affect the voting patterns due to differences in ethnicity, age, gender and experience. However we have no data to predict the impact of these differences.

For social media statistics on Teluk Intan please check this post.

1. Seat Statistics

P76.Teluk Intan is a mixed-ethnicity semi-urban parliamentary seat located in Perak. It was won by the Democratic Action Party (DAP) in GE13 with a majority of 7,313 votes.

Based on our GE13 data there were 60,483 voters of which 393 are early (postal) voters and the rest are regular voters. Due to complexities in our method 387 early voters were not analysed and are excluded from the rest of this report. For age ranges we are using the age of voters as of Dec 2013, not Dec 2012.

Out of the remaining 60,096 regular voters:

• 22,987 (38.25%) are ethnic Malays
• 25,352 (42.19%) are ethnic Chinese
• 11,493 (19.12%) are ethnic Indian
• 264 (0.44%) are from other ethnic groups

29,072 (48.38%) of the voters are men, while 31,024 (51.62%) are women.

The chart below illustrates the breakdown of voters by age group. Youth (ages 21-30) make-up 17.17% of the regular voters.

We do not have the electoral roll being used for this by-election. Expect some differences in our statistics compared to the current roll.

Due to the different voting patterns of Malay, Chinese and Indian voters, we will present separate analyses for these 3 groups in addition to the overall analysis. This will help provide a clearer picture for campaigners to use.

2. Analysis of Support (all ethnic groups)

Predicted GE13 Support (based on GE12 results)

Leaning BN: 15,672 (26.08%)

Fence-sitters: 18,070 (30.06%)

Leaning PR: 26,354 (43.85%)

Estimated GE13 Support (based on GE13 results)

Leaning BN: 15,999 (26.62%)

Fence-sitters: 15,741 (26.19%)

Leaning PR: 28,356 (47.18%)

‘Leaning BN’ means these voters have a >50% probability of voting for BN and ‘Leaning PR’ means these voters have a >50% probability of voting for PR. A >50% probability does not mean a guaranteed vote for either side. Fence-sitters are voters whose leaning could not be determined from the results.

By comparing the 2 datasets we can estimate the swing in support by the voters. For this analysis we will focus on the swing to and away from PR.

The following histograms illustrate the distribution of probabilities among voters. The blue bars represent ‘Leaning BN’ voters. The yellow (50%) bar represents the fence-sitters. The red bars represent ‘Leaning PR’ voters.

The number of fence-sitters has decreased, while the number of voters leaning PR and voters leaning BN have increased. Support for PR has moved further to the right extending to the 90% level.

The average support for PR for each voter has increased by an average of 4.9%. For women the increase was 4.6%, while for men the increase was 5.22%.

When we group the voters by their est.swing, there is a clear pattern of movement towards PR. This histogram illustrates the number of voters by estimated swing to/from PR.

10,774 voters (17.93%) experienced a drop in support. 6,778 voters reduced their support by 1% – 10%, meaning that BN-leaning voters swung further towards BN; PR supporters became fence-sitters; and PR supporters remained supporters.

15,328 voters (25.51%) had little to no change. This covers fence-sitters and leaning PR/BN voters who did not move from their current level of support.

33,994 (56.57%) swung to PR. 16,481 voters increased their support by 11% – 22%. Supporters became stronger; fence-sitters moved towards PR; and BN-leaning voters moved towards fence-sitters/PR.

5,307 voters moved from fence-sitters and leaning BN towards leaning PR.
2,299 voters moved from fence-sitters and leaning PR towards leaning BN.
1,807 voters moved from leaning PR or leaning BN towards becoming fence-sitters.

This chart shows the swing by age, calculated by averaging support values for voters by their age group. Average support for PR increased from voters of all ages, particularly ages 63 years and above.

When we combine this graph with the number of regular voters by age, we can see the impact of the swing.

The trend is voters aged 27 and below, and aged 63 years and above swung further to PR.

This chart illustrates the estimated level of support from voters. With PR getting support from only 47% of the electorate, PR faces a challenge to win the seat.

Next we look at the support for Malay, Chinese and Indian voters which will reveal patterns behind the average swing.

3. Analysis of Support (Malay voters)

Total: 22,987 voters

Predicted GE13 Support (based on GE12 results)

Leaning BN: 13,432 (58.43%)

Fence-sitters: 6,982 (30.37%)

Leaning PR: 2,573 (11.19%)

Estimated GE13 Support (based on GE13 results)

Leaning BN: 13,411 (58.34%)

Fence-sitters: 6,656 (28.96%)

Leaning PR: 2,920 (12.70%)

The following histograms illustrate the distribution of probabilities among voters. The blue bars represent ‘Leaning BN’ voters. The yellow bar represents the fence-sitters. The red bars represent ‘Leaning PR’ voters.

Support for PR remains low. Support shifted from being concentrated within a range of 51% – 60% to being concentrated in a range of 55% – 64%. The number of fence-sitters has had a slight decrease. Support for BN has decreased slightly, resulting in an increase in voter concentration in the 40% – 49% range.

The support for PR for each voter has increased by an average of 0.85%. For women the increase was 0.76%, while for men the increase was 0.95%. Malay women voters appear to be PR’s weak point.

When we group the voters by their est.swing, there is a pattern of movement both away and towards PR. This histogram illustrates the number of voters by estimated swing to/from PR.

6,550 voters (28.49%) experienced a drop in support. The majority of the decrease was 5,503 voters reducing by 1% – 8%, meaning that BN-leaning voters swung further towards BN; PR supporters became fence-sitters; and PR supporters remained supporters.

6,675 voters (29.04%) had little to no change. This covers fence-sitters and leaning PR/BN voters who did not move from their current level of support.

9,762 (42.47%) swung to PR. The majority of the swing was 7,091 voters increasing by 1% – 7%. Supporters became stronger; fence-sitters moved towards PR; and BN-leaning voters moved towards fence-sitters/PR.

1,296 voters moved from fence-sitters and leaning BN towards leaning PR.
396 voters moved from fence-sitters and leaning PR towards leaning BN.
1,087 voters moved from leaning PR and leaning BN towards becoming fence-sitters.

This chart shows the swing by age, calculated by averaging support values for voters by their age group. Average support for PR increased from voters aged 22 – 40 years, with some minor increases from voters aged 51 – 53 years.

When we combine this graph with the number of regular voters by age, we can see the impact of the swing.

The trend is voters aged 40 and below swung further to PR, with the swing increasing with each younger generation. The peak was at 6.27% from voters aged 22 years. Malay voters aged 41 and above however swung further away from PR by as much as 4.86% (and 10.18% for 98 year-olds). It would appear that young Malays support PR while middle-aged and older Malays are on the fence or leaning towards BN.

After the swing, PR has >40% support from Malay voters aged 22 – 35 years. 44% of Teluk Intan’s Malay electorate are aged 21-40 years. If the youth swing continues PR may push support past 50%.

This chart illustrates the estimated level of support from voters. Despite the swing away from PR, PR managed to increase support from 11.19% to 12.7% of Malay voters. However the small shift combined with the weak support indicates that BN may win more votes from the middle-aged and older generation Malays.

4. Analysis of Support (Chinese voters)

Total: 25,352 voters

Predicted GE13 Support (based on GE12 results)

Leaning BN: 1,417 (5.59%)

Fence-sitters: 4,105 (16.19%)

Leaning PR: 19,830 (78.22%)

Estimated GE13 Support (based on GE13 results)

Leaning BN: 3 (0.01%)

Fence-sitters: 2,786 (10.99%)

Leaning PR: 22,563 (89.00%)

The following histograms illustrate the distribution of probabilities among voters. The blue bars represent ‘Leaning BN’ voters. The yellow bar represents the fence-sitters. The red bars represent ‘Leaning PR’ voters.

The majority of support for PR is still above 50%, the number of fence-sitters has decreased by one-third and the ‘Leaning BN’ voters have decreased to 3 voters. Support for PR has shifted far to the right, with a large concentration in the 74% – 91% range. This means support for PR from Chinese voters is very strong and BN’s main target should be the fence-sitters and voters with support from 50% – 60%, a total of 5,019 votes.

The support for PR for each voter has increased by an average of 13.27%. For women the increase was 13.09%, while for men the increase was 13.46%.

When we group the voters by their est.swing, there is a strong pattern of movement towards PR. This histogram illustrates the number of voters by estimated swing to/from PR.

1,013 voters (4.00%) experienced a drop in support. 622 voters reduced support by 1% – 4%, meaning that BN-leaning voters swung further towards BN; PR supporters became fence-sitters; and PR supporters remained supporters.

2,386 voters (9.41%) had little to no change. This covers fence-sitters and leaning PR/BN voters who did not move from their current level of support.

21,953 (86.59%) swung to PR. 12,096 voters increased their support by 11% – 19%. Supporters became stronger; fence-sitters moved towards PR; and BN-leaning voters moved towards fence-sitters/PR.

3,066 voters moved from fence-sitters and leaning BN towards leaning PR.
3 voters moved from fence-sitters and leaning PR towards leaning BN.
431 voters moved from leaning PR and leaning BN towards becoming fence-sitters.

This chart shows the swing by age, calculated by averaging support values for voters by their age group. Average support for PR increased from voters of all ages.

When we combine this graph with the number of regular voters by age, we can see the impact of the swing.

The trend is voters aged 63 and above swung further to PR, with the peak at 18.55% for 93 year-olds. Voters aged 34 years and below also swung further to PR, with the peak at 15.09% for 32 year olds. There is a decline in the swing from ages 42 – 54 years, where the swing value decreases from 13.29% to 11.08%. Swing values stayed within a range of 13% – 15% for ages 59 – 90 years. Unlike the Malay voters, support did not increase with each younger generation.

This chart illustrates the estimated level of support from voters. PR has support from 89% of the Chinese electorate, and this support is strong at 70+% from 17,158 (67.68%) of the Chinese voters.

5. Analysis of Support (Indian voters)

Total: 11,493 voters

Predicted GE13 Support (based on GE12 results)

Leaning BN: 794 (6.91%)

Fence-sitters: 6,771 (58.91%)

Leaning PR: 3,928 (34.18%)

Estimated GE13 Support (based on GE13 results)

Leaning BN: 2,585 (22.49%)

Fence-sitters: 6,035 (52.51%)

Leaning PR: 2,873 (25.00%)

The following histograms illustrate the distribution of probabilities among voters. The blue bars represent ‘Leaning BN’ voters. The yellow bar represents the fence-sitters. The red bars represent ‘Leaning PR’ voters.

Support for PR has decreased, with remaining supporters moving closer to the middle. The number of fence-sitters has had a slight decrease. The ‘Leaning BN’ voters have increased their concentrations in the 24% – 30% range and in the 47% – 49% range. Support for PR from Indian voters is low but still greater than BN. There is a concentration of support for BN close to the 50% mark, meaning there is room for PR to swing those voters over.

The support for PR for each voter has decreased by an average of 5.41%. For women the decrease was 5.67%, while for men the decrease was 5.12%.

When we group the voters by their est.swing, there is a pattern of movement away and towards PR. This histogram illustrates the number of voters by estimated swing to/from PR.

3,188 voters (27.74%) experienced a drop in support. 2,065 voters reduced support by 22% – 33%, meaning that BN-leaning voters swung further towards BN; PR supporters became fence-sitters; and PR supporters remained supporters.

6,055 voters (52.68%) had little to no change. This covers fence-sitters and leaning PR/BN voters who did not move from their current level of support.

2,250 (19.58%) swung to PR. 994 voters increased their support by 8% – 14%. Supporters became stronger; fence-sitters moved towards PR; and BN-leaning voters moved towards fence-sitters/PR.

945 voters moved from fence-sitters and leaning BN towards leaning PR.
1,900 voters moved from fence-sitters and leaning PR towards leaning BN.
237 voters moved from leaning PR and leaning BN towards becoming fence-sitters.

This chart shows the swing by age, calculated by averaging support values for voters by their age group. Average support for PR decreased from voters of all ages, apart from 92 year-olds.

When we combine this graph with the number of regular voters by age, we can see the impact of the swing.

The trend is voters aged 27 – 67 swung further from PR, with the peak at 8.28% for 60 year-olds. There is a pattern of swing values away from PR increasing with age.

This chart illustrates the estimated level of support from voters. After the swing away from PR, PR support has dropped from 34.18% to 25% of Indian voters. If this trend continues then BN stands to increase their share of votes from Indians. However with 53% of the Indian electorate in the fence-sitter category it is hard to predict which way the voters will swing.

6. Polling District Swing

The following is a list of polling districts (daerah mengundi) with their average estimated swing to PR and estimated support for PR based on GE13 results. The list is sorted alphabetically and only covers averages of regular voters. Districts with a negative swing are in bold. We define an ethnic group as the majority if >50% of the voters belong to that ethnic group.

1. BATAK RABIT (48.89% support, -2.6% swing) – Malay majority
2. BATU DUA BELAS UTARA (63.13% support, 12.23% swing) – Chinese majority
3. EASTERN GARDEN (71.12% support, 8.54% swing) – Chinese majority
4. JALAN ANSON (63.05% support, 13.5% swing) – Chinese majority
5. JALAN BATAK RABIT SELATAN (68.08% support, 7.76% swing) – Chinese majority
6. JALAN BATAK RABIT UTARA (73.66% support, 16.59% swing) – Chinese majority
7. JALAN CANAL (63.37% support, 7.41% swing) – Chinese majority
8. JALAN LAXAMANA (50.95% support, 0.98% swing) – Mixed
9. JALAN MARKET BARAT (79.24% support, 15.54% swing) – Chinese majority
10. JALAN MARKET TIMOR (74.93% support, 18.91% swing) – Chinese majority
11. JALAN SPEEDY (68.8% support, 10.09% swing) – Chinese majority
12. JALAN SUNGAI NIBONG (59.18% support, 5.67% swing) – Chinese majority
13. KAMPONG BAHAGIA (35.88% support, 2.65% swing) – Malay majority
14. KAMPONG BANJAR (52.73% support, 2.04% swing) – Mixed
15. KAMPONG BARU AYER HITAM (24.43% support, 0.15% swing) – Malay majority
16. KAMPONG CHANGKAT JONG (24.04% support, -5.23% swing) – Malay majority
17. KAMPONG GURU (66.13% support, 4.85% swing) – Chinese majority
18. KAMPONG PADANG TEMBAK (39.39% support, 1.82% swing) – Malay majority
19. KAMPONG SELABA (18.59% support, -1.56% swing) – Malay majority
20. KAMPONG SUNGAI SAMAK (27.09% support, 4.52% swing) – Malay majority
21. LADANG SELABA (53.28% support, -22.06% swing) – Indian majority
22. LADANG SUNGAI SAMAK (42.74% support, -8.41% swing) – Indian majority
23. LADANG SUSSEX (32.83% support, -27.53% swing) – Indian majority
24. LADANG ULU BERNAM (34.79% support, -11.64% swing) – Indian majority
25. NOVA SCOTIA (48.58% support, -17.68% swing) – Indian majority
26. PASIR BEDAMAR BARAT (79.76% support, 15.52% swing) – Chinese majority
27. PASIR BEDAMAR SELATAN (77.84% support, 14.53% swing) – Chinese majority
28. PASIR BEDAMAR TENGAH (79.64% support, 14.61% swing) – Chinese majority
29. PASIR BEDAMAR UTARA (80.41% support, 14.91% swing) – Chinese majority
30. PEKAN BARU (59.97% support, 8.21% swing) – Chinese majority
31. SUNGAI BUGIS (18.67% support, 3.09% swing) – Malay majority
32. SUNGAI KERAWAI (28.58% support, 0.53% swing) – Malay majority
33. SUNGAI SULI (70.08% support, 26.75% swing) – Chinese majority
34. SUNGAI TEMAH (59.84% support, -2.59% swing) – Indian majority
35. SUNGAI TUNKU (25.35% support, 3.6% swing) – Malay majority
36. TAMAN CECILY (54.52% support, 2.77% swing) – Mixed
37. TAMAN SERI SETIA (56.67% support, 4.02% swing) – Mixed

Analysis:

1. All Chinese-majority districts have strong support for PR
2. All 4 Mixed districts have >50% support for PR, however its closer to the 50% mark
3. All Malay-majority districts have a low level of support for PR, with 3/10 districts swinging further to BN
4. All 6 Indian-majority districts have swung further to BN as well, however LADANG SELABA and SUNGAI TEMAH still have >50% support

6. Summary

In order to win, the candidate needs 30,242 votes (based on the previous constituency size of 60,483 voters). For BN this means getting votes from early voters (387), BN-leaning voters (15,999) and fence sitters (13,856).

That would require a minimum of 88% of fence-sitters (13,856/15,741) to swing to BN, assuming the other groups support BN 100%. A better target would be 50% of fence-sitters (7,870 voters) and 21% of PR supporters (5,986 voters).

This may be difficult for BN to achieve due to the support PR received from all Chinese voters and the swing from Malay voters between 21 – 40 years old. After the swing PR support from Malay voters remains low. However Malay votes alone are not enough for BN to win and they need a swing from the other ethnic groups. The swing to BN from Indian-majority districts is a good indicator and something for BN to build on.

In our projections based on these statistics, 15% – 30% of Chinese voters need to swing over to BN in order for BN to win.

For PR to win, they need to retain support from their existing voters (especially Chinese voters) and get 2,000+ votes from the fence-sitters and Leaning BN voters. To achieve this PR can push on the Malay youth swing and try to win back support from the Indian voters.

7. Summary of Methodology

Voters are assigned to polling lanes based on their age and polling district (the area they reside in). A lane can hold 350-700+ voters, with a common size of 400-650 voters. From the results of the lane, we can assign a probability value to each voter in that lane.

For example if 400/600 votes went to BN, then the average probability of voters in that lane voting for BN is 66.67%. We can take the analysis a step further by adjusting for ethnicity, because each lane will have a fixed maximum number of ethnic voters. If there are 500 Malay voters in our example that means BN obtained at most 400 Malay votes and at least 300 Malay votes. This changes the probability of Malays voting for BN to a value between (300/500) to (400/500), or 60% – 80%. We take the average of this, giving a probability of 70%.

Note that this calculation does assume that turnout is evenly distributed by race. While that is not true, high voter turnout rates and ethnically dominant lanes reduce the impact of error. We tested both the average probability and the ethnicity-adjusted probabilities in our GE13 simulations and the ethnicity-adjusted probabilities yielded greater accuracy. We also had to take into account that voters migrate between constituencies. When that happens they bring their individual probabilities with them. This helps avoid inaccuracies that are introduced by focusing only on polling district results from previous elections.