2009-05-26

Penanti - Racial composition by voting stream

Although we want to move away from racial politics, anticipating and analyzing racial composition of voting return by voting stream can still be useful for positive purposes: to neutralize negative campaign and to balance the information power of federal government versus the opposition/NGOs.

Use: Analysts can compare this table to the voting return, by voting stream (classroom), recorded by all party polling agents, to gauge the effectiveness of their campaign message.

Only the Home Affairs ministry and SPR know the officially-recorded racial composition by saluran (voting stream),

  • The SPR does NOT publish racial data by DM, TM, or saluran (stream),
  • Our algorithm is crude, but ends up with reasonably accurate and detailed results
  • see Table B, where we compare our over estimations with "official" estimations reported by the media and http://undi.info.,
  • Because any estimation bias is likely to be cumulative (not offsetting), a small error at DUN level would mean EVEN SMALLER errors at DM or Saluran level.
Algorithm:
  • We estimate racial breakdown based on the patterns in voter names.
  • The algorithm:
    • first filters for "Malay" names, starting with Bin, Binti, and all its variations, plus a few common Malay names,
    • then it filters for "Indian" names, by by A/L, A/P, D/O, S/O,
    • plus a few common Indian name particles, and spelling and vowel patterns,
    • the remaining are considered "Chinese"
  • Admittedly, this algorithm is imperfect.
  • This algorithm currently cannot handle the Thai and Dayak name pattern (as in Batang Air).
  • Inside Penanti, the "Other" minority ethnic names, such as Thai names, tend to be lumped into the "Malay" category, inflating it by perhaps 0.6%-0.7% in Penanti. In other constituencies, it is often the "Chinese" category that is inflated by about 1%, by absorbing the errors.
  • However, since the official figures show a fluctuation in the "Others" category between GE12 (0.1%) and now (0.7%), we have some doubts about the consistencies of the official figures on the "Others."

Table A: P044 Permatang Pauh N12 Penanti 20090416 By-Election Roll
Ethnic Breakdown by Saluran - Estimation By an Algorithm by http://malvu.blogspot.com.

Note postal voters include non-police, absentee voters.
parl/DUN/DMSaluranNamaDMTM Malay Chinese Indian Total% Malay% Chinese% IndianTotal %
044/12/011Guar PerahuSMK Guar Perahu 346 4 - 35098.9%1.1%0.0%100%
044/12/012Guar PerahuSMK Guar Perahu 521 - - 521100.0%0.0%0.0%100%
044/12/013Guar PerahuSMK Guar Perahu 521 - - 521100.0%0.0%0.0%100%
044/12/014Guar PerahuSMK Guar Perahu 520 1 - 52199.8%0.2%0.0%100%
044/12/021Kuala MengkuangSK Guar Perahu 345 3 2 35098.6%0.9%0.6%100%
044/12/022Kuala MengkuangSK Guar Perahu 596 - 7 60398.8%0.0%1.2%100%
044/12/023Kuala MengkuangSK Guar Perahu 598 1 4 60399.2%0.2%0.7%100%
044/12/024Kuala MengkuangSK Guar Perahu 595 3 5 60398.7%0.5%0.8%100%
044/12/025Kuala MengkuangSK Guar Perahu 602 2 1 60599.5%0.3%0.2%100%
044/12/031Telok WangDewan Orang Ramai Kampung Tun Sardon Ara Kuda 331 183 85 59955.3%30.6%14.2%100%
044/12/041MengkuangSK Mengkuang 309 38 3 35088.3%10.9%0.9%100%
044/12/042MengkuangSK Mengkuang 462 45 10 51789.4%8.7%1.9%100%
044/12/043MengkuangSK Mengkuang 471 39 7 51791.1%7.5%1.4%100%
044/12/044MengkuangSK Mengkuang 482 32 3 51793.2%6.2%0.6%100%
044/12/051Sungai LembuSJKC Perkampungan Sungai Lembu 4 499 4 5070.8%98.4%0.8%100%
044/12/061PenantiSMK Mengkuang 166 169 15 35047.4%48.3%4.3%100%
044/12/062PenantiSMK Mengkuang 249 82 21 35270.7%23.3%6.0%100%
044/12/063PenantiSMK Mengkuang 279 59 14 35279.3%16.8%4.0%100%
044/12/070Kubang UluPos 22 1 - 2395.7%4.4%0.0%100%
044/12/071Kubang UluSMK Penanti 185 153 12 35052.9%43.7%3.4%100%
044/12/072Kubang UluSMK Penanti 261 237 21 51950.3%45.7%4.1%100%
044/12/073Kubang UluSMK Penanti 294 195 31 52056.5%37.5%6.0%100%
044/12/081Sungai SemambuSK Tanah Liat 245 101 4 35070.0%28.9%1.1%100%
044/12/082Sungai SemambuSK Tanah Liat 305 191 7 50360.6%38.0%1.4%100%
044/12/083Sungai SemambuSK Tanah Liat 372 120 11 50374.0%23.9%2.2%100%
044/12/084Sungai SemambuSK Tanah Liat 388 109 6 50377.1%21.7%1.2%100%
044/12/090Tanah Liat Mukim 9Pos 1 - - 1100.0%0.0%0.0%100%
044/12/095Tanah Liat Mukim 9SK Tanah Liat 291 53 6 35083.1%15.1%1.7%100%
044/12/096Tanah Liat Mukim 9SK Tanah Liat 434 72 17 52383.0%13.8%3.3%100%
044/12/097Tanah Liat Mukim 9SK Tanah Liat 419 93 11 52380.1%17.8%2.1%100%
044/12/098Tanah Liat Mukim 9SK Tanah Liat 394 118 11 52375.3%22.6%2.1%100%
044/12/101Berapit RoadSJKC Jit Sin (A) Jalan Berapit 50 276 24 35014.3%78.9%6.9%100%
044/12/102Berapit RoadSJKC Jit Sin (A) Jalan Berapit 80 444 28 55214.5%80.4%5.1%100%
044/12/103Berapit RoadSJKC Jit Sin (A) Jalan Berapit 146 380 27 55326.4%68.7%4.9%100%
Total


11,284 3,703 397 15,38473.3%24.1%2.6%100.0%
Note: SM = Sekolah Menengah, SK=Sekolah Kebangsaan, SMK=Sekolah Menengah Kebangsaan, SJKC = Sekolah Jenis Kebangsaan (China), SJKT = Sekolah Jenis Kebangsaan (Tamil), SAR = Sekolah Agama Rakyat
Note: If you cannot see the full table width, try to click Menu>View>zoom out (or Ctrl-Middle-Mouse-Button-Scroll). Alternatively, click-drag to highlight the table, copy, and paste into Excel to view.
Note: You are welcome to contact us to understand more about this algorithm.


Table B shows that
  • we have compared our estimations against the official ethnic breakdown (from UNDI.info and government figures reported by Bernama).
  • These DUN-level comparisons are based on the older GE12 roll (2009 Feb 5 gazetted, data dated 2007 Dec 31), the Permatang Pauh by-election roll (dated 2008 Jul 31), and this latest Penanti by election (dated 2009 Apr 16).
  • Comparing the latest roll (2009 April) at the DUN level, our estimates are very close to official estimates, with difference of 0.1% for the Chinese and Indian components.
  • The difference in the Malay composition is larger at 0.6%, apparently because our algorithm lumps the "Others," mostly ethnic Thai names, under the Malay group in Penanti.
  • But note there is fluctuation in the "Others" categories in the official figures between GE12 and now.
  • Down at the stream-level (saluran), the error in absolute number should be even smaller, and the estimate should be useful for stream-level analysis.
Table B: Comparison of overall racial composition of Penanti, at different time, by different sources
Other Estimates% Malay% Chinese% IndianOthersTotal
Our estimates for p044n12 ethnicity 090416 By-election for Penanti (This by-election)73.3%24.1%2.6%
100.0%
SPR/Bernama 200904 (Note 1)72.7%24.2%2.4%0.7%100.0%
Our estimates for p044n12 ethnicity 080731 By-election for Permatang Pauh (calculate for Penanti only)73.2%24.2%2.6%
100.0%
Our estimates for p044n12 ethnicity 080205 GE73.2%24.3%2.5%
100.0%
UNDI.INFO 20080205 (Note 2)73.0%24.2%2.4%0.1%99.7%

Note 1: http://www.bernama.com.my/bernama/v5/news_lite.php?id=407049
Note 2: http://undi.info/ax/state/pn/2008/states/N12/P44.html

No comments: