Here’s an analysis of what happened to 900+ missing voters, as charged by PKR. (Sent by Cheah Kah Seng)
[For all tables and data, please see:
Were 900+ Voters Deleted and Replaced in Permatang Pauh?
The numbers are about right – 951 voters were deleted from Permatang Pauh and replaced by another 961 since the 12th GE. But the reasons were technical and not too sinister. Still, the whole confusing affair highlights the needs for SPR transparency and voter registration reform.
Our investigation did find that 951 voters were “deleted” after the 12 GE in Permatang Pauh, while another 961 voters were “added” to the by-election roll dated 2008 Jul.
But we also found that these changes were due to three not-too-cynical reasons and minor SPR mistakes:
1. The technicality of modifying police IC numbering scheme,
2. Regular voter registrations, inward and outward movements, and
3. SPR’s dead-voter deletion overdrive.
Although the reasons are not sinister, there are lessons for the SPR: How can SPR avoid such public confusion and distrust in the future?
1. First, let police and military voters switch to a 12-digit civilian IC to simplify verification, update, and avoid duplicate registrations.
2. Second, make voter registration automatic with MyKad status at the NRIC Department. Meanwhile, make NRIC accountable to the parliament, to prevent Sabah-like IC mess,
3. Third, make SPR more transparent with the gazetted and quarterly electoral rolls. SPR has recently stopped the online publication of the quarterly rolls in PDF format.Given adequate data, NGOs and the media can actually help SPR avoid unnecessary confusion.
What’s so complicated about the electoral rolls?
* First, we need to clarify which versions of electoral rolls are being bandied about.
* We believe the confusion arises from comparing these versions:
o the 2008 Feb 5 gazetted roll used in the 12th GE, with data up to 2007 Dec 31 (=2007Q4),
o The 2008 Jun 19 gazetted roll, with data up to 2008 Mar 31 (=2008Q1),
o The ungazetted supplementary roll with data from 2008 Jun 30 (=2008Q2),
o The 2008 Jul 31 ungazetted roll for Permatang Pauh, based on the 2008 Jun 19 gazetted roll, adjusted for subsequent deletions, but not additions, ie, deletion-type changes between 2008 Apr 1 and 2008 Jul 31.
* PKR members are probably comparing versions (a) with (d), because they are most likely to have the MS Access soft copies for these two. Version (c) is most widely available as PDF files, and reflect a “flow,” or change, data, not a “point-in-time,” cumulative, data.
* In contrast, the SPR probably thinks in terms of versions (b) vs (d), since (b) is a gazetted version. Version (c) is an intermediate version, that is to be gazetted. If and when (c) is gazetted, the Permatang Pauh section can lead to further confusion, because its data are dated 20080630, and is a “prequel” to the by-election version (d).
Finding 1. The difference of only 10 voters between the 12th-GE and by-election rolls hides larger changes beneath.
Finding 2. 951 voter ICs were deleted after the 12th GE roll, while 961 voter ICs were added to the by-election roll, netting the superficial increase of only 10 voter ICs.
Our question is then: Were the deletion of 951 voters and addition of the other 961 voters reasonable and kosher?
Finding 3. Nearly half the changes involved an identical group of police postal voters, who have not actually moved, but have only had their police IC numbers slightly modified. This is enough to confuse the public, even apparently the SPR itself.
* Overall, there were 492 police postal voters in Permatang Pauh as of the 12th GE.
* One of them (G6014602) already had the “most updated” IC numbering format, and hence this IC number was not changed, “deleted” or “added back.”
* Another two (I0013148 and R0071611) were permanently deleted after the 12th GE, probably due to retirement.
* The remaining 489 (=492-1-2) police postal voters had their IC numbers modified slightly, as we will explain later.
* This IC number change causes these 489 postal voters to appear to have gone “missing” en masse.
* The “missing” voters would total 491 if we count the 2 retirement cases mentioned above.
* Meanwhile, the same 489 postal voters’ newly modified IC numbers make them look like 489 newly “added” voters. This likely happens because PKR, and even SPR itself, have no choice but to search the database by relying on the unique IC numbers, known in database parlance as the “primary index”.
* This apparent and illusory “deletion” of 491 and “addition” of 489 postal voters account for about half the 951 “missing” and 961 “replacement” voters in the by-election roll.
Finding 4. The proof is in the detail: Let us look at all 489 pairs of slightly-changed police IC numbers.
From Table C below, we can see that the 489 pairs of old and new police postal IC numbers are nearly identical, having only these 2 feature changes:
1. All police IC # starting with “R0” or “R” in 2008 Feb 5 roll have been changed to “RF” in the by-election roll. For example, R0117812 has been changed to RF117812.
2. All police IC # starting with “G0,” “I0,” or “RT0” have been changed to “G,” “I,” or “RT” eliminating preceeding zeros. For example “RT007013” and “I0013642” have been changed to “RT7013” and “I13642” respectively.
…[long table of police ICs]
Finding 5: SPR did a poor and confusing job in modifying these police IC numbers, and in explaining the changes.
While we now understand that 489 police postal voters really did not move in or out of Permatang Pauh, there remain other questions and expected problems:
* SPR should have expected this IC number confusion, should have been ready to explain the changes, thereby preempting a loss to its own credibility.
* This postal IC number modification occurred in a rush, only after the 2008 Jun 19 gazetted roll, in which we can find the old-formatted IC numbers. Confirming this timing is that we cannot find these “deletion” or “addition” in the 2008Q1 supplementary roll.
* The changes do not look permanent. Any further changes will again invite confusion. Some of the “new format” police IC numbers are too short or too long to be a standardized, permanent, format:
o Most of the IC#s are preceded by one or two alphabets, followed by 6 digits,
o But there are IC#s followed by only 4-5 digits, for example, RT9714, G9929, and I13642.
o There is an IC# followed by 7 digits, ie, RF4068948. With the “RF” prefix, this IC# has 9 characters in all, and is a non-standard occurrence.
o With the above oddities, we would expect SPR at anytime to add back the zeros to the shorter IC#s make them a standard length, reigniting confusion.
* A separate matter is that about 404 of the “Bukit Mertajam” police HQ/station officers have probably been gerrymandered into the “Permatang Pauh” seat. We will deal with the geographical question another time.
Why did SPR risk confusion to modify the police IC numbers? We believe SPR “shortened” the police IC numbers to preempt any embarrassing discovery of duplicate registrations of postal voters, as reported earlier:
* MalaysiaKini.com reported just before the 12th GE that computer programmer Ong Guan Sin had found 278 pairs of such double-registered voters that differ by a leading zero.
* With a complete database, we later found 771 such doppelgängers that differ by a leading zero, or only different by the initial “R0” and “RF.”
Obviously, the SPR didn’t anticipate that this IC-shortening move itself would cause new internal and external confusions.
Finding 6: The deletion of 460 non-postal voters after the 12th GE is more dicey. We can only trace 218 of the 460 deletions – given SPR’s lack of transparency. The remaining 242 could have contained the 5 embarrassing cases of “alive-and-well but deleted-as-dead” voters documented by the media and NGO.
Table D tells us that:
* 460 of the 951 voter ICs deleted were non-postal voters
* Our investigation of the 2008Q1 supplemental roll revealed that 84 voters were deleted because they moved out of P044, and 89 deceased voters were deleted. SPR was going through an aggressive campaign to clean up cumulated dead voters across the country.
* In total, SPR deleted 173 (=84+89) voters from Permatang Pauh in 2008Q1. We have recorded this finding in our earlier blog “Pro Forma Electoral Roll for P044 for 2008Q1.”
* That means SPR deleted the remaining 287 (=460-173) during 2008Q2 (April, May, June), and also in July.
* We cannot trace what happened then because SPR stopped publishing supplemental roll for 2008Q2, and only displayed the hard copies in post offices.
* Further, we can tell by manually checking SPR’s P044 by-election voter verification Web page, that SPR deleted all voters age 100 or more across the board, which account for 25 voter deletion after the 2008 Jun 19 gazetted roll. We have documented this apparent “stroke-of-a-pen deletion of centenarians here.”
* Further, SPR deleted a total of 20 voters aged 86 to 89 years old.
* In all, we can trace 218 (=84+89+25+20) of the 460 deletions of non-postal voters so far.
* That leaves 242 deletions of non-postal voters to be explained.
From recent pattern of a special SPR campaign to delete dead voters, we figure it is reasonable to expect about 200 deletions of moved-out and dead voters per quarter, for April, May, and June, a period that should overlap with the above 86+ voter deletion (25+20).
* Any registration changes for July 2008 (after 2008Q2), however, have not been even public displayed – not to mention gazetted. It would be imprudent for SPR to delete from July changes, although it could account for perhaps 100 or more missing voters.
* SPR’s lack of transparency and unwillingness to make 2008Q2 supplementary roll easily available means we cannot trace the last 242 deletions. We can only make partial, manual, checking and guesses.
* These 242 untraced deletions could easily contain mistakes created in a rush. If so, SPR would have embarrassed itself by trying to avoid embarrassing itself.
* MalaysiaKini.com and MerdekaReview.com have reported a total of 5 cases of voters, who are alive and well, who were told by SPR staff on polling day that they had been considered dead and deleted from the rolls.
* We can confirm one case (Ang Kean Tatt, born 1983) where the voter did exist in the 2008 Feb 5 roll, is alive and well, but is not found in the 2008 Jul 31 roll, and was told he was dead in the voters roll.
* We suspect other cases may involve long-dormant voters who have suddenly found themselves gerrymandered away from Permatang Pauh even before the 12th GE.
* See various reports:
o Four names missing from electoral list
o 选民册上宣报死亡 真正幽灵选民显现？ Machine translation (will contain funny machine translation mistakes).
Finding 7: The “addition” of 961 voters in the by-election roll was also inflated by 489 new-formatted police postal voter IC numbers
Finding 8: The remaining 472 (=961-489) new non-postal voters can be traced to normal new voters and moved-in voters in 2008Q1, gazetted in 2008 Jun 19.
* Still, the difference of 13, between 485 non-postal voters added in 2008Q1 and 472 additional non-postal voters found on the by-election roll, is a small mystery. Some 13 voters were somehow deleted or not included somewhere.