Findings from a study indicate strong possibility of 'phantom voters' in the electoral roll as at September 2003, according to election watchdog Malaysians for Free and Fair Elections (Mafrel) which commissioned the check.

The study, conducted in February, covered three parliamentary constituencies - Lembah Pantai in Kuala Lumpur, Tanah Merah in Kelantan and Gombak in Selangor - with a random sampling of 1,000 registered voters in each.

However, the findings only took into account 'accepted data' from 796 voters in Lembah Pantai, 790 in Tanah Merah and 965 in Gombak.

Based on this, the highest number of discrepancies were recorded in Lembah Pantai, leading Mafrel to contend that 73.12 percent of the registered voters are probably 'phantoms'. The figure in Tanah Merah was put at 20 percent and in Gombak at 50 percent.

Two other categories of flaws covered 'phantom addresses' and 'roll inaccuracies' (incomplete addresses of voters whose names appear in the roll).

Even taking into account an error margin of plus or minus 3.1 percent, the electoral roll contain "fatal defects and irregularities", claimed Mafrel chairperson Abdul Malek Hussin at a press conference in Petaling Jaya today.

He said the study was conducted by 75 individuals who were trained and deployed to conduct ground observation on the randomly generated voter lists. Each researcher was given a list of 40 voters.

He acknowledged that Mafrel was unable to base its findings on the very latest electoral roll - for the period ending Dec 31, 2003 - since the EC itself had failed to make this available, even after two months of its completion. (The list was gazetted last Tuesday, according to press reports.)

"It was no fault of Mafrel," he said, noting that the date of gazette has also left insufficient time to scrutinise the latest roll before polling day on March 21.

Nevertheless, Abdul Malek said that even the probability that phantom voters exist - as played out in various permutations found by Mafrel surveyors - indicate a need for further scrutiny of the roll.

"The whole process of the general election will be flawed and undemocratic if an inaccurate and unauthentic electoral roll is used," he said.

As a remedy, Abdul Malek called on the EC to push back the nomination and polling dates, which it is empowered to do within 60 days of dissolution of Parliament - so that the rolls can be cleaned up first.

'Phantom voters

'

Mafrel defines 'phantom voters' as individuals with untraceable addresses or as unidentified voters - the latter refers to individuals are not found living in the addresses registered, and who are not known to other registered voters living in the area.

The total pool of 'phantom voters' in each constituency combines the percentage of untraceable addresses and unidentified voters.

In the case of untraceable addresses, Lembah Pantai topped the list with more than 37.69 percent of voters, followed by Gombak (26.01 percent) and Tanah Merah (11.01 percent).

For unidentified voters, the breakdown was 35.43 percent in Lembah Pantai, 24.04 percent in Gombak and 9.62 percent in Tanah Merah.

According to the study, one location in Kampung Kerinchi in Lembah Pantai recorded a whopping 142 voters at a single address, which turned out to be a wood-and-zinc shack selling sundry goods. The operators denied that there were 142 people living there ( above ).

'Phantom addresses'

Mafrel also found an average of 11.08 percent probability of dubious, false and unknown addresses, otherwise known as phantom addresses, across all three constituencies.

Again, the highest number - 15.73 percent - was found in Lembah Pantai, with 11.81 percent in Tanah Merah and 1.82 percent in Gombak.

Two households in Kampung Pasir Dalam, Lembah Pantai, were registered as having 156 and 120 voters respectively, but the addresses were found to be non-existent. The site is now part of the New Pantai Expressway, which was opened on March 1 ( right )

Another 34 addresses were located in Kampung Sentosa, which had burned down two years ago, but were still included in the September 2003 electoral roll. The site now accommodates a Fire and Rescue Services complex.

'Roll inaccuracies'

Of the addresses which could be verified, Mafrel still found 22.84 percent inaccuracy in the details - referring to addresses with missing house- or apartment-numbers.

Tanah Merah recorded the most inaccuracies, at 39.12 percent, followed by Lembah Pantai (23.38 percent) and Gombak (6.44 percent).

One incomplete address points to the Ulu Gombak Orang Asli Hospital.

"If voters are legitimately registered at this locality, then there is no reason why the house number should be missing, as all government quarters are clearly marked," Abdul Malek noted.

Responding to a question, Abdul Malek said the EC's recent explanation that incomplete addresses are given in the roll in order to protect voter privacy from unscrupulous marketeers, "does not make sense".

"The argument of privacy does not hold water. If this is the case, then why are complete addresses given for areas with middle- and high-income groups, as in Lembah Pantai? Is the EC saying it is acting to protect the privacy of lower-income groups only?"

The EC's response, he said, is indicative of how the independent body often shirks responsibility in ensuring the legitimacy and accuracy of the voting process.

In the 1999 general election, opposition parties had alleged that 'phantom votes' - using fake identity cards and the identity of deceased voters - had been used to secure victory for Barisan Nasional.

The issue dominated the headlines in 2000 when PAS and Umno pointed a finger at each other for allegedly 'importing phantom voters by the busload' to vote in the Lunas by-election.

Last September, EC chairperson Abdul Rashid Abdul Rahman blamed PAS as responsible for introducing 'phantom voters' into some constituencies, but also denied that there were some 2.8 million phantom voters on the electoral roll, as asserted by then Prime Minister Dr Mahathir Mohamad.