The most commonly cited figure in the election is perhaps 198, the number of parliamentary seats held by Barisan Nasional (BN). A meaningful understanding of the election output however must go beyond this simple figure or that of its comparison to the152 seats in the 10th Parliament.

On the other hand, many vigilant observers pointed out how the opposition suffered from systemic bias and irregularities, but do not aptly differentiate between the two. This analysis hopes to suggest a more structured reading for in-depth and precise discussion.

Seat allocations in any election is the result of three factors: (a) voters' preference, (b) disproportionality and (c) electoral corruption (if any).

When we talk about 'popular sovereignty' as the basis of democracy, we actually mean 'voters' preference, namely what the voters desire. This is shown in the vote share gained by various parties, if there is no 'electoral corruption'. The differences between the vote shares (V%) and seat shares (S%) gained by the parties, is 'disproportionality', which exists in all elections in all systems, but at different levels. A good measure of the disproportionality which we would use hereafter is the Seat-to-Vote Ratio = V% / S%.

Both 'disproportionality' and 'electoral corruption' distort the choice of the citizenry. While both are undemocratic so to speak, they must be distinguished: the former is clearly legal and systemic, while the latter if not unlawful is often due to discretion.

For example, a marginal constituency might be won over, not democratically by winning the hearts and minds of more voters, but technically by (a) changing the constituency boundary, (b) 'importing' (on electoral roll only) own supporters from other constituencies (c) 'exporting' (on electoral roll only) opponents' supporters to other constituencies. The first method, called gerrymandering, is legal and can only be rectified by changing the law or even the constitution, whereas the other two would constitute election offences under existing regulations.

The first two parts of this analysis will focus on separating (a) voters' preference from (b) disproportionality so the actual message from the voters might be better captured. The underlying assumption here that (c) electoral corruption does not happen in any instance is by no means our empirical assessment, but merely for analytical clarity. To bring in irregularities here would only disrupt the revelation of systemic bias. Electoral corruption would be dealt with conceptually in the third part of the analysis on its sources, symptoms and competing explanations.

How big is Abdullah's victory?

Not obsessed with record breaking as his predecessor, Prime Minister Abdullah Ahmad Badawi has made two records in BN's history and outshone the former: (1) the best electoral results (in both vote and seat share) brought by a new premier (2) the highest seat share in Parliament. [Table 1]

We must nevertheless remember the best measure to Pak Lah's (as the premier is fondly known) popularity has to be BN's vote share, not its seat share which has been inflated by disproportionality. In other words, the figure is 63.97 percent. There is definitely a swing towards Pak Lah, by 7.5 percent compared to 1999. Unless we could assume that the intentionality (technology is another question) of electoral irregularities was significantly higher now than in 1999 when BN faced its greatest challenge, such swing cannot be argued away by mere charges of electoral irregularities

With due recognition to the so-called 'Pak Lah Tsunami', let's look at the disproportionality, from the Seat/Vote Ratio (Table 1 & Table 2). If every party is allocated seats proportionate to its vote share, the ratio would be 1. Therefore, BN is therefore now over-represented by 41 percent while the opposition are under-represented by 44 percent to 95 percent. [Table 2]

What would the situation be if the 'one citizen one vote' principle was upheld in 2004 and also 1999? (Table 3) BN should have an increase of 31 seats instead of 50 while PAS an increase of 5 rather than a decrease of 20. DAP that was greatly under-represented in the last election should now get 21 seats, amounting to a decrease of 3 instead of an increase of 2. On the other hand, Keadilan combined with PRM should suffer a worse decrease of 6 seats rather than 4, but at a much stronger position of 19 seats. [Table 3]

This story is nothing new - many who claim to live in reality and laugh at idealism would tell you. Now consider two things.

1 BN voter = 27.3 Keadilan voter

s

First, if one argues that our system has not been proportionate, please compare Abdullah's triumph with BN's best ever achievement in1995. Then, BN was really sweeping over the nation but it won a lower 84.38 percent of seats with the highest ever vote share of 65.2, over-represented by only 29 percent.

Had the disproportionality ratio remained the same at 1.29 (Table 1), BN today would have won only 180 seats, leaving 39 seats to the opposition. Simple arithmetic then tells us that if the systemic advantage enjoyed by BN remained the same since 1995, Abdullah would have won only 28 more seats (from 152 before the dissolution), the additional 18 seats are actually delivered by the 2002 delineation exercise.

Secondly, the over-representation of BN at 1.41 times gives serious implications when compared to the under-representation of DAP at 0.56, PAS at 0.21 and Keadilan at only 0.05 (Table 2). This simply means 1 BN voter = 2.5 DAP voters = 6.8 PAS voters = 27.3 Keadilan voters, by dividing the BN ratio over that of the respective opposition parties. Phrased alternatively, it takes all 616,912 voters who voted for Keadilan to successfully vote in one Keadilan MP, but only 4,441,425 BN voters for all 198 BN lawmakers, hence, only 22,431 BN voters for one BN representative. When a medium-sized party yielding nearly 10 percent of national support can be overwhelmed by 27 times via the electoral system, can we still call Malaysia a democracy? How long shall we continue to turn our eyes away?

What causes disproportionality to this degree? The most well-known culprit is 'gerrymandering'. It refers to the deliberate re-drawing of constituency boundaries t o make a party enjoy plurality support in as many constituencies as desirable. You can easily tell the sign of gerrymandering from the electoral map if you see constituencies with weird shapes that do not respect natural, administrative or community boundaries. The constituency Governor Gerry crafted in his 19th century Massachusetts with partisan calculation looked like a salamander, explaining the origin of the term.

In Malaysia, gerrymandering engineers are given a free hand as they only have to respect the state border. As Janadas Devan ( Strait Times , March 26, 2004) aptly points out, "the profiles of every street, alleyway and lorong - finely sifted according to age, income levels, race, religions and previous voting patterns - can be fed into computers and out can pop constituencies shaped like cicaks, pelanduks, musangs, babi-hutans, tapirs or monyets."

The other partner

The less well-known partner is 'mal-apportionment'. It means deliberate apportionment of constituency size to over-represent certain group of voters and under-represent others. Upon Independence, the limit of constituency disparity was set at +/- 15 percent from the average constituency in each state. The 1962 amendment to the Constitution allows the rural constituency to be of half the size of any urban one. Such limits were totally abolished in 1973. In 2004, the average constituency size is 46,998 (voters), the largest parliamentary constituency P105 Kapar 104,185 (2.22 times the average, 20.51 times the smallest) whereas the smallest P125 Putrajaya only 5,079 (0.11 times the average, 0.05 times the largest).

While gerrymandering often leads to mal-apportionment, these twins could actually act independently. The best example of mal-apportionment without gerrymandering is Putrajaya, the boundary of which is exactly that of the Federal Territory but with only 10 percent of average constituency's voters.

For gerrymandering without mal-apportionment, consider this hypothetical scenario: P114 Kepong and P117 Segambut, with 60,273 and 57,349 voters respectively, have only a minor difference of 2,924 in size. In this election, BN wins the latter with a huge majority of 16,968 but loses the former with a slim majority of 1,854. Suppose a BN-voting electoral district with 3,000 voters can be taken from Segambut to Kepong while a DAP-voting electoral district with 4,462 voters taken from Kepong to Segambut; both constituencies, with zero disparity in electorate size, would likely return a BN candidate.

The third culprit is the electoral system itself. To be fair, all electoral system tends to favour larger parties and penalise smaller parties, but they do so at different levels. The first-past-the-post system (academically known as 'single member plurality') is amongst the most disproportional one. It favours parties with highly concentrated supporters, and in a national context, the largest party.

In this election, the opposition parties were disadvantaged at vastly different degrees, as shown by the 2.5 (DAP), 6.7 (PAS) and 28.2 (Keadilan) ratio. This may not be what some conspiracy theories offer as Umno's concerted efforts (read: irregularities or plots) to wipe out Keadilan, to wound PAS, and to weaken MCA. Technically, it merely suggests DAP was contesting in constituencies with higher supporter concentration.

Should Keadilan and PAS not contest some 'black' areas, they would have suffered lower disproportionality. In other words, if opposition parties want to improve their disproportionality, they should leave more seats to be walked over by BN, hence confining their role as opposition and never a government in waiting. Is that not clear enough how unlikely any change of government could be under the present system?

One may question if the disproportionality figures above is over-estimated by including the number of uncontested seats, 17 or 7.77 percent in this election. To be precise, BN's 63.97 percent vote share and the opposition's 36.03 percent were only competing for the remaining 202 seats. An adjustment will give smaller over-representation of BN at 1.40 and smaller under-representation of PAS, DAP and Keadilan at 0.23, 0.61 and 0.06, but the differences are dismissible. And, since the uncontested seats are generally strongholds of BN, if there is any contestation, the disproportionality would only be higher. The systemic bias is undeniable and simply too serious to ignore.

Note: I thank Mr Tan Teck Chin and Ms Seah Li Ling for sharing the 2004 electoral data they have dedicatedly compiled. Any errors in the table and text above are solely mine.

Trained originally in Economics, WONG CHIN HUAT sees the people as the boss of the government, elections the job interview and parties the job seekers. A columnist in mainstream Chinese dailies up to May 28, 2001 when Nanyang Press was taken over by MCA, he has since been an active member of Writer Alliance for Media Independence (Wami). He now reads comparative democratisation in the University of Essex, UK with specific interest on electoral systems and ethnic politics.