I note with alarm that an increasing number of political leaders, commentators and readers are forming erroneous conclusions from wrong interpretation of numbers in the recent by-elections in Kedah.

This is due to the tendency to use absolute numbers, instead of percentages, to gauge the degree of support a party enjoys among the electorate.

For instance, many have claimed that PAS has gained higher support in the Pendang constituency this time, in spite of having lost the by-election, because it has obtained 22,542 votes (which is 49.7 percent of total votes cast), which is numerically higher than the 22,413 votes (53.5 percent) it obtained in 1999.

This conclusion is wrong, because when we talk about support, we are actually talking about the extent of support. And the 'extent' can only be ascertained if the votes obtained by a certain party are compared to those by another party.

To do so, one has to express the numbers of votes garnered by the respective parties as a percentage of the total votes cast. The respective percentages will then indicate the extent of support each party commands in the constituency.

Thus, in the example quoted above, the correct inference should be: "PAS support in the Pendang constituency has dropped from 53.5 percent to 49.7 percent, slipping 3.8 percent."

And how much does that loss mean to PAS, in percentage term?

The answer to that is not 3.8 percent (53.5 minus 49.7), but 7.1 percent (3.8 divided by 53.5).

Using this methodology, we can take another look at the two election results as follows:

The above tables show that though both parties have won one by-election each, nevertheless in terms of changes to the extent of support, BN's support has gained 10.3 percent and 8.2 percent in Anak Bukit and Pendang respectively, while PAS support has decreased by 8.0 percent and 7.1 percent respectively in the same constituencies.