undi.info have published a neat little tool for predicting the outcome of the 5 May 2013 Malaysian General Elections Parliament seats
Based upon your estimates of the % swing for each Group of voters the tool calculates the individual seats which will change hands. They give a guide of 0 to 15% swings.
What is a swing?
A swing represents the level of change between the two parties from one election to the other.
As an example.
Assume 1,000 valid votes in Seat XYZ in 2008.
The BN candidate got 600 votes and the PR candidate got 400 votes.
The BN candidate has 60% and the PR candidate 40%.
Further assume there were 800 valid votes in 2004 in Seat XYZ.
The BN Candidate got 560 votes and the PR candidate got 240 votes.
The BN candidate has 70% and the PR candidate 30%.
The difference for the BN was an additional 40 votes and the difference for the PR was an additional 160 votes. The BN candidate lost 10% share and the PR candidate gained 10%. The swing is half the loss or the gain which is 5%. When one vote swings the effect is doubled as it is negative on one side and positive on one side.
Assume therefore that in the 2013 elections there are now 1,200 votes. What swings are required for the BN to retain the seat and for the PR to wrest the seat?
To win either candidate requires 601 votes, 50.08% of the 1,200 votes.
So to retain the seat the BN needs 41 more votes than the previous election and the PR needs 201 more votes.
The seat changing swing is (60% - 40%) / 2 = 10%.
So if more than 10% voters switch their votes from BN to PR then the PR candidate will win.
If less than 10% voters switch their votes from BN to PR or if PR votes swing to BN, the BN candidate will win.
What has been the historical swings?
Swings have ranged from -6.8% to 8% in the General elections from 1964 to 2008.
Note that the swing % of votes will differ from the % seats.
This is due to the differences in the % majorities at each election.
What about 2013?
The undi.info tool requires you to estimate the swing for each group of voters. These groupings have historically swung different ways in past elections.
If however you assume that all groups will vote uniformly and input the same % in each Group
you will see that the BN only needs a swing towards it of 0.5% to win back a 2/3 majority. The BN can afford a negative swing of 4.5% before it loses government.
This overlooks the fact that there are other factors that make the swings wildly inconsistent across groups, states and seats.
Other random thoughts.
The votes cast is dependent, upon amongst others:
- seats being contested, (in the past some seats have been won uncontested on nomination day)
- voter turnout , (in some elections voters stay at home, voting with their feet rather than the ballot box)
For instance, in 1974, less people voted than in 1969. They came back in 1978 though! In 2004 there were only 285,044 additional voters than in 1999. In the past 5 elections the higher the additional turnout the the swing against the government was higher.