elections statistics and scenarios

i was inspired to write this by the article presidential election 2005 : outcome revealed by rajiv de silva in lankabusinessonline website. while i will come to similar conclusions here, i think that some assumptions made in the article are way too optimistic (from a unp point of view). it also fails to explain clearly how exactly the figures for north and east were arrived at. i will explain why and how my more conservative assumptions differ from article's as we go along in more detail.

first i will start with the 2004 general election figures taken from election department website.

table 01

party	votes	percentage
upfa	4223970	45.60%
unp	3504200	37.83%
tna	633654	6.84%
jhu	552724	5.97%
slmc	186876	2.02%
ucpf	49728	0.54%
epdp	24955	0.27%
others	86625	0.94%
total	9262732	100.00%

please note that in following scenarios the actual numbers are meaningless and are there for calculation purposes only, it is the percentage that counts.

scenario 01 : a crude add and paste from 2004 results

here i am going to make the following assumptions.

1. nobody has changed their minds from 2004

2. there wasn't any increase in voter registration( please note that except in north east there is no significant increase anyway. as for the increase in north east given the unofficial but real ltte boycott situation i am going to assume(in line with the conservative policy on assumptions here) that will also not matter. more on boycott later.

3. upfa, jhu, and epdp votes will go to mahinda rajapakse

4. unp,slmc,ucpf votes will go to ranil wickremasinghe.

5. tna votes won't go to anybody.

the result in such a situation will be

mahinda	4801649	55.64%
ranil	3740804	43.35%
others	86625	1.01%
total polled	8629078	100.00%
tna	633654
	9262732

of course this is a totally unrealistic outcome but this will be the base.


scenario 02 :

assumptions

1. there will be a 2.5% swing(of total votes) away from upfa to unp. (in lbo article it was assumed that there was a 1% swing in all district except in colombo where the swing was assumed at 3%. i don't understand the difference between colombo and the rest especially considering assumptions they made regarding the jhu vote, see below). a 2.5% swing is realistic and is in fact conservative in comparison to historical record of swings in sri lanka. for instance unp lost about 7.5 from 2001 to 2004 and pa almost the same from 2000 to 2001.

2. there will be an additional 1.5% of total votes for the unp. that a large portion of unp supporters stayed away was the only explanation for reduction(much larger than votes gained by upfa and jhu) in it's vote from 2001 to 2004. all evidence suggests that party base is much more enthusiastic than in 2004. i am also conservatively not assuming a similar phenomenon will affect the upfa this time, though it is more than likely given the internal divisions within the slfp.

3. jhu vote will split 2:1 for mahinda and ranil. in lbo article it was assumed jhu 'majority will accrue to UNP notwithstanding the JHU’s political alliance with the UPFA candidate. This assumption was made after a careful analysis of the GE 2004’s voting patterns.' imo that is too optimistic, though perhaps not unrealistic if recent opinion poll data are correct. anyway i am assuming that only one third of 2004 jhu vote(concentrated mainly in western province) will go unp way.

4. that most north east tamils ignore the unofficial boycott, 60% of those who voted for tna will vote and the vote will split 85:15 between ranil and mahinda. for what will happen if there is a more successful boycott see below. as for the split it is in fact very conservative, mahinda's support among tamils actually do not go over 10 % and please note that i have already added epdp votes to mahinda separately.

5. there wasn't any increase in voter registration, same as 2 on scenario 01

under above assumptions outcome will be :

mahinda	4439183	48.53%
ranil	4622403	50.53%
others	86625	0.94%
	9148211	100.00%

in other words a 2% win for ranil though he goes over 50% by only a whisker. given that third party candidates 's second preferences if counted will probably split equally, 50% rule won't matter in any case.

scenario 03 : a more effective ltte boycott

assumptions

assumptions 1,2, and 3 are the same as in scenario 2.

4. more tamils obey the ltte boycott and only 25% of the tna votes are cast this time( which will split 85:15 as before). unofficial postal voting numbers seems to indicate that about one third voted but i will be more conservative.

5. there wasn't any increase in voter registration, same as 2 on scenario 01. i am in effect assuming that all the new north east voters will obey the boycott.

result:

mahinda	4405917	49.36%
ranil	4433891	49.67%
others	86625	0.97%
	8926432	100.00%

a very marginal victory for ranil.

please note that i do not expect ltte to actively encourage votes for mahinda though they seem to have concluded that a victory for mahinda is in their best interest.(i will post separately on this stupid and seemingly illogical conclusion).

conclusion

in conclusion i am still fairly confident of a unp victory even with a ltte boycott . more so because i have made fairly conservative assumptions in calculating above scenarios.

by far the main and most important assumption is the 2.5% swing against the upfa, without it ranil will not win, period. imo the swing will be larger, but this is the minimum percentage required, anything below this will mean we will have that idiot mahinda in power on 18th..

btw i have been predicting a unp victory here and in other places for some time though i must say i did not expect a ltte boycott, anyway in light of the above i will continue to predict the same.

i have calculated several other scenarios which confirm the above conclusion. if any body wants to find out what will happen on different assumptions please feel free to post them as comments, i will try to calculate and post the results back.

if there are any errors in calculations please point them out.