I used to do an election prediction for every election in TN based on the public mood, newspaper predictions and their cumulative polls but this time, I tried to do the same prediction with the help of data. I have crunched the data from 2014 & 2016 elections and with the help of assumptions, arrived at some numbers. There are 3 factors, which will play an important factor in an outcome of the elections.
- Arithmetic plays important role in making sure that aligned parties of the coalitions getting the votes of the affiliated parties. This arithmetic calculation will ensure minimum votes for the alliance from the core vote bank of the parties.
- chemistry between the cadres/supporters of the individual parties, which may be due to the ideologies , personalities, caste linings etc. in other words, Chemistry ensure the vote transfer of Party A to Party B and vice versa. Without the chemistry, what works on paper may not work in EVMs.
- Last but the most important factor is the overrule mode of the non affiliated or sitting on the fence voters determine the outcome of the elections. Even though the last one is an important factor, unless there is a wave (usually an anti-incumbency) first 2 factors will determine the outcome. And it’s very hard to find the wave in the most of the elections.
Arithmetic
I will be first arriving at the baseline votes for each 39 loksabha constituency for individual parties. So that i can add them to figure out the votes for the alliance.
Baseline
- ADMK. We will take the votes polled for ADMK in 2016 election as the baseline. As the votes it got during the 2014 is part of the Wave( partly due to Modi vs lady rhetoric of Jayalathita and anti-incumbency of UPA 1 & 2), i’m taking 2016 which is much of a no wave election.
- DMK. We will take the 2016 elections as same reason as listed above
- INC. DMK + votes in 2016 includes Congress votes
- BJP, PMK, CPI, CPI(M), KMDK – All these parties participated in the elections without any alliance either in 2014 or 2016. Whenever they contested alone, that is their base data
- VCK,DMDK,MDMK – They were party of DMK alliance in 2014 and MNK in 2016. So VCK = DMK + (2016) – DMK(2014) + INC (2014) but this logic gives us Negative votes. Does VCK brings negative votes? I think more plausible argument is DMK got more votes from unaffiliated in the 2016 than 2014. So we are not assigning any votes for VCK. Same for DMDK & MDMK, not able to get plausible data as they are part of some alliance for last 10 years and I feel DMDK has severely weaken from its peak in 2009. I assume, DMDK = MDMK + VCK. So these should cancel out each other.
- AMMK – This is TTV Dhinakaran’s Splinter group from ADMK. This is their first election ,so we don’t have any base line data and here is where we trend from the usage of strict historical data to projections. There are reports that his party is particular more stronger in Kaveri delta districts and some southern districts but unfortunately we don’t have the data or the ground presence to predict that. So I presume, He will split 5 to 10 % of Admk votes. I will allocating 0% in 1st set, 5% in 2nd set and 10% in 3rd set to his party.
- MAIAM – This is kamal hasan’s new party. How much will he get ? And where will he get those votes only in Urban areas or also in the rural areas. I assume him to get 5 to 8%. There is one more important question is from where he will get his vote share. Will he split the DMK or ADMK votes ? . There are usually disgruntled voters, disillusioned with the existing choices ready to vote for the 3rd alternative in tamil nadu. So most of the voters wouldn’t have voted for DMK or Admk in any point, they were ready to support the 3rd alternative in all elections. I assume 75% of those votes from outside of the 2 parties and 25%(2% of total vote) comes from the those parties. Since there will be an anti-incumbency against the state and central governments, I may get tempted to assume more of the 2% comes from Admk + but his policies or the general perception of the MAIAM and Kamal in seems to pull the voters of the DMK than ADMK. So I assume they will get cut off equal votes from both the combinations and cancel it out or at most it will eat away 1% of the votes from DMK. So, if MAIAM can play a spoil sport, if DMK alliance is losing the constituency by 1% of less of the polled votes in an seat.
- NTK – I don’t think, yet they are in the position to get votes from both these alliances but i believe Seeman will be growing very election.
Chemistry
- I Feel ADMK-BJP and DMK-Congress is a natural alliance, so vote transfer shouldn’t be problem. Only where i will be interested in looking at the vote transfer post poll scenario is DMDK and PMK are contesting. From the moment, Vijayakanth contested in Northern districts (Both parties are stronger here), there is some friction between the parties. Even there are some issues during the 2014 LS alliance in the cadre level.
Before we proceed to the predication. Some of the reasons why these assumptions will go for a toss.
- This is the first election after the demise of both Karunanidhi and Jayalalitha. There are certain percentage of votes for and against these people. We don’t what is going to happen to those Negative and Positive votes. Will the vote bank who were voting for those faces, will keep voting for the same party even after the demise of their colossal leaders? We don’t know.
- We are Considering AMMK,MAIAM & NTK as marginal players. We are discarding them to the 3rd place even before the election has started. Isn’t this unfair? Yes, but that’s how politics have played for 50 years in Tamil nadu. There might be a surprise for us there. TTV Dhinakaran may be able to pull out the RK nagar magic once more in some of the constituencies. He may do the ultimate MGR trick on ADMK. Kamal’s party has put some prominent persons have candidates, they have pull above their weight and get make more inroads than expected ( still I don’t think they will even have a 2nd place finish) . If Seeman’s brothers can transfer all their Social media presence into votes, Sky is the limit.
- There is clear uproar against the TN rulers 2 years back. If that was the case even now, we don’t need any predication or data analysis, it would have been clear 35 + for the DMK + . But the perception has been changed, People have subdued( that’s what I hear), It has pulled off an impressive analysis on arithmetic terms. Is there a wave or not ? We will know on May 23rd
- I’m not in Tamil nadu right now. So all my perception of people mood is from the Family & Friends, Media and Social media. All these things are skewed to our previous bias. So we keep accumulating data supporting our thesis. Popular waves can be picked up when you talk with multitude of people in Grocery shops, barber shops, buses etc. You need to be in ground to feel it. So I’m not adding any percentage for anti-incumbency to DMK alliance, Even though the goBackModi was popular. This is the most important reason, I predict the arithmetic calculation will not work.
Methodology
Before going to the predications, will explain how did i do this. I calculated the votes of the 234 assembly segments from 2014/2016 based on the base line explained above and played around it and arrived at the outcome. I created a console application using C# using the data from the official election commission site as input.
Predictions:
Based on the arithmetic alone and assuming sitting on the fence voters vote as they were doing previously used dataset , Here we go,
- When AMMK gets 5%
- DMK 6 to 10
- NDA 29 to 33
- When AMMK gets 10%
- DMK 9 to 12
- NDA 27 to 30
- When AMMK gets 15%
- DMK 12 to 14
- NDA 25 to 27
If you don’t believe this numbers, you are not alone. Even, it was a quite a shock for me. But what i expect is actual numbers will be largely different due to the unaffiliated votes. and I also believe, number of unaffiliated percentage is steadily increasing and it should be harder & harder to judge the elections based on the arithmetic alone going forward.
Sitting on the Fence:
Sitting on the fence voters may vote for new parties but they wont vote for ADMK/BJP/DMDK/PMK alliance and 5 to 10% of votes should move away from them and good chunk of them should go to the DMK alliance. Based on this numbers, i believe wherever there is less than 1 lakh vote difference (this is from my analysis tool) between the alliances, DMK alliance will win.
Based on my assumption of how unaffiliated voters vote, i predict
- DMK 30 to 32
- NDA 7 to 9.
I have 4 different numbers, each number has solid reasoning behind that. Lets see how much votes AMMK gets and how much unaffiliated voters will prefer DMK.
So i will write one more blog post election results using the votes polled in individual assembly level data and see what happened.
view the world with me 