Elections in India continue to be easily predictable if one does not get too involved with them and get misguided by meaningless statistics. Using fundamental principles, analysts are increasingly able to predict elections as well as polls. Surjit Bhalla is a case in point. He managed to get Bihar very precisely and performed really well on the elections in 2016. The only cheap thrill is that I beat him on both Assam and Kerala where I predicted landslides/ Comfortable wins while he was much more conservative. On Bengal, both of us predicted wins while he predicted a bigger win. On TN, I predicted a hung assembly while he predicted a big win for Jayalaithaa, neither of which happened.
Looking at the structure of the analysis approach, what are these principles that are making it easier to predict elections?
a. Leadership Comparison - As our multiple analysis pieces on Bihar, Bengal and Assam illustrate, leadership ratings alone could determine the result of the election. Across States, parties must constantly track this metric versus other Leaders in the State. Leadership is a function of numerous attributes - Empathy, Trust, Ability of articulate, decisive etc. Swing voters in particular tend to move towards the party with better leadership ratings at the fag end of the campaign. Sonowal, Mamata, Vijayan+Achuthanandan were significantly ahead of their rivals, Jaya was marginally ahead of Karunanidhi/Stalin.
b. Satisfaction with the Govt - Called anti-incumbency by many, It is a simple question - How Satisfied are you with the Government? Governments must score a minimum of 50% +ve and a probably a number higher 60% to be very comfortable about being re-elected. Satisfaction ratings are primarily related to incomes, inflation, perceived corruption, law & order, necessities like power/water/roads, healthcare and numerous other factors that are related to their daily lives. Assam and Kerala Governments had high dissatisfaction while Bengal and Tamil Nadu Governments had average Satisfaction.
c. Campaign Management - How well are campaigns running their election campaign? This is demonstrated by the direction in vote share trends as well as performance in key voting segments. As we indicated in our analysis of the Exit Polls two days ago, one of the reasons we thought that the AIADMK could have beaten (contrary to exit polls) DMK was because it seemed to outperform DMK amongst poor/lower middle class as well as female voters. All of these are big voting segments. It is important for campaigns not just to track top level shares but also on how each individual segments were reacting to the messaging. The best campaigns were run by LDF in Kerala, BJP in Assam and DMK in Tamil Nadu. The DMK campaign managed a massive 9% swing versus the Lok Sabha election though that was not adequate to win the election.
d. Index of Opposition unity - Dr. Roy invented this term and has a whole host of empirical theories around swing and vote share needed to win an election. My view is simpler, higher the opposition unity, better their chances to win an election. But higher IOU alone is not enough if the ruling party is doing well on Leadership, Satisfaction and running a good campaign. IOU was somewhat strong in all the four States.
In my view, these 4 metrics are good enough to estimate the results of an election. However, one must remember that the weights of what is more or less important between the changes by State and in my view is a function of the unique circumstance of the State. Probably, we will have a firmer understanding of this in the future. Assam ticked all the boxes for a BJP win, Kerala and Bengal 3/4 out of the boxes while Tamil Nadu (AIADMK) ticked 2/4 boxes making TN the least predictable of the elections.
Interestingly, the Prashant Kishor campaign in Punjab is working hard on three of these variables (Leadership, IOU and Campaign Managment) and trying to exploit the anti-incumbency in the State.
This analysis is an evolution of my hypothesis a year ago called the 4-laws of parliamentary elections. The emphasis has evolved from just per capita income growth to a wider set of variables, from tenure to leadership ratings and satisfaction. Campaign management is a new variable as campaigns introduce more science and art to their election strategies. Therefore, the current set of variables would also probably evolve even further over time and needs more research.