Assembly Election 2017: Geographic proximity influences voting behaviour reveal poll data of last three elections

The Assembly Elections in the five states of — Goa, Punjab, Uttarakhand, Manipur and Uttar Pradesh — are nearing their conclusion with the final results just a couple of days away. As the results on the counting day of 11 March will sieve out the winners from the vanquished, it would also simultaneously give an idea on voter preferences that would in all likelihood dominate the trend in the run up to the 2019 General Elections to Lok Sabha.

Representational image. Reuters

Representational image. Reuters

As exciting as the exit polls that are going to be unveiled on Thursday are, there's much that can be determined from the voting choices people made in the past elections. We at Firstpost did some number crunching of our own, analysing the poll data from 2002, 2007 and 2012 Assembly elections in Goa, Punjab, Uttarakhand, Manipur and Uttar Pradesh. We studied and analysed victory margins and seats won by different parties in the five states to get an idea of how they voted and realised that Assembly constituencies which are geographically proximate tend to copy each other’s behaviour in so far as voting is concerned.


We studied and analysed winning margins from the preceding three Assembly elections in the five states and categorised them into four buckets with 1 being the high margin of victory and 4 being the lowest margin of victory.

Based on the outcome of the previous three elections, we also calculated average victory margin. The lower the average victory margin, the tighter is the election in that particular constituency. We identified the constituencies where it happened, and grouped constituencies with similar average victory margins in the same colour. Geographically proximate (need not be geographically connected) constituencies form a cluster.

A probability was assigned in the next election based on the average victory margin. These probabilities are party-specific and not individual specific. Using the above method we came to a conclusion that the probability of winning the next election is higher for a party in higher victory margins and vice versa. The above data gives an interesting insight into how voters tend to vote. We realised that voters tend to vote decisively or almost equally. And, if they are geographically proximate (need not be geographically connected), they also tend to copy each other's behaviour. What this means is constituencies are likely to behave similarly within a cluster. We also realised that the chances of them changing their behaviour are high if victory margin is high.

NOTE: Select from visible layers to access maps.

Punjab: Past results and voting clusters

Punjab: Victory Margins

Goa: Past results and voting clusters

Goa: Victory Margins

Uttar Pradesh: Past results and voting clusters

Uttar Pradesh: Victory Margins

Uttararkhand: Past results and voting clusters

Uttararakhand: Victory Margins

Manipur: Past results and voting clusters

Published Date: Mar 09, 2017 02:12 pm | Updated Date: Mar 10, 2017 11:43 pm