These days one can see a passionate debate raging in the media over rigging of opinion polls by political parties to improve their chances of coming to power. While psephologists deliberate over sample sizes and methods used for the latest opinion polls, an interesting incident from the US Presidential elections, 2012, comes to mind. Most political analysts were expecting a closely-contested election and opinion polls were indicating a probability of a very tight finish. However, Drew Linzer, a political science professor from Emory University, predicted results in favour of Obama with a score line 332-206 in June 2012. When the results were out in mid-November, the score line, not surprisingly, read Obama 332 v/s Romney 206!
That was the power of predictive analytics. The US elections used analytics, not only to predict results, but also to influence voters. Obama?s campaign successfully used a multi-pronged digital strategy with an estimated $52 million spent on online ads versus Romney?s $26 million. Obama had an edge over Romney in his social net worth as well; his Facebook fan page had around 33 million ?likes? while his YouTube channel pages had an astounding 240,000 subscribers and 246 million views.
If the US elections are anything to go by, it may even be possible to hazard a guess on the results of the upcoming Lok Sabha elections in India, as extraordinary as that may sound. Having said that, there are, of course, several differences?cultural, demographic, economical, and social?between the Indian and the US scenario, which could make the exercise somewhat foolhardy. But the interesting point here is that of social media and the power it wields, along with the sure but silent emergence of analytics as the one tool that can make or break a prime ministerial candidate?s prospects in the electoral battleground.
The India story
The 2014 Lok Sabha elections are going to be different from all previous elections in the history of India. It would be the first election in India in which digital social media is expected to make a profound impact. The mobile revolution is almost complete and smartphones have become ubiquitous across the length and breadth of India. Today, political discussions don?t just happen at dinner tables and roadside ?addas?, but also on Facebook, Twitter, and Whatsapp.
Giant cut-outs and hoardings of political leaders and parties now bear a few add ons ?details of the official website, Facebook fan pages, and Twitter accounts. Parties also hold Google hangouts and interactive chat sessions on the Internet. In short, the 2014 elections are going to be fought on the digital turf as well, and why not? In today?s era of innovation and technology, how could electoral politics stay out of the purview of the latest advancements in technology?
Big data analytics, widely touted as the technology of tomorrow, seems to be at the very centre of the electoral analytics. But how can it be a deciding factor in the outcome of elections in India, where more than two third of the population still faces repeated power cuts?
Statistics show: The eligible electorate in the upcoming election is expected to be about 80 crore, of which 16% voters would be first timers. More than 65% of Indian population is below the age of 35. This segment has been touched by the Internet revolution in the first decade of the century and is now being influenced by the social media revolution in the second decade. In spite of the fact that urban population (with access to the Internet) often does not vote, this is still a significant proportion of electorate looking at the vote swing data of the last few elections.
The big data impact
Social media sentiment is one of the key focus areas for big data analytics. It is used by tech-savvy enterprises to better connect with their customers by understanding their sentiments and predicting their behaviour. Political candidates use it as a medium to understand and interact with their voters. Similarly, political candidates may set up their own websites or webpages, essentially becoming a channel for interaction between the candidate and the general public. A huge data pool, including social media posts and interactions are collected from several users and is analysed using text analytics algorithms.
Big data analytics can also be used to fetch data from SMSs, videos, Facebook ?Likes?, tweets, and missed calls to the analytics system for deriving insights. So if a candidate wants to understand public opinion in a particular electorate, he/she can directly ask the general public to SMS their views and get a peek into the overall opinion using big data. Analytics can also be used to fetch public opinion under specific demographic categories. For example, candidates can find out what female voters aged 35 and above feel about an issue as opposed to teenaged voters. Using this data, parties can select ad campaigns and advertising channels to be used to reach out to specific audiences. For example, a 21 year old college student who spends roughly an hour on Facebook each day can be reached using Facebook ads, while a housewife can be reached using relevant mobile apps.
The results arrived at can be used to improve surveys by increasing sample size, improving survey quality, and by analysing free form comments. In a country like India, where a majority of the population is not online, such an approach is best used in pockets of high internet penetration.
Election results in several Indian constituencies tend to be competitive, with last couple of percentage points swinging the outcome. Big data analytics can make a big difference by helping to win a Lok Sabha seat against a defeat that entails another long spell of grass root level building exercise. Big Data can easily connect leaders to the general public, and isn?t an electoral victory determined by how well the leader is connected to the masses.
The proverbial elephant in the room in this Lok Sabha could well be ?Hadoop? rather than the one symbolical of a major political party.
Anand Deshpande
The writer is chairman, MD & CEO, Persistent Systems Ltd
