Ivy May 15, 2014 No Comments
As all eyes turn to India and the outcome of the largest “Unreal Elections”, 2014 goes down in history as the year when social media and analytics shaped the elections and voting trajectory in an unprecedented manner.
From the satta model to analytics
This is a country where the satta (odds) market has always played a uniquely independent and parallel role in election and cricket politics. Even as psephologists took over election analysis for the past two decades, they largely relied upon random sampling and regression analysis. For the 2014 elections however, Psephology has been side swept and analytics taken over. From pre-election surveys to exit poll surveys conducted by 7 separate TV News Channel –market research firms initiatives, the political scenario has been fraught with analysis at various levels.
Social Analytics, Sentiment Analysis, Text Analytics, Web Analytics ….
It is all about analytics.
Unstructured data from social media streams (primarily twitter) have been mashed with data mined from blogs, FB ‘likes’, comments across the blogosphere and internet, on a seamless platform, with historical statistics, for predictive analytics like never before! Sentiment analysis became the yardstick of election trends monitoring agencies and text analytics suddenly became the benchmark of social media analytics.
Whats’s more! The analytics is in a real-time environment, with trends being monitored and analysed continuously to gauge the dynamics of people sentiments, political party camapigns and even the impact of political speeches!
Obviously, this means the agencies conducting analysis have been working with Big Data, both unstructured and structured, to produce trends, insights and analysis on an almost 24 x 7 timeline. A remarkable feat for a country housing a 123.5 crores populace and a 81.45 crores electorate!
With the 2009 exit poll analysis having missed out on almost 40 votes plus in favour of the UPA, there have been no stones unturned for the elections analysis of 2014.
The Analytics roadmap in #Elections2014
To begin with, Hyderabad based analytics start-up Modal Analytics was given the job to build the largest repository of voters. It used a dedicated infrastructure that included 64 node Hadoop, PostgreSQL and servers processing over 8 terrabytes of data, through a 12 month period and processed by 10 data scientists.
Pre-poll opinion surveys from News Channels like the NDTV-Hansa Research and ABP-Nielsen together with other political commentators continued their own analysis unabated, January through April. Political parties used a 24×7 analysis for innovating and refreshing political agenda. TV News channels in their competition to log the most viewership and popularity oddly enough took the same path. Like the political parties they took to managing social media activity to increase viewership. For instance the hype on top trending tweets of the Channel Time Now.
Books like the ‘Unreal Elections’ (with amazing insights into the analytic role-playing amongst political party agendas) and the ‘The Accidental Prime Minister’ in a pre-election release – were other social innovations in the run-up to the general elections.
So what we have been seeing was a continuous mish-mash of social analytics and insights being reflected in political campaigns, TV media promotions and see-saw of voter preferences, influenced by the constant hammering of trends and insights, as the voter was meant to be!
The Analytics Role in #ExitPolls2014
Most of the exit-polls had an error margin of + 3%. Herein the survey method and sampling used decides the accuracy of prediction. While random sampling from actual voter lists is the preferred methodology, market research firms use quota sampling that represent various divergent electorate categories, like professionals, farmers, or first-time voters and age-wise demarcated voters, and more. Times Now used a stratified random sampling method based on previous samples.
Bottom-line: The elections of 2014 can be truly tagged as one ruled by analytics. It has also been debated whether the continuous projection of insights and trends, has influenced the 20% undecided voters or the 12 crore first-time voters. (Has anyone done an analysis on the extent to which the surveys have influenced voting?)
In the meantime, the satta market is on an all time high with top bookmakers across the country announcing the bulk of odds late in the evening on March 10 with the punters a little more confident of the NDA reaching 272 seats. Ironically, today even the bookies rely heavily on the predictive analytics insights presented by the seven TV Channel-Polling agencies!
Disclaimer: The above blog is the author’s personal take and analysis. The aim is to bring home the fact that analytics is not only indispensable in elections, but also has the power to change the voting trajectory in a nation.
Leave a Reply