Abstract:

Most Opinion Polls in the West, which attempt to forecast election results, rely on a representative sample of voters and a frank reply to the question: "If the Elections were held tomorrow, who would you vote for? It is the author's contention that when this methodology is followed in India, it often leads to very erroneous results. The problems peculiar to India are as follows : 1) India is a vast country with 75% of the people residing in rural areas - across 600 villages. It is virtually impossible to obtain a representative sample of voters in the conventional sense. 2) India’s democracy is peculiar in that from its independence in 1947 the country has been ruled without interruption for 30 years by one party, the Congress. When asked the party the respondent would vote for, the Congress is almost like a default option among those who have not given much thought about the election and/or tire unlikely to exercise their franchise. Only 50-60% actually vote though almost everyone gives an opinion to the interviewer. Another reason biasing the response to the simple Voting Intention question is what has been termed as the lying factor. In several parts of India, the political parties are known to have a nexus with criminal elements, and the respondent sometimes mentions what he feels is the most diplomatic answer, although he may vote quite differently under the secrecy of the ballot box. This paper sets out a methodology that takes these factors into account to develop a forecasting model that has been validated to be more accurate under these circumstances.