The findings presented here are based on a Nation-wide survey conducted by Lokniti a research program of the Centre for the Study of Developing Societies (CSDS) Delhi. The data for the survey was collected through interviews amongst 19,062 voters spread across 1120 locations of the 267 randomly selected Parliamentary constituencies in 18 state of India. In each of the states were survey was conducted the Parliamentary constituencies were randomly selected using Probability Proportionate to Size sampling technique (PPS). In each of the selected Parliamentary Constituency, one assembly segment was randomly selected using systematic random sampling technique. Using the same method four polling stations were selected in each selected assembly segment. In each of the 1120 sampled locations, voters were selected randomly from the most updated electoral rolls using systematic random sampling technique. A total of about 39,000 randomly selected voters were approached for the interview of which 19,062 could be successfully interviewed. The interview was conducted face to face at the place of residence of the respondent using a standard structured questionnaire in the language spoken and understood by the respondent. The fieldwork for the survey was conducted during last week of June and first week of July. The investigators could not meet the rest of sampled voters due to various reasons like their non-availability at home during survey days, error in electoral roll, unwell or simply unwilling to be interviewed.
The sample size was not uniform in all the 18 states where the survey was conducted. The sample in big states like Uttar Pradesh (3284), Maharashtra (1514), Andhra Pradesh (1681), West Bengal (1397), Bihar (1735), Madhya Pradesh (1439) is bigger compared to some smaller states like Delhi (301), Haryana (246) Chhattisgarh (337), Jharkhand (631) and Assam (575). In Rajasthan the sample (1691) is bigger compared to the proportionality of the size of the state in terms of population. In Tamil Nadu (786) and Kerala (508), the sample is lesser compared to the proportionality of the size of the state in terms of population. While reading the table of vote shares, it may be appropriate to remind the reader that there is greater confidence on the vote share estimates for parties in States where the sample is bigger, compared to states where the sample is small.
The voting question was asked using a dummy ballot paper and dummy ballot box to ensure secrecy. The estimate of vote shares for different political parties are based on a careful analysis of the respondents’ stated preference of voting for a party as marked on the ballot paper, which carried the elections symbols of all the major political parties in the State. Since all surveys suffer from the problem of underestimation for smaller parties, the estimate of vote shares was made after carefully adjusting the vote share of smaller parties and independents.
The sample is representative at the national level as well as for those states for which estimates for votes and seat is being presented day by day. In the national sample Women comprise 44 per cent, 20 percent of the sample consists of Scheduled Caste respondents and 10 per cent of the sample are Adivasis. The Muslims constitute 11 per cent of the sample and respondents from rural areas are 76 per cent of the sample. These numbers, when compared with actual Census figures by and large reflect the representative nature of the sample, although there is an over representation of Dalits and rural respondents and a slight under representation of Women and Muslims. (see Table)
The survey was designed and analysed by a team of researchers at the Centre for the Study of Developing Societies, Delhi. The team included Anuradha Singh, Dhananjai Kumar Singh, K.A.Q.A Hilal, Kanchan Malhotra, Jyoti Mishra, Nitin Mehta, Shreyas Sardesai and Vibha Attri. The work of computer analysis of the data was completed by Himanshu Bhattacharya. The conversion of vote share estimate into seat for alliance and parties both at national and state level was done by Prof. Rajeeva Karandikar of Chennai Mathematical Institute, Chennai. Prof. Suhas Palshikar and Prof. Sandeep Shastri provided their suggestions during the entire exercise. The survey was directed at the national level by Sanjay Kumar.
The survey was coordinated by scholars from the Lokniti Network: E Venkatesu (Andhra Pradesh), Subhrajeet Konwar (Assam), Rakesh Ranjan (Bihar), Anupama Saxena and Shamshad Ansari (Chhattisgarh), Biswajeet Mohanty (Delhi), Bhanu Parmar and Mahashweta Jani (Gujarat), Kushal Pal (Haryana) , Harishwar Dayal (Jharkhand), Sandeep Shastri, Veena Devi and Reetika Syal (Karnataka), Sajad Ibrahim (Kerala), Yatindra Singh Sisodia (Madhya Pradesh), Nitin Birmal (Maharashtra), Prabhat Mohanty (Orissa), Jagroop Singh Sekhon (Punjab), Sanjay Lodha and Nidhi Jain (Rajasthan), Ramajayam (Tamil Nadu), A.K. Verma, Asmer Beg, and Sudhir Kumar (Uttar Pradesh) and Suprio Basu (West Bengal) .