The Team CVoter Corona Tracker: Breaking down the survey methodology
In an attempt to capture the country’s sentiment on the Coronavirus crisis, Team C-Voter has been conducting a daily tracking poll from March 16th 2020 among 18+ adults statewide, including every major demographic. Apart from this, the Economic battery of the poll attempts to capture Indian’s sentiments about the fear of job loss, the extent of job loss, personal savings, overall optimism and other important factors affecting their day to day life.
All the data is weighted to the known demographic profile of every state, including age group, social group, income, region, gender and education levels. Team Polstrat today breaks down the methodology used by Team CVoter in collecting opinion data which ensures an accurate image of the entire geography and demography of India.
How is the survey carried out?
In order to carry out the survey, Team C-Voter uses Computer Assisted Telephone Interviewing (CATI) technology. This is an interactive front-end computer system that aids interviewers to ask questions over the telephone.
Due to the Coronavirus lockdown implemented in the country, Team C-Voter decentralized the entire CATI process to enable all researchers to carry out interviews from their homes. The survey is run in 11 national languages including Hindi, Punjabi, Gujarati, Marathi, Kannada, Malayalam, Tamil, Telugu, Oriya, Bangla and Asamiya. This enables the survey to cater to the local language requirements of the survey in urban as well as rural parts of the country.
(Note: Officially there are 29 languages in India spoken by more than a million people each. However, these 11 languages cover virtually most of India.)
How is random sampling carried out?
In order to understand how the samples are designed, one needs to know the structure of the Indian telephone system. India’s national numbering format is 10-digit for both landline phone as well as cellular phone services.
Firstly, a list of existing telecom circles of the area is made. After this, the first five prescribed digits of all the specified circles are used and then they are extended by five digits randomly. This creates a list of a large number of possible telephone numbers in an area, which are then randomly selected and dialled. Finally, researchers sitting on the system start the process as soon as a call is connected.
This ensures that no particular section of the universe is left out. This is a single frame of surveying and is highly random in its selection of the respondents. India has a very high ratio of teledensity. India has the world’s second-largest mobile phone user base with over 1.16 billion users as of May 2019. Additionally, the CATI process also allows for 100% quality back-check. All the voice calls are digitally recorded in the server and the quality of the calls and the data collected is carefully and continuously monitored throughout the sampling process.
How do we ensure the sample is representative of the entire population of India?
Team CVoter generally takes a nationwide representative sample size of 1200 and for the Corona series as well, the requirement was to call in all states in India to ensure proper representation.
The predictive dialer picks up the number to be dialled from the list of randomly generated numbers based on various telecom circles and digital exchanges in India; covering all the mobile telephonic service providers in the country, ensuring all of India is covered, both geographically and demographically.
The above map shows a typical coverage routine of the Team C-Voter India Omnibus survey on CATI, whereby, each dot marks one respondent. (An omnibus survey is a method of quantitative research where data on a wide variety of subjects is collected during the same interview) Each change in colour marks a different state with a different language/dialect interview. As mentioned earlier, the CATI data covers 100% of the country demographically as well as geographically.
In India, there are 543 Lok Sabha Areas and more than 4000 Assembly Areas across different states. For the tracker, 1200 randomly selected respondents cover more than 1000 random Assembly Segments in the country. In the Gold Standard Quality scenario of n>N, where the number of sampling locations almost equals the number of samples covered. This is as diverse a spread as can be technically possible.
You can read all the infographics and posts based on the Team CVoter #CoronaTracker and the economy battery survey here.
From Polstrat, a non-partisan political consultancy which aims to shift the narrative of political discourse in the country from a problem-centric to a solutions-oriented approach. Team CVoter (Center For Voting Opinions and Trends in Election Research) is a leading international stakeholder research organization with its roots in India.