(The author is the founder-director, CVoter International. He tweets @YRDeshmukh. Views expressed are the author’s own. The Quint neither endorses nor is responsible for them.)
*Survey Methodology: Current projections based on CVoter daily tracking poll conducted during the last seven days among 18+ adults statewide, including likely voters, details of which are mentioned right below the projections as of today. The data are weighted to the known demographic profile of the states.
Sometimes the table figures do not sum to 100 due to the effects of rounding. Our final data file has the socio-Economic profile within +/- 1% of the demographic profile of the state. We believe this will give the closest possible trends. The Tracking Poll Fieldwork covers random probability samples during the last 7 days from the release date. The sample spread is across all assembly segments in the poll-bound states. MoE is +/- 3% at macro level and +/- 5% at micro level with 95% confidence interval.
The data collection started on July 1st and continued until November 9th. Technically speaking, we Conducted 17 weekly waves of 7000 respondents each. But by splitting that sample into 1000 interviews every day, and making one rollover file of the last 7 days every day, we were able to analyze 40 waves of 7000 samples each.
In other words, this gave us an opportunity of analyzing fresh data daily for the remainder of the campaign through Election Day. For the analytics, we are using our proprietary algorithm to calculate the Provincial and Regional Vote share based on the Split-Voter phenomenon.
We have used this algorithm to correctly predict many demographic measurements across many democracies. As mentioned; this is a thorough random probability sample, and we are ensuring a proper representative analysis by statistical weighing of the data to make it representative of the local population as per the latest census figures.