Exit polls, wrong predictions

Methodological Rigour
• CSDS-Lokniti uses a nuanced approach, conducting face-to-face interviews with sampled voters the day after voting.
• The method ensures a close match between the demographics of the sample and actual demographic trends for major groups and communities.
• The approach is based on the probability proportional to size (PPS) method, which adjusts the probability of choosing a particular constituency according to its electorate.

Misleading Euphoria
• All three elections — Chhattisgarh, Madhya Pradesh, and Haryana — shared a common perception of a Congress wave, leading to exaggerated predictions of Congress victories.
• CSDS-Lokniti does not make seat-by-seat predictions and the real gap was not in Congress’s predicted performance but in underestimating the BJP’s ability to secure votes.

BJP’s Winning Formula
• The BJP counter-mobilised voters in response to perceived anti-incumbency and built voter coalitions away from the limelight.
• Differential voter turnout may have complicated the assessment of vote shares.

Are voters lying to pollsters?
• The theory that voters lie to pollsters is far-fetched; if voters were misleading surveyors, pollsters would not have accurately assessed vote share estimates in other elections.
• There are potential factors such as social desirability bias and voter fatigue, leading to skewed responses.

In conclusion, the divergence between vote share estimates and actual vote shares in recent elections indicates the need for caution in election polling and forecasting.