1. Home
  2. Blog
  3. UPSC

Exit Polls Miss the Mark: Understanding Methodological Challenges and Voter Behaviour


Why is it in the news?

  • On June 4, 2024, several exit polls predicted a major victory for the Bharatiya Janata Party (BJP)-led National Democratic Alliance (NDA) in India’s national elections, with some predicting over 300 seats for the BJP alone. However, the actual results showed these projections were far off.
  • For example, Today’s Chanakya predicted up to 400 seats for the NDA, while Axis MyIndia projected 381 seats. The final tally, however, saw the NDA winning 292 seats.

An Analysis

Understanding Exit Polls vs. Opinion Polls

  • Exit polls and opinion polls differ in timing and methodology. Opinion polls gather voter preferences before elections, while exit polls ask voters about their choice immediately after casting their vote.
  • Some pollsters, like CSDS-Lokniti, conduct “post-poll surveys” at voters’ homes after they have voted. These different approaches can lead to variations in accuracy.

Methodological Mistakes

  • Accurate exit polls depend on several factors, including sample size, randomness of selection, and the survey process.
  • A sample size of around 20,000 respondents, if representative, is usually sufficient for Indian elections. For example, CSDS-Lokniti’s sample size of 19,663 respondents yielded better results than Axis MyIndia’s much larger sample of 5,82,574 respondents.
  • Proper sampling should be random, stratified, and based on the electoral rolls, while respondent weighting should reflect the population’s demographic composition.

Respondents’ Willingness to Disclose

  • Exit polls rely on respondents revealing their choices, but some voters, particularly marginalized groups, may be reluctant to do so. Pollsters who fail to build trust with respondents may end up with inaccurate data.
  • Additionally, undecided voters pose a challenge, and how these voters’ responses are weighed can significantly affect the results.

Weighting and Seat Share Conversion

  • Once surveys are completed, demographic weighting helps align the sample with the population’s composition. For example, if Dalits make up 15% of a population but only 12 respondents in a sample, their responses must be weighted accordingly.
  • However, misrepresenting demographic groups can lead to skewed results, as seen with Axis MyIndia, which underrepresented women voters.
  • Converting vote shares into seat shares is a complex process that depends on regional political dynamics and past election trends. The most common method uses vote share swings from previous elections to predict seat changes.
  • However, this conversion process often fails when elections are close or when pollsters do not fully account for local political factors.

Difficulty of Predicting Close Elections

  • Exit polls tend to be more accurate when elections have a clear winner. However, in closely contested elections like the 2024 Lok Sabha polls, predictions on vote shares and seat tallies tend to be off.
  • Pollsters need to provide detailed methodology, including sample size, margin of error, and demographic representation, for their surveys to be taken seriously. Without this transparency, their accuracy remains questionable.
Subject:

Get free UPSC Updates straight to your inbox!

Get Updates on New Notification about APPSC, TSPSC and UPSC

Get Current Affairs Updates Directly into your Inbox

Discover more from AMIGOS IAS

Subscribe now to keep reading and get access to the full archive.

Continue reading