9 reasons why exit polls in India often go wrong

Exit polls have become an integral part of the Indian political landscape, providing a sneak peek into the potential outcomes of elections. However, despite the attention and anticipation they garner, most exit polls in India often miss the mark — the Lok Sabha election of this year as well as the recently announced Haryana assembly election result are glaring examples.

This inaccuracy is not only a result of the complexities of Indian elections but also due to several practical challenges faced by surveyors. In this article, we explore the key reasons why exit polls in the country frequently go wrong, examining the underlying issues faced by media houses, political parties and voters.

Founder-editor of The Squirrels Bhupendra Chaubey and senior journalist Sudeep Mukhia deciphered the reasons for forecasts going awry right after the results of the Haryana and Jammu and Kashmir elections were out. The following passages cover a few additional points.

1. Financial strain of conducting accurate exit polls

The first and perhaps most critical challenge that impacts the accuracy of exit polls in India is the cost involved. Conducting a reliable exit poll requires a large sample size due to India's enormous voter base. The country has nearly a billion voters and even a modest sample size in the hundreds of thousands can cost media organisations a fortune. Surveyors typically charge between Rs 150 and Rs 500 per surveyed person, making it an expensive endeavour.

For most media houses, especially those with limited budgets, this financial burden is difficult to bear. As a result, many media outlets either reduce the size of their sample or outsource their surveys to smaller firms that might not follow rigorous methodologies. This compromises the quality and accuracy of the data collected, leading to skewed results that do not accurately reflect voter sentiment.

2. Complex social engineering by parties

In India, political parties, particularly the Bharatiya Janata Party (BJP), engage in complex social engineering to maximise their electoral success. They often form alliances with smaller caste groups, creating intricate vote banks that are difficult to predict through traditional polling methods. While surveyors typically assess the vote share of parties, they often overlook the nuanced alliances that play a significant role in determining election outcomes.

For example, in many states, a political party might secure the support of smaller, marginalised communities, which can have a decisive impact in constituencies where these groups make up a substantial portion of the population. Predicting these intricate voting patterns requires a deep understanding of regional and caste-based dynamics, which is often lacking in the country's statistical formulae. As a result, voter examination may fail to capture the full extent of the influence that these social alliances wield, leading to inaccuracies.

3. Questionable credentials of some surveyors

There are psephologists and pollsters that generally make it to the media space or the media is enamoured with them despite their embarrassing track record of proving wrong more often than not. Some of them dabble in poll predictions while being members, associates or force multipliers of certain political parties. Anything they claim must be taken with a bucket full of salt.

View this post on Instagram

A post shared by thesquirrels (@thesquirrelsin)

Then there are others for whom the business of psephology is not their breadwinner. Not having a skin in the game makes their hunches go wrong.

4. Insufficiently probing questionnaires

Another major issue in the country is the design of the questionnaires used by surveyors. Many questionnaires lack depth and do not include probing questions that can help gauge the true intentions and motivations of voters. A superficial approach often leads to respondents giving generic answers that do not reflect their actual voting preferences. The inability to ask follow-up questions also adds to the problem.

In a highly competitive and complex political environment like India's, voter preferences are influenced by several factors, including caste, religion, regional issues and political alliances. A well-designed questionnaire should probe deeper into these factors to ensure that the data collected is comprehensive and representative of the voters’ actual choices. However, most surveys fail to ask the necessary questions that can yield accurate insights.

5. Lack of competence among surveyors

The human element involved in conducting exit polls is equally crucial. Such exercises often rely on poorly paid young men and women employed as field surveyors to gather data. However, these surveyors may lack the experience or expertise required to ask follow-up questions or interpret nuanced responses from voters. The training provided to these surveyors is often minimal, which can lead to misinterpretation of voter responses.

In regions where local customs, languages or caste dynamics play a significant role in voting behaviour, surveyors who are not adequately trained can fail to capture important details. This lack of competency, coupled with an insufficient understanding of local contexts, results in flawed data collection, further contributing to the inaccuracies.

6. Close contests: Issue of marginal constituencies

Elections in the country are known for producing razor-thin margins in several constituencies. For instance, in the recent Haryana assembly elections, many seats were won by less than 500 votes. In such scenarios, the surveyor faces a nearly impossible task. Predicting a victory or loss in these constituencies, where the difference is so marginal, is a challenge that even the most sophisticated polling models may not be able to overcome.

When the vote margin is this narrow, even the smallest discrepancy in the polling sample can lead to significant errors in the predictions. Surveyors need to account for these close contests, but doing so requires extremely granular data collection and analysis, which is often not feasible in large-scale voter studies.

7. Problem with vote share and seat correlation

One common misconception in such exercises is the assumption that a party’s overall vote share will directly translate into a proportional number of seats. However, the outcome of elections in the country depends heavily on how a party’s vote share is distributed across different constituencies. A party that enjoys widespread support across a state but fails to concentrate its vote share in specific areas may end up losing more seats than anticipated.

For example, Party X may receive more votes across the state than Party Y, but if Party Y’s votes are concentrated in certain constituencies, it will win those seats while Party X's distributed vote share may not be sufficient to win many seats. This makes seat predictions based on vote share highly unreliable, as the geography of the vote matters as much as the total number of votes cast.

8. Voter intimidation in certain regions

While not applicable to Haryana, in some regions of the country, exit polls are affected by voter intimidation. The ruling party or local political groups sometimes resort to coercion, warning voters of scary repercussions of voting against them. This is particularly problematic in states where political violence is common or where there is a history of authoritarian control over the electorate.

Voters who fear retribution from the ruling party or local strongmen may either refuse to participate in the surveys or provide misleading answers. As a result, the data collected in these regions is often not representative of the actual voter sentiment, leading to erroneous predictions.

9. The silence of the underprivileged

This might have been true in certain pockets of Haryana where the so-called lower castes did not want to antagonise the Jats who appeared to be overwhelmingly supporting the Congress. This is a key factor that skews the accuracy of exit polls in the country — the reluctance of marginalised communities to speak out against the preferences of the dominant caste in their area. 

In many states, if lower-caste groups decide to vote against the candidate supported by the upper castes, they may fear consequences like ostracism. This fear leads them to either abstain from political inquiries or provide answers that align with the dominant group's preferences, even if they voted differently.

This phenomenon further complicates the data collected in regions with strong caste hierarchies, as the actual voting behaviour of the underprivileged may remain hidden from the surveyors.

Exit polls in the country face numerous challenges that make accurate predictions difficult. From financial constraints and complex social engineering by political parties to flawed questionnaires and voter intimidation, the hurdles are significant. To bring reliability to what is now a laughable indulgence of news television after every election, media organisations must invest more in training surveyors, designing better questionnaires and understanding the intricacies of regional voting patterns. Only then can the country see more accurate exit polls that reflect truly the will of the electorate.