That family group chat will fight about exit polls from now until the last EVM battery is unplugged. The 2026 Indian state assembly cycle is not one election. It is four states and one union territory—Tamil Nadu, Kerala, West Bengal, Assam, and Puducherry—with 824 seats between them and roughly 188 million people on the rolls when those places last voted. The assemblies complete their terms in the first half of 2026, so polling has to happen before the current governments' clocks run out. One rule shapes the exit poll part of the story immediately: Section 126A of the Representation of the People Act, 1951 forbids conducting an exit poll and publishing or disseminating its result during the period the Election Commission of India (ECI) notifies. That period runs from the start of the first poll until 6:30 pm on the final polling day. I have spent years explaining randomised controlled trials to relatives; explaining this statute to my family group is the same job, except louder.
Exit polls are not the election. They are a sample of people who have already voted, asked on their way out of a polling station, then weighted, modelled for who refused to answer, and converted into an estimated vote share. The polls are useful because India's first-past-the-post system makes seat conversion brutal and unstable. A small error in vote share can move dozens of seats. In West Bengal in 2021, the Trinamool Congress won 213 of 294 seats and the BJP won 77; a poll that missed the scale of that gap still might have correctly named the winner. That distinction—direction versus magnitude—matters in a family chat at 7 pm on counting day.
Who Votes, When the Polls Close, and What the Law Does With Exit Polls
Any Indian citizen aged 18 or older on the qualifying date can register in the constituency where they ordinarily reside. Elections in this cycle will be held in phases set by the Election Commission. Tamil Nadu, Kerala and the union territory of Puducherry have recently gone to the polls in a single phase. West Bengal and Assam have been multi-phase exercises; West Bengal voted in eight phases in 2021 and Assam in three. The reason for that difference is simple: paramilitary forces and central police units cannot be in every district at once, and the ECI tries to put enough personnel on the ground before it opens a single polling booth. The consequence for viewers is also simple. Nobody may broadcast a legitimate exit poll after West Bengal's first phase and then claim it is merely a preview. The legal embargo keeps the field dark until the final phase is done.
| State/UT | Seats | Incumbent government heading into 2026 | Recent phase pattern |
|---|---|---|---|
| Tamil Nadu | 234 | DMK-led alliance under M.K. Stalin | Single phase in 2021 |
| Kerala | 140 | Left Democratic Front under Pinarayi Vijayan | Single phase in 2021 |
| West Bengal | 294 | All India Trinamool Congress under Mamata Banerjee | Eight phases in 2021 |
| Assam | 126 | BJP-led alliance under Himanta Biswa Sarma | Three phases in 2021 |
| Puducherry | 30 | All India N.R. Congress-led alliance under N. Rangasamy | Single phase in 2021 |
In Kerala, 2021 was exceptional: voters returned the Left Democratic Front for a second straight term, breaking a long cycle of alternation between the LDF and the Congress-led United Democratic Front. In Tamil Nadu, the DMK-led alliance won 159 of 234 seats in 2021; the incumbent AIADMK finished well behind. Those result patterns are the baseline, but a baseline is not a forecast. Each state has its own contest over local government performance, urban services, health, school education, land and taxes—the things voters actually meet at the kitchen table.
The Election Commission of India enforces the ban under Section 126A. The penalty for breaching it is imprisonment for up to two years, a fine, or both penalties at once. The legal logic is not censorship for its own sake: an exit poll published after an early phase in West Bengal or Assam can influence what voters do in later phases. In single-phase Tamil Nadu, Kerala and the union territory of Puducherry, the same rule produces a cleaner timetable: polls can run after 6:30 pm on polling day, once every booth has closed. Multi-phase states simply have to wait longer.
Here is the part most broadcasts don't spell out. Exit polls are not the final result, and neither are post-poll surveys. A post-poll survey uses a larger questionnaire, often through interviews at home after voting day, while an exit poll is a short interview outside the booth. Both are statistical estimates. The Lokniti-CSDS post-poll surveys publish sample sizes, survey dates, question wording and methods for good reason: without those details, a number on a TV screen is just a number. India's multi-party alliances also make vote share a poor predictor of seats. In a three-cornered contest, a candidate can win a constituency with far less than half the vote, and vote share can be spread so unevenly that a 2-3 percentage point polling error produces a completely different assembly.
If you are reading exit poll coverage in real time, ask these five questions. They are the same questions a decent analyst asks before putting a number in a table.
- What is the sample size, and how were polling stations selected?
- What is the margin of error, and does the graphic say so?
- Was the poll conducted across all phases or only the earlier ones?
- How are undecided and refusing voters handled?
- Does the organisation have a published methodology and a track record across the previous three elections?
My kitchen-table version is less elegant. An exit poll is a slice of the family's food preferences gathered at the front door after breakfast, then used to predict what the house will eat for dinner. The slice matters only if you know who opened the door, how many people you missed, and whether the person who answered has changed their mind by 7 pm. The error is not a sign the thing is useless; it is a sign the model deserves less reverence than the final count.
Counting day follows the final phase, and the ECI usually sets a single counting date across a state. Postal ballots are counted first. Then the electronic voting machines are opened and the votes tallied. At randomly selected polling stations, VVPAT paper slips are verified against the electronic counts. Results are updated round by round on the ECI's results portal, which is a better primary source than a news channel's ticker when two channels are disagreeing about the same constituency. The returning officer declares the winner only after the VVPAT verification for that constituency is complete.
These five elections will not just determine chief ministers; they set the party arithmetic for the next general election. If the BJP holds Assam and makes inroads in Tamil Nadu or Kerala, that is a different national story than if it holds only Assam. If Mamata Banerjee's Trinamool Congress survives another eight-phase election in West Bengal, it strengthens the only state-level party of that size outside the two national alliances. But the immediate consequences are more local: budgets, recruitment for state services, school management, public health, and the speed of disaster relief all shift with the result. Voters who care about those things should check candidate affidavits and past performance through the Association for Democratic Reforms, not rely on exit polls to do that work for them.
In my family, the only prediction anyone believes is the one made after the result. That is not a bad policy. The exit poll can tell you where the evening is leaning, and the law tells you when you may be told. The rest is waiting, watching the Election Commission's official figures, and remembering that 824 seats are not a poll. They are an election.
Photo by Big G Media on Unsplash