One seat, a short schedule and far more attention than a general election gives it
Bharat Survey is a by-election survey agency for candidates, parties, researchers and newsrooms who need reliable numbers from a single constituency in a few weeks. A bye-poll is usually held within six months of a vacancy, often while the rest of the state is not voting, so every campaign concentrates on one seat. We design the sample for that seat, interview with GPS-verified field teams and report with the margin of error stated.
A by-election is not a smaller general election. Turnout, candidate choice and the reasons people vote all shift when only one seat is in play.
Under Section 151A of the Representation of the People Act, 1951, the Election Commission generally holds a bye-election within six months of a seat falling vacant, unless less than a year of the term remains. That gives campaigns a compressed calendar and a seat that suddenly receives state-level leaders, media crews and money. A survey plan borrowed from a general election misses those effects.
We treat the constituency as the whole universe. Polling stations are grouped into clusters by area and past turnout, households are selected at random within them, and quotas are checked against Census 2011 and the published electoral roll totals, never against personal voter data. The result is a seat estimate with a stated margin of error and clear sub-group sizes.
We add questions that a general-election survey rarely needs: whether the voter knows why the seat is vacant, how much the ruling government's record matters for a single seat, and whether the voter plans to turn out at all when the result cannot change who governs. Turnout intent is often the most useful number in a bye-poll.
Schedules vary, so treat this as a typical shape, not a promise. We fix the dates against the ECI notification for your seat.
Map polling stations into clusters, draft the questionnaire and run a baseline read on issues and candidate recognition before tickets are final.
Once candidates are known, a full pre-poll round measures preference, second choice and turnout intent across the seat.
A shorter tracking round shows movement. Fieldwork and any publication follow the Section 126 silence period and ECI directions.
Interviewers work outside a random sample of stations. Results stay embargoed until the Section 126A window notified by the ECI closes.
Each study answers a different question. Many clients commission two of them for the same seat.
Use this to match the study to the decision you have to take.
| Study | When it runs | Decision it supports | Typical sample for one seat |
|---|---|---|---|
| Winnability survey | Before candidate selection | Which name to field | Sized per shortlist, often 800 to 1,200 |
| Opinion poll | During the campaign | Where to campaign and on what | 600 to 1,500 |
| Pre-poll tracking | Two or three rounds | Whether the campaign is moving voters | 400 to 800 per round |
| Exit poll | Polling day | Early read on vote share after voting | Depends on stations sampled |
Sample ranges are planning guides, not quotations. Final sizes depend on the margin of error you need and how many sub-areas must be reported separately.
Campaign volunteers can use our offline Android app for booth-level tasks: assigned areas, daily visit counts, polling-day status posts and issue alerts to the campaign office. It runs without voter-roll uploads and does not track individual voters.
The same app runs your own short surveys when you want volunteers to record doorstep issues. We mark volunteer-collected data separately from our sampled survey, because the two answer different questions.
Where a candidate wants more than a survey, we run door-to-door canvassing coverage and a small war room dashboard for the seat. Rallies, advertising and social media are outside our scope.
All political research and field services.
Learn moreSampled pre-election polls with method notes.
Learn morePolling-day interviews and post-poll analysis.
Learn moreField partner for newsroom polls.
Learn morePolitical survey services in Hindi.
Learn moreTell us the seat and the polling date.
Learn moreStraight answers to the questions people ask most about this topic.
Once the seat, questions and sample are agreed, fieldwork usually starts within a week. The questionnaire needs translation and a short pilot, and polling-station clusters have to be mapped. Trying to skip those steps saves a day or two and costs far more in accuracy.
For a seat-level estimate most studies use between 600 and 1,500 interviews. If you need separate figures for blocks, wards or social groups, each of those needs enough interviews on its own, so the total rises. We share the calculation before quoting.
Not while the exit poll ban notified by the Election Commission under Section 126A is in force. For bye-elections held alongside other polls, the window can stretch until the last phase ends. We hold results until the notified time.
We take projects from candidates, parties, media and researchers on a neutral basis, keep each client's data separate and never share one client's results with another.
Send the constituency name, expected polling window and the decision you need to take. We will reply with a sample design and timeline.
