Methodology
This page documents how BigBro Insights collects, processes, verifies, and publishes political data. Every published figure is traceable to an official government source. This page is updated whenever our process changes.
Individual reports also include a report-specific methodology section, data source list, and limitations statement.
Data Philosophy
BigBro Insights operates on a single foundational rule: every data point we publish must be traceable to an official government document. Not to a news report, not to a database that aggregates from unknown sources, not to a figure cited in a Wikipedia article — to a primary official document.
This constraint makes our data narrower than what many political analytics platforms offer. We do not publish vote-share projections, swing models, or demographic estimates. What we do publish is more reliable: official results as recorded by the authorities responsible for conducting elections.
We believe that accurate, boring data is more useful than compelling, unreliable data. Our reports are not designed to generate headlines — they are designed to be correct and verifiable.
Source Selection Criteria
A data source is accepted for use in BigBro Insights reports only when it meets all four of the following criteria:
Published by a government body
The data must originate from the Election Commission of India, a State Election Commission, a municipal authority, or another statutory government body. Third-party aggregators, news archives, and unofficial transcriptions are not used.
Available in official document form
We work from primary documents — Form 20 result sheets, official gazette notifications, Delimitation Commission orders, or published ward result PDFs. We do not rely on news reports or secondary summaries of official data.
Publicly accessible
Every source we use is publicly available without requiring a paid subscription, FOI request, or special permission. We link to the source or note its public access path in every report.
Unambiguous in scope
The document must clearly state what data it covers — the election, the date, the constituency or ward boundary. Documents with unclear scope or provenance are excluded.
Data Collection and Processing
Collection. Data is collected directly from primary government sources — the ECI results portal, official gazette PDFs, GHMC published records, and Delimitation Commission documents. We do not scrape unofficial aggregators or use automated tools on government portals without review.
Transcription. Figures from PDF or image-based documents are manually transcribed and cross-checked against at least one independent reading. Structured digital data (CSV or Excel files published by ECI) is imported programmatically and validated against official totals.
Validation. After transcription, constituency-level vote totals are summed and checked against the official aggregate totals stated in the source document. Any discrepancy of more than ±10 votes (a known rounding artefact in some ECI documents) is flagged and investigated before publication.
Formatting. Data is structured into a consistent schema for each report type (e.g., assembly constituency results use a fixed column structure: candidate name, party, votes received, margin). Formatting conventions are documented in each report's methodology section.
What We Do Not Do
These are not value judgements about other research approaches — they are the specific boundaries of our methodology:
- We do not use exit polls, opinion polls, or survey data of any kind.
- We do not use historical vote-share projections or modelled estimates.
- We do not use unofficial "results night" tallies from news channels.
- We do not impute or estimate data for constituencies where official data is unavailable — we note the gap instead.
- We do not publish voter-level data. All published data is aggregated at the constituency or ward level.
- We do not use social media data, search trends, or any form of sentiment analysis.
Acknowledging Limitations
Every report that has known data gaps, transcription uncertainties, or methodological constraints includes a Limitations & Caveats section. We do not omit limitations because they are embarrassing or because we believe they are minor — we disclose them because our credibility depends on our readers being able to calibrate their trust in our data.
Common categories of limitation in Telangana political data include: historical records from before 2014 (when Telangana was part of undivided Andhra Pradesh) that use different administrative boundaries; ward boundary changes between GHMC election cycles; and minor vote-count corrections issued by ECI weeks or months after initial results.
Update Policy
When reports are updated
A report is updated when (a) an official correction is issued by the originating government body, (b) a figure in the report is found to be incorrectly transcribed from the source document, or (c) a newer official document supersedes the data in the report. Updates are noted with an "Updated [date]" label on the report page.
How updates are versioned
Significant updates increment the report version (v1, v2, etc.). Minor corrections (typographical errors, formatting) are made without a version bump but are noted in the report's revision log. The original published figures are preserved in the revision log for reference.
What triggers a new report rather than an update
If the scope of new data is substantial — a new election cycle, a redistricted constituency, a new ward boundary order — we publish a new report rather than updating an existing one. The older report remains published with a note pointing to the newer version.
Reporting Errors
If you believe a figure in one of our reports is incorrect, we want to know. Please contact us with: (1) the name of the report, (2) the specific figure or statement you believe is incorrect, and (3) the official source you believe contradicts it.
We investigate all reported errors within five business days and publish a correction if the error is confirmed. We do not silently correct errors — confirmed corrections are noted in the report's revision log with the date.