A scorecard of living-conditions indicators for India's Muslim
population, with Hindu and all-India comparison baselines on every metric.
Take a quick tour
New here? This short walkthrough shows how the dashboard is organised and how
to read each card: its Muslim figure, the Hindu and all-India baselines, and the
deep-dive tabs.
Why this exists
The Sachar Committee report (2006) remains one of India's most-cited
assessments of Muslim socio-economic status. Nearly two decades later, few
public resources keep that lens current: the data exists, scattered across the
Census of India, NCRB, NFHS, AISHE, PLFS, and other primary sources, but it
is not assembled in one place, with comparison baselines, in a form that a
non-specialist can read.
muslimdata.in is an attempt at that single place. Every number is sourced
from a primary public dataset, traced back to its original file, and shown
alongside the Hindu and all-India comparison so the reader can judge the gap
without having to compute it.
What's on the dashboard
23 indicators across six themes (population, health, education,
employment, representation, justice), drawn from 25
primary sources. Every card covers one indicator. It shows the latest
figure for India's Muslims, where that figure ranks across religious
communities, and how it has shifted over time when the source has more
than one survey round.
Click any card to open a larger view with the full methodology notes for
that indicator, including period covered, cut-offs, sample design, and known
caveats. Every card also links to the downloadable CSV of the underlying
data.
How the data is handled
Every published value follows the same four-stage pipeline:
Source. The original file (Census table, NFHS chapter PDF, NCRB
Crime in India volume, AISHE all-India report, PLFS unit-level data,
candidate-affidavit compilations, etc.) is downloaded from the publishing
agency, fingerprinted with a SHA-256 checksum, and archived in this repo's
/sources directory so any reader can reproduce the extract.
Extract. A per-source script reads the original file and produces
a cleaned table.
Canonicalise. Cleaned tables are reshaped into a long
schema-validated CSV per indicator (one row per geography × religion ×
year), stored in /canonical. These are the files linked from
every card footer.
Publish. The dashboard you are reading is generated from those
CSVs, with no live API calls. The build re-runs whenever the canonical
files change, and the deployed site is rebuilt automatically on every push
to the main branch.
The point of this discipline is that any number on the page can be
independently verified, end-to-end, by anyone with a browser and a
spreadsheet program.
Primary sources
The 25 sources currently feeding the dashboard, with their
publishers and release cadence:
PRS Legislative Research (analysing ECI affidavit data) · cadence: per election
Using and citing this data
The data is free to reuse. Every card has a Download CSV link for that
indicator, and the complete machine-readable dataset lives in the
open-source
repository under /canonical. The code, the manifest and the
compiled CSVs are released under the MIT licence.
If you quote, chart or build on a figure, please credit two things:
The original source. Every number comes from a primary government
dataset, named in that card's methodology and in the list above (the Census of
India, the National Family Health Survey, the Periodic Labour Force Survey, NCRB,
the National Sample Survey, and so on). That underlying data belongs to the
publishing agency, so credit it.
muslimdata.in, for compiling, computing and presenting it, with a link
to the specific indicator page so a reader can check the method and the
provenance for themselves.
A citation might read:
"Muslim share of population", muslimdata.in, 2026,
muslimdata.in/m/pop-share/.
Compiled from Census of India 2011.
Two small requests. Link to the specific
/m/<indicator>/ page rather than the homepage, so the method,
the year and the caveats travel with the number. And read the card's methodology
note before quoting a figure: it states which way the metric reads (lower is not
always worse), the survey's limits, and how current the data is.
Ask an AI assistant
The dashboard also answers questions through an MCP server (Model Context
Protocol), so an AI assistant can query these numbers directly and cite the
right page. In Claude: Settings, then Connectors, then Add custom
connector, and paste https://muslimdata.in/mcp (no sign-in
needed). Assistants that support MCP by URL can use the same address. The
server is read-only and serves the same canonical figures as this site;
machine-readable summaries also live at
/api/catalog.json.
Limitations
The most recent census is 2011. The 2021 round has been indefinitely
delayed, so every Census-derived indicator (population share, sex ratio,
literacy, urban share, district concentration) is dated and will remain so
until the next round is released.
Several Muslim-disadvantage indicators commonly cited in news coverage,
including share of central civil services intake, share of judicial
appointments, and share of corporate-board seats, are not yet on the
dashboard because the data is not publicly released in religion-disaggregated
form. Where civil-society compilations exist, we are evaluating them for
inclusion.
NCRB's "communal incidents" series is a count of police-registered
rioting cases under IPC Sec.147-151. Several states stopped recording
'communal' as a separate sub-category after ~2017, which deflates the
official totals. Civil-society compilations report substantially higher counts.
For metrics like the MLA share, religion is not officially tabulated
by the Election Commission. Numbers rely on candidate-affidavit
classification by journalists.
Contributing or reporting an error
The repository is open source on
GitHub.
Issues and pull requests are welcome:
If a number looks wrong, open an issue with the metric, the page URL,
and the number you expected. The pipeline preserves the full source chain
so a correction is traceable to its origin.
If you want a metric added, open an issue with the indicator definition
and the publicly accessible primary source. We prioritise metrics where
religion-disaggregated data is officially released.
If you spot a stale Census date or out-of-date methodology note,
flagging it as an issue is the fastest fix.