About Data Pandas

Public data, made comparable, cited every time.

An independent reference for the statistics that describe the world: built from primary sources, kept up to date, and free for anyone to read.

Independent No advertisers, no paywalls
License CC BY 4.0
Updated Against the issuing agency's cadence

A reporter's shortcut, kept in the open.

Data Pandas was founded in 2023 by a small group of data journalists who kept reaching for the same public datasets in their reporting, and kept running into the same wall: the online write-ups were years out of date, and the government sources they pointed back to were dense, fragmented, and unfriendly to navigate.

So we started doing the work ourselves, pulling the figures straight from the issuing agency, normalizing the units, and writing up plain-language rankings that linked back to the original source. The first pages were for our own reference. Other journalists, researchers, and curious readers started using them too.

Since then we've maintained those datasets release after release, expanded into new topics, and begun publishing our own original research: surveys and analyses produced in-house when the public record doesn't cover the question we want to answer.

"

If a figure is worth quoting, it's worth citing the agency that issued it. That's the whole charter.

§ 02 · What we publish

Two kinds of pages.

Maintained rankings drawn from public data, and original research we run ourselves when the public record doesn't cover the question.

01

Maintained public datasets

Rankings built from national statistical offices, intergovernmental organizations, and peer-reviewed registries. We pin each release, refresh on the issuing agency's cadence, and keep the previous edition in the archive.

02

Original research

Surveys and analyses produced in-house when the public record doesn't cover the question. Methodology and raw responses are published alongside the writeup, under the same citation rules we hold public data to.

§ 03 · Editorial principles

Five rules. We will not break them.

The editorial standards every page is held to, every time. Posted in the open so you can hold us to them too.

01

Primary sources or it didn't happen.

Every figure is traceable to the agency that issued it. We don't aggregate aggregations; we don't paraphrase a press release. If the original source goes offline, we mark the value as withheld, never silently fill it.

02

Methodology in the open.

Each ranking page tells you exactly how the figure was computed: the sample, the year, the unit, the source, and the assumptions. If two reasonable methodologies disagree we publish both and note where they diverge.

03

Slow data beats fast data.

We update against the publication cadence of the issuing agency, not the news cycle. A figure published annually gets updated annually. We will never invent a daily refresh just to look fresh.

04

Independent of advertisers and sponsors.

Data Pandas is reader-supported and bootstrapped. We accept no sponsored rankings, no underwritten datasets, and no editorial input from outside the masthead.

05

Free to read, cite, and reuse.

All figures are licensed CC BY 4.0 with proper attribution to the upstream agency. The only thing we ask is that you cite where the number actually came from.

§ 04 · Methodology

From release notice to ranking page.

A typical dataset takes about a week. Here's what happens in between.

  1. 1

    Identify the question and the issuing authority

    We start with the question a reader is likely to ask, then identify the single agency most authoritative for it: USDA NASS for U.S. crop production, FAOSTAT for international agriculture, WHO GHO for global health.

  2. 2

    Pull, parse, and version-pin the source data

    Each dataset is fetched directly from the agency's published release (CSV, API, or parsed table), pinned to a specific edition, and stored alongside its release notes. We never modify the underlying numbers.

  3. 3

    Normalize units and reconcile geographies

    Different agencies use different units, vintages, and geographic codes. We map every figure to a canonical unit (metric, USPS, ISO 3166) and document every conversion.

  4. 4

    Editorial review and contextual writing

    An editor writes the explanatory passages: what the metric measures, who leads, why, and what's worth knowing. Every claim links back to an agency-published reference.

  5. 5

    Publish, monitor, and re-issue

    Pages publish with a visible edition stamp. When the agency issues a new release, the page re-runs the pipeline; the previous edition is kept in the archive.