Skip to content
SVERIGEFAKTA

Our method

How we work — and what you can hold us to

SVERIGEFAKTA compiles official statistics. The editorial team currently does not publish its names, for privacy reasons — but everything we show is traceable to public primary sources. You don't have to trust us. You can verify us. That's the whole point.

Our principles

We commit to the following — and you can hold us to it.

Principle 1

Primary source, always

Every figure links to its origin — SCB, Brå, NIER, ESO, the Police, the Swedish National Agency for Education, the Election Authority, OECD. We never cite a second-hand site as the source.

Principle 2

We disaggregate rather than lump together

An average often hides more than it shows. We report subgroups (e.g. immigration by reason) when the data allows, because the whole would otherwise be misleading.

Principle 3

We separate measurement from interpretation

What a number is and what it means are two different things. We keep them apart, and we write it out when something is an interpretation rather than a measurement.

Principle 4

We don't set the threshold for you

We show where something stands — compared with the country, over time, ranked — but we don't decide whether it's "good" or "bad". That judgement is yours.

Principle 5

We show what contradicts

When the data contradicts a common claim — including among our own readers — we report it just as clearly as the data that confirms it. A site that only shows what it wants is not a facts site.

Principle 6

We report uncertainty

Margins of error, short time series, definition differences and data breaks are written out, not away. Where research disagrees we show multiple perspectives.

Principle 7

Correlation is not causation

That two things move together doesn't mean one causes the other. We remind readers where it's easy to draw hasty conclusions.

Principle 8

We correct openly

When we're wrong, we change it — and write it in our changelog, dated. Correcting isn't a weakness; it's what makes a facts site worth trusting.

Method rules

The technical translation of the principles above — this is how we build every page.

A. Data & sources

  • Every data point cites its primary source (Statistics Sweden, Brå, NIER, ESO, the Police, the National Agency for Education, the Election Authority, OECD). Official databases (Kolada/RKA) are acceptable. Private aggregators must never be cited as the source — we go to their primary source.
  • Every figure is shown with its value, unit, year and source. Comparable figures also show the national average, direction (▲▼→) and ranking (position X of 290).
  • Flagged or uncertain figures are verified against the primary table before publication. If reliable data is unavailable, we state ‘No data’ — we never guess.
  • Definitions are stated where they may be confused. In particular, ‘foreign-born’ (born abroad) versus ‘foreign background’ (foreign-born or born in Sweden to two foreign-born parents — always a larger group).

B. Reporting & honesty

  • Break down subgroups where the data permits; avoid misleading totals.
  • Separate measurement from interpretation in the text; label interpretations as interpretations.
  • Report counter-evidence: data that contradicts a proposition is shown as clearly as data that supports it.
  • Make uncertainty visible: margins of error, short time series, definition breaks and a note beneath charts.
  • Correlation is not causation: caveats appear next to the data, not in a distant footnote.
  • Contentious figures are shown in pairs where that gives a more honest picture (for example merit scores with and without newly arrived pupils).

C. Economy-specific

  • All SEK amounts over time are shown in both current and constant prices (nominal and inflation-adjusted).
  • Net public-finance contributions already include schools, healthcare, care services, Swedish-language tuition and interpreting — we do not add these costs again. The only unallocated item is the justice system.

D. Graphics & colour

  • No isolated figure in a data block — key-figure cards show the source and year in the same card.
  • Visual aids: ranking gauges (1–290 with a national-average marker), sparklines for trends and a full chart on selection.
  • Colour rule: evaluative colour (red=bad/green=good) is used only where the source defines an objective direction (for example, a higher merit score is better). Figures without an objectively better or worse direction — such as foreign-background share, foreign-born share and demographics — use neutral colours (blue/gold/grey); only the direction arrow shows change.
  • Colour must never be the only signal — always an arrow/text alongside (colour blindness).

E. Structure & SEO

  • FAQ blocks with schema.org FAQPage markup on municipality pages.
  • Internal links: neighbouring municipalities and related chapters.
  • A ‘Last updated’ date on each data page (from metrics.json).
  • Urban area versus municipality: we explicitly state which geography a figure covers.

F. Accessibility

  • Keyboard navigable; ARIA labels on charts, tables, gauges, sliders and maps.
  • Contrast of at least WCAG AA (4.5:1 for body text, 3:1 for large text).
  • No meaning conveyed by colour alone (see D).

G. Privacy & security

  • No source maps, local paths, author names or .map files in production.
  • No platform badge, project ID or preview domain in HTML, og:image or asset URLs.
  • EXIF metadata removed from all images.
  • Security headers enabled (CSP, HSTS, nosniff, referrer policy and permissions policy).
  • No user tracking; no data sold.

H. Transparency & corrections

  • Every data page must be traceable to the primary source by an outsider.
  • Errors are corrected openly and dated in our changelog.
  • Where research disagrees: multiple perspectives, no conclusion is drawn for the reader.

See also: Method & disclaimers · Sources and reports · Changelog

Common page structure

Consistency over variation. Every data/chapter page follows the same seven building blocks:

  1. Opening section — heading, short introduction and source line (‘Data from: …’).
  2. Sections revealed while scrolling.
  3. Every figure and chart — value, unit, year and source; national average, direction and ranking where relevant.
  4. A ‘What the data does not say’ block wherever uncertainty or limitations exist.
  5. Source footer with primary sources.
  6. A ‘Last updated’ date.
  7. Related pages and, on municipality pages, neighbouring municipalities.

Principle 0 — overrides everything

If a rule clashes with what the data shows — follow the data.

The rules serve the truth, not the other way round.

Editorial workflow

From primary source to published page — six traceable steps with version history.

    Step 1
    Data source

    The primary source is identified: public agency, research institute or official database. Secondary sources such as opinion writers and editorials are never accepted as the source.

    Step 2
    Data verification

    Definitions, year, unit and scope are verified against the original document. If anything remains unclear, the agency is contacted or the figure is not published.

    Step 3
    Processing

    When we calculate something ourselves (cohorts, projections or indices), the formula, assumptions and sensitivity analysis are documented. In-house calculations are labelled explicitly.

    Step 4
    Visualisation

    Charts follow the principles in D. Colour must never carry meaning alone. Cards show the value, unit, year and source.

    Step 5
    Publication

    The page is published with a source line, a ‘Last updated’ date and a link to the primary source. Metadata (title, description and JSON-LD) is generated.

    Step 6
    Version history

    Every change is recorded in the changelog with its date, page and description. We never overwrite earlier content without a trace.

See also: Changelog · All sources · Datasets

Found an error?

A number that doesn't match its source is the most important thing you can show us. We correct openly and date it in the changelog.

Report an error