Are Swedish media biased?

There is research that measures journalists' personal values (JMG's journalist surveys) and research that measures actual coverage. Systematic studies of coverage identify patterns in selection and framing but rarely go all the way to proving intentional bias. We report documented patterns and refrain from evaluative judgements about ‘bias’ without primary studies.

What does the Delmi study show about media and migration 2015?

Delmi 2017:6 (Strömbäck et al.) analysed print-newspaper news articles from 1 January 2010 to 30 June 2015. Negative framings of immigration were more common than positive ones, coverage was dominated by refugee immigration, and differences between newspapers and years were generally small. The authors judge the picture to be more negative than research and official statistics warrant. Autumn 2015 is not included.

How is coverage volume measured?

By systematic searches in media archives (Retriever) where you define keywords, sources (e.g. daily press + evening press + public service) and time period. Hit counts are then aggregated per week or month. Method choices affect the result and must be disclosed.

Why did trust in morning press fall 2020–2023?

SOM Institute time-series show a decline after the 2020 peak. Researchers point to several interacting factors: political polarisation, increased criticism from certain political camps, new alternative media channels, general trust shifts. A single causal explanation does not exist.

How high is Swedish media trust internationally?

In the Reuters Digital News Report, Sweden is year after year among the countries with the highest news trust. Absolute levels differ across surveys (SOM and Reuters ask different questions), but the pattern is stable: higher than most comparable countries.

What separates news selection from editorial opinion?

News selection — which events become news, how they are prioritised and framed — is a publicistic task governed by news-value criteria. Editorial opinion (editorials, columns) is evaluative text separated organisationally and graphically from news reporting.

Is a headline-vs-statistics comparison fair?

If both are correctly rendered and dated, the comparison is a valid data point — but it does not on its own prove a pattern. That requires multiple paired observations under explicit selection rules. We show examples and disclose the rules, so the reader can weigh them.

What is the Media Ethics Board?

MEN (from 2020) rules on complaints against press-ethics rules and can censure a publication. The Media Ombudsman (MO) prepares cases. The system is self-regulatory — not governmental — and applies to affiliated media.

How can I check a news item against reality myself?

Go to the primary source. If an article refers to a Brå report, a court ruling or a Statistics Sweden extract: find the original document, check period, definition and sample. Headline and lead are often simplified — the essential information is further down in the article or in the underlying source.

Culture & values

Press archive

Headlines beside concurrent statistics (2015–2024)

Three headlines from SVT and DN plus Brå's own 2024 press release, beside named statistics from the same year. All four have been checked against the publisher's original; a headline is not a quote from any interviewee. The pairing shows contemporaneity; it tests a headline only when both address the same measurable proposition.

Sources & method

Data source

Definitions

— ‘Dödat med skjutvapen’ = Brå/Polisens definition av dödligt våld med skjutvapen i kriminell miljö och utanför.
— ‘Fullbordade detonationer’ = Polismyndighetens redovisning av faktiska sprängningar (exkl. försök och andra brand-/hot-händelser).
— ‘Anmälda spräng-brott’ = Brås statistik för allmänfarlig ödeläggelse genom sprängning inkl. försök — ligger systematiskt högre än Polisens fullbordade-serie.
— ‘Inkomna asylansökningar’ ≠ ‘beviljade förstagångsbeslut’ — 2015-raden avser inkomna ansökningar (Migrationsverket), inte beviljade beslut.

Uncertainties

— Rubrikurvalet är illustrativt, inte slumpmässigt — för systematisk analys, se Strömbäck/Andersson/Nedlund (Delmi 2017:6).
— Tidigare exempelrubriker från DN 2000, DN 2005 och Expressen 2010 kunde inte återfinnas hos publicisten och är borttagna; urvalet täcker därför bara 2015–2024 och säger inget om rapporteringen före 2015.
— Sambandet mellan rubrik och samtida siffra säger inget om medvetenhet eller skuld — det visar bara samtidighet.
— När rubriken och statistiken mäter olika propositioner kan statistiken inte användas för att bevisa eller motbevisa rubriken.

Sources for this page

The data comes from official primary sources.

SOM-institutet · Annually
SOM-institutet (Göteborgs universitet) — Svenska förtroendetrender 1986–2025
Förtroende för samhällsinstitutioner och medier (SOM-institutet)
Latest data point 31 Dec 2025 · Source published 30 Mar 2026

Read the full investigation of how Sweden has changed.

Back to overview →

The question this page answers

How well does Swedish news coverage mirror the statistical situation — in volume, timing and word choice?

At a glance

Key figures — the most important numbers on this page

The media and migration (Strömbäck et al.)

Delmi 2017:6

Content analysis of print newspapers 1 Jan 2010–30 June 2015; autumn 2015 is not included.

Trust in morning press: 36 %

SOM 2023

Share reporting fairly/very high trust. The 2020 peak (46 %) has turned downward.

Trust in Swedish Radio: 62 %

SOM 2023

Still high in international comparison, but below the 2020 peak (74 %).

Sweden is a top country for news trust globally

Reuters DNR

Reuters Institute Digital News Report — annual international comparison.

SVT + SR + TV4 reach a large share of the population daily

≈ 4 in 5

Nordicom, Media Barometer — public service still has broad reach.

Standard tool for media-archive searches

Retriever

Volume analyses at article level require an explicit selection of sources and search string.

Brå 2021:9

Difference between what public authorities measure and what media choose to highlight.

Headline ≠ article ≠ reality

Source critique

A headline is a selection from a text that is itself a selection from reality.

What does the data show?

Objective observations — not interpretations

The VOLUME of media coverage is measurable via archive searches (e.g. Retriever). The gap between coverage and reality often depends as much on what is not reported as on what is.
The TIMING of coverage can be compared with concurrent statistics. Delmi 2017:6 does not measure timing around autumn 2015 — its sample ends on 30 June 2015.
WORD CHOICE in coverage can be analysed systematically (frequencies, semantic fields, sources cited) — but individual headlines cannot, without selection criteria, be generalised as ‘the media’.
Isolated examples of divergence between headline and statistics do not prove general bias. Systematic studies with defined sample, search string and time period are required.
Sweden's population continues to report high trust in traditional news media in international comparison (Reuters DNR), while the SOM series show a decline from the 2020 peak.
Trust is a subjective perception measured through self-report — it does not directly say anything about the truthfulness of reporting.

Method note
We use three types of evidence: (1) published media research with stated method (Delmi, JMG, Institute for Media Studies), (2) systematic volume analyses in media archives with documented search string, source selection and period, (3) paired comparisons of concurrent headline and official statistic — both linked in the source. We avoid isolated examples presented without a selection rule, and we always distinguish volume (measurable), timing (measurable) and ‘bias’ (interpretation). Headlines are cited as primary material, not as evidence of intent.

Definitions

How the numbers are counted — and what they do not cover

News selection: Editorial decisions about what becomes news, how it is prioritised and how much space it gets. Studied within news-value research.
News values: The criteria that govern what becomes news: timeliness, consequence, proximity, deviance, elite involvement, personification, etc. (Galtung & Ruge; Harcup & O'Neill).
Framing: How an event is presented — which angle, which actors and which consequences are emphasised. Can change without any change in the underlying facts.
Content analysis: Systematic method where text or image material is coded on pre-defined variables to measure patterns quantitatively or qualitatively.
Volume analysis: Counting the number of published articles/segments about a theme over time, typically based on a search string in a media archive.
Media archive (Retriever): Commercial service collecting published articles and broadcasts from Swedish media. Standard tool for volume and content analysis in research and PR.
SOM Institute: Research institute at the University of Gothenburg that annually measures opinion, trust, media habits and values. The basis for Swedish trust time-series.
Reuters Digital News Report: Annual international comparison of news trust, news habits and willingness to pay in 40+ countries. Reuters Institute, Oxford.
Nordicom: Nordic Information Centre for Media and Communication Research at the University of Gothenburg. Publishes the Media Barometer and key figures on the Swedish media market.
Institute for Media Studies: Independent Swedish foundation that publishes journalism and media research, often focused on news selection and reporting.
JMG (University of Gothenburg): The Department of Journalism, Media and Communication. A central Swedish environment for media research.
Delmi: The Migration Studies Delegation — a public committee commissioning and publishing independent research on migration, including media coverage of migration.
Publicistic task: The editorial responsibility to select, prioritise and scrutinise. Distinct from advertising and opinion journalism.
Opinion journalism: Editorials, columns, debate — evaluative texts editorially separated from news reporting.
Press ethics: Rules on publication and professional ethics, supervised by the Swedish Media Ethics Board (MEN) and the Media Ombudsman (MO).
MEN / MO: Media Ethics Board (from 2020) and Media Ombudsman review publications against press-ethics rules. Replaced the earlier PON/PO system.
Correction: A published correction of incorrect information. Required by press-ethics rules where a publication risks being perceived as harmful.
Source in original: Going to the primary source (public report, court decision, dataset) instead of relying on a journalistic account.
Selection bias: Systematic skew that arises when the sample is not representative — applies to both research and editorial selection.
Headline writing: Editorial decision on how a text should be summarised in headline/lead. The headline is often written by someone other than the article author.
News aggregator: Service that lists headlines from multiple sources (e.g. Omni). Headline exposure can there become more important than the article body.
Media consonance: The phenomenon that independent newsrooms nonetheless choose similar angles — explained by shared news values, common primary sources and competition.
Verification rule: Journalistic core rule: check information against more than one independent source before publication. A publicistic method, not a legal requirement.
Source protection: Constitutionally protected right to provide information anonymously to newsrooms. Regulated in the Freedom of the Press Act and the Fundamental Law on Freedom of Expression.

Primary sources

Agencies and research institutions behind this page

agencies/institutions

reports & studies

primary datasets

Frequently asked questions

Short answers to what is most often discussed

It means the comparison between (a) how much and in what way media cover a phenomenon and (b) what published official statistics show about the same phenomenon at the same time. Differences may reflect news values, timing, wording or focus — not necessarily bias.

Logical next steps if you want to understand the background

Headlines beside concurrent statistics (2015–2024)

Sources for this page

At a glance

What does the data show?

Definitions

Primary sources

Frequently asked questions

Government language

Promises vs statistics

Headlines beside concurrent statistics (2015–2024)

How to read this

Human consequence

Sources for this page

At a glance

What does the data show?

Definitions

Primary sources

Frequently asked questions

Related topics

Government language

Promises vs statistics