Framing

What it looks like

A frame is the set of things a story makes salient and the set it leaves out. Every story has one, because no account can include everything. That is why framing is not a synonym for lying, and why fact-checking a framed headline almost always returns "true."

The choices are usually small and grammatical. Which side’s reaction appears first. Whether a number is called a cost or an investment. Whether the sentence has a person in it at all. A headline can be entirely accurate and still do most of the work of an argument.

Why it works

The classic demonstration is Tversky and Kahneman’s 1981 experiment, in which the same choice described in lives saved and lives lost produced opposite preferences from statistically identical groups.1 That is equivalence framing: two logically identical restatements. Headlines mostly do something subtler, which Druckman calls emphasis framing — both versions are true, and they differ in what they make salient.2

The sharpest measured effect is grammatical. Fausey and Boroditsky changed nothing but the sentence structure — whether a person or an object was the subject — and moved both blame ratings and the dollar amounts participants assigned in damages.3 A sentence with no agent in it is not a neutral sentence. It is a sentence that has removed somebody.

This matters more than it should because most people never get past the headline. An analysis of over 35 million Facebook posts containing links found that roughly three quarters of news links were shared by people who had not clicked through.5 For most of the audience, the headline is the article.

Documented cases

One report, two true headlines, October 2015

When the International Agency for Research on Cancer classified processed meat in October 2015, coverage split along a line that had nothing to do with accuracy. The classification describes how confident researchers are that something causes cancer. It does not describe how much cancer it causes. A headline can report either and be correct.

Cancer Research UK published the distinction explicitly on the day, along with the underlying numbers: a lifetime bowel cancer risk of roughly 66 in 1,000 for the highest consumers against 56 in 1,000 for the lowest.6 The World Health Organization’s own question-and-answer page said the same thing — that sharing a classification category "does NOT mean that they are all equally dangerous."7 Both framings were available from the same document on the same morning.

The grammar of a traffic report

Ralph and colleagues coded 200 US local news articles about pedestrian and cyclist crashes from early 2018. In 81 percent, the grammatical subject was the vehicle rather than a driver; in a third of sentences there was no agent at all.4 The follow-up experiment found that rewriting the same events with the driver as the subject shifted where readers assigned blame.

No reporter in that sample invented anything. The pattern is what a newsroom style habit looks like from the outside, and it is a useful example precisely because nobody chose it as an argument.

What to do about it

• Find the grammatical subject. Ask whether swapping it changes your reaction. "A car struck a pedestrian" and "a driver struck a pedestrian" report the same event.
• Notice when there is no agent at all. "Shots were fired." "Mistakes were made." Somebody has been removed from that sentence, and it is worth asking who.
• Separate the strength of evidence from the size of the effect. "In the same category as" is a statement about confidence, not about magnitude. They get conflated constantly, in both directions.
• Read a second outlet’s headline on the same event before forming an impression. Where the two diverge is exactly where the editorial choice was made, and it is visible in about fifteen seconds.
• Open the article. On a 75 percent no-click baseline, this alone puts you in a minority of readers.

Sources

Every figure on this page comes from one of these. Each entry names the publisher, the date and the sample size, so you can check it rather than take our word for it.

1. Tversky, A. & Kahneman, D. — "The Framing of Decisions and the Psychology of Choice"Science 211(4481):453–458 · 30 January 1981 · n=152 and n=155https://www.csc2.ncsu.edu/faculty/mpsingh/local/Social/f24/wrap/readings/Tversky+Kahneman-framing-1981.pdf
2. Druckman, J. N. — "The Implications of Framing Effects for Citizen Competence"Political Behavior 23(3):225–256 · 2001 · theoreticalhttps://faculty.wcas.northwestern.edu/jnd260/pub/Druckman%20Political%20Behavior%202001.pdf
3. Fausey, C. & Boroditsky, L. — "Subtle linguistic cues influence perceived blame and financial liability"Psychonomic Bulletin & Review 17(5):644–650 · 2010 · n=236, 179 and 589https://www.uolearninglab.com/wp-content/uploads/2014/11/FauseyBoroditsky_LinguisticBlame_PBR2010.pdf
4. Ralph, K., Iacobucci, E., Thigpen, C. & Goddard, T. — "Editorial Patterns in Bicyclist and Pedestrian Crash Reporting"Transportation Research Record 2673(2):663–671 · 2019 · n=200 articleshttps://static1.squarespace.com/static/51a15902e4b022d36442072d/t/5e58148b355ace7fe8aee4bd/1582830841202/Editorial+Patterns+in+Bicyclist+and+Pedestrian+Crash+Reporting
5. Sundar, S. S. et al. — "Sharing without clicking on news in social media"Nature Human Behaviour · 2025 · 35 million+ posts, 2017–2020https://www.nature.com/articles/s41562-024-02067-4
6. Dunlop, C. — "Processed meat and cancer — what you need to know"Cancer Research UK · 26 October 2015https://news.cancerresearchuk.org/2015/10/26/processed-meat-and-cancer-what-you-need-to-know/
7. World Health Organization — "Cancer: Carcinogenicity of the consumption of red meat and processed meat"WHO question-and-answer page · 26 October 2015https://www.who.int/news-room/questions-and-answers/item/cancer-carcinogenicity-of-the-consumption-of-red-meat-and-processed-meat