# What did the MIT study on false news find?

> Written with AI · [Communications: misinformation and how to check it](https://lessoons.com/en/misinformation-and-verification) · Grade 12 · 2026-11-14 · https://lessoons.com/en/misinformation-and-verification/mit-study-on-false-news

The MIT study on false news, published in Science in 2018 by Soroush Vosoughi, Deb Roy and Sinan Aral, followed about 126,000 stories shared on Twitter from 2006 to 2017. It found that false news spread farther, faster, deeper and more broadly than true news, was 70% more likely to be retweeted, and that people, not bots, mainly drove the difference, probably because falsehood seemed more novel.

## The topic explained

### The question and the data

Do lies really travel faster than the truth, or does it only feel that way? Three researchers at the Massachusetts Institute of Technology, Soroush Vosoughi, Deb Roy and Sinan Aral, set out to measure it. Their article, “The spread of true and false news online,” appeared in Science in March 2018 and is one of the most cited studies of misinformation.

They began with stories that six independent fact-checking organizations, among them Snopes and PolitiFact, had already rated true or false; the six agreed in more than 95% of cases. Then they found every time those stories had been posted on Twitter between 2006 and 2017: about 126,000 cascades, shared more than 4.5 million times by about 3 million people. Politics was the largest category, with about 45,000 cascades. Using fact-checkers' verdicts meant the researchers did not decide what was true themselves, an important protection against their own bias.

### Measuring a cascade

A cascade begins when someone posts a story without retweeting anyone, and grows as others retweet it and others retweet them. The researchers measured its size (how many people shared it), its depth (how many steps of retweets from the original), its breadth (how many people at the widest step) and its speed. Depth matters: a story that reaches depth 10 has passed from person to person ten times, far beyond the original poster's followers.

On every measure falsehood won, “farther, faster, deeper, and more broadly,” in many cases by an order of magnitude. False stories were 70% more likely to be retweeted than true ones, and reached a cascade depth of 10 about 20 times faster. The effect was strongest for political news. A common misreading is that every false post goes viral; the finding is about averages across many thousands of cascades, not a rule for every post.

### Why? Novelty, emotion and people

Many people assumed automated accounts, bots, were to blame. The researchers ran their analysis again with bots removed, and the difference remained: humans, not bots, mainly spread false news. Their main explanation is the novelty hypothesis: false stories tend to be newer and more surprising than what users had seen, and novel information is more attractive to share, perhaps because passing on something new makes a person seem well informed. The authors were careful to say they cannot claim that novelty causes retweets; they found an association.

The replies told a similar story. Words in replies to false news expressed more surprise and disgust; replies to true news expressed more sadness, anticipation and trust. Falsehood, in other words, is often designed, or simply free, to be more shocking than reality.

### What the study cannot show

Every study has limits, and naming them is part of a strong answer. It covers one platform, Twitter, in the years 2006 to 2017, before later changes to how platforms rank and label content. It includes only stories that fact-checkers chose to check, which are likely to be the more prominent and contested ones. It shows association, not cause. Still, its main lesson holds: the accuracy of a story does not drive its spread, so correcting false news depends on people who pause before sharing.

## Practice: the 2018 MIT study

For 12th grade communications students. Where an example is made up, it says so. Answer from what is printed on the page, then check yourself against the answers.

1. **Who wrote the study, where and when was it published, and what period of Twitter did it cover?** — Soroush Vosoughi, Deb Roy and Sinan Aral of MIT; Science, March 2018; tweets from 2006 to 2017.
2. **How did the researchers decide which stories were true and which false, and why was that a good choice?** — They used the verdicts of six independent fact-checking organizations, which agreed in more than 95% of cases. It meant the researchers did not judge truth themselves, which protects against their own bias.
3. **Look at the diagram. What is the missing label, and what does it mean?** — Depth 3: the third step of retweets from the original tweet, people who retweeted someone who had retweeted a retweet.
4. **In a made-up illustration, 10 in 100 readers retweet a true story. If a false story is 70% more likely to be retweeted, how many in 100 would retweet it? Use the chart.** — 10 × 1.7 = 17 in 100. The chart shows the same ratio as an index: 170 against 100.
5. **What did the researchers find about bots, and why does it matter for solutions?** — Removing bots did not remove the difference: people mainly spread false news. So removing bots alone will not solve the problem; people's sharing habits matter.
6. **Fill in the missing cell, and explain how it supports the novelty hypothesis.** — Surprise, disgust. Surprise fits the idea that false news seems newer and more unexpected than what people had seen, which makes it more attractive to share.
7. **A classmate says: “The study proves that novelty causes people to share lies.” What is wrong with this?** — The authors themselves wrote that they cannot claim novelty causes retweets. The study found an association between novelty and sharing, not a proven cause.
8. **Name two limits of the study, and explain one of them.** — For example: it covers only Twitter in 2006 to 2017; it includes only stories fact-checkers chose to check; it shows association, not cause. The second means the sample may over-represent prominent, contested stories, so the results may not hold for all news.

## Common questions

### How much faster does false news spread?

In the 2018 MIT study, false stories were 70% more likely to be retweeted than true ones and reached a cascade depth of 10 about 20 times faster. The effect was strongest for political news.

### Do bots spread most fake news?

Not according to the MIT study. When the researchers removed bots from the data, false news still spread faster than true news, so humans were mainly responsible.

### What is a retweet cascade?

The chain of shares that starts from one original post and grows as people retweet it and others retweet them. Researchers measure its size, depth, breadth and speed.

### What is the novelty hypothesis?

The idea that false news spreads more because it is more novel and surprising than what people have seen, and novel information is more attractive to share. The MIT researchers found an association but did not claim it was the proven cause.

### Is the MIT study still relevant?

Its data end in 2017 and cover only Twitter, so platforms and habits have changed since. But its central finding, that accuracy does not drive spread and people do, remains a basis for later research on misinformation.
