Brexit misinformation statistics sit at the intersection of a close referendum, contested campaign claims and a rapidly changing media environment. The figures below cover the June 2016 EU referendum, the Oxford Internet Institute’s Twitter study, and Ofcom research conducted in 2024. They measure different things, so campaign spending, online activity and public attitudes should not be treated as interchangeable evidence.
Contents
- The referendum and confidence in campaign fairness
- Campaign spending rules and enforcement
- What the Twitter sample measured
- The concentration of Brexit-related activity
- Automation across hashtag groups
- How UK adults encounter misinformation
- How people check information
The referendum and confidence in campaign fairness
The Electoral Commission’s report on the UK’s membership of the EU recorded 17,410,742 Leave votes in the June 2016 referendum, equal to 51.9% of ballots. Remain received 16,141,241 votes, or 48.1%. The narrow difference matters when interpreting later arguments about campaign influence: the result was large in absolute numbers but close as a share of votes cast.
The same Electoral Commission report found a substantial gap between public perceptions of campaign fairness and balance. Fifty-two per cent of respondents disagreed that the referendum campaigns were fair and balanced. Thirty-four per cent agreed, including 12% who agreed strongly. The figures describe opinion about the campaigns; they do not establish that any particular claim was false or that misinformation changed the result.
The Commission received just over 1,000 public complaints about campaign material during the referendum campaign. Complaints are a measure of public concern and regulatory workload, not a count of proven misinformation cases. The report also noted that the referendum period overlapped with regulatory timeframes for the May 2016 elections, meaning campaigners had to allocate part of their spending to referendum or election rules.
Campaign spending rules and enforcement
The Electoral Commission’s referendum guidance set different spending limits for registered campaigners. The designated lead campaigns, The In Campaign Ltd and Vote Leave Ltd, each had a £7,000,000 limit. Political parties registering as referendum campaigners with more than a 30% general-election vote share also faced a £7,000,000 limit.
The limits then stepped down by previous vote share. Parties with more than 20% and up to 30% faced a £5,500,000 limit; those with more than 10% and up to 20% faced £4,000,000; and those with more than 5% and up to 10% faced £3,000,000. Parties with under 5% vote share faced a £700,000 limit. Other registered referendum campaigners were also limited to £700,000, while an unregistered campaigner could spend £10,000 before registration was required.
The enforcement figures come from the Electoral Commission’s Report of an investigation in respect of Vote Leave Ltd, Mr Darren Grimes, BeLeave and Veterans for Britain. The Commission fined Vote Leave £20,000 for failing to produce a complete spending return that omitted £236,501.44 of payments. The same investigation identified £676,015.87 in undeclared common-plan spending and found that Vote Leave overspent the statutory limit by £449,079.34.
The overspending offence brought another £20,000 fine. Vote Leave was fined a further £1,000 for failing to include required invoices and receipts for eight payments, and another £20,000 for failing to comply with an investigation notice. The investigation also recorded that Veterans for Britain expected a £50,000 donation in mid-June 2016 and later asked Vote Leave for a £100,000 donation to pay Aggregate IQ directly.
The Commission said Vote Leave and BeLeave were working under a common plan. It also said Vote Leave and Veterans for Britain did not appear to be collaborating on a campaign, despite prior contact. These are findings attributed to that investigation and should not be expanded into broader claims about every referendum message or campaigner.
| Referendum campaign category | Spending limit |
|---|---|
| Designated lead campaign; or party above 30% vote share | £7,000,000 |
| Party above 20% to 30% vote share | £5,500,000 |
| Party above 10% to 20% vote share | £4,000,000 |
| Party above 5% to 10% vote share | £3,000,000 |
| Party under 5% vote share; other registered campaigner | £700,000 |
| Unregistered campaigner before registration was required | £10,000 |
What the Twitter sample measured
The Oxford Internet Institute study, Bots, #StrongerIn, and #Brexit: Computational Propaganda During the UK-EU Referendum, examined more than 1.5 million tweets collected from 5 to 12 June 2016. The dataset contained 313,832 distinct Twitter user accounts and sampled 25 specific Brexit-related hashtags across Leave, Remain and neutral categories.
Across the sample, there were 1,831,626 tracked hashtag occurrences. Leave-oriented hashtags accounted for 993,176 uses, or 54%; Remain-oriented hashtags accounted for 363,217, or 20%; and neutral hashtags accounted for 475,233, or 26%. The most frequently appearing named hashtags in the supplied figures were #voteleave, with 341,839 appearances, and #StrongerIn, with 110,653.
These figures describe the selected hashtag sample during one week, rather than all online discussion during the referendum. They therefore show the scale and composition of a research dataset, not the total number of Brexit posts or the reach of every claim.
The concentration of Brexit-related activity
The OII study found that activity was highly concentrated among a relatively small group of accounts. The top 100 users generated more than 121,000 tweets in one week, described as about 8% of all StrongerIn-Brexit traffic. Accounts posting at least 100 times during the week generated 32% of all Brexit Twitter traffic.
Fewer than 2,000 users produced that 32% share out of more than 300,000 users in the sample. In the study’s wording, fewer than 1% of accounts generated almost one-third of all content in the sample. This concentration is relevant to how political narratives can appear prominent online, but tweet volume alone does not measure persuasion, readership, accuracy or voter behaviour.
The OII paper’s Table 2 classified 662,745 tweets as exclusively Brexit, representing 51.8% of the table’s total. Exclusively StrongerIn tweets numbered 186,279, or 14.6%, while exclusively neutral tweets numbered 234,170, or 18.3%.
The mixed categories were smaller: Brexit-neutral tweets numbered 69,322, or 5.4%; StrongerIn-neutral tweets numbered 35,412, or 2.8%; Brexit-StrongerIn tweets numbered 49,556, or 3.9%; and tweets combining Brexit, StrongerIn and neutral hashtags numbered 40,926, or 3.2%.
Automation across hashtag groups
In the full Table 2 sample, 179,083 tweets, or 14.0%, were generated by heavily automated accounts. The same table counted 1,531 disclosed-bot tweets, or 0.8% of the total. The study’s automation measure is narrower than the everyday use of the word “bot”, so these results should be read using the OII paper’s terminology.
Automation varied by hashtag combination. The exclusively StrongerIn category contained 28,075 heavily automated tweets, equal to 15.1% of those tweets. The exclusively Brexit category contained 97,431, or 14.7%, while the exclusively neutral category contained 13,436, or 5.7%.
Among mixed categories, the Brexit-neutral group contained 11,667 heavily automated tweets, or 16.8%. The StrongerIn-neutral group contained 5,099, or 14.4%, and the Brexit-StrongerIn group contained 9,735, or 19.6%. The three-way Brexit-StrongerIn-neutral group had 13,640 heavily automated tweets, equal to 33.3% of that category.
The study also reported that only 5.7% of exclusively neutral-hashtag tweets came from heavily automated accounts, compared with one-third of tweets using a mix of all hashtag types. Those differences indicate that automation was not distributed evenly across the sample. They do not, by themselves, identify which messages were misleading.
How UK adults encounter misinformation
Ofcom’s Understanding misinformation: an exploration of UK adults’ behaviour and attitudes surveyed 4,269 politically representative UK adults aged 16 and over. YouGov conducted the fieldwork from 27 June to 2 July 2024. In that research, four in ten UK adults said they had encountered misinformation or deepfake content in the previous four weeks.
Of those who encountered false or misleading information, 71% said they had seen it online, 43% on television, and 21% through print newspapers or their associated websites or apps. Thirty-nine per cent of respondents had seen false or misleading information about UK politics, including the General Election, during the prior four weeks. The comparable figures were 33% for international politics and current affairs and 25% for health and medical information.
Ofcom’s Adults’ Media Use and Attitudes Report 2024 provides a broader 12-month measure: 45% of adult users said they had seen a deliberately untrue or misleading news story on social media in the past 12 months. Among those adults, 41% did not take any action after seeing a misleading story. Thirty-four per cent reported the story to the site or app, while 14% shared it online to let followers know it was incorrect or misleading.
The two Ofcom measures use different time periods and research descriptions. The four-week figures should not be directly added to or substituted for the 12-month social-media figure.
How people check information
Ofcom’s 2024 media-use research found that 15% of adults rarely or never consider the truthfulness of information they read on social media. Among those who do consider truthfulness, 91% said they make specific checks to verify information. Fifty-one per cent check who posted the information, 46% look at comments to see what others have said, and 16% use fact-checking websites such as Full Fact.
Trust in mainstream news was also uneven. Sixteen per cent of adults said they do not use mainstream news outlets at all, and 6% avoid mainstream news specifically because they do not trust it to be accurate. Among adults who use mainstream sources, 41% said they sometimes question the information they receive from them and 22% said they always question it.
People’s exposure to disagreement varied as well. Fifteen per cent of users rarely see views they disagree with on online communications platforms. The figure was 20% among adults aged 16–24. At the other end, 21% said they often see views they disagree with online.
The evidence from Brexit-era campaign enforcement, the 2016 Twitter sample and 2024 Ofcom surveys measures different stages of the information cycle: campaign rules, online publishing behaviour, exposure and checking habits. Keeping those stages separate is essential when using Brexit misinformation statistics to assess what was said, how widely it circulated and how audiences responded.