Does the Political Context Shape How “Due Impartiality” is Interpreted? An Analysis of BBC Reporting of the 2019 UK and 2020 US Election Campaigns

ABSTRACT

Balance and impartiality are central principles in journalism, but this study argues their conceptual application in news reporting should be subject to more academic scrutiny. In the UK, the way “due impartiality” has been applied and regulated by broadcasters has raised concerns about promoting a ‘she-said-he-said’ style of reporting, which constructs balance but not scrutiny of competing claims. In this study, we analyse how the UK’s “due impartiality” was applied by journalists in different political contexts by assessing how the BBC dealt with competing party-political claims. We develop a nuanced quantitative analysis of BBC journalist interactions (N = 967) with claims made by the four main party leaders during the 2019 UK and 2020 US elections. Overall, we found BBC reporting robustly challenged claims by US politicians, whereas coverage of UK politicians often only conveyed claims and counterclaims with limited journalistic intervention, particularly on television news. We argue that impartiality should be viewed more as a fluid than fixed concept given that the context shapes how it is applied. As concerns about misinformation have grown over recent years, we conclude that more finely tuned studies are needed to understand how journalists apply concepts about balance and impartiality in political reporting.

KEYWORDS:

• Misinformation
• impartiality
• broadcast news
• election reporting
• public service broadcasting
• political journalism

Introduction

The normative model of democracy presupposes an electorate sufficiently informed to be able to exercise their franchise. Hence, one frequently applied measure of the health of democracy in a society is whether journalists can deliver the necessary information to the public. To safeguard standards in reporting, professional journalists draw on concepts such as impartiality, balance, and objectivity to ensure they accurately inform the public and hold power to account. But how these concepts have been applied across different media systems has not been subject to a great deal of empirical scrutiny (Hopmann, Van Aelst, and Legnante 2011; Sambrook 2012). Broadly speaking, as Sambrook (2012) has outlined, balance can be defined as an allocation of equal space to opposing views, with objectivity interpreted as an effort to exclude subjective judgement, while impartiality is defined as an attempt to reflect different ideas, opinions, interests, or individuals with detachment. But, as this study explores, these are difficult concepts to define and apply in journalism. We offer a UK case study that examines how the BBC applied its rules about “due impartiality” in its domestic and foreign reporting of the 2019 United Kingdom (UK) and 2020 United States (US) election campaigns. Since BBC impartiality is applied to the same high levels of editorial standards across all of its news output, in theory the application of “due impartiality” should be practised with the same rigour and robustness irrespective of the geographical political context. We also considered whether the political context shaped how impartiality was interpreted by examining two high profile politicians—Boris Johnson and Donald Trump—who have been associated with debates about spreading misinformation.

We focussed on the BBC in the UK because it remains one of the most prominent and trusted media outlets worldwide (Nielsen, Schulz, and Fletcher 2020). Despite an increasingly high choice media landscape, the BBC continues to be by far the most watched, read and listened to information source, particularly during national events such as election campaigns.

Currently in the UK, Ofcom is the body in charge of regulating communication services, including television and radio broadcasting. But as well as being regulated by Ofcom, the BBC also has its own editorial guidelines¹, including how “due impartiality” should be interpreted during election campaigns. Section 10.3 of the guidelines, for example, encourages editorial judgements rather than explicit mandates, stating that “The BBC should make, and be able to defend, editorial decisions on campaign coverage on the basis that they are reasonable and carefully reached, with due impartiality” (BBC 2019, 198). A bullet point acknowledges that coverage may differ between platforms: “10.3.16: The way in which due impartiality is achieved among parties will vary, depending on the format, output and platform” (BBC 2019). However, the same point goes on to explain that “content producers must take responsibility for achieving due impartiality in their own output without necessarily relying on other BBC content or services” (BBC 2019). This suggests internal pluralism should operate at the item/article or programme level rather than at an organisational level. Also pertinent to this discussion is section 10.3.18, which states that “The principles of fairness and due impartiality that underlie the BBC’s coverage of UK votes should also inform reporting in other countries” (BBC 2019, 199). In other words, how impartiality is achieved will be different across news output, but it should be applied with the same rigour domestically as well as internationally. Our study offers one of the first ever studies to assess how impartiality is interpreted across different platforms and formats (television and online sites), and during both US and UK election campaigns.

To date there has been limited academic attention paid to understanding how journalists apply impartiality across domestic and foreign reporting. And yet, Presidential elections in the US have long been of interest to UK audiences and many people in the US consume UK media, notably the BBC. US elections typically receive extensive coverage in the UK and the Trump presidency carried even greater news values for UK media organisations, including of course the BBC. According to the corporation’s own guidelines, the public service broadcaster should employ the same principles of impartiality in reporting a US Presidential Election as a UK General Election.

However, Nossek's (2004) research has suggested journalists are more likely to adopt a professional frame when they cover foreign rather than national news. For example, if journalists reported foreign elections it would be defined as “their” events, whereas national elections are “our” events. In Nossek's (2004: 343) own words, "the more ‘national’ the report is, the less ‘professional’ it will be, i.e. the closer the reporters/editors are to a given news event in terms of national interest, the further they are from applying professional news values". This is because journalists are often citizens of the country on which they are reporting, with greater personal concern as to the outcome of elections in their own country, and thus a national frame on "our" news may subdue professional ideals. Nossek’s (2004) work implies that differences may be found between how journalists interpret impartiality between a UK (domestic) and US (foreign) election campaign. Viewed through this prism, the adoption of a professional frame could result in journalistic output closer to a normative conception of impartiality than in instances where a national frame is adopted.

Conversely, however, there are both theoretical and practical reasons which argue that impartiality is easier to aim for in a domestic setting. Indexing theory (Bennett 1990) would predict that election reporting would frequently cover the disagreements in policy between competing major parties, in both domestic and international settings. Yet, it is likely journalists have closer links with domestic rather than international political parties, meaning that domestic party information subsidies (Gandy 1982) will more frequently arrive in journalist’s inbox (and nowadays DMs, Twitter feeds etc.). This facilitates the writing of impartial stories given a range of sources can be reflected in coverage.

Understanding the Concept of Impartiality and Balance and Applying It in Journalism and to the BBC

Hopmann, Van Aelst and Legnante’s (2011) review of academic studies about the application of balance in news reporting concluded that it is a difficult concept to operationalise. Given the potential for media influence on political knowledge and behaviour (Sparks 2010), they also observed that it is a concept frequently, yet imperfectly, examined by media scholars.

Balance has been typically operationalised within two-party systems (notably and most frequently the US) by comparing the quantity and, at times, quality and valence of coverage of the two parties, with quantification of measures of difference being seen as evidence of “imbalanced” coverage. Yet, even in this most simplistic two-party model, using a basic visibility operationalisation of balance, a 50:50 coverage split would not necessarily entail balance. Operationalising the concept affording accurate claims of (im)balance is complicated by multiple factors addressed in prior work. An incumbency effect, where the present government can be, perhaps justifiably, expected to receive greater coverage than the opposition, complicates visibility measures. Measuring coverage valence is a logical extension to volume, but an “obviously more complex” (Hopmann, Van Aelst, and Legnante 2011, 248) measure. Issue coverage is another measure frequently included in measures of balance; whether coverage focusses on issues owned by (Petrocik 1996), or higher up the agenda (Brandenburg 2005) of, one party is also a germane factor in considering the balance of coverage.

Meanwhile, applying impartiality in news reporting has often included the construction of balance (with its challenges in operationalisation) alongside measures to show that coverage is unbiased, even- handed, open-minded and not favouring one side over another (Cushion 2011). Hartley (2019) notes that impartiality should take account of a full range of views, the relative weight of those views and how these both may alter over time. Thus, it is a concept difficult to operationalise, to measure in empirical studies and also to successfully embed within journalistic practice. Yet balance and impartiality have been seen as central tenets of the particular form of journalism which developed in the early part of the twentieth century. This form—“new American journalism” (Høyer and Pöttker 2005)—adopted an intra-story pluralism model and was the form which came to dominate also in the UK, notably shaping the style of BBC news output when it first began brodcasting. In his 1924 book, Broadcast Over Britain (Reith 1924), the first Director-General of the BBC, Lord Reith, outlined his vision for the corporation, arguing that “It will not be easy to persuade the public of an absolute impartiality, but impartiality is essential” (112). This vision of impartiality included the notion that part of the mission to “inform and educate” was to tell the public both, and multiple, sides of a story—i.e., balance. This ideal was tested two years later with coverage of the 1926 General Strike. According to Higgins (2014), history records that Reith was more concerned with an appearance of rather than an achievement of impartiality; though Reith was seemingly open to allowing opposition opinion to be covered on the BBC, he was ultimately more than happy to acquiesce to government requests to not do so.

Over time the ethos of impartiality and balance shaped BBC output, whether, as argued by critics, in appearance only, or in genuine principle. But whatever the rationale, the trope of covering "both sides" of an issue shaped much of twentieth-century journalism practice (Mindich 2000). The continued dominance of this model was recognised by the 2007 Bridcut Review of BBC output which suggested that the “seesaw” metaphor of reporting “both sides” (balance) should perhaps be replaced by a wagon wheel metaphor allowing more and greater nuance of opinion (impartiality). The ethos of multiple “balanced” perspectives as a means of informing the public in an impartial manner has of course been extensively critiqued. Work from the Glasgow University Media Group (1976, 1980) notably and consistently showed how the range of viewpoints aired was limited to a narrow and elite choice of sources rather than a full range of perspectives. The BBC acknowledged that concerns over balance were detrimental to its reporting of the climate crisis. A strict adherence to balance with reporting on climate change led to instances where the scientific consensus on Anthropogenic Climate Change was incorrectly balanced with climate denialism (BBC Trust 2011), or, more succinctly, truth was balanced with untruth. In this instance, balance allowed for consideration of the impartiality requirement by taking account of a full range of views, but not that the relative weight of opinion had changed over time.

Though the model of journalism in the UK can be seen to have been developed largely from a media routine perspective (Hopmann, Van Aelst, and Legnante 2011), we would argue that a political system perspective (Hopmann, Van Aelst, and Legnante 2011)—comparing, say, domestic and foreign election reporting—is also needed to understand if impartiality is applied differently. If we consider our UK case study in more detail, the “due” in the BBC’s impartiality rules relates to journalists exercising their judgements about editorial matters, such as source selection. In other words, journalists construct their sense of balance and evenness when reporting on political actors and events (Cushion, Lewis, and Callaghan 2017). In outcome terms, particularly at UK election times, impartiality was historically often interpreted by broadcasters as offering equal, "stop-watch" airtime to the two major parties of Labour and the Conservatives. In practice, this meant party representatives were paired with an opposition spokesperson to argue their case directly or indirectly on the issue of the day, resulting in a "she-said-he-said" style of political reporting. As many more parties compete during a campaign, this binary opposition was no longer sustainable as the two-party dominance lessened; broadcasters had to reflect the diversity of party-political opinions. Thus, two-party opinion is now occasionally augmented with limited coverage of other parties when salient to an issue or election campaign (Hughes 2016); perhaps what we may call a '"she-said-he-said-they-also-said" style. However, despite this, the old and simpler formula still holds primacy. The Wahl-Jorgensen et al. (2017) study of sources over time found that, regardless of statements by the BBC to embrace going beyond simple opinion binaries, the “paradigm of impartiality-as-balance” (781) endures.

Despite the extensive criticism of both-sideism and its clear limitations in affording democratic norms, it is generally held up as a model which at least aims for impartiality. However, the "he-said-she-said" style of reporting is premised on an understanding that both sides represent factually accurate positions. But, in the current political landscape, increasingly characterised by misinformation, the credibility of what politicians say has come under greater scrutiny. In this study, we examine two actors—Donald Trump and Boris Johnson—that have well-documented records of making false or misleading statements. The Washington Post Fact-Checker detailed over 30,000 statements from Trump during his presidency which they labelled as false (Kessler 2021). Meanwhile, Oborne (2021) is one of many who have extensively detailed episodes of Johnson making deceitful statements.

UK Case Study: Introducing the BBC’s Reality Check

A recent tool seen as an important buttress against misinformation, and increasingly employed by media organisations, are specific fact-checking departments (such as Channel 4’s FactCheck or the Washington Post’s Fact Checker) or stand-alone organisations (such as PolitiFact or Full Fact). While fact-checking is, of course, part and parcel of all news reporting, dedicated fact-checking services aim to examine in detail claims made by actors such as politicians and reach conclusions about the veracity of the claim (Cushion et al. 2022a; Cushion et al. 2022b). Typical fact-checking articles use independent expert opinion, reliable sources of data and numerous triangulating sources to enable them to reach a conclusion (see Graves 2016, for a detailed account, and Soo et al. forthcoming). BBC Reality Check was established in 2017, and, according to the Head of BBC News at the time, the service would be “weighing in on the battle over lies, distortions and exaggerations” (cited in Jackson 2017). Birks (2019a, 2019b) examined the extent to which fact-checkers, including Reality Check, weighed in during the 2017 and 2019 UK General Elections. This work found a greater embedding of fact-checking journalism within main news coverage, yet also detailed the public’s concerns about what they perceived as a subjective form of reporting being inherently biased. Nonetheless, recent studies suggest the public are receptive to greater fact-checking in routine journalism (Kyriakidou et al. 2022; Morani et al. 2022).

Fact-checking as a form of explanatory journalism conforms well to the wagon wheel metaphor outlined in the Bridcut Review and is therefore an influencing style one might expect to now see in BBC news output beyond Reality Check. Yet, as noted by Graves (2016), fact-checking takes time, and often does not finish at a deadline, leading to one interviewee remarking to Graves that, “[journalists] don’t have time or they’re on deadline and … so it becomes ‘he said, she said’” (97). Our study will examine the degree to which claims, and counterclaims are reported by Reality Check, and consider how they interpreted “due impartiality.”

We therefore hypothesise that, despite BBC guidelines, the operationalisation of impartiality may change dependent on the electoral context (reporting domestic or foreign politicians), the specific politician being reported on, and the media platform (reporting online, in a fact-checking article or television news bulletin).

Methodology

Through content analysis, a sample of election-related news items was examined on the BBC News website (including Reality Check) before the 2019 UK General Election (28th November–11th December 2019) and the 2020 US Presidential Election (1st September–3rd November 2020). The sampling was systematic: all the standard news articles within the two timeframes were included excluding opinion pieces and editorials, live updates and video-only pieces. The different time ranges (10 weeks of coverage for the US sample and 2 weeks for the UK) were necessary to obtain two consistent and comparable datasets with approximately the same number of political claims in them. The BBC News online articles reporting the US 2020 Presidential Election were retrieved in December 2020 by using the link to the devoted US 2020 section (https://www.bbc.com/news/election/us2020) of BBC News. The UK 2019 General Election articles were accessed through the archived content of the Internet Archive organisation (https://web.archive.org/) because those published in the Election 2019 section of the BBC News website were no longer available (https://www.bbc.co.uk/news/election/2019).

The online UK election sample contained 128 election items of which 47 featured political claims by one or both main party candidates (Boris Johnson and Jeremy Corbyn). The US sample included 173 election items with 46 items having political claims by one or both of the two presidential candidates (Donald Trump and Joe Biden). Items in the online sample refer to individual stories on the BBC news website. Although Reality Check is hosted on the BBC Online site, they were subdivided as BBC Online or Reality Check. We therefore analysed these sections of the BBC website separately.

The broadcast sample consisted of all election-related content from the BBC News at Ten programme. For the UK sample, shows broadcast between 28th November and 11th December 2019 were examined. For the US sample, a longer period was required to obtain a similar amount of news items. This meant all programmes between 30th September and 3rd November 2020 were examined. For the broadcast sample, the unit of analysis was the news item categorised by the type of convention rather than by the topic of the story.² The UK broadcast sample generated 72 election items with 24 items featuring political claims by one or both candidates. While the US sample generated 49 items with 27 items featuring political claims by either candidate.

A claim was operationalised as any instance where one of the four politicians made any politically relevant statement where the statement’s veracity could be challenged.

Therefore, manifesto pledges of future intentions were not included. Politically relevant statements encompassed all policy areas broadly conceived, electoral processes and personal statements with potential electoral influence. Interactions were defined as instances where an element within the story explicitly or partially/implicitly challenged a politician’s claim (Corrective Interactions—CIs) or completely or partially validated the claim (Validating Interactions—VIs). Null Interactions (NIs) were also recorded—these were instances where a claim was published but received neither a corrective nor validating interaction.

The variables coded for both the online sample and broadcast sample were split into two categories—item level and interaction level. For the item level, the unit of analysis was the whole online news article or broadcast news item. For the interaction level, the unit of analysis was each individual interaction with a claim. Variables coded for at the item level were as follows (see Appendix 1 for the full lists of codes):

• resence of political claim—this was whether the item contained a political claim.

• Item topic (e.g. Brexit, healthcare)—the most dominant topic of the item was coded from the list in Appendix 1.

• Fact-checking tool—whether it was a Reality Check item or an item which mentioned Reality Check.

The following were coded at the interaction level:

• Author of claim—which politician made the claim (Johnson, Corbyn, Biden, Trump).

• Claim topic—what topic was the claim about (e.g. Economy, Healthcare). The same list in Appendix 1 as used for Item topic.

• Interaction manner—(Corrective Interaction/ Validating Interaction/ Null Interaction). CIs and VIs were also further coded as either Explicit or Partial/Implicit categories to capture the degree of correction/validation being undertaken.

• Interaction source category—this was a list of categorisations of who interacted with the claim to supply a corrective or validation. The journalists may themselves interact with text, such as “that is incorrect,” (coded as “BBC journalist”) or another source (e.g. “according to a Labour Party spokesperson” would be coded as politician/party). The full list of categories is in Appendix 1.

• Interaction type—journalist text were instances where the text spoken or written by the journalist directly interacted with the claim, e.g. “that’s not right.” Direct quote were instances where the journalist quoted another source interacting with a claim. Indirect quote was where the journalist paraphrased a source. Instances of the use of charts or tables were coded as Figure. Any use of numerical information was coded as Numerical data. Hyperlinks to other web pages were followed and that page given an overall evaluation as to how it interacted with the claim: these were coded as Hyperlink. Image types were instances where a claim was interacted with via an image—typically this would be an enlarged pull quote in online stories. In broadcast, these were instances where B-roll imagery was manifestly interacting with a claim. For example, audio from a Donald Trump speech claiming that “[immigrants at the US-Mexico border] are so well taken care of. They are in facilities that are so clean,” being juxtaposed with visuals of the centres showing the opposite.

• Interaction placement (Immediate/Delayed). Immediate corrections online were classified as those which came within three paragraphs of the claim or within three paragraphs of another interaction which was classified as immediate. For the broadcast sample, delayed interactions were classified as those where a substantive point was broadcast between claim and interaction.

In addition, broadcast data was coded for the following variables:

• Item convention—the type of broadcast convention used conforms to prior studies (e.g. Cushion and Lewis 2017) of televised news such as pre-recorded package or standalone anchor piece.

• Claim made directly or not by author (Direct/Indirect)—this coding specifies if the politician’s own words were broadcast, or they are quoted by the journalist. The Author of indirect claim variable (Reporter/Anchor/Source) specifies who was quoting the politician.

• Source of interaction (Direct/Journalist) clarifies if the person or body interacting with the claim is broadcast directly or quoted by a journalist. These latter two were interaction level variables.

Two coders each separately coded approximately half of the online and broadcast items. An inter-coder reliability test was undertaken on a random sample of 10% of items (sample n = 43) and of interactions (sample n = 100). All variables achieved over 90% agreement rates and high Cohen’s Kappa scores—ranging from 0.9 to 1.0 (see Appendix 2 for details).

Findings

Overall, we found journalists applied different levels of scrutiny when covering political leaders during the US and UK elections (see Table 1). In total, n = 967 interactions (either a corrective interaction (CI), validating interaction (VI) or null (NI)) with claims were coded, including 435 interactions with Trump, 303 with Johnson and far less with Corbyn (119) and Biden (110). It is not unusual for an incumbent to receive more coverage than a challenger because of government announcements and policy initiatives having an intrinsic news value. However, the periods examined here immediately preceded the respective elections and were overwhelmingly campaign related stories across the two major parties in both the UK and the US. This meant that although the stories covered certain policy areas, they were claims or criticisms of records on that policy area. Therefore, a simple incumbency effect does not adequately account for the differing levels of coverage.

Table 1. Interactions by politician.

	Biden	Corbyn	Johnson	Trump	Total
N interactions	n = 119	n = 110	n = 303	n = 435	n = 967
Interaction manner
Corrective (CI)	38%	50%	70%	73%	65%
Validation (VI)	24%	18%	7%	15%	14%
Null (NI)	38%	32%	22%	12%	21%
Corrective Strength	n = 42	n = 60	n = 213	n = 315	n = 399
Explicit Corrective (ECI)	50%	65%	53%	71%	63%
Implicit Corrective (ICI)	50%	35%	47%	29%	37%

Note: Table column percentages may not total 100% due to rounding and/or omitted rows.

We also discovered the level of journalistic interactions varied when reporting different political leaders. For Trump, 73% of journalistic interactions with his claims were CIs, with Johnson having a similar figure of 70%. Corbyn (50%) and Biden (38%) had far lower proportions of corrective interactions. Of the CIs, 71% of Trump’s were explicit compared to 65% for Corbyn, 53% for Johnson and 50% for Biden. An example of a typical Explicit Corrective Interaction appeared on Oct 27th when a Reality Check article covered a claim made by Trump about Covid-19 that “[the US has] one of the lowest mortality rates.” It was immediately followed by the journalist stating: “Verdict: That’s not right. The US ranks high globally in terms of covid deaths per person.” This form of abrupt correcting by a journalist was reserved almost exclusively for Trump. When UK politicians were explicitly corrected, even on Reality Check, such an abrupt form was not utilised, and any particularly strong rebuttals were usually delivered by quoting other political operatives—invariably a Labour politician countering a Conservative claim or vice-versa, e.g. an 8th December Reality Check article contained a claim by Johnson that there would not be any checks for goods travelling from Northern Ireland to Great Britain under the Brexit deal. This claim was followed by a journalist stating: “Labour said the PM's claims about his deal with the EU were ‘fraudulent’.”

Overall, Trump made 187 unique claims, which received 315 CIs. This means that for every claim he made, he was corrected on average 1.7 times. Several of Trump’s statements, such as the claim to have built the greatest economy in the history of the US, received detailed rebuttals with numerous CIs. For Boris Johnson, his 168 unique claims received 213 CIs, meaning the average corrective was 1.3 for every claim.

Although extensive correcting of Trump did occur, there were still instances where unsubstantiated claims were not challenged. Following his stay in hospital with Covid-19, for instance, Trump’s return to the campaign trail was covered by News at Ten on 13th October. The anchor introduced the item with “The president told his supporters that he was now ‘immune’, and that he felt ‘powerful’.” The claim of immunity went unchallenged. Yet, official World Health Organisation guidance and the National Health Service position at that time was that it was unknown whether having Covid-19 granted natural immunity to further incidence of catching the virus. In the same news item, Trump also made false statements against Joe Biden, saying, “We have somebody running that’s not 100%. He’s not 80%. He’s not 60%.” This completely unsubstantiated claim that Joe Biden was mentally unfit to hold the office of president was also reported on BBC news online on 14th October, with Trump saying, “He's shot, folks. I hate to tell you, he's shot … can you imagine if I lose to a guy like this? It's unbelievable.” Again, this falsehood went unchallenged. How far journalists should challenge or even ignore false claims has been subject to fierce debate over recent years. While it could be argued that repeating falsehoods may help promote and spread misinformation, research has suggested that reporting falsehoods in the context of a corrective can correct misunderstanding (Porter and Wood 2019). However, reporting it is still an editorial choice, a choice which, without rebuttal, likely adds a layer of legitimacy to the claim (Lewis & Cushion, 2019).

Table 1 revealed that we found a high level of journalistic interacting with, and scrutiny of, political claims made by all four leaders. However, as illustrated in Table 2, there were clear differences in the way that BBC journalists applied them in US and UK election campaign coverage. On stories related to the US election, BBC journalists were the source of just over half of CIs with claims by Trump (51%) and Biden (52%) but far fewer for Corbyn (21%) and Johnson (27%) in UK election coverage. Rather than issuing personal correctives, in UK election coverage journalists typically edited coverage so other politicians or parties corrected the claims for them—a classic "she-said-he-said" style of reporting. On closer inspection, we found it was far more common for journalists to draw on direct quotes as correctives in UK election coverage than in US election coverage. The US coverage conversely contained much higher correctives with journalist’s own intervention into a claim being the main type of interaction. We also found correctives of Trump and Biden were more likely to appear immediately following a claim than those for Johnson and Corbyn, which was often included later in a news story, typically following contextual paragraphs.

Table 2. Number, type, and placement of corrective interactions by politician.

Corrective Interactions (CI)	Biden	Corbyn	Johnson	Trump	Total
	n = 42	n = 60	n = 213	n = 315	n = 631
Source of CI
BBC journalist^a	52%	21%	27%	51%	40%
Politician/Party	14%	57%	38%	13%	25%
Other^b	5%	13%	17%	14%	14%
Government^c	24%	8%	8%	16%	13%
Non-BBC media	5%	0%	2%	6%	4%
Type of CI (selected)
Direct quote	15%	55%	38%	19%	29%
Indirect quote	7%	13%	17%	29%	25%
Figure	10%	0%	2%	7%	5%
Numerical data	14%	5%	8%	9%	9%
Journalist’s text	40%	24%	24%	32%	25%
Hyperlink	14%	2%	11%	17%	13%
Placement of CI
Immediate	95%	70%	59%	81%	74%
Delayed	5%	30%	41%	19%	26%

^a All interactions without any source attribution are coded as BBC journalist.

^b This category covers sources not covered by the other listed categories. Sources such as NGOs, charities, academics, think tanks and experts.

^c Any government source or agency such as law enforcement and statutory agencies. Party spokesperson is included in party/politician category.

Overall, the data suggests BBC journalists were far more reluctant to directly correct UK politicians than US politicians despite, in theory, applying the same “due impartiality” to news output.

Table 3 shows the policy topics on the claims interacted with by journalists. For UK election coverage, they broadly fall in line with the issues both leaders were championing. With Boris Johnson, journalists most frequently interacted with his claims on Brexit (almost all being correctives) and crime and law and order issues. By contrast, most interactions with claims made by Corbyn were on healthcare, Brexit (illustrative of this issue’s dominance in the campaign), and the environment. In the US election, Covid-19 understandably dominated the coverage, but more so in the reporting of Biden, with coverage of Trump more balanced across topics. However, it was notable that Trump’s statements about immigration were frequently covered and corrected.

Table 3. Corrective interactions and validating interactions by politician.

	Biden	Corbyn	Johnson	Trump	Total
Interactions (CI or VI)	n = 68	n = 81	n = 235	n = 382	N = 766
Brexit	0%	19%	36%	0%	13%
Economy/business/tax	15%	11%	8%	12%	10%
Environment	7%	15%	0%	12%	8%
Healthcare	57%	22%	17%	25%	25%
Immigration	1%	0%	0%	12%	6%
Law and order	9%	10%	18%	13%	14%
Defence/terrorism	7%	4%	5%	8%	7%
Other	3%	19%	15%	18%	16%

Table 4 illustrates the different ways journalists interacted with claims according to the BBC platform: either the BBC News at Ten broadcast, BBC online or the BBC Reality Check pages of the website. All 134 Reality Check interactions were either correctives (76%) or validations (24%). Of course, since Reality Check aims to provide a fact-checking service, this degree of scrutiny of claims and holding politicians to account is perhaps to be expected. But it does confirm the BBC’s fact-checking services deliver a more robust approach to challenging claims than television and online reporting. Many online stories provided a link to a Reality Check fact-check, allowing readers to check the veracity of claims made either directly in the article being read, or more generally in the area of policy being reported.

Table 4. Interactions by media type.

	Online	Broadcast	Reality Check
Interactions	n = 665	n = 168	n = 134
Corrective (CI)	66%	55%	76%
Validating (VI)	14%	4%	24%
Null (NI)	20%	40%	0%
Corrective Interactions (Type)	n = 436	n = 92	n = 102
Direct quote	32%	43%	1%
Indirect quote	15%	15%	1%
Figure	6%	3%	1%
Numerical data	8%	5%	16%
Journalist’s text	20%	30%	68%
Hyperlink	17%	0%	13%
Corrective Interactions (Strength)
Explicit (ECI)	65%	59%	62%
Implicit (ICI)	35%	41%	38%

However, on television news, Reality Check was not mentioned once. It is worth pointing out of course that viewers for the BBC News at Ten on television are vastly greater than the number who would be exposed to a Reality Check article (Kyriakidou et al. 2022).

While Reality Check corrected or validated all claims, the non-null interactions with claims online (80%) were notably greater than in broadcast (59%). The 40% null interaction figure for broadcast meant that many statements made by the politicians were broadcast without any guidance to viewers about the veracity of a claim, or even exposure to a counterclaim by an opposition politician. We did note instances where a claim was met with a counterclaim which though related and on topic did not directly address the original claim—a "he-said-she-said-something-related" style.

In short, Reality Check applied “due impartiality” differently with a far more robust approach to challenging claims and counterclaims. Most Reality Check correctives were in the form of journalist’s own interpretation, and frequently these were clear-cut verdicts such as, “that’s not right” (e.g. in the September 30th Reality Check story, this is the wording which follows Trump’s claim to have “built the greatest economy in history”). This compared to the far greater use of sources providing quotes as correctives in other online stories and on broadcast (e.g. a 6th December News at Ten item examining the leader’s debate included the claim that the Conservatives were building 40 new hospitals—a claim balanced by a rebuttal from Corbyn saying the true figure was six).

To help illustrate the differences between reporting across BBC platforms, it is worth reflecting on how this specific claim about building new hospitals was comparatively dealt with by television news, online news and Reality Check. Prior to, and during the election period, Conservative politicians repeatedly claimed that the Conservative government was building 40 new hospitals.³ This claim was covered in the 6th December News at Ten item. Here, apparently responding to previous criticism of the claim, Johnson explained in more detail and ended with a diluted claim of, “there will be forty new hospitals.” Corbyn then argued that the claim of building 40 hospitals became twenty and later six. It was presented as a “he-said-he-said” story. It is pertinent that government documents, available at the time, showed that over half of the planned builds were to existing facilities, and therefore not “new” hospitals by any reasonable interpretation of the term. The original claim of building 40 hospitals was extensively rebutted on Reality Check and was clearly labelled as misleading and inaccurate. Amongst other correctives, numerous NHS Trust sources were quoted as stating that there were no plans for, let alone active construction of, new hospitals. The detailed contradiction concludes with, “so it's not correct to suggest that 40 new hospitals are currently being built.” It is a well-formed fact-checking article which may be summarised as “he-said-he’s-wrong.” BBC online featured a story on December 3rd with the following claim: “Boris Johnson is constantly championing the 40 new hospitals he wants to see built.” The underlined phrase is a link to a Reality Check page which summarises the full Reality Check article discussed above, including the same blunt conclusion. The journalist here uses the hyperlink facility as a form of corrective rather than including a judgment within text, thus effectively outsourcing the corrective to a BBC colleague.

Television news coverage, by contrast, simply balanced competing claims by first featuring the claim by Johnson about 40 new hospitals and then offering a rebuttal from Corbyn. The potential dangers of giving false equivalence to competing claims when the weight of evidence significantly supports only one claim has been widely discussed. The weight of evidence here, provided by one part of BBC news, was that 40 new hospitals were not being built, but perhaps due to concerns of appearing impartial in the much more high-profile flagship television news bulletin meant that this conclusive judgement was not conveyed to audiences, but only limited to those either directly reading Reality Check or those motivated enough to click through to the online story to find the fact-check. We drew on this claim in detail to illustrate how television news simply balanced soundbites on an issue, whereas online news provided a far more robust and nuanced examination of this policy commitment because of the fact-checking role played by BBC Reality Check. In the UK government's Health Infrastructure Plan (Department of Health & Social Care, 2019), then Health Secretary Matt Hancock wrote, “We’re giving the green light to more than 40 new hospital projects across the country, six getting the go-ahead immediately, and over 30 that could be built over the next decade.” A claim of “forty new hospital projects … that could be built” is different to a claim of “will be forty new hospitals,” and very different to, “we are building forty new hospitals.” But it was only the BBC Reality Check's fact-checking service that provided the necessary degree of scrunity to challenge a key manifesto promise.

It is perhaps to be expected that online news has a greater level of journalistic interaction with claims than broadcast news given the limited airtime on television. However, to investigate this assumption further we worked out the comparable length of time it took to read broadcast and online coverage. We found the average duration of broadcast items was 130 seconds, with online items on average taking around 2.5 times that duration to read. On average, 1.3 claims were made per online item and 0.9 per broadcast item. This meant that, standardising for duration, approximately double the rate of claims were made in broadcast news than online. Yet, each inaccurate claim in broadcast was corrected half as frequently as online. This meant, after controlling for the different formats of television and online platforms, politicians were allowed to make far more claims on television that were far less likely to be contested. In other words, the medium is the message: online BBC reporting applied “due impartiality” far more robustly than television coverage. Different story archetypes perhaps provide a neat, if imperfect, summation of these findings. On broadcast, the stories are typically “he-said-she-said.” Online they are more akin to, “he-said-she-says-he’s-wrong,” whereas Reality Check has a model closer to “he-said-he’s-wrong.”

Towards More Comparative Scrutiny of how Journalists Construct Balance: Why the Political Context Shapes “Due Impartiality”

This study developed a new systematic research design and conceptual understanding of how impartiality was interpreted and applied across different political contexts and media platforms. Our UK case study examined BBC news output which is required to be duly impartial when reporting all election campaigns, whether domestic or foreign. Yet our findings illustrated clear differences in how impartiality was applied in the UK and the US context. When covering UK elections, BBC journalists tended to resort to a more 'she-said-he-said' style of reporting compared to the reporting of the US election where political claims were more directly challenged. Nossek’s (2004) work has suggested there are differences in how journalists report foreign and national issues—and our case study about election reporting in the UK and US has reinforced this perspective. But our research has added more empirical weight and detail to precisely how journalists' apply impartaility when reporting a "foreign" (US) and "national" (UK) election campaign. Put simply, BBC journalists were more emboldened when reporting a foreign than domestic election, directly questioning statements by Donald Trump.

In the UK election campaign, however, journalists appeared more concerned with constructing balance when interpreting rules about impartaility, drawing on other parties to challenge the claims of political leaders. Without interviewing journalists or editors, we cannot establish whether this was due to external influences, such as the greater party-political pressure they may receive when reporting domestic rather than foreign news issues.

But in our view, the differences between the BBC’s domestic and foreign coverage of elections were not only influenced by the greater scrutiny they would have received in their UK coverage. The editorial construction of US and UK coverage articles would have been influenced by the closer relationships which journalists have with domestic rather than foreign political actors. For example, UK journalists would have more frequent contact and access to better-established relationships with UK party communications teams than US counterparts. This would facilitate a greater supply of “information subsidies” (Gandy 1982) which help shape the construction of stories and perhaps explains why more UK than US leaders and parties were quoted in domestic election coverage.

This sociological influence and the domestic pressure to report impartially also helps explain specific instances where US and UK election coverage was different. For example, Trump’s egregious claims about Biden’s mental acuity were reported without any journalistic challenge. It may be the case that the sheer volume of correctives required of Trump carried concerns of an appearance of bias and therefore ignoring claims of lower perceived importance, or perhaps credibility, was perceived as restoring some balance.

BBC editorial guidelines state that due impartiality should be employed throughout the organisation, but they acknowledge how it is achieved may vary, “depending on the format, output and platform” (BBC 2019). Our study has revealed such variation. In broadcast news, a limited conception of the range of views (i.e. he-said-she-said) was given primacy. This reinforces Wahl-Jorgensen et al.’s (2017) analysis of BBC news that argued an “impartiality-as-balance” approach shaped coverage. The BBC’s Reality Check meanwhile drew on a far greater range of perspectives and offered an evaluation of the weight of those range of views. BBC news online coverage was typically somewhere in-between—offering a range of views with less consideration of their relative weight. The pull towards the perhaps perceived safety of "she-said-he-said" in broadcast reporting particularly is possibly a reflection of its more prominent status, with coverage scrutinised more closely by the public and politicians. "He-said-she-said" is a long-established and comfortable form stretching back to the Reithian ideal discussed earlier. It is therefore more embedded in the construction of broadcast news than in newer digital platforms. Reality Check was specifically designed to produce a form of news that would help to counter misinformation. It is a style of journalism, as this study shows, which considers both the range and relative weight of opinions and thus a closer embodiment to the wagon wheel envisioned by Bridcut. Yet, reporting in other parts of the BBC is still more accurately encapsulated by the seesaw.

The relevance of our findings goes beyond our case study of the BBC. Fact-checking journalism has risen in prominence in recent years, an implicit acknowledgement that the "he-said-she-said" approach to reporting does not work effectively if one, or perhaps both sides, regularly make dubious or false claims (Cushion et al. 2022b). The previously reliable employment of opinion balance is perhaps now an impartiality marker mitigated by the (un)reliability of the sources of information. Moreover, recent research about how audiences believe misinformation should be countered by broadcasters has demonstrated there is strong support in favour of more fact-checking style journalism (Cushion et al. 2022; Kyriakidou et al. 2022), echoing findings by Barthel, Gottfried, and Lu (2016). We would argue that more finely tuned studies are needed to understand how journalists apply concepts about balance and impartiality in political reporting. Our study went beyond analysing coverage at a macro level (story) by assessing reporting on a micro level (political claim), establishing whether a claim was present, if it was challenged or not, and how explicitly it was corrected within a news item. We would encourage journalism scholars to more creatively fine tune studies that empirically reveal editorial choices in how impartiality is constructed across different media platforms and online sites.

A follow-up qualitative analysis, in this respect, would likely reveal more nuanced differences than those which could be captured at the scale of data here. For example, there were several instances in our sample where the same claim covered across the different media was coded as an Explicit Corrective Interaction, yet the language used in broadcast seemed softer than that online, which in turn was softer than that used on Reality Check. Our study was deliberately designed to capture data at election periods as these are the periods of perhaps greatest importance in terms of possible democratic consequence, and the time when media receive extra scrutiny for impartiality and balance. There is no strong theoretical reason to believe that findings for a non-election period would alter from our direction of findings, but it is acknowledged that the strength of findings could be different; lesser scrutiny may embolden more BBC journalists to embed greater and stronger fact-checking practices.

This case study, in examining the BBC, questioned how the concept of impartiality was operationalised. We found efforts to achieve due impartiality altered depending on the domestic or international context of the story, the politician involved in the story and the platform producing the story. Viewed from this perspective, impartiality appears to be a fluid rather than fixed concept and applied differently according to the political context.

Impartiality was even interpreted differently between and within departments of the same news organisation. This prompts the need for more empirical inquiries about how impartiality is interpreted more widely. The paradigm of “impartiality-as-balance” (Wahl-Jorgensen et al. 2017) is still often the default position of journalism, evidenced here as being employed particularly in broadcast and more in a domestic than international setting. At a time characterised by misinformation, we would argue that rather than simply reporting what "she- said" and what "he-said", journalists must reflect the relative weight of evidence when reporting political claims, and apply more robust scrutiny and interrogation about what "she said" and what "he said".

Acknowledgements

The authors would like to say diolch yn fawr to the anonymous reviewers and the special edition editors for the time they took to read the work in progress and provide useful and constructive feedback to help improve the work.

Disclosure Statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

The authors would like to acknowledge the Arts and Humanities Research Council (grant award AH/S012508/1), who provided the funding to enable this research.

Notes

1 As online content is beyond its remit, Ofcom does not have a direct oversight and enforcement role of BBC online content. However, under the Digital Economy Act of 2017, Ofcom has a responsibility to opine as to whether BBC online material conforms to its own editorial standards (Ofcom 2020).

2 See Appendix 1 for the list of conventions. Anchor intros were coded as separate items if the duration was over 30 seconds. Otherwise, anchor intros were included as part of the main convention of the item.

3 The precise Conservative manifesto wording was that they are proud, "to have begun work on building 40 new hospitals across the country." Johnson used phrases such as "we are building 40 new hospitals." In the November spending review, Chancellor Rishi Sunak spoke of “building 40 new hospitals.”

Appendices

Appendix 1. Coding Variables

Table

Variables—online and broadcast
Item topic	Fact-Checking Tool
Brexit	Reality Check story
Healthcare	Reality Check mentioned
Taxation
Economy/business	Author of Claim
Education	Boris Johnson
Housing	Jeremy Corbyn
Poverty/inequality	Donald Trump
Terrorism/defence/foreign affairs	Joe Biden
Employment/wages
Welfare/benefits	Claim Topic
Social care	[same as item topic]
Pensions
Transport	Interaction manner
Environment/climate change	Explicit corrective
Legal/Constitutional reform	Implicit/partial corrective
Crime/law and order	Implicit/partial validation
Prisons/sentencing laws	Explicit validation
War/conflict	Null interaction
Immigration
Tech companies/digital tools	Interaction Type
Race Relations	Direct quote
Media Regulations	Indirect quote
Manifesto (i.e., multiple policies)	Figure (e.g., chart, table)
Opinion polls/ Horse Race	Numerical data
Voting process	Journalist comment
Political tensions/leadership issues	Hyperlink
Campaign trail/rallies/strategies	Image
Party Spin/ PR/News Management	Tweet
Campaign funding /party accounts
Political scandal/malpractice	Interaction Placement
Personal life	Immediate (within three paragraphs of the claim or prior immediate interaction)
TV debates	Delayed
Other
Interaction source category
BBC journalist
Journalist/Media (external, not BBC)
Politician/Political Party	Variables—broadcast only
Non-UK/EU Politician/Political Party
Economist	Item convention
Think Tank	Anchor only
Charity	Reporter package on location
Campaigner/Pressure Group	Anchor/reporter live two-way
Trade Union	Anchor/reporter live studio interview
Business	Studio discussion (live)
Entertainment	Studio discussion (pre-recorded)
Academic	Interview with guest (live)
Education	Interview with guest (pre-recorded)
Healthcare/NHS	Package studio
Government Department/Statutory Agency	Video diary
Non-Statutory Agencies
Law Enforcement Agencies	Claim made by
EU Institution/regulations/MEP	Direct by claimant
Citizen	Indirect
Pollster/Opinion polls
IGO/NGO	Supplier of Indirect claim
Lawyer/Legal Organisation	Anchor
Scientist/health/medical Expert	Reporter
Expert Other (e.g., data analyst)	Source
BBC Expert
Reality Check	Supplier of interaction
Other	Source direct
No interaction	Source indirect (journalist)

Appendix 2. IRR Scores for Individual Variables

Table

Variable No	Variable description	Level of agreement, with Cohen’s Kappa (CK) in brackets
Database Variables
1	TV Convention	100%
2	Is there a political claim (yes/no)	100%
3	Item topic category	100%
4	Fact-checking tool/feature	97.7% (0.85 CK)
5	Disinformation/fake news/conspiracy theory mentioned (yes/no)	97.7% (0.88 CK)
Interactions Variables
1	Author of claim	100%
2	Topic of claim	98% (0.98 CK)
3	Source of interaction	97% (0.97 CK)
4	Manner of interaction	96% (0.94 CK)
5	Type of interaction	98% (0.98 CK)
6	Placement of interaction	99% (0.98 CK)