Studying statistical manipulation

Manipulation is an important but somewhat confusing concept. Social psychologists have long investigated manipulation, but they differ among themselves about its essential characteristics, whether it involves deceit or is a particular form of persuasion (Buss, Gomes, Higgins, & Lauterbach Reference Buss, Gomes, Higgins and Lauterbach1987; Gass & Seiter Reference Gass and Seiter2018). Some critical discourse analysts have viewed manipulation as a key, critical concept that identifies illicit persuasive techniques (Fairclough Reference Fairclough, Mey and Asher1998; Chilton Reference Chilton, Saussure and Schulz2005; de Saussure & Schulz Reference de Saussure and Schulz2005; van Dijk Reference van Dijk2006; Maillat Reference Maillat2013). The attempts to produce an agreed upon definition, let alone theory, of manipulation have not been successful. People can manipulate in different ways, and therefore rather than seeing manipulation as a single phenomenon, it might be more pragmatic to distinguish between different ways of manipulating (Whitfield Reference Whitfield2020). For example, manipulating people directly to believe or do something may be very different from manipulating information, which involves illicitly changing the meaning of information (Billig & Marinho Reference Billig and Marinho2014, Reference Billig and Marinho2017). Manipulating statistics is a form of manipulating information. Although manipulating information involves deceit, it is not usually the same as outright fabrication: in ordinary conversation a distinction is often made between fabrication and manipulation (Huma, Stokoe, & Sikveland Reference Huma, Stokoe and Sikveland2021). Those who are said to manipulate statistics do not usually invent numbers, but they might collect, analyse, or report statistics in knowingly distorted ways, while dishonestly presenting the data as being objective and unbiased (Billig Reference Billig2021).

The centrality of statistical data in modern society has given rise to the constant possibility that governments might regularly manipulate their country's statistics. Most nation-states have created specialized bureaucracies for producing official, national statistics. France, for example, has the Institut national de la statistique et des études économiques (INSEE) and the USA has the Federal Statistical System. In the United Kingdom, on which this article concentrates, there is the Office of National Statistics. During the Covid-19 pandemic such organizations have been responsible for compiling official national figures relating to the disease, with international organizations such as the World Health Organization using nationally based data to compile international databases (Billig Reference Billig2021).

As Espeland & Stevens (Reference Espeland and Stevens2008) have remarked, professional statisticians are better placed to manipulate statistical data than the statistically untrained. Given that macro-economic variables can be measured in different ways, governments may put pressure on their official statisticians to collect and describe statistical data in politically favourable ways. Prewitt (Reference Prewitt2010) discusses such pressures, but does not specifically call them ‘manipulation’. By contrast, Aragão & Linsi (Reference Aragão and Linsi2020) have outlined different ways politicians have overseen the production of biased macro-economic statistics, and they describe these as statistical manipulations. Indeed, it might be easier to manipulate macro-economic data than simpler data, for ordinary members of the public might not have views about how gross domestic products should be measured, but they will have a firmer grasp about the topic that we are examining.

In this article, we examine an exchange of letters between the Conservative government's health minister, who was manipulating Covid-19 data, and the head of the UK Statistics Authority (UKSA), an official organization for preventing statistical manipulation and misuse. We examine how the minister was challenged about the way that the number of tests for Covid-19 were calculated at first diplomatically, and how he responded to that challenge. The issue at stake is one that the public could understand. One does not need statistical training to know what it means to take a test for a disease like Covid-19, and the public is likely to suspect that something fishy is going on if politicians include tests that no one has taken in their total figures for tests carried out.

In our work, we have generally taken a bottom-up, psychological approach that begins with examining examples extending over time, rather than with theory or definitions. The aim is to observe social action as directly as possible in order to understand what is going on, rather than using episodes to serve or illustrate theory (Billig Reference Billig2013, Reference Billig2019; Billig & Marinho Reference Billig and Marinho2017). In the present article, we did not approach the analysis of the exchange with a strictly determined methodology because, in our view, what is being analysed should largely determine how it is analysed (Billig Reference Billig2013). We are certainly not the first analysts of public letters to approach their material in this way: David Kaposi took this stance when he analysed the famous public exchange of letters between Hannah Arendt and Gershom Scholem (Kaposi Reference Kaposi2008, Reference Kaposi2009).

This does not mean that we approached the material completely free of assumptions. We have adopted a premise from critical discourse analysts. There may be a variety of ways to analyse discourse critically (Wodak & Meyer Reference Wodak and Meyer2009; van Dijk Reference van Dijk2012), but whichever type of analysis is used, critical analysts emphasise that it is not sufficient merely to describe linguistic characteristics, but these should be related to wider social, political, and ideological processes. Similarly, we assume that people perform actions with language (e.g. Edwards & Potter Reference Edwards and Potter1992; Potter Reference Potter, Wetherell, Taylor and Yates2001; Wodak Reference Wodak2009), and that this means, if one wants to know why the writer of a public letters is writing in a particular way, then we must try to understand what that letter-writer is doing. As we see in the analysis of our example, it is sometimes not immediately apparent what the letter-writer is doing and that close examination of the text and knowledge of the wider background are required.

UK Statistics Authority

The United Kingdom might resemble other countries in having a national institution for producing official statistics, but unlike many other nations it invests this institution with an independent regulatory section, responsible for ensuring good statistical practice, especially in relation to official UK statistics. The UK Statistical Authority was established in 2007 by the Labour government and its legally constituted duties include formulating, and regularly updating, a Code of Practice for Statistics. The latest revision was published in 2018 (UK Statistics Authority 2018). This Code asserts its own authority, stating that it ‘should be used by all those in government who produce and use statistics’ (2018:7, emphasis in original). If ever there was an official institution devoted to combatting statistical manipulation, this is it, although the Code does not specifically use the word manipulation.

Given that monitoring the government's statistics is one of the Authority's central duties, it is unsurprising that UKSA constantly stresses its independence from government. For instance, its Code describes UKSA as ‘an independent statutory body’ that ‘operates at arm's length from government’ (UK Statistics Authority 2018:2). UKSA regularly uses the same self-description on its website (e.g. UK Statistics Authority 2020a). In these self-descriptions, we can see the importance of managing what social psychologists have called ‘communicator credibility’ (e.g. Hovland & Weiss Reference Hovland and Weiss1951), and rhetorical theorists term ‘ethos’ (Sullivan Reference Sullivan1993; Bruss & Graff Reference Bruss and Graff2005). To be a credible organization only concerned with trustworthy statistics, UKSA needs to manage its reputation as being independent from government and from all political parties.

Anyone appointed as its chair can expect to clash from time to time with governments and their ministers. The current chair of UKSA, Sir David Norgrove, took up his position in early 2017. He is an economist, not a statistician, and has a long and distinguished career in the civil service. When he was much younger, he served as the personal secretary of the then Prime Minister, Margaret Thatcher. By his professional background, Norgrove is very much an insider, well versed in the ways that the higher levels of the civil service operate.

UKSA has published a brief guide to its policies on intervention, especially in cases where public bodies were failing to meet the desired standards of statistics (for the latest updated version of this guide, see UK Statistics Authority 2021). In many cases, the Authority's statisticians, including its chief statistician, would try to work with the statisticians of governmental and other bodies in order to correct any failings. The Authority might decide that it does not need to make public statements, especially if the failings are speedily and satisfactorily rectified. By contrast, the UKSA guide specifies that if the issue relates to ‘broader political use, especially by Ministers and elected representatives’, then the chair of UKSA, rather than the chief statistician, will respond particularly by writing a public letter.

In an interview with the magazine Civil Service World, Norgrove discussed the thinking behind this policy (Dunton Reference Dunton2017). He said that Ed Humpherson, UKSA's Director General of Regulation and chief statistician, would write most of the letters about statistical failings, but if the misuse was ‘particularly bad, or if it's a very senior person or a minister who's misused data’, then he, as the Authority's chair, would write a public letter which would be published on the Authority's website. Norgrove emphasised that ‘letters from me should be the last resort’.

Norgrove also said that UKSA had limited powers because parliament had not given it the right to censor misleading statistics: for that would be, he said, ‘almost an undemocratic way of behaving’. The Authority could only point out politicians’ errors and ‘hope that either they respond or that the public pressure forces them to respond’. Here, we see Norgrove's twin strategies: he wants to work behind-the-scenes to settle statistical problems, but also, as a last resort, he wants to be able to expose misleading statistics publicly. As we see below, these contradictory strategies can lead to the chair to use politely diplomatic language and also language that is more directly critical.

Comparatively early in the pandemic, there was an exchange of four public letters between Norgrove and Matt Hancock, the Minister of Health, concerning government statistics about the number of tests conducted for Covid-19. Norgrove wrote the first letter on May 11, 2020 with Hancock replying on May 27. Norgrove responded on June 2, and Hancock's response followed on June 11. That was the end of their public exchange of letters.

Background and manipulation

At the start of April, Hancock had announced his target of administering a hundred thousand tests a day by the end of the month. On May 1 he claimed to have met his target (Billig Reference Billig2021), and, five days later, opposition politicians, including the leader of the Labour party, claimed in parliament that the figures did not support Hancock's claim. They offered two main reasons: the hundred thousand tests had been achieved only on one day, not daily; and, secondly, the numbers of tests included tests that had not actually been administered. Labour's new health spokesperson actually claimed that ‘testing figures are now being manipulated’. We have shown elsewhere that the minister was indeed manipulating the testing figures (see Billig Reference Billig2021).

Daisy Cooper, a Liberal Democrat Member of Parliament, wrote to Ed Humpherson, expressing concern that the government's claims about testing were ‘a misrepresentation of the statistics’. She asked UKSA to check the validity of Hancock's claims (UK Statistics Authority 2020b). Humpherson passed the letter to Norgrove who, when he wrote his first public letter to Hancock, forwarded a copy of the letter to Cooper (UK Statistics Authority 2020c).

The Code of Practice devotes a whole subsection to the virtues of clarity, so one might expect that the chair, when writing to a government minister, would express himself as clearly as possible. By constrast, if Norgrove wanted to work with the minister to rectify statistical errors, then he needed to maintain relations with Hancock, not putting him in a position where to save face he would defend poor statistics. Therefore, as we see below, Norgrove used the sort of language that is often called ‘diplomatic’—language that operates by hints and unexpressed implications. We hear in this public exchange of letters the voice of a manipulative politician who writes plainly, and a diplomat who unlike Lewis Carroll's Alice neither quite says what he means nor means what he says.

Beginning of first letter

Analysts who seek to interpret political language usually have to make selections. In the present case, we cannot give equal attention to all of the passages of all of the letters. It might be thought that a simple rule would suffice: select the most politically dramatic passages and analyse those. Sometimes, however, passages of discourse that initially seem unremarkable can reveal more about what a speaker or writer is doing than the dramatic passages (Billig & Marinho Reference Billig and Marinho2017).

The start of Norgrove's first letter to Hancock contained nothing to attract the headline writers (UK Statistics Authority 2020d). Following the formal address ‘Dear Secretary of State’, Norgrove's opening paragraph was:

The footnote for the superscript 1 comes at the end of the letter. It is a link to the parliamentary report of Prime Minister's Question Time for May 6, 2020, when Boris Johnson announced the government's target of 200,000 tests a day (for details see Billig Reference Billig2021).

What is unsaid, or unwritten, can sometimes be more revealing than what is (Billig & Marinho Reference Billig, Marinho, Murray and Durrheim2019; Murray & Durrheim Reference Murray and Durrheim2019; S. Scott Reference Scott2019). As Sherlock Holmes said in The Adventure of Silver Blaze, the dog that didn't bark is curious. Norgrove's opening contains no aggressive barking—nothing to warn Hancock.

Significant absences should be understood in relation to routine presences, especially those that are so routine that they often pass unnoticed. Norgrove's opening paragraph breaks standard conventions for a formal letter. According to the BBC's How to write a formal letter (n.d.), ‘Your opening sentence should clearly state why you are writing the letter—get to the point straight away’. After ‘Dear Sir/Madam’ should come: ‘I am writing to apply for the position of…’; ‘I am contacting you about the poor service I received …’, and so on. The rest of the letter, then, should be rhetorically constructed to support the stated purpose.

By contrast Norgrove's opening paragraph does not inform Hancock why he is writing. The three clauses are phrased as statements of fact about the government's goals for testing Covid-19 and the media's reporting of these goals—all of which Hancock can be presumed to know. There is a significant omission: Norgrove does not write that the government's target and its supposed ‘progress’ (a strange word, given that much of the media coverage had been critical) had been recently debated in parliament, with opposition politicians making strong statistical criticisms.

Most of Norgrove's public letters begin conventionally. When replying to letters sent to UKSA, Norgrove will first thank the sender and then indicate how he will proceed. Sometimes, Norgrove initiates correspondence by referring to a letter that the Authority had received from someone else. In December 2020, he began a letter to the leader of the Scottish Liberal Democrats: ‘I am writing about correspondence the UK Statistics Authority received in relation to the First Minister of Scotland's reply in the Scottish Parliament to your oral question on 17 September’ (UK Statistics Authority 2020e). Norgrove could have begun his letter to Hancock similarly, by saying he was writing about correspondence received from Daisy Cooper MP concerning the figures for Covid-19 testing. Norgrove did not mention Cooper or her letter. It is not immediately obvious why Norgrove is writing, but his purposes become clearer when we consider the linguistic construction of that opening paragraph, and what happens in subsequent letters.

The impersonal opening

Norgrove's opening paragraph is surprisingly impersonal. Conventionally formal letter-writers use first and second person pronouns and/or adjectives at the start. The opening sentence of Norgrove's letter to the leader of the Scottish Liberal Democrats contains two second person words and one first person: ‘Thank you for your letter of 7 May… which has come to my notice’. These words establish why I, the letter-writer, is writing to you, the recipient.

Norgrove's opening paragraph to Hancock contains no first or second person words. The grammatical subject of the first two clauses is ‘the Government’ that announces the goals for testing. There is nothing remarkable in depicting the Government as an actor, taking decisions, making announcements, and so on. In this case, it is curious because the person in the Government who made the first announcement was the person to whom Norgrove was writing. Hancock not only announced the goal but, when he did so, he described it as ‘my goal’ (Billig Reference Billig2021). The opening paragraph carries a footnote, which only references the announcement of the second goal. There is no footnote for the first goal: that would have referenced Hancock's announcement or, in the context of the letter, your announcement.

The phrasing and the omission are too much of a coincidence to be accidental, especially in a formal, carefully worded letter. By his apparently factual opening paragraph, Norgrove would seem to be establishing that the topic of his letter is the government's Covid testing statistics. However, Norgrove does not state that there might be something wrong with those statistics nor that his addressee might be responsible for any such errors. Both the letter-writer and the letter-recipient know who announced the goal of achieving a hundred thousand tests a day. Both also know that the same person announced that the goal had been achieved, declaring it to be ‘an incredible achievement’ (see Billig Reference Billig2021 for details of Hancock's May 1, 2020 press briefing). That announcement is unmentioned in the opening paragraph.

We can presume that Norgrove is doing something deliberate by avoiding first and second person words. By not linking Hancock to the first announcement of government aims, and by making the paragraph impersonal, he seems to be writing diplomatically, not threatening Hancock's face in front of the public. Norgrove does not write: ‘I am writing to you because there is something seriously wrong with the numbers of tests that you are using to claim that you have met your goals’. Then again, he does not specifically deny that this is why he is writing.

The paratactic voice of diplomacy

The conventional start of a formal letter rhetorically sets up the letter's argument. An unwritten ‘because’ joins opening and subsequent paragraphs: having initially stated that you are applying for a post or complaining of the service you received, you then give reasons why you should be selected for the post or be given compensation for the poor service. By contrast, Norgrove's initial paragraph conveys no explicit sense of argumentation.

In their New Rhetoric (1969) the great rhetorical analysts Chaim Perelman and Lucie Olbrechts-Tyteca distinguished between argumentative and non-argumentative discourse. They drew upon the grammatical distinction between hypotaxis and parataxis, which the German grammarian Friedrich Thiersch introduced into the study of language in the nineteenth century (Dymarsky Reference Dymarsky2014). Perelman & Olbrechts-Tyteca wrote that the ‘hypotactic construction’ is the ‘argumentative construction par excellence’, whereas non-argumentative structure is paratactic (Reference Perelman and Olbrechts-Tyteca1969:158). The terms parataxis and hypotaxis draw attention to contrasting relations between clauses. In paratactic discourse, the clauses are arranged, as it were, side-by-side, with none being given importance over the others. The rhetorical theorist Richard Lanham (Reference Lanham2003:29) describes parataxis as ‘syntactic democracy’ with each clause equal, unlike hypotaxis where writers syntactically rank their clauses for the benefit of their readers, often using words such as because or therefore, thereby indicating how one clause might rhetorically serve another (see also Fahnestock Reference Fahnestock2011).

The three clauses in Norgrove's opening paragraph are paratactic, for they are set side-by-side with none syntactically being put above the others. Nor does Norgrove outwardly use the rest of his letter to support his opening paragraph, because it appears to be purely descriptive. There is no dispute that the announcements were made and that the press covered what happened. However, it is possible to construct hypotactic discourse without using words like because or therefore: argumentative connections between statements can be semantically implied rather than outwardly stated (Halliday Reference Halliday2003; Bertinetto & Ciucci Reference Bertinetto and Ciucci2012).

The distinction between parataxis and hypotaxis is important for understanding Norgrove's letter for a specific reason. Biljana Scott, a social scientist with a background in linguistics and an interest in studying diplomacy, has used the distinction to highlight the characteristics of diplomatic language. She claims that diplomatic language often conveys argumentative meaning while being syntactically paratactic (B. Scott Reference Scott2013, Reference Scott2018). Diplomats generally seek to keep diplomatic channels open, and Scott suggests that they can do this by using underspecified, indirect language, thereby avoiding the sort of direct arguments that might offend, threaten face, or rupture relations. Diplomats convey meanings without stating them, and they expect other diplomats to understand what they are doing: ‘Parataxis, the juxtapositions of phrases and sentences with no indication of the link between them, is a rhetorical device that invites us to jump over the gap by inferring plausible connections’ (B. Scott Reference Scott2018:55). In contrast to nineteenth-century linguists who assumed parataxis to be the mark of childish and/or ‘primitive’ language, Scott sees diplomatic parataxis as the mark of a highly sophisticated form of communication.

To discover what Norgrove was conveying, but not specifying, in his syntactically paratactic opening paragraph, we must look beyond that paragraph, which was conveying an unsaid message to Hancock: ‘I am writing diplomatically and politely, but we have found problems with your statistics about testing’. Lurking between the lines of the written words lie unformulated threats that the recipient (and the analyst) must decode.

Coercive praise

Following the opening paragraph, Norgrove switches immediately to personal pronouns: ‘I know you are a strong supporter of the proper use of statistics’. He is praising his letter's recipient, but there is also unexpressed coercion. He cannot disagree: ‘Oh no, I don't support the proper use of statistics’. The coercive element is that if Hancock strongly supports the proper use of statistics, then, by implication, he should support what Norgrove has to say, for the Statistics Authority is the independent UK arbiter on the proper uses of statistics. If Hancock rejects the Authority's authority, he would reveal himself as less than a strong supporter of the proper use of statistics.

As often happens, the small words in a diplomatic statement can carry a message to insiders. We should not presume that the phrase ‘I know’ operates discursively in a carefully phrased public letter as it does in the sort of informal interaction studied by Herder, Berenst, de Glopper, & Koole (Reference Herder, Berenst, de Glopper and Koole2020). Norgrove will have had expert advisers helping him draft his public letter, selecting each word with care. What he does not write can be significant in helping to decode what he does write. Norgrove does not write ‘I believe’ you are a strong supporter or simply state ‘you are a strong supporter’. Instead, he writes unequivocally: ‘I know you are a strong supporter’.

We might ask what Norgrove knows about Hancock's strong support for statistics. Since diplomatic language operates on the recipient knowing what the speaker/writer hints at, we should be asking if Norgrove knows that Hancock knows that he knows about Hancock's strong support. It might sound a complex question whose intricacy cannot be untangled by any analyst lacking private, psychological knowledge of Hancock and Norgrove. In point of fact, the untangling is not difficult.

This was not the first time the UKSA had queried Hancock's use of statistics (Matthews-King Reference Matthews-King2018). An incident occurred in 2018 not long after Hancock had been appointed Minister of Health. Again, Hancock was boasting about numbers, claiming in a tweet that there had been a ‘terrific’ increase of a thousand National Health doctors in just three months. Doctors and others claimed that Hancock's figure was misleading, because the vast majority of the thousand were students, not qualified doctors. One doctor contacted UKSA, which publicly responded: ‘We have discussed this matter with the Secretary of State's office, and they have removed the tweet’.

There seems to have been behind-the-scenes pressure on Hancock, most likely coming from Norgrove, given the Authority's policy of passing matters involving ministers to the chair. The tweet had been Hancock's and it is unthinkable that officials in his department could have removed it without the minister's approval. The indirect, diplomatic mode of expression lessened the chance of the new minister losing face.

The meaning of Norgrove's ‘I know’ is becoming clearer: it is a diplomatic reminder of an incident that Hancock would not like to be publicly reminded of. We know that Norgrove knows that Hancock is liable to exaggerate numbers for political purposes. Norgrove also knows that Hancock backed down when confronted by the Authority. And Norgrove knows that Hancock knows that he knows this. Norgrove is not complimenting the minister innocently; he is criticising him diplomatically.

Hinting towards the point

In the main part of his letter, Norgrove does not overtly discuss the two charges that Cooper had raised with the Authority: that the number of tests had been boosted inappropriately to meet Hancock's target; and that the figures for the daily rate of tests did not meet the target. Norgrove does not directly say that the government's statistics are inaccurate. Using carefully hedged, indirect language, he raises questions about their presentation.

Regarding the government's targets, Norgrove writes that ‘for the sake of clarity and confidence’, it is important that the government clarify what precisely its target refers to: whether it is ‘testing capacity; tests that have been administered; test results received; or the number of people tested’. Combining these different things into a ‘total national number of tests could mask helpful operational detail’. The verb ‘could’ hedges the statement, suggesting that operational detail might not be masked. Norgrove does not say that the overall total is misleading–but diplomatically he conveys the possibility.

Norgrove does not just say that the extra information should be given ‘for the sake of clarity’, as if he is merely recommending a clearer presentation of basically sound data. He adds ‘and confidence’, without stating whose confidence might be affected. Again, he sounds as if he is trying to be helpful by recommending ways of presenting statistics so that people in general might accept them with greater confidence.

Behind Norgrove's expressed words stands an unexpressed general principle: one cannot (even, should not) have confidence in data whose meaning is unclear. Norgrove does not clarify this general principle. However, he writes that ‘it would support trustworthiness for the testing data to be more straightforward to find, with detailed breakdowns and richer commentary’. Again, he uses a noun (‘trustworthiness’), rather than saying ‘people would trust the testing data more’. And again, he does not say that Hancock and his department might be deliberately withholding information in their presentation of data.

Significantly Norgrove adds no words to the effect ‘but, of course, I absolutely trust the numbers that you are presenting unclearly’. He implicitly includes himself and the statisticians of UKSA among those who might not have confidence and trust in the testing statistics. By implication, rather than by direct assertion, the purpose of his letter is becoming clearer. His diplomatically hedged language implies that the current testing statistics are not to be trusted.

In his final paragraph, Norgrove asserts without diplomatic hedging: ‘We urge Government to update the Covid-19 national testing strategy to show more clearly how targets are being defined, measured and reported’. Norgrove has not entirely abandoned diplomatic language: he writes that because Covid-19 data is ‘inevitably complex’, it is all ‘the more important that publications should meet the standards set by the Code of Practice for Statistics’. He is not directly claiming that the government's testing figures fail to meet the Code's standards, but by mentioning the Code in this way, he is implying it. Why is it so important that the government defines its targets precisely? Norgrove is not yet saying.

Hancock's reply May 27, 2020

Diplomatic language possesses a fundamental weakness: it can be easily misunderstood or ignored. A person speaking diplomatically can only successfully communicate their intended meaning to those who can, or who choose to, recognize unexpressed messages. A diplomat cannot force recipients to read between the lines. If they try to force their meaning on recipients, then, as likely as not, they will end up speaking non-diplomatically. Some recipients may be unfamiliar with the sophisticated codes of diplomatic speakers. Others may deliberately choose to take diplomatic messages literally, especially when it is in their interest to do so.

Hancock waited sixteen days before replying to Norgrove and then he broadly ignored Norgrove's diplomatically expressed criticisms (UK Statistics Authority 2020f). Unlike Norgrove, Hancock wrote in plain, unhedged language. Grammatically plain language, however, is not the same as honest language, for a speaker/writer can use plain language to tell untruths or when illicitly manipulating information (Billig & Marinho Reference Billig and Marinho2014, Reference Billig, Marinho, Harris and Fahnestock2022).

Hancock claimed that he was already doing everything that Norgrove was asking him to do. In his first paragraph, Hancock affirms that he ‘strongly’ supports clear and open presentation of statistics, and he uses grammatically plain language: ‘As you know I have authorised publication of the most full and complete data about our COVID-19 response’. Note the lack of hedging: ‘the most full and complete data’; and also the active voice: ‘I have authorised…’. Note also Hancock's phrase ‘as you know’: Hancock knows that Norgrove does not accept that the data on testing is full and complete, otherwise Norgrove would not have written his letter.

In his second paragraph Hancock absorbs Norgrove's diplomatic praise, taking it as if it had been warmly intended without any coercive element: ‘I welcome your positive comments about the information we publish about testing each day’. Again, the verbs are in the active voice, plainly telling what I welcome and what we are doing.

As well as writing about what I/we have been doing and are doing, Hancock also writes what I/we are about to do: ‘We are today publishing a clear definition of how our target of capacity to perform 200,000 tests a day by the end of May will be measured and reported’. It sounds clear and reassuring, as if it should satisfy Norgrove and UKSA. Nevertheless, grammatically plain language can suggest more (and less) than it appears to. Hancock does not indicate what this very soon-to-be-published clear definition might be. It is curious that Hancock withholds the information on the very day that it is supposedly to be published.

If Hancock felt that he could not publicly let Norgrove know the definition before its publication, then he could have waited a few hours before sending his letter to Norgrove. Having already waited seventeen days before replying to Norgrove, a few more hours would make little difference. Hancock seems to have timed his letter so perfectly that he is able to avoid informing Norgrove what the clear definition of the target was, while reassuring him that there was one imminently on its way. It is as if he was trying to hide something. This suspicion is increased because no new definition appeared later that day, or the following day, or the day after. When Norgrove replied on June 2, he complained that he was still waiting for the new methodology to be published.

We should not expect a match between the structures of language and motive. Devious motives and devious speech acts can result in clear messages, even clear messages that are defending unclear statistics.

Becoming less diplomatic: Norgrove's second letter

The public letters between Norgrove and Hancock resemble interactional talk in that they are sequential. Discursive psychologists have stressed that sometimes the meaning of an utterance only become clear after others have reacted to the utterance and the original speaker defends, clarifies, or repairs their original utterance (e.g. Edwards Reference Edwards1997:100ff). Similarly, to understand the meaning of Norgrove's first letter, we do not need to know about his state of mind as he was writing the letter. We need to examine his public response to Hancock's response.

Norgrove's first letter had failed to persuade Hancock that he needed to change how Covid-19 testing was being calculated and the data was being presented. Norgrove's second letter is much changed from his first. The event that made Norgrove alter his style was not the publication of new data with greater inadequacies, nor was it the promised new methodology. It was Hancock's dismissive reply. If Norgrove was going to have an effect on the minister, he needed to scale back the diplomatic approach and either put public pressure on Hancock more directly or use the sort of language that would catch the attention of journalists whose reports would then put pressure on the minister. In his second letter Norgrove reduced the amount of hedging, upgraded the strength of his critical terms and, above all, pieced together his argument more directly. As such, he was moving from the parataxis of the first letter towards an argumentative, hypotactic style.

Diplomatic language is not an all or nothing style but it is graded so that one can speak, or write, more diplomatically or less diplomatically. Because Norgrove's expressed strategy was to work with the minister to improve the quality of the data, he was not going to write completely undiplomatically. He continued to write impersonally, using a roundabout way to criticise Hancock's actions while avoiding associating those actions with Hancock personally.

Norgrove's second letter, unlike his first, has a conventional beginning, which features first and second person singular pronouns, and gives reasons for the letter. It begins: ‘Thank you for your letter of 27 May, in which you described some welcome, though limited, additions to the official data on COVID-19 tests, including a proposed note on methods (not yet published at the time of writing)’ (UK Statistics Authority 2020g). Politely he welcomes the changes that the minister mentioned, but without any diplomatic hedging or minimising he calls them ‘limited’. They are not ‘somewhat limited’, ‘a bit limited’, or ‘could be thought to be limited’: they are limited. Also, his opening sentence mentions that the note on methods, which Hancock had promised, remained unpublished. Specific criticism was unnecessary: merely mentioning the non-publication in his opening sentence was rhetorically sufficient.

The opening sentence functions to explain why Norgrove is writing: the minister should not think that his letter had solved the problems that Norgrove had diplomatically raised. The recipient can expect the rest of the letter to argue for this. The only other sentence in the first paragraph underlines this: ‘I am afraid though that the figures are still far from complete and comprehensible’. Again, there is no diplomatic modifying to lessen the strength of criticism. Quite the reverse, the criticism is augmented: the figures are “far from” being complete and comprehensible.

Conversation analysts use the term upgrade to indicate when a speaker increases a prior evaluative assessment, for instance, giving an even more positive, or more negative, assessment than a previously expressed positive or negative assessment (Bilmes Reference Bilmes2019). We might borrow the term upgrade and say that Norgrove's second letter upgrades the critical assessments of his first letter. In his first letter, he suggested that the government's way of presenting the data on testing casts doubt on its ‘trustworthiness’ and the ‘confidence’ people might have in the data. In the second letter the delicately expressed critical language is upgraded, as Norgrove uses more directly critical words such as ‘misleadingly’, ‘inadequate’, and ‘mistrusted’.

When a speaker upgrades the assessments of a previous speaker, they frequently are not demonstrating that they hold a more intense, and thus different, assessment than the previous speaker (Pomerantz Reference Pomerantz, Atkinson and Heritage1984). They tend to upgrade for interactional reasons, as a non-upgraded response might be understood as offering only lukewarm agreement or even dissent with the previous speaker (Edwards Reference Edwards1997:145f). Norgrove is upgrading his own assessments, not those made by someone else. After receiving Hancock's letter, which complacently dismissed, or knowingly overlooked, Norgrove's diplomatically expressed concerns, Norgrove had strategic and, thus, interactional, grounds for upgrading his previous language. The upgrading would make his dissatisfaction clear and, thus, interactionally unavoidable.

Norgrove: Making the implicit explicit

In his second letter, Norgrove makes explicit some of his implicit criticisms, as he ties together argumentative themes that had previously been left disconnected. Norgrove had written that the overall total number of tests needed to be broken down into separate categories in order to understand better the nature of the disease. Now, in his second letter, Norgrove makes his reasoning devastatingly clear. The first purpose of the testing statistics is ‘to help us understand the epidemic’, but ‘the way the data are analysed and presented currently gives them limited value’. Norgrove writes that he will mention ‘just a few issues in relation to the data as currently presented’, thereby implying that there are many more problems that he could discuss.

Significantly, the first problem that he mentions is that the total number of tests carried out includes ‘tests posted out’ but not ‘carried out’. This was one of the charges that Daisy Cooper made in her letter to UKSA and that opposition politicians made to Hancock in parliament (Billig Reference Billig2021). The evidence for the charge was to be found in the Department of Health's own methodological notes—notes, which despite Hancock's claim in his letter, were still in operation unchanged. Those notes made clear that testing subcontracted to private companies would be counted in terms of the number of tests that the companies mailed out, whether or not these resulted in completed tests (Billig Reference Billig2021). Thus, the problems with the data were not merely presentational, but they contained serious methodological flaws.

Norgrove did not stop there. He claimed that that the methodological difficulties were compounded by presentational problems. The distinction between tests ‘carried out’ and those that had only been ‘posted out’ was ‘too often elided during the presentation at the daily press conference, where the relevant figure may misleadingly be described simply as the number of tests carried out’. Norgrove's language is strong and unequivocal, but there is still a vestige of diplomacy.

Norgrove uses the passive voice, which permits him to avoid identifying who might be describing the data in a misleading manner (Billig Reference Billig2008, Reference Billig2013). He certainly does not identify the person whom he is addressing. Yet, Norgrove might assume that Hancock would recognize himself and might even recognize that Norgrove was being tactful. One of the most illustrious pieces of critical discourse work connected the use of passives with the exercise of power, as commands expressed passively are made to appear as if they are ‘objective’ rather than the work of particular named individuals or groups (Fowler, Hodge, Kress, & Trew Reference Fowler, Hodge, Kress and Trew1979). The context of power is central to Norgrove's use, but his diplomatically passive constructions follow from his lack of power over a government minister.

Towards the end of the letter, Norgrove writes a paragraph that seems to summarise his main criticisms. It starts with a sentence of praise:

Again, Norgrove's praise is not innocent, but it delivers an unconcealed kick. Not only do the testing statistics (not just their presentation) fail to meet the standards of the Code but they ‘still fall well short’. Then the rhetorical boot strikes its target with upgraded firmness: ‘It is not surprising’ that such data is ‘so widely criticised and mistrusted’. He implies that critics are correct to criticise and mistrust such inadequate figures.

Amongst the criticisms of methodology and presentation, Norgrove includes a comment that makes sense of his argumentative strands, which are still not drawn together in tight hypotactic order. Norgrove comments on what the government is doing when presenting its figures for the total number of tests. Its errors appear not to be accidental: ‘The aim seems to be to show the largest possible number of tests, even at the expense of understanding’.

If we convert Norgrove's paratactic style into a hypotactic argument, we can see why his first letter began so strangely: its opening sentence was a description of the government's targets for testing without this being a major theme in what followed. In his second letter, Norgrove comes closer to stating directly that the official statements of aims were in some way related to the inadequacy of its statistics: namely, that the government had political motives for trying to show ‘the largest possible number of tests’. To use the sort of language that Norgrove, ever the diplomat, would not use: the government had motives to manipulate its statistics.

Norgrove's primary aim, however, is to prevent manipulation. To achieve this, he needs to work with the producers and presenters of statistics, especially those who might be tempted to manipulate numbers. Norgrove did not end his second letter to Hancock with accusations. He publicly welcomed the willingness of Hancock's department to work with the Authority ‘to discuss how the data and their presentation could be improved’. He looked forward to future cooperation, finishing on a positive note of agreement: ‘I am sure you would agree that good evidence, trusted by the public, is essential to success in containing the virus’. These words, like similar words in Norgrove's first letter, were reaching out to the minister in a warm, coercive embrace.

Final letter

Hancock's reply, which he sent on June 11, was the shortest of the four letters (UK Statistics Authority 2020h). Basically, he expressed his willingness to work with UKSA. At one level, this represented a success for Norgrove's diplomatic approach. He had not alienated the minister and now they would cooperate out of the public eye. Norgrove did not reply with a further public letter—he had achieved a seemingly happy ending, or perhaps a cooperative beginning. As so often is the case in politics, what Hancock did not write is as significant as what he did. A happy ending for Hancock may not have been quite the same as a happy ending for Norgrove.

From the start of this fourth letter, it was clear that something had changed. Hancock had begun his first letter to Norgrove with the conventionally formal ‘Dear Sir David’, but his second letter began ‘Dear David’. This was not a sign that Hancock was rudely ignoring Norgrove's title. Quite the reverse, it was a display of informal friendliness. Hancock hand-signed his letter, as he had his first one, ‘Matt’. He was publicly conveying that the minister and the chair were on good, first-name terms.

Matt began by thanking David for his previous letter, which had set out ‘your concerns with our presentation of statistics on testing’. As if to emphasise that he had not experienced any change of heart between first and second letters, Matt wrote: ‘I said in my first letter that I am keen for us to work closely with your team on developing the presentation’. He emphasised that the Matt-David cooperative show was already underway. They had spoken on the day he had received David's last letter. While his department continues ‘to publish daily updates on testing numbers, I've asked my officials to continue to work with you on improving the presentation of these’ (emphasis added).

Nowhere in this letter does Hancock deal with any of Norgrove's specific statistical points. Nor does he mention the Code of Practice which, according to Norgrove, the statistics on testing had failed to meet. Hancock writes as if Norgrove had been only concerned with the presentation of statistics, rather than the methodology that produced those statistics. But something had changed.

We should not assume that Hancock had been directly persuaded by the arguments in Norgrove's second letter—arguments he had ignored when they were expressed in the diplomatic hints of the first letter. In the worlds of politics and diplomacy the psychological assumption that changes of outward position necessarily represent changes of inner attitude is far too simple. As is well-known, politicians will often be evasive when asked whether they have changed or not changed their minds (Clayman & Heritage Reference Clayman and Heritage2002; Bull Reference Bull2003, Reference Bull2008).

Hancock had reasons for changing his position and also had reasons for not wishing to appear to have changed his position. His reasons for changing his position are simple. Only two British national newspapers had reported Norgrove's diplomatically phrased first letter and both were left-of-centre broadsheets. Hancock could well believe that he could withstand the criticisms of the Guardian and Independent, neither of which supported his government. What had changed with Norgrove's second letter was that the next day, British newspapers, whether tabloid or broadsheet, whether left-wing or right-wing, picked up the story that the chair of the Authority was criticising the minister (for details see Marinho & Billig Reference Marinho, Billig and Ilie2023).

As Norgrove indicated in his interview with Civil Service World he could not force politicians to change their statistics, but he could try to shame them. A technical analysis tucked away on the webpages of the Statistical Authority have virtually zero effect. Stories in the press, especially in the parts that are normally supportive, could put a politician in a difficult position. We have good grounds for supposing that Hancock was discomfited by the publicity provoked by Norgrove's second letter.

If Norgrove's strategy had been to attract the interest of the press, then he had judged his upgraded terminology well. Practically all of the national papers reported that Norgrove had accused Hancock of ‘misleading’ the public and/or that the testing statistics could not be ‘trusted’ (Marinho & Billig Reference Marinho, Billig and Ilie2023). Some upgraded Norgrove's upgrades. The right-wing tabloid, The Sun, wrote that the ‘watchdog’ accused Hancock of ‘fiddling’ the data. Norgrove, of course, had not used ‘fiddling’, a slang term which implies deliberate dishonesty. When the friendly parts of the press use this word, politicians should fear for their reputation.

So, Hancock sends a co-operative reply, and his possible political crisis passes. Norgrove has achieved his stated aim of working behind the scenes to correct statistical errors. Diplomacy combined with undiplomatic threat seem to have won the day, but there was another day to come.