Sage Journals: Discover world-class research

Abstract

Statistics are a central part of political communication, yet little is known about how they are used rhetorically by politicians. This article therefore develops a rhetorical understanding of statistics in political debate and explores how they function primarily as strategies of argumentation. Through an analysis of how British politicians use numbers in debates on the National Health Service ‘winter crisis’, it is argued that four tropes underpin the use of statistics as a rhetorical device. The trope of dehistoricisation is said to engender consensus over the facticity of statistical arguments, while the tropes of synecdoche, enthymeme and framing are said to enable contestation over their presentation and meaning. The article concludes that a rhetorical understanding of statistics is vital to elucidating the selective, contestable and strategic ways in which numbers function in political debate, thereby challenging the notion that quantification can be an objective or value-free means of establishing political claims.

Keywords

Crisis NHS politics rhetoric statistics

Introduction

There has been a resurgence of interest in rhetorical approaches to the study of politics. At the heart of this research effort is a belief that strategies of argumentation are an important prism through which we can understand political language and the political process more broadly. In particular, the development of Rhetorical Political Analysis (RPA) has encouraged a number of scholars to identify and analyse the different rhetorical devices mobilised by political actors and to situate these in relation to wider political phenomena. Many different rhetorical strategies, such as narrative, hyperbole, anecdotes and quotation, have been related to wider political concepts, like strategy, problem definition, institutions, tradition and ideology (Atkins, 2015; Atkins et al., 2014; Crines and Happell, 2017; Finlayson, 2006; Finlayson and Martin, 2008). While this research effort means that we know a lot about how politicians mediate the political process with their words, we still know very little about how they do this with numbers.

This is a significant omission, especially when we consider the central role that statistics play in political debate (Prevost and Beaud, 2012; Simon, 2008). Discussions over important political phenomena like public policy regularly trade on statistical accounts of public reality, where numerical indicators are leveraged by politicians to define problems, to argue for political change and to critique and defend government records. This signals a pressing need to better understand how these accounts function as strategies of argumentation, and to incorporate such an understanding into existing rhetorical approaches to politics.

Our aim in this article, therefore, is to develop a rhetorical understanding of statistics in political debate, using the National Health Service (NHS) winter crisis as an example. We begin by outlining the privileged position that statistics have in political debate, which in part rests upon their perceived objectivity. Despite this, we cast doubt on the notion that statistics can provide a neutral and undistorted means by which we can conduct political argument. The interpretive and open-ended nature of political debate, we argue, means that the role of statistics in the establishment of political claims is rarely impartial or apolitical. Instead, we suggest that a rhetorical approach can help us to better conceptualise the role that quantifiable information plays in political debate.

We then apply this understanding to an analysis of four Prime Minister’s Questions (PMQs) debates from 2016 to 2018, which all focus on the NHS Winter crisis. We identify and discuss four predominant rhetorical tropes that pervade these debates and, at a wider level, are central to the way statistics function to underpin political arguments. We argue that the trope of dehistoricisation serves to engender consensus between orators over the facticity of statistical arguments, whereas the tropes of synecdoche, enthymeme and framing enable contestation over their presentation and meaning. Crucially, we contend that these tropes illuminate the selective, contestable and strategic ways in which numbers function in political argument. Following our analysis, we conclude by discussing the implications of these findings. Emphasis is placed on normative concerns regarding political truth and the importance of this argument in the context of the coronavirus pandemic.

Numbers, politics and rhetoric

Since their inception in the late 18th century, statistics have been inextricably linked to the nation state (Desrosieres, 1998). Their connection is evident in the word itself: statistic is derived from the German statistika, which itself is derived from a combination of Italian and New Latin terms for ‘the state’. Few governments would initially embrace the logic of quantification, with many policymakers retaining a healthy suspicion towards the political interests and methodological variation involved in the production of supposedly impartial statistical information (Poovey, 1993). After the so-called ‘avalanche of numbers’ in European society from the 1820s to the 1840s, however, statistics became a mainstay in Western politics (Hacking, 1990). As Coleman (2018) puts it, ‘from the 1830s onwards, governments began to base truth claims upon statistical inferences’ that could be ‘re-presented as unassailable facts’ (p. 160). Some have suggested that the last three centuries have seen a further intensification of this process, with numbers taking on even more significance in governance (Kurunmaki et al., 2016). This can be observed most explicitly in the contemporary push towards ‘big data’, where states seek to manage phenomena through large-scale data collection, publication and communication (Dunleavy, 2016).

In this centuries-long process, numbers have come to attain a privileged position as an allegedly apolitical and objective mode of socio-political description. Theodore Porter (1995) argues that we trust numbers because of their perceived objectivity. This objectivity stems from what Desrosieres (2001) terms ‘metrological realism’, a viewpoint that ‘computed moments (averages, variances, correlations) have a substance that reflects an underlying macrosocial reality, revealed by those computations’ (p. 348). In other words, statistics are said to reveal an objective ontological reality or truth that was previously hidden. To approach numbers in such a fashion, however, is to ignore the way quantitative information is highly contingent on a series of decisions: what should be counted, how it should be counted, how phenomena are categorised, what analysis should be applied and how this information should be communicated (Best, 2008). In this way, quantitative information does not reveal an ontological reality but functions as a particular representation of the world around us (Latour and Woolgar, 1979).

Therefore, when statistics are used in public discourse, they function rhetorically, depending ‘on a conversation, an argument among human beings about what they are going to take seriously’ (McCloskey, 1987: 487). There is now a growing body of work dedicated to outlining the various ways in which statistics function rhetorically in public discourse, with much of this work rooted in journalism studies. Particular attention has been paid to the ways in which statistics can confer a certain authority, legitimacy and objectivity on the people and institutions using them. Roeh and Feldman (1984), for example, examined how statistics in news media function as ‘agents of a rhetoric of objectivity’, creating an ‘impression of nothing-but-the-facts-journalism’ (p. 347). Because statistics can bestow trustworthiness and an aura of impartiality on those who use them, Koetsenruijter (2018) has suggested that they are a product of ethos, the branch of rhetorical argument geared towards character. Statistics, in this way, are primarily geared towards establishing the credibility and reliability of speakers.

This apparent impartiality and systematicity can mask the often selective and strategic ways in which statistics are produced and disseminated. The rhetorical power of statistics is that they can seem like authoritative, mirror descriptions of tangible social facts rather than a partial description of reality. According to Lugo-Ocando and Lawson (2018), one problematic domain in which this plays out is the mediation of official poverty statistics. These numbers are created and presented by organisations to fit specific media requirements, with the aim of establishing public support for already determined policies and solutions. This happens, they argue, because ‘these numbers are seen as neutral entities that underpin journalistic “truth,” not as contested tools that help to construct social reality in specific ideological terms’ (Lugo-Ocando and Lawson, 2018: 74).

Aviles (2016) has observed a similar process in public policy reform, drawing on the specific case of a Puerto Rican Pension Report. The authors who drafted the report, he argued, were ‘good in the art of producing a statistical rhetoric that appeals to a lay general public that bestows credibility to numbers’, adopting a ‘realist style of political rhetoric that facilitates representing themselves as disinterested scientists that interpreted what the data assert and explained their incontestable conclusions’. Rather than revealing anything objective or necessarily true about pension reform, the report only revealed ‘what its authors thought to be appropriate evidence and persuasive arguments for public policy debate and deliberation’ (Aviles, 2016: 49).

When it comes to public discourse, then, the enduring power of statistical arguments lies in the way they can ‘seemingly expel private interests and ambiguity’, and ‘offer the possibility of systematically undistorted communication’ (Peters, 2001: 436). Yet, as the above examples show, this hides the fact that they are ultimately ‘strategies of communication’ (Porter, 1995: viii), developed, presented and communicated according to some partial political interests. The promise of an objective and undistorted politics, underpinned by rigorous methods of quantification, is fundamentally at odds with a rhetorical view of political communication, where there is said to be no political argument that can be removed from the influence of certain political ideas, beliefs and strategic intentions (Finlayson, 2004).

To see political argument as partial or strategic is not to devalue it, or to, as some critics might do, ‘subsume rhetoric into the category of illegitimate coercion’ (Finlayson, 2004: 538). Instead, it is to address concepts like political ‘truth’ and ‘reality’ in interpretive terms. Democratic debate involves a variety of voices, ideologies and value-systems, each bringing with them conflicting versions of truth that defy any single statistical description of political reality (Coleman, 2018). Since most political issues hinge upon interpretive and culturally contingent questions of meaning and justice, they often cannot be empirically verified or resolved. For this reason, political truth is never finally described, but is instead an ongoing rhetorical endeavour, one in which the competing visions of socio-political reality emerging from these different subject positions are negotiated and articulated through various modes of ‘proving, pleasing, and persuading’ (Finlayson, 2014).

In this way, what is problematic in rhetorical terms is the way in which the process might be stymied by the undue faith placed in particular rhetorical forms, which can in turn close the space for a more inclusive and diverse rhetorical contestation and democratic engagement. Here, the current role that statistics play in political debate presents a particular problem for a rhetorical understanding of political truth. The notion that objective political reality is ‘out there’ and discoverable through techniques of statistical calculation dangerously overstates the role that quantification can play in the search for political truth. Statistics are just one form of representation and argument in political debate, one that has both affordances and drawbacks in the description of socio-political reality, prioritising knowledge that is generalisable, aggregated and systematic, while tending to disregard anecdotal knowledge that is experiential, personal and emotional (Peters, 2001; Porter, 1995). However, in contemporary politics an ‘epistemological hegemony of quantification’ (Coleman, 2018: 160–161) has emerged, where faith in the discourse of statistical objectivity has reduced the space for the voicing of non-quantifiable counter arguments, which come to be treated like ‘expressions of empirical irresponsibility’.

Yet, as discussed, little work has sought to examine how statistics function rhetorically in political debate. This is despite the fact that a rhetorical approach to political interpretation has experienced a recent resurgence with the development of RPA (Atkins et al., 2014; Finlayson, 2007; Martin, 2015). RPA is primarily concerned with the devices, proofs and justifications that political actors bring forward when making claims and how this can be connected to the ‘formation, propagation, development and change of ideas in politics’ (Atkins and Finlayson, 2013: 162). More than just techniques or devices of political persuasion, however, these practices of political rhetoric are said to exist in a mutually constitutive relationship with the wider political process (Martin, 2014). They reveal and create power structures, and indicate historically and culturally situated modes of communicating politics, thereby ‘legitimating certain actors and ideas at the ontological level’ (Price-Thomas and Turnbull, 2018: 212). While a number of rhetorical practices have now been investigated using this understanding, quantification and statistical argument has so far been overlooked by proponents of RPA.

Given the ‘epistemological hegemony’ of quantification in politics, however, there is a pressing need to pay more attention to the way statistical arguments are connected to claims about the political world, and to further advance conceptual frameworks through which these arguments can be better understood and critiqued. This is particularly the case in politics, where quantification underpins a range of arguments about political issues, evaluating performance, defining problems and prescribing solutions. To this end, we explore the way statistics function as rhetorical devices in political debate. In doing so, our aim is not to suggest that statistics should not play an important role in the way we contest and discuss political reality. But rather, our aim is to demonstrate their rhetorical nature, calling into question their perceived objectivity. In the following sections, we will explore the unique ways in which statistics underpin political claims and identify some of the common rhetorical tropes that emerge in statistical arguments.

Methodology

As we were focusing on the political communication of the NHS winter crisis, we chose to analyse PMQs. We covered two NHS winter periods: December 2016–January 2017 and December 2017–January 2018. All of these PMQs were conducted between Theresa May and Jeremy Corbyn, allowing for a level of consistency in our analysis. Within these 12 PMQs, we only used discussions that focused on ‘the NHS’ and ‘the winter’. This resulted in four sets of debates: 11 January 2017, 20 December 2017, 10 January 2018 and 24th January 2018.

Once we had our corpus, we first highlighted all statistics as defined by Mallows (2006). This includes inferential statistics – a piece of information from a portion of the population, which is then extrapolated – and descriptive statistics – information about the whole of the sample set. For example, the following statement would be highlighted ‘Virgin Care got £200 million-worth of contracts in the past year alone’ (Hansard, 2018a) as this number represents the cumulative total of all the Virgin Care contracts in the previous year (a descriptive statistic). This excluded other numbers, such as measurements (e.g. ‘12 hours’), unless they were attached to a statistical claim. After doing this, we used the existing literature on statistics and rhetoric (Aviles, 2016; Kilyeni, 2013; Koetsenruijter, 2011; Roeh and Feldman, 1984), as well as an understanding of other rhetorical devices, to identify how these statistics were used in these debates. Our analysis focuses on four rhetorical strategies: dehistoricisation, enthymeme, synecdoche and framing.

Context: The NHS winter crisis

A ‘crisis’ within the NHS over the winter months (December, January and February) has become a perennial issue. A rise in seasonal-specific medical problems, including respiratory diseases, the flu and norovirus, means more patients attend Accident & Emergency (A&E) wards with complex medical issues (The Health Foundation, 2019). Many of these patients are then admitted to hospital for treatment rather than being discharged, placing a considerable degree of strain on the healthcare system. However, many argue that to place too much emphasis on disease would be misleading as it ignores the consistent under-funding of the NHS by the Conservative government from 2010 (Ham, 2017; Iacobucci, 2018). In this way, the NHS winter crisis should not be understood as simply a national health crisis but instead as a crisis in the provision of healthcare.

The crisis can be measured using a range of statistical accounts of NHS performance, mostly derived from a set of publicly accessible databases published by NHS Digital. The most common yardstick is the ‘four-hour wait’: how many patients are seen, treated and admitted, transferred or discharged within 4 hours of checking in at the A&E desk. The lower the number, the more severe the ‘crisis’. The cause of the crisis is often framed numerically too, whether it is A&E attendance, funding for social care or excess bed days (the number of days that a bed is occupied by a patient that could be discharged from the hospital) (Blunt et al., 2015; NHS England, 2017). Such are the long-term data trends that some argue that this seasonal crisis is not ‘explosive . . . but predictable, and predicted’ (McCartney, 2018: 360).

The wealth of statistical information is reflected in the highly contested political debates that emerge over the crisis. Politicians use certain statistics to underpin their position on whether the NHS is experiencing a crisis, how severe the crisis has become and how the crisis could be resolved. Thus, the NHS winter crisis is not only an excessively quantified phenomenon, but a highly politicised one too.

Findings

Some 1.8 million people had to wait longer than four hours in A&E departments last year. The Prime Minister might not like what the Red Cross said, but on the same day the British Medical Association said that ‘conditions in hospitals across the country are reaching a dangerous level’. Jeremy Corbyn (Hansard, 2017a)

The fact is that we are seeing more people being treated in our NHS: 2,500 more people are treated within four hours every day. Theresa May (Hansard, 2017a)

If a ‘typical’ exchange between Jeremy Corbyn and Theresa May could be identified from our corpus, the extract above is exactly that. It displays how statistics are strongly associated with an ontological ‘reality’ – most explicitly stated by May when she connects ‘the fact is that’ with ‘2,500 more people are treated’.

We argue that dehistorisation underpins this connection between truth and statistics. By omitting the production (or history) of a statistic in political communication, the number is rendered as an ever-present fact. In presenting numbers in such a way, both political actors can then connect quantified information with ‘reality’ because the quantified information is simply describing reality.

Whereas dehistoricisation functions to secure the relationship between numbers and reality, three rhetorical devices are deployed to contest which numbers best define this reality. Synecdoche involves using the part to represent the whole – taking one of the many statistical articulations about the NHS and positioning this piece of knowledge as defining the entirety of the ‘crisis’. In a similar vein, framing occurs when the two political actors refer to the same data but present it in different ways – most commonly achieved by presenting a total or a relative number. In almost every case within our corpus, these two devices were deployed without justification for why that particular statistic better represented the reality of the crisis than another one. Such a style of rhetoric is termed enthymeme – where an argument is based on at least one premise being unstated. So, one could argue that the NHS is in crisis (conclusion) based on one stated premise (a particular statistic) and one unstated premise (why this particular statistic means the NHS is in crisis). To understand how all four rhetorical devices function individually, we provide in-depth explanations of each.

Consensus: Dehistoricisation

Dehistoricisation works to erase the production of statistics from rhetoric. In doing so, a statistic is presented as an ever-present fact. Thus, ignoring the epistemological history of all statistics: people (often experts) decide whether to count something, how to go about counting and how to summarise the results of this counting process (Best, 2008: 2).¹ We identified two stages of production where this device of ‘omission’ – what is ‘unstated’ rather than what is ‘stated’ – was used.

Statistics rely on definitions of phenomena in order to function. If one wants to statistically claim that something is ‘bad’ or ‘good’, one needs to specifically define what ‘good’ and ‘bad’ mean in that context. The process of creating these definitions, and the definitions themselves, are excluded from PMQs:

‘Care Quality Commission figures suggest that nearly a quarter of care homes need improvement’. Jeremy Corbyn (Hansard, 2018a)

Here, Corbyn treats ‘need improvement’ as an eternal category that predates the statistic itself. This is not the case. An expert(s) has defined what ‘needs improvement’ means in order to produce a classification system that can comparatively rank care homes across the United Kingdom. Such a process requires taking a pre-existing phenomenon – of care for the elderly – and retrospectively applying a categorisation system. In doing so, the ‘expert(s)’ rely on a range of cultural, economic, bureaucratic and social norms to establish an acceptable level of care for the elderly, which then allows them to determine which care homes need improvement and which do not. Thus, these definitions are far from the absolutes that the act of dehistoricisation presents.

Once categories are defined, data need to be collected to fulfil the classification system. Most of the statistics used in PMQs are drawn from the NHS database:

Why are there some hospitals where the percentage of patients waiting more than 30 minutes is zero and other hospitals where the percentage of patients waiting more than 30 minutes is considerably higher? Theresa May (Hansard, 2018b)

The claim above refers to a comparison between hospitals using the same metric of waiting times. What the strategy of dehistoricisation excludes here is the nature of the database and the data collection required to sustain it. The NHS database is a collection of information that quantifies the entire population about which it makes claims. In other words, the database does not take a sample of 10 hospitals, extrapolate the data from these hospitals and make claims about all hospitals. Each individual hospital collects comprehensive data on what they do and who they do it to.

Thus, the dataset is not just colossal in size yet also involves a colossal amount of labour to produce it. The information is created by nurses, doctors, porters, administrative staff, technical personnel and so on. Thus, NHS statistics are highly reliant on a range of people correctly entering information into computers all year round. Mistakes, most often unintentional, are to be expected. References to statistics in the PMQs, however, erase (or dehistoricise) this part of the production process.

Not only does dehistoricisation mask the different levels of production a statistic passes through but it also disguises a basic truth about quantification. That is, statistics have captured something that has happened, not something that is happening. Take the extract below:

We see now 7 million more diagnostic tests than seven years ago, 2.2 million more people getting operations, and survival rates for cancer at their highest ever level. (Hansard, 2017b)

At the time of this statement, 20 December 2017, these numbers are not an accurate representation of those ‘getting operations’. Under the guidelines for late 2017, the NHS publishes data roughly 1 week after the final data point has been collected. So, they are a snapshot of the amount of people who were ‘getting operations’ at one point about 1 week before they were used in PMQs.

Dehistoricisation serves to omit certain fundamentals of statistics – they have a history, which is often nuanced, and they capture a moment that occurred, not one that is occurring. Such a strategy results in statistics being presented as ever-present facts. In his study of how journalists reported science statistics, Tony Van Witsen (2019) identifies a similar discursive pattern. Numbers are often used alongside ‘certainty markers’, such as ‘clear’, ‘plain’ or ‘overwhelming evidence’, to emphasise the certainty of quantitative information. While Van Witsen points to the prevalence of specific words or phrases rather than the omission of information, both strategies achieve the same end goal: to emphasise the facticity of numbers. As Merry (2016) explains, such a process removes numbers from epistemological debate. It is from this rhetorical foundation that synecdoche and framing function.

Conflict (i): Synecdoche

The use of synecdoche in political rhetoric has traditionally been associated with stories and anecdotes. Both Atkins and Finlayson (2013) and Oldenburg (2015), for example, have shown how the stories politicians tell function synecdochally, providing individual examples of situations, problems and issues through which wider generalisations about their nature can be ascertained. So far, the possibility that politicians may use statistics in similar ways has not been considered. Yet, when political actors use statistics to describe political reality, they too are only necessarily drawing upon ‘selections of reality’ (Burke, 1969: 59). Statistical information, like NHS staffing and funding levels, only represent individual aspects of diverse phenomena, such as the health service and the economy. Like with stories, what often emerges in the rhetoric of statistics is synecdoche: a use of these individual aspects, or a group of these individual aspects, to embody the phenomena of which they only represent a part. Staffing and funding levels, for example, might be used to symbolize the overall health of the NHS. This is a rhetorical accomplishment, one that requires the synecdochic description of a phenomenon with one or a group of its constituent statistical parts.

Statistical debates over political phenomena thus often revolve around competing synecdochic accounts of which individual statistics best embody a particular issue or concept. A good example of this emerged in Jeremy Corbyn and Theresa May’s debate over the NHS’s preparedness for winter. There are a range of phenomena that are associated with how ‘prepared’ the NHS is during winter: staffing, funding, hospital beds, accident and emergency waiting targets, mental health facilities and so on. Yet, upon attacking the government’s winter NHS performance, Corbyn used one statistic in particular:

I know it seems a long time ago, but just before Christmas, I asked the Prime Minister about the 12,000 people left waiting more than half an hour in the back of ambulances at A&E departments. She told the House that the NHS was better prepared for winter ‘than ever before’. What words of comfort does she have for the 17,000 patients who waited in the back of ambulances in the last week of December? (Hansard, 2018a)

Where Corbyn saw the number of people waiting more than half an hour in the back of ambulances in the last week of December as symptomatic of the NHS’s lack of preparedness for winter, May’s statistical rhetoric suggested that the NHS’s preparedness for winter could be gleaned from the number of beds available and flu vaccinations administered:

Yes. It might be helpful if I let the House know some of the things that were done to ensure that preparedness. More people than ever before are having flu vaccines, and 2,700 more acute beds have been made available since November. (Hansard, 2018a)

A similar contestation emerges when the two politicians discuss whether the NHS is experiencing a ‘humanitarian crisis’ during the winter of 2016/2017. Jeremy Corbyn referred to the number of people who spent more than 12 hours on trolleys in hospital corridors as evidence of the Red Cross’s claim:

Last week, 485 people in England spent more than 12 hours on trolleys in hospital corridors. The Red Cross described this as a ‘humanitarian crisis’. (Hansard, 2017a)

The claim of a humanitarian crisis was rebuffed by Theresa May as she referred to the absolute increase in the number of people treated in A&E compared to 6 years ago:

He also refers to the British Red Cross’s term, ‘humanitarian crisis’. I have to say to him that I think we have all seen humanitarian crises around the world, and to use that description of a national health service that last year saw 2.5 million more people treated in accident and emergency than six years ago was irresponsible and overblown. (Hansard, 2017a)

As with ‘preparedness’, there are a range of statistics that could be used to corroborate or dispute the claim of ‘humanitarian crisis’. Of this quantitative whole, Corbyn and May elected for different parts to represent the entire entity. In doing so, they provided alternative synecdochic accounts of ‘humanitarian crisis’. While synecdochic conflict occurs over different statistical datasets and their ability to capture the ‘reality’ of a political issue, framing involves contestation over how to present the same dataset.

Conflict (ii): Framing

In January 2017, Corbyn and May discussed patient treatment at A&E centres. Instead of arguing from different datasets, both political actors used the same part of the NHS dataset concerning 4-hour wait times to make their arguments. The conflict occurred in how they both presented the data. Corbyn used information on wait times to criticise the government, whereas May used the same information to defend the government:

Some 1.8 million people had to wait longer than four hours in A&E departments last year. Jeremy Corbyn (Hansard, 2017a)

The fact is that we are seeing more people being treated in our NHS: 2,500 more people are treated within four hours every day. Theresa May (Hansard, 2017a)

This exchange implied agreement between May and Corbyn that data regarding the 4-hour wait limit is an acceptable barometer for measuring NHS performance – that is, a synecdochic consensus. But conflict arose in how the data were presented: Corbyn decided to use absolute numbers of people that waited longer than 4 hours in the previous year (2016), while May opted to compare how many more people were being treated on this day compared to the previous ones.

Both types of framing have their own affordances (and omissions). The absolute number emphasises the ‘total magnitude’ of the problem at hand yet lacks temporal context – was 1.8 million people a decrease or increase from previous years? While the relative number uses temporal context to emphasise progress – we are improving day by day – yet ignores how a rising population, and an expanding NHS service, could account for part (or all) of the so-called ‘progress’. The persuasive effects of this kind of framing, often referred to as equivalency framing, has been well documented. Minor changes to the presentation of the same numerical data has been shown to exploit various psychological biases, such as the tendency to prioritise loss aversion, numerators instead of denominators in ratios and left-most digits, along with a tendency to respond favourably to positive framings and unfavourably to negative framings (Olsen, 2013; Pedersen, 2016; Tversky and Kahneman, 1981).²

Conflict (iii): Enthymeme

Both framing and synecdoche operate with similar rhetorical logic. That is, they both present their statistics (either from different datasets or the same) yet rarely provide an explanation concerning why their statistic is valid – that is, can this statistic be used to make this claim? In doing so, they make use of enthymeme: arguments where a premise in left unstated (Finlayson, 2012: 762). Such a rhetorical technique is described by Morrell and Hewison (2013) as ‘perhaps the natural form of argument in politics, because of the complexity and inescapable dilemmatic nature of governance’ (p. 66). Within the NHS policy in particular, their research showed that enthymeme is a central rhetorical device (Morrell and Hewison, 2013). A typical example of this can be observed below in Theresa May’s defence of the government’s record on the NHS:

This Government are putting more money into the national health service. We see more doctors and nurses in our NHS, more operations taking place in our NHS, and more people being treated in accident and emergency in our NHS. (Hansard, 2018a)

This argument makes use of enthymeme. The claim that the NHS is in a satisfactory (or at least improving) state has two premises. The first are the statistical claims she makes: increase in funding, improved staffing levels and rises in operations and treatment. The second involves an explanation of how these statistical claims are evidence of a satisfactory NHS. The first premise is stated yet the second one is not.

A similar pattern can be identified when we revisit the previous two sections. Synecdoche functioned by making conclusions on the basis of two premises, one that is stated (e.g. patients are spending more time waiting in the back of ambulances) and one that is implied but unstated (e.g. the time patients spend waiting in the back of ambulances is a good indicator of NHS performance). Whereas framing involves an unstated premise about why an absolute or relative presentation of the same information is more or less suitable to describe the crisis at hand, there are cases where the premise is stated yet these tend to fall into a rhetorical tautology. Take the case below as an example:

We see now 7 million more diagnostic tests than seven years ago, 2.2 million more people getting operations, and survival rates for cancer at their highest ever level. Those are figures, but what does that mean? It means more people getting the treatment they need. It means more elderly people getting their hip operations. And it means that today there are nearly 6,500 people alive who would not have been if we had not improved our cancer care. (Hansard, 2017b)

Theresa May states the first premise (the three statistics) and then, through a rhetorical question, sets up an argument as to why these three statistics best encapsulate the reality of the crisis. Her second premise, however, involves repeating the first premise. For May, ‘2.2 million more people getting operations’ means ‘more elderly people getting their hip operations’. So, even in cases where the premise is stated, no logical link between the statistic and the claim is made.

Conclusion

We appreciate that the use of statistics in this manner may be a product of the format of PMQs – an antagonistic, relatively short exchange between two party leaders – but we also argue that dehistoricisation reflects a wider consensus about what type of knowledge statistics are considered to be. That is, numbers occupy the space of immutable fact – standing in stark contrast to other forms of representation. It is upon this epistemological foundation that enthymematic arguments are made, functioning through the rhetorical tropes of synecdoche and framing, which persuade and convince during political arguments about the NHS winter crisis. These are just a few of the tropes that underpin statistical arguments in political debate, but they represent a starting point towards a richer understanding of how politicians connect statistics to wider claims about the political world.

In the development of this rhetorical understanding of statistics, we have not aimed to challenge or denounce their usage in political debate. Statistical information can and should play an important role in the way our politicians negotiate and try to persuade us on issues like the NHS winter crisis. However, what should be challenged is the privileged role that statistical argument can sometimes possess and the notion that it is sound basis for empirically verifiable and universal notions of political truth. While numbers may at times allow us to verify and universalise basic truths (i.e. that a certain number of people waited longer than 4 hours last night in an A&E department), what often makes truth political is the way in which these kinds of statistics are connected to broader arguments about what they mean (is the NHS underfunded and in crisis? Is the NHS coping well with increased winter demand from an ageing population?).

This point has become even more pertinent with the emergence of COVID-19, in which statistics have formed a central part of the way politicians have debated the crisis. Whether being used to explain government guidelines, justify policy interventions or defend and critique national responses, the prevalence of statistics throughout the pandemic has underlined both their political significance and their complex relationship with political reality. The debates that have emerged around this crisis, much like the ones discussed above, illustrate not only the importance of ‘quantifying crisis’, but also the dangers of overlooking the contradictory, selective and strategic ways in which this is often done. As such, a rhetorical understanding of statistics can help to illuminate the partial and contestable ways in which numbers operate in political argument.

Footnotes

Funding

The author(s) received no financial support for the research, authorship and/or publication of this article.

ORCID iD

Brendan Lawson

Notes

References

Atkins

(2015) Narrating one nation: The ideology and rhetoric of the Miliband Labour Party. Politics 35: 19–31.

Atkins

Finlayson

(2013) “. . . A 40-Year-Old Black Man Made the Point to Me”: Everyday knowledge and the performance of leadership in contemporary British politics. Political Studies 61: 161–177.

Atkins

Finlayson

Martin

(2014) Rhetoric in British Politics and Society. Basingstoke: Palgrave Macmillan.

Aviles

(2016) The art of public policy statisticians: The case of the Puerto Rico pension reform report. Caribbean Studies 44: 47–68.

Best

(2008) Birds – dead and deadly: Why numeracy needs to address social construction. Numeracy 1: Article 6.

Blunt

Edwards

Merry

(2015) What’s Behind the A&E Crisis? London: Nuffield Trust.

Burke

(1969) A Rhetoric of Motives. Berkeley, CA: University of California Press.

Coleman

(2018) The elusiveness of political truth: From the conceit of objectivity to intersubjective judgement. European Journal of Communication 33: 157–171.

Crines

Happell

(2017) Rhetorical style and issue emphasis within the conference speeches of UKIP’s Nigel Farage 2010-2014. British Politics 12: 231–249.

10.

Desrosieres

(1998) The Politics of Large Numbers: A History of Statistical Reasoning. Cambridge, MA: Harvard University Press.

11.

Desrosieres

(2001) How real are statistics? Four possible attitudes. Social Research 68: 339–355.

12.

Dunleavy

(2016) Big data and policy learning. In: Stoker

Evans

(eds) Evidence-Based Policy Making in the Social Sciences: Methods That Matter. Bristol: Policy Press, pp. 143–168.

13.

Finlayson

(2004) Political science, political ideas and rhetoric. Economy and Society 33: 528–549.

14.

Finlayson

(2006) ‘What’s the problem?’: Political theory, rhetoric and problem-setting. Critical Review of International Social and Political Philosophy 9: 541–557.

15.

Finlayson

(2007) From beliefs to arguments: Interpretive methodology and rhetorical political analysis. The British Journal of Politics and International Relations 9: 545–563.

16.

Finlayson

(2012) Rhetoric and the political theory of ideologies. Political Studies 60: 751–767.

17.

Finlayson

(2014) Proving, pleasing and persuading? Rhetoric in contemporary British politics. The Political Quarterly 85: 428–436.

18.

Finlayson

Martin

(2008) ‘It ain’t what you say . . .’: British political studies and the analysis of speech and rhetoric. British Politics 3: 445–464.

19.

Hacking

(1990) The Taming of Chance. Cambridge: Cambridge University Press.

20.

Ham

(2017) Political Crisis in the NHS. Available at: https://www.bmj.com/content/356/bmj.j218.full

21.

Hansard (2017a) Engagements: 11th January 2017. Available at: https://hansard.parliament.uk/commons/2017-01-11/debates/4139811D-0756-4C58-827D-E00D1CCD970E/Engagements

22.

Hansard (2017b) Engagements: 20th December 2017. Available at: https://hansard.parliament.uk/commons/2017-12-20/debates/9B501034-E35A-45F8-8834-18BAA9EFF507/Engagements

23.

Hansard (2018a) Engagements: 10th January 2018. Available at: https://hansard.parliament.uk/commons/2018-01-10/debates/4D4FFA34-F478-4434-BDEC-92B3F92A7D85/Engagements

24.

Hansard (2018b) Engagements: 24th January 2018. Available at: https://hansard.parliament.uk/Commons/2018-01-24/debates/156B95FC-2467-42BD-BFD5-464656276D9C/Engagements

25.

Iacobucci

(2018) NHS Cancels Planned Surgery and Outpatient Appointments in Response to Winter Crisis. Available at: https://www.bmj.com/content/360/bmj.k19.full

26.

Kilyeni

(2013) The rhetoric of numbers in print advertisements for cosmetics. Scientific Bulletin of the Politehnica University of Timisoara. Transactions on Modern Languages/Buletinul Stiintific al Universitatii Politehnica din Timisoara. Seria Limbi Moderne 12: 17–26.

27.

Koetsenruijter

AWM

(2011) Using numbers in news increases story credibility. Newspaper Research Journal 32: 74–82.

28.

Koetsenruijter

(2018) Numbers in the news: More ethos than logos? In: Nguyen

(ed.) News, Numbers and Public Opinion in a Data-Driven World. London: Bloomsbury Publishing, pp. 177–190.

29.

Kurunmaki

Mennicken

Miller

(2016) Quantifying, economising, and marketising: Democratising the social sphere? Sociologie Du Travail 58: 390–402.

30.

Kuypers

(2010) Framing analysis from a rhetorical perspective. In: D’Angelo

Kuypers

(eds) Doing News Framing Analysis: Empirical and Theoretical Perspectives. New York: Routledge, pp. 286–311.

31.

Latour

Woolgar

(1979) Laboratory Life: The Construction of Scientific Facts. Princeton, NJ: Princeton University Press.

32.

Lugo-Ocando

Lawson

(2018) Poor numbers, poor news. In: Nguyen

(ed.) News, Numbers and Public Opinion in a Data-Driven World. London: Bloomsbury Publishing, pp. 62–77.

33.

McCartney

(2018) When a crisis is the predictable outcome of poor policy making. BMJ 360: k90.

34.

McCloskey

(1987) Rhetoric within the citadel: Statistics. In: Wenzel

, et al. (eds) Argument and critical practice: Proceedings of the fifth SCA/AFA conference on argumentation. Annandale, VA: Speech Communication Association, pp. 485–490.

35.

Mallows

(2006) Turkey’s paper after 40 years. Technometrics 48: 319–325.

36.

Martin

(2014) Politics and Rhetoric: A Critical Introduction. New York: Routledge.

37.

Martin

(2015) Situating speech: A rhetorical approach to political strategy. Political Studies 63: 25–42.

38.

Merry

(2016) The Seductions of Quantification: Measuring Human Rights, Gender Violence, and Sex Trafficking. Chicago, IL: The University of Chicago Press.

39.

Morrell

Hewison

(2013) Rhetoric in policy texts: The role of enthymeme in Darzi’s review of the NHS. Policy & Politics 41: 59–79.

40.

NHS England (2017) A&E Attendances and Emergency Admissions. NHS England. Available at: https://www.england.nhs.uk/statistics/wp-content/uploads/sites/2/2017/06/Monthly-AE-Report-April-17.pdf (accessed 1 July 2019).

41.

Oldenburg

(2015) Re ‘Characterising’ the anecdote: Synechdoche and ethotic argument in Presidential debate rhetoric. Communication Studies 66: 103–120.

42.

Olsen

(2013) Leftmost-digit-bias in an enumerated public sector? An experiment on citizens’ judgement of performance information. Judgement and Decision Making 8: 365–371.

43.

Pedersen

(2016) Ratio bias and policy preferences: How equivalency framing of numbers can affect attitudes. Political Psychology. Epub ahead of print 25 September. DOI: 10.1111/pops.12362.

44.

Peters

(2001) ‘The Only Proper Scale of Representation’: The politics of statistics and stories. Political Communication 18: 433–449.

45.

Poovey

(1993) Figures of arithmetic, figures of speech: The discourse of statistics in the 1830s. Critical Inquiry 19: 256–276.

46.

Porter

(1995) Trust in Numbers: The Pursuit of Objectivity in Science and Public Life. Princeton, NJ: Princeton University Press.

47.

Prevost

J-G

Beaud

J-P

(2012) Statistics, Public Debate and the State, 1800-1945: A Social, Political and Intellectual History of Numbers. London: Routledge.

48.

Price-Thomas

Turnbull

(2018) Thickening rhetorical political analysis with a theory of distance: Negotiating the Greek episode of the Eurozone crisis. Political Studies 66: 209–225.

49.

Roeh

Feldman

(1984) The rhetoric of numbers in front-page journalism: How numbers contribute to the melodramatic in the popular press. Text. Interdisciplinary Journal for the Study of Discourse 4: 347–368.

50.

Simon

(2008) The choice of ignorance: The debate on ethnic and racial statistics in France. French Politics, Culture & Society 26: 7–31.

51.

The Health Foundation (2019) NHS Winter Pressures. Available at: https://www.health.org.uk/news-and-comment/featured-content/nhs-winter-pressures

52.

Tversky

Kahneman

(1981) The framing of decisions and the psychology of choice. Science 211: 453–458.

53.

Van Witsen

(2019) How daily journalists use numbers and statistics: The case of global average temperature. Journalism Practice. 14: 1047–1065.

Towards a rhetorical understanding of statistics in politics: Quantifying the National Health Service ‘Winter Crisis’

Abstract

Keywords

Introduction

Numbers, politics and rhetoric

Methodology

Context: The NHS winter crisis

Findings

Consensus: Dehistoricisation

Conflict (i): Synecdoche

Conflict (ii): Framing

Conflict (iii): Enthymeme

Conclusion

Footnotes

Funding

ORCID iD

Notes

References