The article ‘The Muddle of the “Westminster Model”: A Concept Stretched Beyond Repair’ reflects many institutional, cultural and sociopolitical shifts, but does the approach adopted by Meg Russell and Ruxandra Serban (Reference Russell and Serban2021, hereafter ‘R&S’) further muddy the concept rather than clarifying what the term covers and why it matters? The narrative about ‘a concept stretched beyond repair’ and conclusion that ‘it is time for the “Westminster model” to be retired’ poses a powerful, potent and provocative challenge that we would like to meet.
The aim of this article is to stress-test the internal logic and reliability of this thesis. We step back to reflect on the intellectual framework used to come to this conclusion. We re-contextualize the argument against a set of broader perspectives. We also rely on the tools and insights of conceptual political analysis but arrive at a different conclusion: The Westminster model was never designed or intended to act as a positivist concept facilitating ‘scientific’ research. The ‘muddled’ qualities that R&S identify reflect the fact that it is, and always has been, a looser ‘organizing perspective’ or ‘sensitizing concept’ that draws attention to several themes and institutional characteristics.
In short, we suggest that R&S hold the notion of the Westminster model up to (and against) a standard of conceptual precision that it was never intended or designed to fulfil. Viewed from this perspective, it was always ‘destined to disappoint’. The risk of engaging in such academic debates is that it overlooks the relevance of the concept, in the governing traditions and everyday practices of practitioners. This point is critical.
What R&S provide is a text-based analysis which, they suggest, reveals the ‘muddled’ mechanics of the Westminster model to such an extent that the concept should be retired from the lexicon of political analysis. Few scholars would disagree that the Westminster model is and always has been something of a ‘fuzzy concept’ and the argument that it has become ‘over-stretched’ (Sartori Reference Sartori and Sartori1984) has been made before (see Matthews Reference Matthews2011). We interrogate the new evidence that R&S offer to justify their conclusion that the concept is now ‘beyond repair’. We suggest that when assessed from a broader perspective, the Westminster model is ‘stretched but not snapped’ (Barber Reference Barber2014).
Our core contention is that R&S's argument is over-stretched. The evidence they provide demands that we ask to what extent did their epistemological and methodological approach muddy an already muddled concept? To develop this ‘stretched-not-snapped’ position, we provide an account of R&S's research and conclusions, and we argue that the conceptual, institutional, comparative and interpretive perspectives add value to any reconsideration of the Westminster model. Finally, we seek to establish a more balanced and nuanced debate about ‘managing’ the unavoidable messiness of the Westminster model.
The muddle of the Westminster model
‘The term “Westminster model” appears often in both the academic and practitioner literatures. It will be familiar to many specialists in comparative politics, public administration and law.’ R&S (Reference Russell and Serban2021: 744) ask in the opening line of their article: ‘But what precisely does it mean, and is there consistency in its application? If put under the microscope, can any clear meaning actually be discerned? This article suggests that while ostensibly serving as a “model” in the comparative literature, the term instead risks inducing muddle and unclear thinking’ (R&S Reference Russell and Serban2021: 744, emphasis added). The aim of this section is to provide a brief account of the theory, methods and data that support R&S's suggestion that the Westminster model had become stretched ‘beyond repair’.
R&S approach the analysis of the Westminster model through a positivist lens which utilizes Giovanni Sartori's (Reference Sartori1970) well-known work on the use of concepts in the social and political sciences. At the heart of this approach is a focus on ‘conceptual stretching’ and ‘conceptual travelling’ derived from an awareness of the potential trade-off between ‘extension’ and ‘intension’. Fuzzy concepts are bad concepts. Seeking salvation for concepts means, according to this Sartorian approach, ‘[F]irst, collect a representative set of definitions; second, extract their characteristics; and third, construct matrices that organize such characteristics meaningfully’ (Sartori Reference Sartori, Collier and Gerring2009 [1984]: 121). ‘Hence the key is “reconstructing a concept from its literature”' (Sartori Reference Sartori, Collier and Gerring2009 [1984]: 121, italics in original). R&S (Reference Russell and Serban2021: 748) ‘take our cue from this suggestion’.
As a result, R&S undertake an extensive review of books, articles and university reading lists which, setting aside questions about the criteria for selecting their data, led to them identifying 239 relevant texts (see R&S Reference Russell and Serban2021: table 1).These texts were then assessed against three questions: Was the Westminster model explicitly defined? What attributes were associated with the model by these authors? To which countries was the model applied?
This desk-based approach discovered that only 38 (or 16%) of the texts reviewed offered an explicit and full definition. A further 40 (17%) offered a partial definition. Finally, 90 (38%) had an implicit definition based on common attributes. As the basis for an article on the ‘muddle of the Westminster model’, a finding that 71% of relevant texts did include at least some form of definition could be seen as a problem. This problem is compounded by the inconvenient fact that Arend Lijphart's comparative work provides a common root text. R&S (Reference Russell and Serban2021: 751) tackle this challenge head-on by exploring the ‘diversity of attributes’ found within each of the 168 texts ‘following Sartori, to investigate the “intension” of the term’. The results of their second-stage review are set out in R&S table 2. Contrary to R&S's core thesis, the main definitional features of the Westminster model appear relatively clear, consistent and understood (discussed below). Some core characteristics, such as executive dominance or adversarial culture are, as the authors suggest, difficult ‘to define or measure precisely’ (R&S Reference Russell and Serban2021: 752) but the evidence also suggests that this has not prevented detailed analysis from being undertaken (e.g. Curtin Reference Curtin2014; Kenny and Casey Reference Kenny and Casey2021; Tsebelis Reference Tsebelis, Ganghof, Honnige and Stecker2009). Even R&S (Reference Russell and Serban2021: 752) find it necessary to concede that ‘Despite the imperfect nature of the categories, the table provides some clear indications of attributes most commonly associated with the model’.
We might expect an argument about the ‘muddle of the Westminster model’ and ‘a concept stretched beyond repair’, which calls for comparative scholars to ‘drop this term’, to find more muddle. Undeterred, R&S (Reference Russell and Serban2021: 755) stick to their positivist position: ‘But whether they [the list of common attributes] describe a “model” with real comparative application remains distinctly open to doubt.’
Turning to the countries most commonly discussed as examples of the Westminster model, the extent of ‘muddle’ is again overstated (see R&S table 3). The analysis reveals a relatively tight comparative field that reflects a widely acknowledged colonial heritage. To make this point is not to deny the existence of long-recognized patterns and processes of ‘majoritarian modification’ (e.g. Bevir Reference Bevir2008; Flinders Reference Flinders2005, Reference Flinders2009; Gamble Reference Gamble2006). Nor do we deny Lijphart's early (Reference Lijphart1984: 19) admission that ‘in nearly all respects, democracy in New Zealand is … a better example of the Westminster model, than British democracy’. But we do claim the continuing existence and influence of the ‘family resemblances’ underpinning Westminster (see below). Paradoxically, R&S draw attention to this family. By focusing on observable, measurable characteristics – on operational variables – R&S underplay both the ideational or cultural relevance of the Westminster model. They also underestimate the intellectual dexterity of the comparative political science community. The next section develops our ‘stretched-not-snapped’ position by demonstrating this dexterity; by exploring the institutional, comparative and interpretive organizing perspectives.
‘Muddled’ or ‘muddying’ concepts?
A conceptual perspective
This subsection makes three main arguments: (1) it is possible to argue the Westminster model is not a concept; (2) if it is a concept, then it is ‘an essentially contested concept’; and therefore (3) the ‘family resemblance’ thesis demands reconsideration. We suggest that the Westminster model is best understood not as a concept but as an ‘organising perspective’ (Greenleaf Reference Greenleaf1993). As Andrew Gamble explains:
[Organising perspectives] provide a framework for analysis, a map of how things relate, a set of research questions. Such perspectives are not normally testable or falsifiable, yet it is hard to imagine how the study of politics could proceed very far without them … Organising perspectives have to be distinguished from theories which present distinct hypotheses that may in principle be falsified. (Gamble Reference Gamble1990: 405, emphasis added)
Gamble (Reference Gamble1990: 406) goes on to note that ‘an organising perspective will always be implicit in the way the research is framed and conclusions are drawn’. He notes, as do R&S, the influence of John Mackintosh's The Government and Politics of Britain (Reference Mackintosh1970) as an important source on the Westminster model's core characteristics. The core insight that Gamble offers is the manner in which the Westminster model was itself imbued with a particular Whig interpretation of British history – that is, a commitment to flexibility, pragmatism and ‘the good chap theory of government’ (Hennessy Reference Hennessy1995: 57, 64). Our point is that the Westminster model did not originate within academe. It was not intended to be a neat, precise, specific idea suitable for scientific analysis. It possesses a ‘void for vagueness’ (see Cooter Reference Cooter1994) which inevitably invites argument and means that ‘[f]or works following a more positivist tradition in political science this seems problematic’ (R&S Reference Russell and Serban2021: 747).
We suggest that it is the ‘positivist tradition in political science’ that is more a problem than the Westminster model because it seeks to impose such a narrow, quantifiable and foundational interpretation on the study of mankind (see the section on the interpretive perspective below). For R&S, and in comparative politics more generally, Sartori's work on concepts has been hugely influential. So it is important to recognize that he was promoting a more positivist approach at a specific point in the evolution of political science. It was an approach but it was often promoted as the approach. Its political and procrustean qualities have been subject to swingeing critique (see Crick Reference Crick1959; Mills Reference Mills1959; see also Almond Reference Almond1990). We explore different epistemologies below and give examples of the concept's usefulness in comparative interpretive research.
The purpose of setting out this disciplinary history is to highlight that, by putting the Westminster model under the microscope and aiming to identify precisely what it means, R&S are judging it against an evidential level that it is destined never to reach. Arguably, any common currency concept (‘the state’, ‘democracy’, ‘citizenship’, etc.) would fail if judged against such restrictive criteria. Therefore, if the Westminster model is to be defined in conceptual terms, then it is the work of W.B. Gallie (Reference Gallie1956) on ‘essentially contested concepts’ that provides the most fitting intellectual reference point. ‘An essentially contested concept … is one that is so permeated and surrounded by values that reasonable men [sic.] … may argue interminably without ever reaching agreement on the true meaning and implications of the concept’ (Birch Reference Birch1993: 8). Put simply, an essentially contested concept is an idea or organizing perspective (Gamble Reference Gamble1990) that has no clear, unambiguous meaning. Any definitions will be of little universal help as they will be contested by those who draw different boundaries around the concept. Several scholars have accused Gallie of naively promoting conceptual relativism (Collier et al. Reference Collier, Hidalgo and Maciuceanu2006). However, in many ways his argument was not that concepts could not be studied empirically but simply that the conclusions arising from any analysis would inevitably reflect the author's organizing perspective.
This brings us to this subsection's third and final argument and R&S's rejection of the ‘family resemblance thesis’ (see Goertz Reference Goertz2006: 36) in which blurred boundaries are of less importance than the presence of common attributes. The findings of R&S are a product of their method which focused on ‘key attributes observed’ (R&S Reference Russell and Serban2021: table 4) in selected books, articles and university reading lists. That means key cultural traditions and historical beliefs and practices are not captured because they are not easily ‘testable or falsifiable … yet it is hard to imagine how the study of politics could proceed far without them’ (Rhodes et al. Reference Rhodes, Wanna and Weller2009). Rod Rhodes et al. offer primary research to develop a broader approach to the ‘family resemblance’ argument:
Where we find some local identification with Westminster precepts, these countries we suggest appear as a grouping of ‘most similar’ nations, related by history, belief systems, and inclination. They form a loose family of governments, not clones of one another. Westminster was, and remains, useful shorthand to identify their origins and some of their basic characteristics. We consider it unnecessary to ‘box’ countries. Instead, we prefer to analyse the family of ideas. (Rhodes et al. Reference Rhodes, Wanna and Weller2009: 23)
Given that Rhodes et al.'s book concludes with a section entitled ‘Beware the Search for Precision’, it might possibly have formed more of a cautionary intellectual reference point. But our point is that R&S's threshold level for defining (and measuring) ‘the reasonable expectations for a useful political science term’ (R&S Reference Russell and Serban2021: 745) is defined by a certain formalism and deductive logic that we suggest may miss far more than it captures. To develop this point, the next subsection adopts an institutional perspective.
An institutional perspective
It should be of no surprise that the UK is ‘almost universally mentioned’ as the country where the term ‘Westminster model’ is said to apply (R&S Reference Russell and Serban2021: 755) because the original geographic coordinates of political Westminster were located in the UK's capital. The origins of the model are rooted firmly in the historical-political context of Britain, and the ‘descriptors’ of a distinctive British ‘institutionalized arrangement’ of representative and responsible government inform the original model. R&S (Reference Russell and Serban2021: 745, 750, 755) are, of course, aware of these spatial and historical sources. It therefore comes as something of a surprise to find that the UK, seemingly by default, should be subject to R&S's (Reference Russell and Serban2021: 761) blanket verdict that ‘it is time for the “Westminster model” to be retired’. The ‘conceptual confusion and real-world change’ (R&S Reference Russell and Serban2021: 760) which are seen to have stretched the model ‘beyond recognition’ elsewhere are automatically assumed to have afflicted the concept's core territory as well.
Indisputably, there has been change since the inception of the model in the UK and it has itself fuelled almost constant argument. But it is too large an intellectual leap to suggest the Westminster model should be discarded. This section offers three arguments against the approach and conclusion of R&S. First, the Westminster model's origins lie within an idealized and normative view of legitimate political power and it has never been a static notion. Second, it continues to provide not only a dominant narrative and common vocabulary but also the foundation against which counter-concepts are forged. Finally, the historical analysis of British politics reveals a relatively successful process of conceptual evolution and re-modelling.
There is a basic agreement, as R&S (Reference Russell and Serban2021: 745, 755) note, that the origins of the term, if not its actual designation, are to be found in the historical-political context of 19th-century Britain (see Judge Reference Judge1993). What the Westminster model sketched was a convergence between the prescriptions of a liberal view of the constitution and the practice of liberal government. In a profound sense, it was an iterative model tracking episodic changes in, for example, the configuration of political institutions, inter-institutional dependencies and state interactions with civil society during the latter part of that century.
By the end of that period, as the model became seemingly ossified in seminal academic writings, it was not an exact descriptor. It resembled more of an ‘idealised view’ (Birch Reference Birch1964: 74; Mackintosh Reference Mackintosh1970: 31): a view that conflated liberal theories of representation with a Diceyean view of liberal government (see Judge Reference Judge2014: 111). Its core ‘comprised … four distinct but interrelated doctrines’: a theory of democratic representation; the doctrine of parliamentary sovereignty; the accountability of ministers to parliament; and the legal principles of the rule of law (Birch Reference Birch1964: 65). Although often conceived simply as an empirical description of governing practice, in fact the Westminster model was infused with normative theories of representation and accountability, as well as legal theories of the sources of legislative supremacy and constitutional theories of the interactions of political institutions. In sum, as an organizing perspective in the UK, it ‘amounted to a theory of legitimate power’ (Birch Reference Birch1964: 65).
The continuing salience of this frame, and its clarity of meaning for UK state institutions, is evident in contemporary official descriptors of UK governance. The Cabinet Manual (Cabinet Office 2011: 2–4) and the Ministerial Code (Cabinet Office 2019: 1–3), for example, in listing the core principles that ministers ‘should be governed by’, draw heavily upon the distinctive institutionalized arrangement subsumed within the Westminster model. The UK Parliament, on its official website (UK Parliament 2020), attests to the continuing centrality of the principles identified above by Anthony Birch. It is unequivocal in proclaiming that ‘parliamentary sovereignty’ in the 21st century remains ‘the most important part of the UK Constitution’. It remains a core constitutional feature. Consequently, developments that are thought to threaten the integrity of parliamentary sovereignty (i.e. devolution, EU membership, the Human Rights Act 1998, establishment of a UK Supreme Court, etc.) are explicitly framed and interpreted by Parliament – despite the constitutional contortions needed – as working within rather than beyond the principle of parliamentary sovereignty.
Moreover, the Westminster model ‘remains the dominant narrative’ among central actors in Whitehall and Westminster (Rhodes Reference Rhodes2011: 306). As the work of Mark Bevir (Reference Bevir2010: 125), Patrick Diamond (Reference Diamond2014: 61), Dave Richards (Reference Richards2008: 199) and Alastair Stark (Reference Stark2010: 2–12, Reference Stark2011: 1151–3) – to mention just a few leading scholars – has revealed, the concept that R&S call to be jettisoned continues to define, shape and legitimize the behaviour of political actors in Whitehall and Westminster. In which case, as Matthew Hall (Reference Hall2011: 17) concludes, ‘the importance of the [Westminster model] lies in relation to how actors understand the British political system, rather than the accuracy of the description it offers’.
The Westminster model has also, perversely perhaps, provided an analytical root for the subsequent development of alternative models of conceptualizing governance (‘policy communities’, ‘networks’, ‘(multilevel) governance’, ‘the differentiated polity’, ‘asymmetric power model’, etc.) in that their shared starting premise was that they provided alternative and more accurate counter-concepts (e.g. Bevir Reference Bevir2010: 82; Bevir and Rhodes Reference Bevir and Rhodes2003: 198–9; Marsh Reference Marsh2012: 46–8; Rhodes Reference Rhodes1988). Even among those analysts most intent on demonstrating that the modern practices of UK governance are far removed from an idealized Westminster model, there is a recognition of the model's continuing salience as a theory of legitimate power. Its standing is reflected in the ubiquity of assertions such as the model is a ‘legitimizing tradition’, ‘legitimating tool’, ‘legitimating framework’ and ‘legitimizing mythology’ (e.g. Diamond Reference Diamond2014: 43; Flinders Reference Flinders2011: 25; Rhodes Reference Rhodes2011: 306; Richards Reference Richards2008: 199; Richards and Smith Reference Richards and Smith2016: 512).
It is important to acknowledge that R&S were mainly seeking to make an argument about the contemporary value of the Westminster model in comparative political science. However, this section is suggesting that in making this argument they miss fundamental insights relevant to comparative analysis. Thus, geographic coordinates, historical-political roots and the configuration of British state institutions and institutional interactions are evidence of something akin to an iterative model where changes in political practices led to the evolution and recalibration of the model.
Thus, from its mid-19th-century origins, the Westminster model's four defining doctrines identified by Birch developed to encompass notions of a professionalized, non-partisan bureaucracy (associated with models of ‘Whitehall’), and of ‘meta-constitutional orientation of power-hoarding’ (associated with models of majoritarian democracy). R&S (Reference Russell and Serban2021: 752, 761) come close to acknowledging a process of iteration, by referring to ‘Lijphart's version of the model’. Nonetheless, they continue to seek an essentialist set of common attributes ascribed to a complete meaning of the model (R&S Reference Russell and Serban2021: 751). Indeed, some of the ‘conceptual confusion’ discerned by R&S may arise from failing to specify which version of the Westminster model is being analysed.
Lijphart (Reference Lijphart1984: 5), for example, noted a distinction to be drawn between ‘the British version of the Westminster model’, as the ‘original’ and narrow model focused upon ‘British parliamentary and governmental institutions’, and his more widely focused majoritarian model (with the term ‘Westminster model’, appended in parentheses, after the term majoritarian model (Reference Lijphart1984: xiii, 5). Lijphart initially identified nine characteristics of the Westminster model that ‘together make the model thoroughly majoritarian’ (Reference Lijphart1984: 9). Yet, this was a model with notable ‘add-ons’ from the core characteristics identified earlier by Birch (Reference Birch1964), and later by Gamble (Reference Gamble1990) and Rhodes et al. (Reference Rhodes, Wanna and Weller2009).
In the decades after publishing Democracies (Reference Lijphart2012), Lijphart proceeded to provide iterations of the Westminster model – to reflect more accurately ‘changes in British politics’ (Lijphart Reference Lijphart2012: 20). In this process, the defining institutional elements of the first Lijphart model were recalibrated: with some dropped, some refined and some added (Reference Lijphart2012: 9–29). While Lijphart (Reference Lijphart2012: 20) maintained that such modification did not constitute a ‘basic shift from the Westminster model’, the process of modification that he undertakes might be conceived as the analytical equivalent of iterative conceptual (re)modelling. R&S, in acknowledging ‘Lijphart's version of the model’, could have investigated how the UK has provided iterated versions of the Westminster model and how the model itself has evolved. Moreover, the precision attributed by R&S (Reference Russell and Serban2021: 760) to Lijphart's modelling of Westminster misses his diachronic adaptations and reformulations. Equally, it obscures the partial institutional modifications and deviations acknowledged by Lijphart (Reference Lijphart1984: 5; Reference Lijphart2012: 10, 20) when comparing the UK and those other states that ‘closely approximated’ to the Westminster model. While Lijphart (Reference Lijphart2012: 20, 26, 29) acknowledges flexibility – in approximations, deviations, modifications and ‘various other qualifications’ – needed to compare states with ‘predominantly Westminster-style institutions’, R&S appear committed to their search for comparative analysis founded on precise attributes and precise terms (Reference Russell and Serban2021: 744, 745).
A comparative perspective
Writing from a comparative perspective, the sheer prevalence of recent analyses that utilize the concept of the ‘Westminster model’ suggests that not all scholars find the term problematic. Adopting the same methodological approach (i.e. a databases-based search of keywords such as ‘Westminster model’ or ‘Westminster system’) for the most recent period not covered by R&S (i.e. January 2017–January 2021) yields a rich seam of comparative scholarship on ‘Westminster traits’, including bicameralism, legislative practice, cabinet governments, pluri-constitutionalism, intelligence oversight, etc. (e.g. Collier and Raney Reference Collier and Raney2018; Defty Reference Defty2020; Fleming Reference Fleming2019; Helms Reference Helms2020; O'Brien Reference O'Brien2019; Wincott et al. Reference Wincott, Davies and Wager2020). Moreover, as Julian Bernauer and Adrian Vatter (Reference Bernauer and Vatter2019) have illustrated, the concept of the ‘Westminster model’ has guided sampling strategies in both small-N and medium-/large-N comparative studies. With this comparative body of work in mind, it is possible to advance three main arguments against the core conclusion made by R&S. First, they fail to distinguish between theoretical and empirical concepts. Second, they adopt a narrow sampling strategy. Finally, they do not acknowledge that ‘conceptual stretching’ is (and to some extent inevitably) at the heart of any comparative endeavour.
Taking these arguments in turn, R&S (Reference Russell and Serban2021: 748) set out the requirements for concepts in political science. This opening part of their article adopted a Sartorian approach with its emphasis on taxonomical hierarchy (for a discussion see Collier and Mahon Reference Collier and Mahon1993: 145) while dismissing the more expansive and flexible ‘family resemblance’ approach that assumes valid cases will share a set of properties while not all cases share them all. However, in comparative politics, concepts are not necessarily bound to a theoretical connotation or a collection of abstract properties. There are also empirical concepts equipped with observable attributes and some theoretical meaning (see Sartori Reference Sartori and Sartori1984).
When it comes to concept formation in comparative political science, the more fundamental insight highlights the role and value of ‘ideal types’ or ‘polar types’. ‘The advantage of the ideal type is that it captures the full richness of a type’, as Matthijs Bogaards eloquently explains (Reference Bogaards2000: 397–8), ‘empirical phenomena will never fully correspond to it, nor should they be expected to do so’. The polar type is another possibility. Polar types come in pairs, as they mark the extremes along a continuum. Real-world cases will fall somewhere in between, in greater or lesser proximity to one of the poles. Ideal types are almost by definition crafted as ‘pure’ concepts, just as polar types (the most obvious of which are Lijphart's consociational versus majoritarian democracies) are designed to be extreme, not to reflect reality, but as heuristics that facilitate the analysis of complexity (for a discussion, see Swedberg Reference Swedberg2018 or Palonen Reference Palonen2018: 7). Our suggestion is simply that the logic-building processes through which comparativists construct conceptual frameworks are flexible enough to cope with complexity. It is therefore symptomatic rather than problematic if certain features of the ideal type are absent or even ‘borrowed’ from their opposite incarnation.
The same logic of constructing theoretically maximalist ‘extrema’ to compare empirical variety can be found elsewhere in comparative politics (e.g. Bernauer and Vatter Reference Bernauer and Vatter2019; Gerring and Thacker Reference Gerring and Thacker2008). R&S's epistemological and methodological approach would be unlikely to capture the full richness of type – to paraphrase Bogaards (Reference Bogaards2000) – irrespective of the focus of analysis because its contours are too narrow and rigid. If there is an epistemological and methodological gap to fill, then it is about greater transparency and precision when distinguishing between theoretical and empirical concepts (see Swedberg Reference Swedberg2018).
This focus on conceptual types flows into the second argument and the implicit suggestion in R&S's approach that there is but one sampling strategy in comparative politics. R&S (Reference Russell and Serban2021: 745) outline the origins of the Westminster model which, at one time, served as a ‘convenient label’ to denote the ‘English’ constitution (Bagehot Reference Bagehot2001 [1867]). What R&S (Reference Russell and Serban2021: 745) criticize is that the Westminster model started to live outside its original heartland: ‘it became increasingly used in a comparative context to describe countries influenced by the British system’. The significant confusion about its definition and its ‘over-extension’ to ever-new cases lead R&S to claim that the Westminster model is falsely used to select cases for comparative analyses. Eventually, such a sampling strategy feeds ‘flawed descriptive and causal inferences’ (R&S Reference Russell and Serban2021: 747).
However, there are different rationalities underpinning case selection in comparative political studies (for a discussion see Seawright and Gerring Reference Seawright and Gerring2008); the Westminster model is not only used as an ex ante ‘purposive’ sampling strategy in which countries with ‘Westminster legacies’ (Patapan et al. Reference Patapan, Wanna and Weller2005) are put together (see also the section on the interpretive perspective and Rhodes et al. Reference Rhodes, Wanna and Weller2009: 230; Boswell et al. Reference Boswell, Corbett and Rhodes2019: ch. 4). A distinction needs to be drawn between small-N (qualitative) and medium-/large-N (quantitative) research designs. In the former, cases may be selected on joint colonial heritage, although such studies are generally focused on a single specific (institutional) trait. Examples include Derek O'Brien's (Reference O'Brien2019) work on bicameralism in small states, Cheryl Collier and Tracy Raney's (Reference Collier and Raney2018) comparative study of parliamentary sexism and sexual harassment, Thomas Fleming's (Reference Fleming2019) analysis of partisan dealignment and committee powers, Florence So's (Reference So2018) exploration of leadership turnover in opposition parties, or Ludger Helm's (Reference Helms2020) scholarship on ‘heir apparent’ prime ministers in Westminster democracies. Although such small-N comparative analyses can only claim limited generalization beyond the set of selected cases, they are highly valuable since, as a ‘most likely case’ of (former) Westminster models-design, they offer unique evidence of how ‘transplanted’ (or colonially imposed) role models (see Vatter et al. Reference Vatter, Freiburghaus and Arens2020) ‘may produce different results and evolve in distinctive ways’ (Altman Reference Altman2008: 483).
In major medium-/large-N studies, the ‘Westminster model’ is not used as a sampling strategy on its own. Rather, broader categories such as OECD countries (Roller Reference Roller2005) or consolidated democracies with a minimum population size (Lijphart Reference Lijphart2012), for example, are employed. Then countries that share the ‘Westminster’ heritage can be sub-selected from the full sample to identify and investigate certain commonalities. Through this approach, a ‘Westminster’ cluster may result as an ex post empirical result rather than constituting an ex ante subset of cases.
What becomes obvious, then, is how medium-/large-N studies may guide and refine theoretical samples of (former) members of the ‘Westminster family’ (Eichbaum and Shaw Reference Eichbaum and Shaw2007: 613) anew. Using ‘running examples’ (Bernauer and Vatter Reference Bernauer and Vatter2019: 10) does ‘enrich the quantitative-comparative analyses presented with some qualitative illustrations on the origins, dynamics and consequences of empirical patterns of democracy’. Depending on the (institutional) trait that is of interest, the Westminster model is thus not an arbitrary, but indeed a valuable ‘purposive’ sampling strategy. To critique comparative scholarship for utilizing the Westminster model as a ‘convenient cloak to imply that single-country studies have comparative application’ as R&S suggest (Reference Russell and Serban2021: 761) risks painting an inaccurate picture of the existing research base. This narrow view also fails to recognize that not only generalization but also describing and analysing complex specificity in context (Geertz Reference Geertz and Geertz1973) is a valuable and valued goal in comparative studies (see Boswell et al. Reference Boswell, Corbett and Rhodes2019, and the ‘interpretive’ section below).
This point brings us to this section's third and final argument and our suggestion that R&S's approach does not recognize that ‘conceptual stretching’ is at the heart of comparative work. Or to put the same point slightly differently, the boundary between ‘conceptual travelling’ and ‘conceptual stretching’ is opaque. So, whenever a concept ‘travels’ to new terrain, in all but a tiny fraction of cases there is likely to be some form of ‘stretching’. In fact, conceptual stretching and conceptual flexibility are at the heart of comparative research. R&S, in contrast, proffer a fixed view of what a concept in comparative politics can or should provide.
We argue that concepts with a real comparative potential cannot have prescribed, essential elements, just as they cannot be fixed in time. Otherwise, comparativists would not be able to make sense of an increasingly complex world. We confront ‘the challenge of dealing conceptually with a great diversity of post-authoritarian regimes’ (Collier and Levitsky Reference Collier and Levitsky1997: 430). This diversity is especially true for empirical cases around the ‘democratic universe’, ‘patterns of democracy’ or ‘democracy with adjectives’ (see Lijphart Reference Lijphart2012) or highly divergent ‘levels of power diffusion’ (Bernauer and Vatter Reference Bernauer and Vatter2019). Hence, to serve the comparative goal, concepts must be flexible (on this see Vatter et al. Reference Vatter, Freiburghaus and Arens2020). If serious conceptual concerns have arisen (e.g. intermingling of ideal and empirical types, over-stretching, etc.), or if misguided sampling strategies have fed flawed inferences, then comparative scholars have been willing to engage in open conceptual debate. In fact, consociationalism – the majoritarian Westminster model's polar type – is a striking example of how the community has come to terms with fundamental disagreements on both the concept's core features and its merits for the comparative enterprise. For example, Arend Lijphart and Ian Lustick have engaged in an epic fight about alleged ‘logical, theoretical, and empirical’ defects (Lustick Reference Lustick1997: 88; see also Barry Reference Barry1975). Later, other comparativists – for example Matthijs Bogaards (Reference Bogaards2000); Soeren Keil and Allison McCulloch (Reference Keil and McCulloch2021) – joined in. For more than half a century, consociationalism has been well and alive and continues to inspire scholars of comparative politics and beyond (see Bogaards et al. Reference Bogaards, Helms and Lijphart2019), because it has prompted criticism and debate. Today, scholars even go as far as dropping essential cornerstones altogether (e.g. consociationalism's sociopolitical ‘pillarization’; see Freiburghaus and Vatter Reference Freiburghaus and Vatter2019) to make it fit rather than forcing it to ‘retire’ (R&S Reference Russell and Serban2021: 745, 761). The evolution of comparative political sciences has been shaped by ‘great punctuations’ that destroy faith in coherent theories and concepts just to see the subfield reconstruct itself around new assumptions (see Blyth Reference Blyth2006). We are confident the scholarly community will similarly ‘self-regulate’ the Westminster model as a valuable concept in a broad analytical toolkit.
An interpretive perspective
An interpretive perspective offers a viable alternative comparative research strategy to those concerned with variables, models and generalization (e.g. Boswell et al. Reference Boswell, Corbett and Rhodes2019; Rhodes et al. Reference Rhodes, Wanna and Weller2009; Weller et al. Reference Weller, Grube and Rhodes2021). Interpretive approaches also capture the role and relevance of the beliefs of practitioners. At its simplest, if people believe it to be true, the consequences are real, so they act as if there is a Westminster model. The Westminster tradition continues to shape beliefs and practices (see Rhodes Reference Rhodes2011) and the aim of this subsection is to develop this argument.
R&S (Reference Russell and Serban2021: 747, 759) adopt a naturalist perspective or the ‘positivist tradition’. Explicitly, they follow Giovanni Sartori on concept formation. The case against Sartori's approach is put forcibly by Bevir and Kedar (Reference Bevir and Kedar2008: 504), who argue that Sartori treats concepts as ‘social objects as if they were akin to physical ones’. His is an example of ‘strong essentialism’ because he requires a commonality of attributes for a concept so that ‘when a concept is applied to new cases, then, if those cases do not share the core features that are shared by previous cases, the validity of the concept diminishes’. Finally, for Sartori concepts are ‘tools over which the social scientist should seek perfect control’ (for a more extended discussion see Bevir and Kedar Reference Bevir and Kedar2008: 509–10).
By contrast, an interpretive approach focuses on ‘meanings, contingency, the situatedness of the scholar and the dialogical principle’. Mark Bevir and Asaf Kedar (Reference Bevir and Kedar2008: 508, 514) argue that such ‘anti-naturalist premises are the most appropriate for social science’. They conclude that the positivist tradition of Sartori is another example of the ‘discredited naturalism’ pervading ‘the unreflective practice of many social scientists’. This criticism flows over into R&S's (Reference Russell and Serban2021: 759) discussion of ‘family resemblances’ and especially their argument that, ‘In strict social science terms Table 4 demonstrates that the concept does not even meet the formal, albeit loose, requirements of family resemblance – in terms of a minimum number of clearly definable features being shared’. Their table is a list of the attributes of the Westminster system. Two points are immediately apparent. The phrase ‘in strict social science terms’ means they espouse the naturalist tradition, and the list of attributes is an exercise in ‘weak essentialism’ (Bevir and Kedar Reference Bevir and Kedar2008: 508). Paradoxically, although R&S recognize the messy diversity, they proceed to reject it for not meeting their predilection for strong essentialism.
To tease apart this criticism, it is helpful to distinguish between tight and loose notions of ‘family’. As with real life, some families are tightly cohesive in the traditional sense of a nuclear family (two-up, two-down, two-sexes), whereas other families are less cohesive and function more like a loosely coupled network. Clearly, R&S are using ‘family’ in the more cohesive sense. However, Wittgenstein used the phrase ‘family resemblance’, ‘to convey a sense of indeterminate plurality, not common membership’. So, his ‘looser sense of a “family” is clearly anti-essentialist’ (Bevir and Kedar Reference Bevir and Kedar2008: 513–14: Wittgenstein Reference Wittgenstein2009 [1953]: §68–9). R&S shy away from the loose notion of family because they are looking for concepts that can be used to generate general, causal explanations.
R&S (Reference Russell and Serban2021: 747) devote one paragraph to the interpretive approach to the Westminster model, describing it as ‘a set of beliefs and a shared inheritance’ before moving on quickly by noting that ‘for works following a more positivist tradition in political science this seems problematic’. Our point is simply that it might be the narrowness of their approach that is a problem.
Bevir and Rhodes (Reference Bevir and Rhodes2003), for example, look at how ideas are constituted (and reconstituted) through human action against the backcloth of inherited traditions. Their focus is on meaning and its historically contingent nature; on complex specificity in context, so they focus on the Westminster family in ‘the natural language sense – of individuals sharing a common ancestor – rather than in formal social science terms’ (R&S Reference Russell and Serban2021: 759). It is precisely that common ancestry that Rhodes et al. (Reference Rhodes, Wanna and Weller2009: 230) take as their rationale for comparing Westminster. They explore ‘a loose family of governments’ which are related by history, belief systems and inclination. They analyse ‘the family of ideas’. They argue that, ‘if the test of Westminster is whether any country … fits a prescribed set of institutional characteristics, … the critics have a point, though a trivial one’. Such a test rests on essentialism and ‘assumes a historical standard against which comparison can be made. They [the critics] miss the point’ (Rhodes et al. Reference Rhodes, Wanna and Weller2009: 222).
Comparative interpretive research is different because it explores complex specificity in context, not generalizations (see Boswell et al. Reference Boswell, Corbett and Rhodes2019 for a more detailed discussion). It seeks to raise new questions by ‘shaking the bag’ and edification or ‘new, better, more interesting, more fruitful ways of speaking about’ politics and government (Rorty Reference Rorty1980: 360). Interpretive research provides detailed studies of social and political dramas and of the performative dynamics through which institutional relationships will inevitably play out. Their commitment to ‘complex specificness’ means that naturalist political scientists often write off the value of interpretive research as idiographic. Critics claim that it is not possible to deduce laws and predict outcomes from fieldwork; that is, it is not possible to generalize. Of course, it does not seek to make statistical generalizations and produce ‘laws’, but it does make general statements. John Boswell et al. (Reference Boswell, Corbett and Rhodes2019: 28–30) prefer the phrase ‘plausible conjectures’ to either ‘generalizations’ or ‘general statements’ because the phrase better communicates the provisional nature of knowledge.
Plausible conjectures are to interpretivists what generalizations are to naturalists. They are general statements that are plausible because they rest on good reasons and the reasons are good because they are inferred from relevant information (see Boudon Reference Boudon1993). What interpretivists cannot do based on these data is make generalizations and propound laws. The aim is for ‘small facts to speak to large issues’ (Geertz Reference Geertz and Geertz1973: 23). In a similar vein, Michael Burawoy (Reference Burawoy1998: 5) suggests the task is to extract ‘the general from the unique, to move from the “micro” to the “macro”’. Interpretive research can uncover and deliver plausible conjectures that speak to large issues. Variables, models and generalization are not the only game in town.
This brings us to a fourth and final critique: the Westminster model is not just an academic notion living in our theories and research seminars (textbooks and student reading lists). It exists in a world of practice: the notion of the Westminster model is alive and well in the everyday practices of civil servants and ministerial departments (Rhodes Reference Rhodes2011). Inherited Westminster beliefs inform today's practices. So, the claim that the expression ‘Westminster model’ ‘perpetuates a false impression among practitioners that a meaningful “family” of political systems exists’ might err on the side of overstatement. In the conclusions to Comparing Westminster, Rhodes et al. point out that:
While academics may bemoan the use of the term, in one form or another it continues to thrive. It is used by prime ministers and leaders of the opposition; by frustrated back benchers and senior public servants; by judges and constitutional lawyers. References often occur in commissions of inquiry and in newspaper editorials. If they all continue to use the term and regard it as useful, as part of the political dialogue, then the issue for us is not whether they are right or wrong, but why they do so. It is of minor concern whether the term is still used fruitfully by other academics. (Rhodes et al. Reference Rhodes, Wanna and Weller2009: 222)
In sum, jettisoning the Westminster model is premature. For interpretivists, the concept offers theoretical purchase and analytical leverage to those wishing to understand the past, present and future of democracy in many parts of the world. If the naturalist approach does not provide such purchase then it is arguably a reflection of the limits of that approach, not of the Westminster model. The interpretive approach, which views Westminster as an inherited tradition, not a strict model with fixed characteristics, does provide that purchase and power by helping scholars explain not only the ways in which the tradition has evolved but also the everyday beliefs and practices of politicians and public servants.
Conclusions: stretched but not snapped
‘[W]hile the “Westminster model” may once have had a meaning (or meanings), a combination of conceptual confusion among authors and real-world change have seen it stretched beyond recognition.’ R&S (Reference Russell and Serban2021: 760) conclude, ‘By now the term has gone well beyond individual cases of what Sartori referred to as “individual ambiguity” to achieve a damaging state of “collective ambiguity” in the literature’. We would, of course, agree that a dead concept deserves a decent funeral but we have sought to question the claims and evidence on which R&S make their terminal diagnosis. We have suggested their approach risks further muddying the concept rather than clarifying what the term encapsulates, how it is utilized and why it still matters for comparative research.
The four discrete but interwoven conceptual, institutional, comparative and interpretive perspectives have each questioned the contribution of R&S's approach and conclusion while also revealing the Westminster model's continuing relevance. ‘Stretched but not snapped’ rather than R&S's ‘stretched beyond repair’ provides, we suggest, a far more accurate, subtle and sophisticated phrase for managing the muddle of the Westminster model.
Our critique of R&S's approach can be summarized in the suggestion that the Westminster model was never designed or intended to act as a positivist concept facilitating scientific research. The ‘muddled’ qualities that R&S identify reflect the fact that it is, and always has been, an ‘organizing perspective’ or ‘sensitizing concept’ that draws attention to several themes and institutional characteristics. It is in ‘managing the muddle’ – a challenge that the comparative section suggested scholars were more than capable of doing – that political science is required to demonstrate its creative intellectual qualities.
This emphasis on creativity stands in stark contrast to the schema that R&S adopt. Our fear is that an essentialist approach served only to muddy the waters. R&S's theory and method might be interpreted as setting an intellectual trap that the Westminster model was always going to fall into by nature of its fuzzy qualities. The main aim of our scholarship over the years has been to expose and critique the inner logic and constraining shadow of the Westminster model. Through either interpretive insights, or at the very least a more pluralistic approach to theory and methods, we offer a more rounded account of the concept and its potential for comparative research.
The opening decades of the 21st century have been tumultuous. Populism and pandemics, technological shifts and population flows all point towards an increasingly ‘liquid world’ (Bauman Reference Bauman2011) in which the capacity of concepts forged within more stable times to provide intellectual anchorage and shared understandings will almost certainly diminish. The tension between acknowledging the existence of complexity while longing for the certainty and clarity provided by some overarching unitary truth will inevitably increase in ways which place new challenges at the door of political science. Viewed from this perspective, one of the core contributions of R&S may well have less to do with ‘muddying’ or ‘muddling’ and far more to do with the debate they have initiated on what we expect from our concepts.
