Re-Structuring Parliamentary Roles

Prominent extant definitions of, and approaches to, parliamentary roles conflate roles and behaviour and, consequently, contain a latent behaviouralism that enfeebles the role that institutions and other structural features play in outcomes. To overcome such issues, this article makes the case for a historical institutionalist approach to the study of parliamentary roles, premised on a critical realist ontology and the figure of homo sentiens . Such an approach defines parliamentary roles in terms of sets of expectations impinging on incumbents of the social position of Member of Parliament and has a number of consequences for how we study parliamentary roles. Namely: the focus of research is squarely on legislative roles ; attention is shifted away from focusing (so much or solely) on what mp s think; patterns of behaviour become the starting point for identifying parliamentary roles; and normative questions concerning the ‘goodness’ of parliamentary roles and attendant parliamentary institutional architecture gain prominence.


23
From early beginnings in structural-functionalism and symbolic interactionism, through the behaviouralist and rational choice revolutions and the new institutionalist counter-revolution, to the interpretivist and ideational turns: to learn about role theory in the sub-discipline is to learn about the history of the broader discipline to which it has made an important contribution. Yet reading about parliamentary roles is also a slightly awkward affair because there often appears both a recurring lack of confidence and a certain defensiveness about the contribution it has made. For example, Donald Searing begins Westminster's World by saying: Who now writes about roles?… [P]olitical scientists have done without it for nearly a decade. Articles continue to be produced but they aren't much discussed. There aren't any major research projects under way on political roles. Nor are there any significant innovations in theory and method. It seems difficult to believe that studies on political roles actually dominated our research on politicians during the 1960s and early 1970s (1994,1).
This was perhaps understandable in 1994 but, nearly two decades later, Kaare Strøm was able to give the same conclusion in his updated chapter in the Blomgren and Rozenberg collection on parliamentary roles that he gave in his original 1997 essay: "[In] contemporary political science role analysis has fallen somewhat out of favor" (2012,99). Similarly, after beginning their edited collection by stating that the role of parliamentary roles has been unfashionable since the 1980s (2012a, 1), Rozenberg and Blomgren conclude by saying "Hopefully, this volume has confirmed that roles are coming back into legislative studies" (2012,211).
The source of these uncertainties seemingly stems from a failure to develop a homogenous role theory, and disappointing attempts to link mp's views about their roles to their activities. Blomgren and Rozenberg state, "[I]t was difficult to identify sets of attitudes associated with being an mp that were at once comprehensive, coherent, consequential, and mutually exclusive" (2012a, 1). It may be too crude a characterisation but there seems to be a lack of confidence among those who wish for homogeneity (firmly ensconced within a rational choice tradition) due to the disappointing outcome of empirical analyses, and a certain defensiveness among those located outwith the rational choice tradition who wish to assert their work's usefulness, despite not adhering to a deductive logic, nor aspiring to universalism and generalisability.
In making their argument that roles are coming back into legislative studies, Rozenberg and Blomgren make a series of claims: (i) "It is illusory to call for a re-structuring parliamentary roles International Journal of Parliamentary Studies 1 (2021) 22-46 single conception of parliamentary roles"; (ii) "studies of legislative roles require data about mp s' opinions"; (iii) that the comeback of parliamentary roles is closely linked to the rise of neo-institutionalism;1 and (iv) "the come-back… is characterized by the idea that roles are less interesting to study per se, but that they say something of other phenomena" (2012, 211-220; 2012b).
I wish to offer wholehearted support to only two of these four claims. I agree that studying roles allows us to highlight important issues concerning other social and political phenomena and that this is the main contribution of parliamentary role analysis, even if identifying and explaining (repertoires of) parliamentary roles is interesting and important in and of itself. I also think it is true that new institutionalism is important for understanding parliamentary roles. However, I wish to move away from rational choice institutionalism, sociological institutionalism and the strand of historical institutionalism that seeks to combine the micro-level, calculus logic of the former with the macro-level, cultural logic of the latter that too often has dominated debates on parliamentary roles hithertofore. Instead, I wish to draw attention to work within historical institutionalism that makes use of critical realism (for example Bell 2011Bell , 2012Delbridge and Edwards 2013;Marsh 2009), as well as more general critical realist work on roles.2 Critical realism3 has not been discussed in parliamentary roles literature. This is a shame because there is a long tradition within this philosophical approach of thinking about roles, as well as about explanation, causality, the relationship between structure and agency, and how science proceeds in 1 Neo-or new institutionalism is an umbrella term for a number of analytical frameworks that reject (at least aspects of) both the 'old' institutionalism, which underpinned much early political science, and the behavioural and rational revolutions of the 1950s and 60s. At the heart of all strands of new institutionalism are two very simple ideas: that outputs cannot be determined simply by inputs into the political system; and that, therefore, institutions matter and make a difference to outcomes (see Peters 2005). 2 E.g. Archer 1995Archer , 1996Archer , 2000aArcher , 2000bArcher , 2003Bhaskar 1998;Elder-Vass 2010;Layder 1981;Luke and Bates 2015. 3 Critical realism is a movement in philosophy that seeks to provide a framework within which the study of aspects of society can take place. It combines a depth ontology (the idea that there is a real world beyond our understanding of it, not all of which reveals itself to our senses), an epistemic relativism (the idea that "all beliefs are socially produced, so that all knowledge is transient, and neither truth-values nor criteria of rationality exist outside historical time" (Bhaskar, 1998, 62-63)), and a judgmental rationalism (the idea that there are ways of adjudicating between different beliefs and theories and that not all competing knowledge claims should be regarded equally). For those unfamiliar with critical realism, excellent introductory material can be found in Archer et al. 1998;Collier 1994;and Danermark et al. 2002.  Within critical realism, a role is best defined as "a set of expectations impinging on an incumbent of a social position" (Thornton & Nardi, 1975, 872). Using this definition as a starting point, I make a fourfold argument. First, it is not illusory to call for a single conception of parliamentary roles (even across markedly different kinds of parliaments and political systems), although the specific forms, or repertoires, that roles take in specific legislative contexts will differ.4 Second, a sharp distinction should be drawn between parliamentary roles and the patterns of behaviour that are shaped, at least partly, by that role. Making this distinction avoids both the conflationism that besets the most prominent definitions of parliamentary roles and the subsequent problems of ascribing causality and downplaying the role of structural features of society, including sometimes institutions.
Third, many roles identified in the literature -in particular, what Rozenberg and Blomgren label as representative roles -are not really roles at all but are best understood as stances. This is not to say that these stances are unimportant for understanding the roles and activities of parliamentarians and the functioning of parliaments. However, mp s' stances should be seen as only a part of what makes up a role. Therefore, through the incorporation of representative stances, the focus of research on parliamentary roles should be squarely on legislative roles and the question of "how mp s organize their activity" (Rozenberg and Blomgren 2012, 211) or, perhaps more accurately, how mp s' activity is organised.
Fourthly, this critical realist conceptualisation of parliamentary roles shifts attention away from focusing (so much or solely) on what mp s think. This is especially the case during the first step of research put forward by Searing (1991Searing ( , 1255 concerned with the mapping of parliamentary roles, but also during the second explanatory step of research. I, therefore, only offer tepid support for Rozenberg and Blomgren's claim about the necessity of data about mp s' opinions, and believe that the emphasis placed upon mp s' opinions in many approaches is detrimental to understanding more fully the structural and institutional factors that shape parliamentary roles. To make this argument, the article contains six further sections. First, I summarise the most prominent extant conceptualisations of, and broad approaches to the study of, parliamentary roles: those proposed by Donald 26 Searing and Kaare Strøm. Second, I criticise both Searing's and Strøm's conceptualisations of parliamentary roles, as well as other definitions within the literature, on the basis that they conflate roles with behaviour and that their work contains a latent behaviouralism that enfeebles the role that institutions and other structural features of society play in outcomes. Third, I offer an alternative conceptualisation underpinned by critical realism and developed from the definition of roles as 'sets of expectations' introduced above. Fourth, I reflect on the role of new institutionalism within parliamentary roles scholarship and argue for a form of historical institutionalism premised on a critical realist ontology and the figure of homo sentiens. Fifth, I outline the ramifications of my alternative conceptualisation and its attendant institutionalist framework for how we might study parliamentary roles. In particular, I contend that we should adopt a threefold research process (when treating parliamentary roles as the dependent variable) that begins with divining parliamentary roles through the analysis of mp s' behaviour, moves on to analysing the structural and institutional factors which shape (repertoires of) parliamentary roles, and only then moves on to analysing thinking by and also, crucially, about mp s. In this research process, mp s' opinions become much less central to the study of parliamentary roles and, on some occasions at least, they may not even be necessary to say something of import. The necessity of such a shift in emphasis lies in the attempt to move away from the dangers of a static, indexing or cataloguing of role expectations that arguably infects many approaches to the study of parliamentary roles, and towards a dynamic, structurally and institutionally-situated analysis of the concrete interactions of role-incumbents. In the sixth concluding section, I re-emphasise the importance of studying roles, both those in parliament and elsewhere in the political domain, on the basis that they are meaningful social features that make a difference to outcomes. This then allows me to return to the claim made by Rozenberg and Blomgren (2012) that the main contribution of parliamentary roles research should be in terms of what it helps us say of other phenomena. I argue that a critical realist-informed, historical institutionalist approach to the study of parliamentary roles is well placed to take on this task because it foregrounds the situatedness of role incumbents both within parliament and society more broadly. As such, it allows us to reflect upon state-societal relations and evaluate the ability of (certain groups of) mp s to access and perform particular parliamentary roles and undertake their representative, scrutiny and legislative functions. This, in turn, allows us to pose normative questions about the 'goodness' of certain institutional arrangements and the necessity of parliamentary and societal reform.

Searing and Strøm on Parliamentary Roles
Despite the continued prominence of Wahlke et al. (1962) and Eulau et al. (1959), most recent scholarship on parliamentary roles takes the work of Searing or Strøm (or sometimes both) as its starting point, as Searing himself notes (2012, xxi). I do similar here.

Searing and Roles as 'Patterns of Goals, Attitudes and Behaviours'
In his work on the UK House of Commons, Searing (1987Searing ( , 1991Searing ( , 1994Searing ( , 1995 employs a motivational approach to conceptualise mp s' backbench and leadership roles. Underpinned by both interpretivism and March and Olsen's new institutionalism (1984), he argues this approach is able to incorporate insights from both sociological and economic traditions by "recognizing that the roles of politicians are embedded in the institutional contexts while at the same time treating the role players as purposive actors with independent standpoints" (1991, 1252).
Searing believes that his approach is able to bring together homo sociologicus and homo economicus in order to study homo politicus by offering a synthesis of rules and reasons which understands their interactions in role creation. Searing rejects a universalistic role theory, rational choice theory, and what he terms naturalistic explanation and attendant law-like generalisations, prediction and deductive reasoning. Instead, he embraces of a more modest framing theory which openly acknowledges the local, contextually-bound nature of roles with its value lying "in the perspective they provide for particular explanations of particular cases" (1994, 25). Searing's alternative interpretive explanation of roles is based upon understanding their players' motivations and purposes. Promoting semi-structured interviews as a preferred research method, he says "if one wants to know why politicians do something, the most sensible way to begin the investigation is to ask them and listen carefully to what they say" (1991, 1244; 1255-1256).
At the heart of this motivational approach, therefore, is the encouragement of "the reconstruction of political roles as they are understood by their players" (Searing, 1991(Searing, , 1252. Searing argues, "by directing our concepts and measures toward the roles that exist in the minds of politicians [rather than substituting assumptions about operant conditioning or egoistic utility maximisation], we will be in the best possible position to explain the consequences of these roles for political behavior" (1991, 1250; see also 1994, 389).
This then leads Searing to conceptualise (political) roles as: "particular patterns of interrelated goals, attitudes, and behaviors that are characteristic of people in particular positions" (1994, 18; original emphasis; see also 1991, 1255), 'behaviours' being included because "this is how politicians themselves see it" (1991,1254). For Searing, analysing motivational roles requires two steps: a descriptive mapping exercise of roles within an organisation; and, subsequently, an explanation of the components, consequences and origins of those roles (1991,1255). Mapping requires reconstructing roles at a slightly more abstract, compositional level than the accounts provided by individual players themselves, which can then be supplemented with "quantitative probing of the interpretations as they develop" (Searing, 1991(Searing, , 1255. The second step seeks to build upon this mapping exercise by offering an interpretive explanation of the roles uncovered and, in particular, such things as the relationship between motivational cores and the attendant attitudes and behaviours, the relationship between roles and other non-typical attitudes and behaviours to which they are sometimes related, and why people take up one role rather than another. Using this approach and through a team undertaking 521 interviews with mp s, Searing identifies or reconstructs four backbench preference roles (Policy Advocate, Ministerial Aspirant, Constituency Member, and Parliament Man (sic)) and four leadership position roles (Parliamentary Private Secretary, Whip, Junior Minister, and Minister) that, together, he believes, form the roles that characterise Westminster's World.

Strøm and Roles as 'Strategies' and 'Regularized Patterns of Behavior'
Strøm's work on parliamentary roles (1997; 2012) is less extensive than that of Searing but no less influential. As with Searing, Strøm offers a general critique of previous approaches to (parliamentary) roles: [R]ole analysis has not always been quite clear about what it can do or about its own limitations. The concept itself has seemed to subsume individual beliefs, common expectations, actual behaviours and even institutional functions without clear demarcations or causal stipulations between these different components (2012, 99).
Strøm also agrees with Searing that the motivational approach, as opposed to the structural or interactional approaches, is the most promising way to study parliamentary roles and that new institutionalism is important for such an endeavour (1997,(157)(158). Strøm departs from Searing in two main ways. First, he supports rational choice as an alternative motivational approach, stating that "its parsimony and deductive rigor make it the most plausible vehicle for such a theoretical project" (Strøm, 2012, 86). Second, Strøm draws a clearer demarcation between roles and preferences to that provided by Searing.
Strøm defines roles in general as "regularized patterns of behavior that individuals display in different social circumstances, most typically in common holden bates International Journal of Parliamentary Studies 1 (2021) 22-46 and repeated activities on which others depend" (2012,85). He defines parliamentary roles in particular as routinized strategies, or behavioral strategies, that are "driven by reasons (preferences), and constrained by rules" (2012,87).
It is important to Strøm not to conflate roles with either reasons, preferences or rules. Roles should be thought of as strategies or game plans that are "endogenous prescriptions as to how actors… may most successfully and efficiently act to maximize the likelihood of whatever outcomes they favor" (Strøm, 2012, 87). Yet these strategies only make sense, Strøm says, when we understand the preferences/reasons that drive them and the institutions that shape them (2012,87).
For Strøm, preferences are "exogenously given 'tastes' that… parliamentarians have over the outcomes that affect their political fortunes" (2012, 87). Although he recognises that cultural expectations and idiosyncracies may shape routines, Strøm states roles are "most likely to flow from reasoned and deliberative pursuits in which parliamentarians engage". Within a parliamentary context, the most important preferences are reselection, reelection, party office, and legislative office (2012, 87 & 90-98; see also 1997, 158).
Strøm defines institutions as the rules that constrain reason/preferences (2012, 86). The most important institutions enabling or constraining parliamentarians to reach the four preferences outlined above are the electoral system (especially with regard to reselection and reelection), party rules (with regard to all four but especially party and legislative office), and parliamentary procedures (especially with regard to party and legislative office) (Strøm, 2012, 91).
Where real-world parliamentary roles lie on a spectrum of position roles ('fully institutionally determined strategies') at one end and preference roles ('institutionally unconstrained strategies') at the other depends on the interplay of these preferences and institutions (Strøm, 2012, 86). The different roles undertaken by parliamentarians reflect the different strategic choices made by those parliamentarians about the allocation of scarce resources in pursuit of those preferences, and the conditioning effects of institutions (Strøm, 2012, 87-88). Roles are not directly observable but can be inferred from "the patterned behavior that parliamentarians display" (Strøm, 2012, 88).

Criticising Searing and Strøm for the Conflation of Roles and Behaviour and its Consequences
Although emerging from different traditions, there are a number of similarities between Searing's and Strøm's work and how they conceptualise parliamentary roles. They both agree on the strengths of a motivational approach for re-structuring parliamentary roles International Journal of Parliamentary Studies 1 (2021) 22-46 studying roles and Strøm himself argues that he is building on a number of Searing's insights (1997, 156-7). Moreover, as Andeweg notes, the typology of parliamentary roles theorised by Strøm is congruent with Searing's typology (2014, 279).
Another similarity is that their conceptualisations of parliamentary roles contain the same weakness: there is a conflation of roles and behaviour, at least at some points in their work. This is problematic because it muddies the water when identifying which causal mechanisms explain the organisation of mp s' activity, not least by offering, albeit implicitly, an enfeebled understanding of structure. In order to avoid these problems, I propose an alternative critical realist-inspired conceptualisation of parliamentary roles.
The criticism that Searing conflates roles with the behaviour to which they help give rise is not new but it is a criticism that Strøm usually does not face. For example, while Andeweg draws attention to Searing's definition of roles including behaviour, he states that Strøm's definition avoids this problem by aligning roles with strategies (2014, 269). However, I believe that this reading is incorrect and that we can and should criticise Strøm for conflating roles and behaviour, at least sometimes. Although Strøm does say that parliamentary roles are "behavioral strategies" and "routinized strategies", he also says that they are "routines, regular patterns of behavior" (2012, 87, emphasis added; see also 2012, 85).
Such conflation, even if sporadic, is problematic because, as Andeweg states: "[I]f behavior is part of a role we can still seek to explain the emergence or change of a role orientation, but we can no longer use an mp's role orientation to explain his or her behavior" (2014, 269). Thus, by invoking such a conceptualisation, we cannot avoid offering anything other than tautological explanations of mp s' behaviour. So, while there is nothing inherently redundant "about finding that politicians who express certain self-conceptions behave in a manner that is consistent with those self-conceptions" (Searing, 1994, 135), as Searing does with the behavior of those mp s who think of themselves as Constituency Members, there is redundancy if those self-conceptions contain the very behaviour being studied. Similarly, while Strøm's statement that "Legislative strategies are not directly observable, but instead we infer them from the patterned behavior that parliamentarians display" (2012, 88) is correct, if roles are defined as strategies and elsewhere as patterns of behaviour, then not only do strategies become directly observable but also roles become both the explanans and the explandum.
Another related problem with such conceptualisations is that a structural feature of society -in this case roles -is being defined in behavioural terms. The problems of such an approach is made clear by Porpora (1989 such an understanding reduces a structural feature of society to an abstraction which, by its very nature as such, cannot be seen as an independent variable which exhibits causal force (1989, 197; 203-206). In this sense, roles become epiphenomena of individual human behaviour and have no independent, irreducible causal mechanisms that help give rise to behaviour. Roles, in this understanding, become enfeebled in their contribution to outcomes.
To overcome these problems, there is a need to define roles both without recourse to behaviour and with a clear recognition that roles help give rise to behaviour, whether patterned or otherwise. There are a number of possibilities already available in the parliamentary roles literature: Wahlke defines a legislative role as "a coherent set of 'norms' of behavior which are thought by all those involved in the interaction being viewed, to apply to all persons who occupy the position of legislator" (1962,8); Blomgren and Rozenberg say that "legislative roles refer to the norms (obligations and expectations) attached to being an mp" (2012b, 8); and Best and Vogel state that roles are "systems of boundary conditions to which the actors who play such roles are supposed to conform, and as the corresponding rights that these boundaries confer" (2012, 37). However, I wish here to draw on critical realist work on roles, not least because of the associated wider ontological and theoretical architecture.

Parliamentary Roles as 'Sets of Expectations'
Founded upon the more general definition offered by Thornton and Nardi (1975), I define parliamentary roles as sets of expectations impinging on incumbents of the social position of Member of Parliament. Although this starting point is similar to some definitions noted above, particularly that of Blomgren and Rozenberg, once placed within a wider critical realist conceptual framework a number of differences emerge.5 First, the stress on sets of expectations establishes the idea that it is misplaced to put too much, or all, emphasis on what mp s either individually or collectively think their roles are. The expectations surrounding being an mp emanates not only from mp s themselves but from (different groupings of) their constituents, the wider electorate, party managers, party members, clerks, research assistants, constituency managers, the media, and a host of civil society organisations. While the impact of constituents, the electorate and party managers especially is often recognised within the literature, this is most often solely in terms of institutional factors which are external to roles and which constrain or enable mp s performing those roles, rather than also as (groups of) actors whose expectations contribute to the sets which comprise the roles attached to the position of mp.
Second, the stress on sets of expectations also establishes the idea that a parliamentary role cannot be reduced to a single attitude, norm or decision rule. This then rules out identifying 'delegates' , 'trustees' and such like -what Rozenberg and Blomgren (2012) label representative roles -as roles.6 Rather, the attitude of an mp should be viewed as contributing to their stance (or non-stance7) within parliament, which can then influence certain expectations that help comprise a role. Archer defines stances as "basic orientations of subjects to society" (2003, 343) with a parliamentary stance then becoming the basic orientation of an mp to parliament and politics. This concept, then, helps us to differentiate between an mp's general approach to being a representative and the parliamentary role they occupy. Stances will surely contribute to some of the expectations that help comprise a role and they will surely contribute meaningfully to the reproduction or transformation of roles and role sets within a legislature. However, stances and roles are not synonymous and should not be viewed as such.
Third, roles need not necessarily be perceived accurately by all those performing such roles. It is the case that roles are intersubjective, cultural elements of society that depend for their existence on at least the tacit acknowledgement of participating agents (Porpora, 1989, 202). However, it is also the case that roles are constructed and re-constructed by a wide range of actors and shaped by a series of structural and institutional factors. As such, the role that an mp performs does not necessarily have to coincide with the role that an mp thinks they are performing.8 For example, it is unlikely that the role, identified by Jenny and Müller (2012), of Spectator -an mp who undertakes little 6 This is not to say that studying these attitudes is unimportant, merely that studying them is studying something other than roles. 7 Not all mp s will necessarily have stances. Some may have lost the ability or inclination to exercise directional guidance over their political activity. These passive agents are people "to whom things happen rather than people who exercise some governance over their lives" (Archer, 2003, 343). 8 In addition to the fact that, as described below, much of the activity of an mp is now very often a team effort of which the mp may not be fully aware, this is for at least three further, parliamentary activity -would be one recognised by those performing such a role (except, perhaps, during the darkest of nights). Indeed, mp s may not even associate themselves, except post hoc, with the more attractive roles identified by Jenny and Müller, such as Exemplary mp and Workhorse, or indeed roles such as Constituency Member, or Policy Advocate, which are widespread among the different typologies on offer. For instance, unlike in the 1970s when Searing and his team undertook their interviews, UK mp s all now have both a significant staff comprising constituency office managers, caseworkers, researchers and the like, and significant resources to support such a team. Therefore, not only may the stance taken by an mp and the choice of role an mp is trying to perform be the result of a collective decision (perhaps also including party members and managers), but also the execution of a parliamentary role may be down to the particular skills and interests of an mp's staff. An mp may well have little or no idea, involvement or interest in the social media strategy put in place by their constituency manager that identifies them as a Local Promoter; they may only be viewed as a Welfare Officer because of their caseworkers' hard work, dedication and care; and they may only be considered a Generalist Policy interrelated reasons. First, quite simply, mp s, like the rest of us, are not infallible, and their perception of their role may not coincide exactly, or even approximately, with the role that they are playing, even if they do direct and closely monitor the work of any support team. Moreover, human powers -including self-perception -are, for a variety of reasons, differentially allocated across actors (Layder, 2009, 177) and, thus, mp s cannot be treated as a homogenous group, all of whom display a similar degree of self-awareness of the role they play and their performance of it. Indeed, in the case of mp s who adopt a non-stance towards their vocation, they may not even be particularly interested in the role they fulfil. Second, roles "do not form in a vacuum, but rather emerge within asymmetrical power relationships and from resources bound to structures (both material and cultural)" (Luke and Bates, 2015, 346-7; see also Bates, 2010). Therefore, organisations, such as parliaments and political parties, can bring to bear downward causation upon the roles performed by mp s, which may not at the time, if ever, be realised by the mp s themselves, in part because of the asynchronicity of attempts to implement control mechanisms by the organisation and periods of reflection for mp s. Third, roles pre-exist particular incumbents, they have a greater durability over time than those particular incumbents, they have a capacity to endure despite considerable changes in the personal features of successive incumbents, and they contain relatively autonomous powers of constraint and enablement which do not remain attached to an incumbent when the incumbent stops performing a particular role (Archer, 2000a, 283). Roles have a reality and their own emergent properties that cannot be reduced to the behaviour, perceptions, preferences or characteristics of their incumbents. Moreover, they do not impact behaviour simply by being filtered through the perceptions of an mp in a linear causal chain of succession. Rather, they make a difference by being part of a contingently-related ensemble of structural and agential factors that, concurrently, make possible a particular pattern of behaviour (Sayer 1984(Sayer , 1998 Advocate because the knowledge and skills of their researcher(s) broaden the range of subjects on which the mp is able to speak convincingly. Fourth, a role does not necessarily have to be coherent to be a role. This is not only because a number of (groups of) actors have differing expectations about what an mp should do, some of which may conflict when brought together in particular coagulations. It is also because of how structures, institutions and organisations intersect and sometimes clash with each other when helping to determine social positions and the plurality of expectations and subsequent variety of roles attached to those positions (Luke & Bates, 2015). This structural and, especially, cultural side of role formation, reproduction and transformation raises, once again, the contribution of new institutionalism to more recent iterations of role theory.

New Institutionalism and Parliamentary Roles: the Promise of a Critical Realist-Informed Historical Institutionalism
New institutionalism's positive contribution to parliamentary role literature is widely acknowledged. For example, Blomgren and Rozenberg argue that the neo-institutionalist turning point in political science has allowed for the elaboration on the relationship between structure and agency and the individual's role within a wider set of institutional mechanisms; this interplay becoming the key scientific endeavour (2012a, 2 and 2012b, 20).
As a consequence, role performance and differentiation are not just agential phenomena, guided by preferences or strategies -something recognised by both Searing and Strøm. For example, Searing argues: For although individual desires and beliefs shape these roles, the fact is that backbenchers are at the same time adjusting to roles that already exist in an already-existing institutional context… Institutional constraints work by setting rules and by setting up situations. They also work by shaping the desires and beliefs of the politicians who encounter the rules and pass through the situations (1994,402).

And Strøm writes:
Institutions affect roles in part by defining the range of behaviors available to members of parliament. Institutions do so by enabling and constraining behavior, that is to say, by making such forms of behavior feasible and others infeasible, and by shaping the incentives that differ- As noted above, Searing positions himself in the new institutionalist tradition of March and Olsen (1984), which Strøm identifies as sociological institutionalism (1997,172), whereas Strøm positions himself within rational choice institutionalism (Strøm, 1997, 156).
Despite this different positioning, Blomgren and Rozenberg suggest that Searing's and Strøm's approaches could be conciliated within the same framework (2012b, 30-31). They state: "Both frameworks share the idea that roles are produced both by institutional rules and personal preferences, and that they have consequences for behavior" (2012b, 30). Blomgren and Rozenberg go on to argue that, by recognising that mp s both understand and play their parliamentary roles, it opens up the possibility of understanding the systematic behavioral aspect of roles "both as a strategic cost-saving device and as a sociological process of diffusion of collective norms" (2012b, 31). They contend that, whether or not these views of Searing and Strøm are complementary or contradictory is up for debate, but that this debate has not yet happened (2012b, 31).
I wish to argue, however, that this debate has happened, both explicitly elsewhere in political science and social science more broadly, and implicitly within parliamentary studies and, especially, the pages of Searing's work (for example 1994, 369-402; and 1991, 1252-1256). This is the debate about the possibility of combining (the study of) homo sociologicus and homo economicus within the same (new institutionalist) framework (Archer 2000a(Archer , 2000bTaylor 1996, 1998;Hamlin 2002;Hay and Wincott 1998;Hollis 1987). Taylor (1996, 1998) argue that historical institutionalism plays a pivotal role in the new institutionalism because it is able to combine what they label as the calculus approach (of rational choice institutionalism) and the cultural approach (of sociological institutionalism). However, as Hay and Wincott argue, the calculus and cultural approaches and hence the rational choice and sociological institutionalisms are founded upon mutually incompatible social ontologies (1998,951): methodological individualism and methodological holism respectively. This intractable divide, as Hay and Wincott call it, has "profound implications for any attempt to fashion a synthetic institutionalism capable of spanning the divide, or even for a less ambitious cobbling together of institutional insights from differently-informed institutionalism[s]" (1998,953).
This incompatability is something that appears to be (implicitly) recognised by Searing when he attempts to bring "homo sociologicus and homo economicus into a flexible framework suitable for studying homo politicus" (1991,1252). Searing argues that the economic definition of rationality is too narrow for the study of political roles and that, if the new institutionalism is to be truly a synthesis, it needs an understanding of rationality as "the exercise of reason in the service of desire" (1991,1253).
Searing, then, modifies homo economicus but, while doing so, leaves the oversocialised homo sociologicus more-or-less intact. Moreover, whatever modifications introduced, Searing does not go far enough in that an attempt to synthesise the economic and sociological approaches (across that intractable divide) remains.
Such synthetic attempts are doomed to unravel because, as Archer notes, the settlement between homo sociologicus and homo economicus requires a moral agent to make it work; one that must be introduced from outside (2000b, 52). No hybrid of the two can provide this moral character, Archer writes, because: On the one hand, 'rational man' is only induced to contract into social norms because they are in his enlightened self-interest: he does not accept them as binding but endorses them calculatively… On the other hand, over-binding [persons] so that they become normatively dopey does not do the moral trick either, since it produces socially conventional behaviour rather than moral sensitivity which bridges the gaps in conventions or copes when they clash (2000b, 52).
We therefore need to find a moral agent (and, subsequently, a form of new institutionalism to accommodate them) who is able to judge discerningly, act reflexively, participate in authentic ethical behaviour, and whose presence means that roles are not reduced either to suits that can be changed at the whim of the wearer or to irremovable straightjackets. Archer's solution is the figure of homo sentiens.9 9 This is a figure borne out of the long-held critical realist desire to overcome the upwards, downwards or central conflationism of much work on the relationship between structure and agency, which reduces explanations to structural causes or agential reasons, or which elides the social and the individual without recognising their distinct, emergent properties, or which fails to recognise fully how structure and agency are ontologically distinct but necessarily empirically related phenomena (for example, Archer 1995, Bates 2006, Bhaskar 1998, Layder 1985. It is a figure borne out of the desire to avoid telling one of two stories where we begin with agential reason or social causes and then perhaps try to fill in the rest as best we can from the other side. As Archer says: [T]here is only one story to tell. Society enters into us, but we can reflect upon it, just as we reflect upon nature and upon practice. Without such referential reality there would be nothing substantive to reflect upon; but without our reflections we would have only a physical impact upon reality (2000a, 13; see also 306-319). Homo sentiens is a person who has "ultimate concerns which are expressive of who they are, and therefore are not a means to some further end" (Archer, 2000a, 83, original emphasis). They are capable of making moral commitments and have a reason for keeping them which derive from their involvement in society (Archer, 2000b, 53). Thus, what an mp wants to achieve in politics (as well as other political actors), due to their ideological and personal commitments and their emotional involvement, are ends in themselves, extensions of the actors themselves, irreducibly social, and which cannot become instrumentally rational public means to further private ends. As Hollis says, those social (and political) relationships to which we are most committed as our deepest concerns are "not the means to [our] flourishing but its constituents" (cited in Archer 2000b, 54). Moral commitments of this kind are neither calculative nor socialised but they are both reasoned and social (Archer, 2000b, 54).
This view, then, leads to the possibility of a more expansive political lifeworld within which Weber's substantive-or axiological-rationality10 and Boudon's cognitive rationality11 can retain a place, alongside that of instrumental rationality (Archer, 2000b, 54; see also Hamlin 2002). It also allows for choice within the personification of roles. As Archer says, our human powers are "not confined to the confines of the role array" and we can reflect upon and, if desired, seek to change both the role array and its wider institutional and systemic context (Archer, 2000a, 314). Therefore, while role-taking could "conceivably be a thin hermeneutic tale told from within a 'form of life"' , role-making "must be a thick account of the objective intertwining with the subjective, thus combining two sets of causal powers which are embedded within the same world" (Archer, 2000a, 314). In other words, and this takes us back to where this section started, the identification and explanation of parliamentary roles are intimately related to that key scientific endeavour of the new institutionalism: the elaboration of the relationship between structure and agency.
Yet, whereas the section began by considering the possibility of both combining Strøm's and Searing's approaches and synthesing calculus and cultural logics, we are now in a position to acknowledge the undesirability and impossibility of such a project. However, we also need to be careful not to throw out the baby with the bathwater: we should not reject historical 38 institutionalism in toto, even if we should dismiss the historical institutionalism of Hall and Taylor. Historical institutionalism remains the most promising analytical framework for identifying, not only the institutional, but also the agential and structural factors involved in the performance, differentiation and transformation of parliamentary roles. This is particularly the case when drawn into a close relationship with critical realism. Critical realism can provide historical institutionalism with a social ontology that is better able to repel oft-made criti-cisms12 directed towards historical institutionalism of structuralist tendencies, an inability to account for endogenous change, and a vagueness when defining institutions, as well as differentiating sharply between the various structural and cultural elements of society and their attendant causal powers (Bell 2011(Bell , 2012Newman 2019). I finish the section, then, by sketching the outlines of such a critical realist-informed historical institutionalism.
As is widely established (for example Peters 2005), at the heart of historical institutionalism are the ideas of: path dependency (i.e. courses of action, policies or, indeed, role repertoires, once introduced, tend to persist); that institutions -defined here as "systems of established and embedded social rules that structure social interactions" (Hodgson, 2007, 96) -play a crucial role in that tendency; and that political interactions should be studied "sequentially, as life is lived" (Sanders, 2008, 39). Institutions are crucial -they matter -because they are real, have their own emergent causal powers, and thereby contribute meaningfully to outcomes (Archer, 1996;Bhaskar, 1998).
However, although crucial to outcomes, institutions do not exhaust the structural features of society (Thelen and Steinmo 1992, 3; see also Strøm 1997, 173, endnote 22).13 Porpora's work on structures is, once again, helpful here, allowing us to differentiate between, on the one hand, institutional structures as (inter)subjective and cultural social rules and, on the other hand, relational structures as objective and material systems of human relations among social positions (1989,(198)(199)(200); see also Bell, 2011Bell, , 2012. These webs of social relations, such as those concerning class, gender, race, the family and, indeed, those between mp s and voters cannot be left out of any analysis of parliamentary roles because "although rules and relationships go together, they are different" (Porpora, 1989, 206). Relational structures are also real and have their 39 own emergent causal powers that cannot be reduced to institutional elements or, indeed, any other aspect of society.
Thus, albeit in different ways, both institutional and relational structures "enable, constrain and, crucially, compel and promote particular behaviour, choices and role expectations" (Luke & Bates, 2015, 334-342). They both, therefore, should be seen as part of the ensemble of factors that contribute to change, whether endogenous or exogenous. Change should not be viewed solely in agential terms but, of course, neither should institutions (and relational structures) be seen as totally determining. As stated by Thelen and Steinmo: [P]olitical actors [are] objects and… agents of history. The institutions that are at the center of historical institutional analyses… can shape and constrain political strategies in important ways, but they are themselves also the outcome (conscious or unintended) of deliberate political strategies, of political conflict, and of choice (1992, 10).
For the reasons outlined above, it is homo sentiens with their moral character and irreducibly human causal powers, rather than homo economicus, homo sociologicus or a hybrid of the two, who allows for the political actor to be this agent of history, not least in shaping (the repertoire of) parliamentary roles.

Studying Parliamentary Roles
The historical institutionalism and the restructuring of parliamentary roles along critical realist lines as outlined above have ramifications for how we identify and study (repertoires of) parliamentary roles and, in particular, the place and prominence within research of mp s' opinions about their roles. For example, there is now no need to adhere to Searing's rule that "if one wants to know why politicians do something, the most sensible way to begin the investigation is to ask them and listen carefully to what they say" (1991,(1255)(1256). Indeed, the application of Searing's rule in the process of studying parliamentary roles can be seen as problematic for at least three reasons.
First, as already argued, the stance an mp takes and the role they attempt to fulfil may well, at least in any system which provides staffing support, be a collective decision. In this sense, the performance of a role may well be a collective effort that is a reflection, not solely of the mp, but also of the ability (and resources) of their staff. Furthermore, the set of expectations that comprise a role emanates from a range of (groups of) actors involved in parliamentary and political processes, which means there is a danger of producing a partial view by placing too much emphasis on the views of one albeit important group.
Second, by starting with and placing too much emphasis on the views of mp s, there is the danger of subcontracting out the work of the social scientist. While there may well be an issue of explaining mp s' behaviour through (deductively-derived) constructs that exist in the minds of social scientists but not in the minds of politicians (see Searing 1994, 13), there are also potential methodological and theoretical problems in accepting politician's views on their roles as (near) gospel. For example, the problems raised above concerning the conflation of roles and behaviour in Searing's work arguably stems from too much emphasis being placed on mp s' opinions: "Specific behaviors are included as integral parts of the roles because that is how politicians themselves see it" (1991,1254). Such an approach stymies the social scientist's job of separating out the wheat from the chaff when analysing what mp s say they do.
Third, by starting (and perhaps finishing) with an examination of mp s' role expectations and the diversity which is found at this level, there is also a danger of reducing the 'thickness' of the analysis by downplaying the influence of cultural and structural factors in role (trans)formation. While the attraction of starting with mp s' views is evident, as Luke and Bates argue (2015, 345-346), such an ethnomethodologically-flavoured approach often falls into the trap of flattening the analysis into cataloguing the different expectations of different groups. This 'static' indexing then comes at the expense of a 'dynamic' structurally and culturally-situated analysis of concrete interactions between role-incumbents which recognises both that powers are differentially located within society (and within parliaments), and that some groupings are probably better placed to succeed within particular social positions and who are, therefore, more likely to take on particular (more influential) roles.
None of the above is to argue that interviewing mp s has no place in the analysis of parliamentary roles. Rather, it is to argue that: interviewing mp s should not be the starting point for identifying parliamentary roles; the opinions of mp s should be ascribed much less prominence in the explanation of parliamentary roles than hithertofore; and, when opinions are sought, the views of political actors who are not mp s should be included when identifying the set of expectations that comprise a role.
The starting point, although not also the end point, of research on parliamentary roles should be what mp s do. As Strøm notes in his discussion of roles (2012, 88), roles are not directly observable but they can be discerned from the patterned behaviour that mp s display (see also Jenny and Müller 2012). This belief, alongside the wish to avoid the dangers of producing a flat, static holden bates International Journal of Parliamentary Studies 1 (2021) 22-46 focus on roles brings into sharp relief that key scientific endeavour of the interplay of structure and agency.
This then takes us back to the fourth, important claim of Rozenberg and Blomgren (2012, 220). Although the identification and explanation of roles and role repertoires are important and interesting in and of themselves and are a necessary part of the research process, the main contribution of the study of parliamentary roles should be seen in terms of what it helps us to say of other phenomena.
Such a contribution is helped by an approach to the study of roles informed by both critical realism and historical institutionalism, not least because of the prominence of their normative and evaluative bent. Sanders argues that, "If [historical institutionalism] teaches us anything, it is that the place to look for answers to big questions about class, power, war, and reform15 is in institutions, not personalities" (2008,53). A critical realist-informed historical institutionalist analysis of parliamentary roles would not only help us to understand parliamentary roles and, subsequently, how parliaments operate better. It would also help point us towards questions of whether, how, and to what extent incumbency within parliamentary roles for particular groupings of mp s is shaped by institutional and structural factors and what impact this has on outcomes, whether in terms of career trajectories for the (groups of) mp s themselves, or in terms of issues of representation and scrutiny, etc. Answering these questions might then raise questions about the 'goodness' of particular roles and the institutions and structures within which they are embedded (Sanders, 2008, 42) and point us towards the evaluation and, potentially, critique and restructuring of those parliaments and the societies of which they are a part. It is to this 'something of other phenomena' which, I believe, is best served by the approach to the study of parliamentary roles outlined here.
The proof of the pudding is, of course, in the eating and the strengths of this critical realist-informed approach can only be made fully apparent in applied work beyond this article's scope. Hopefully the soundness of the theoretical and methodological foundations of such a project has been demonstrated here through the elucidation of three main points.
First, roles need to be defined as distinct from behaviour in order to avoid both tautological explanations and enfeebled conceptualisations of the structural features of society. Such a need is satisfied by moving away from prominent definitions in the field and, instead, conceptualising parliamentary roles in terms of sets of expectations that impinge on incumbents within the social position of mp.
Second, the study of roles benefits enormously from being placed within a new institutionalist framework. However, such a framework needs to be founded upon a social ontology that avoids synthetic attempts to overcome intractable divides between over-and under-socialised persons and that allows us both to explain the social, reasoning and motivated character of the individual, and to recognise structures and institutions as causally powerful and sources of change and indeterminacy in social life. This foundation is provided by a historical institutionalism premised on the figure of homo sentiens and undergirded by critical realism.
Third, there is a danger in placing too much emphasis on the views of mp s about what their roles are. This is not only because it sidelines the contributions of other groups of political actors to the sets of expectations that comprise a role, but also because it increases the possibility of producing a static, flat cataloguing of different expectations of different groups that fails to recognise sufficiently the situatedness of political actors and the determining role of structures and institutions in the shaping of parliamentary roles. This danger can be avoided by, firstly, identifying parliamentary roles through the exploration of (patterns of) mp s' behaviour and then, secondly, producing a dynamic, evaluative analysis that foregrounds the structural and institutional context of role performance, (re)production and transformation and situates accordingly political actors and their concrete interactions.