Should Deliberative Democrats Eschew Modernist Social Science?

Modernist Approaches vs Normative Concepts

Let us begin by looking at the way empirical research redefines deliberation, ignoring its normative dimensions, to fit it within modernist social science. We are not claiming that all deliberative democrats understand themselves to be modernists or take up modernist commitments explicitly. Rather, we are claiming that demands such as Mutz’s have led deliberative democrats to take on modernist assumptions as they operationalize deliberation. The most prominent varieties of modernism here are structuralism and functionalism. Both varieties encourage deliberative democrats to explain actions in ways that ignore the agents’ reasoning. They explain action instead either by its structural conditions or by its functional outcomes.

Deliberative democrats typically make modernist assumptions when they try to operationalize deliberation. As we have argued, for deliberation to have legitimating force, participants have to employ reasoning in justifying their political decision and responding to competing justifications. This requires us to look at the contingent intentionality embodied in their practices. Deliberative democrats with modernist influences have tried to bypass the need to interpret the contingent intentionality of participants by operationalizing deliberation either in terms of its structure or its functionality. They sometimes recognize that they may not be working with deliberation per se. Instead, they see themselves as engaging in what Mutz calls the ‘middle-range approach’: they bypass the question of what deliberation is and study directly the hypothesized causal links between the necessary requirements for deliberation and the desirable outcomes linked to deliberation (Mutz, 2008: 530).

Structuralists operationalize deliberation by identifying structural conditions that are understood to be necessary for deliberation. They view deliberation as an institutional structuring of talk (Warren, 2007: 273). This view allows the deliberative democrat to avoid the question of what kind of talk counts as deliberation. It also allows them to abstract from individual behaviour during the discussion. Game theorists have broadened their definition to any communication that happens with those structural requirements in place (Landa and Meirowitz, 2009: 429; Austen-Smith and Feddersen, 2006; Iaryczower et al., 2018). Whereas early deliberative democrats argued that certain structural conditions are prerequisites of deliberation, structuralists take the presence of these structural conditions as an indication that deliberation has taken place.

Functionalists operationalize deliberation by identifying the functions deliberation is supposed to achieve. Deliberative systems scholars are particularly attracted to this kind of functionalism as a way of defining a system and identifying which practices count as being within a system (Bevir and Chan, 2021). Functionalism allows them to claim that a system can count as deliberative, even if no actual deliberation occurs, as long as it fulfils deliberative functions (Owen and Smith, 2015: 219; Bächtiger and Parkinson, 2019: 107). Functionalism also allows them to claim that a practice should be included in the system, even if it is not itself deliberative, as long as it serves deliberative functions. For example, Neblo and White (2018: 448) argue that even if practices are not ‘obviously deliberative on their face’, they still can be part of a deliberative system by virtue of their role in fulfilling the functions of awareness, translatability, receptivity, and flexibility. Whereas early deliberative democrats argued that deliberation can help fulfil these functions, functionalists suggest that all that matters is that these functions obtain.

Both structuralists and functionalists operationalize deliberation in ways that leave inadequate space for intentionality and so, by implication, for deliberation itself. Structuralists operationalize deliberation by identifying its structural conditions. As they thereby suggest conditions largely determine practice, so they elide creative reasoning and agency. We would suggest, in contrast, that no matter how deliberative institutions are structured, the behaviour of participants – whether and how they deliberate – depends on their choices and reasoning. Functionalists operationalize deliberation by reference to its outcomes. They thereby elide questions about the intentionality of the behaviour that leads to these outcomes in any given case. They play down the importance of reasoning and agency as constitutive of deliberative outcomes. We would suggest, in contrast, that the outcomes of a deliberation depend not only on the fact that a deliberation took place but also on the choices and reasoning of the participants in the deliberation.

We suspect that empirical researchers try to operationalize deliberation because they are seeking modernist explanations. They want their explanations to be formal, that is, to avoid the contingency associated with the recognition of agency. Certainly, much of the empirical research offers causal claims that once again ignore reasoning and agency. Here, the structuralist approach leads empirical researchers to specify independent and dependent variables and their logical connections. Typically, researchers imply that the outcomes of deliberation are determined by how the way the conditions of deliberation are shaped, thereby bypassing the intentionality of participants. For instance, many studies have attempted to show that by making the participants more diverse, deliberation produces better epistemic outcomes (Goodin and Spiekermann, 2018: 140; Landemore, 2020). It is assumed that by increasing the diversity of participants, participants are required to justify their beliefs to those who do not share those beliefs and, hence, are required to improve their reasoning in their effort. The outcome thus appears to be a result of the institutional design rather than the reasoning and agency of the citizens. It is true that functionalists often do not want to be trapped by these institutional variables (Neblo, 2015: 25); instead, they first stipulate what the deliberative functions are and then look around to see what practices can realize these functions. Nonetheless, functionalists do not significantly differ from their structuralist counterparts. They attribute the successful realization of a function to the configuration of practice and the relationship between different practices rather than to the contingent intentionality of the practitioner. For instance, Almeida and Cunha (2016) argue that connectivity between different discursive spheres is a functional need of a deliberative system, and they then proceed to argue that this functional need is met by properly configured institutional designs and the circulation of the participants. Functionalists too ignore questions about intentionality and reasoning, including the ways in which the participants understand the practices.

More generally, when modernists study deliberation, they more or less have to separate it from reasoning and agency. The problem they then face is, of course, that, as we have argued, the very idea that democratic deliberation has legitimating force depends on our defining deliberation, in contrast to other forms of human communication, as being all about participants’ reasoning. The empirical turn cannot respond to the challenges facing deliberative theory if it is not about deliberation in the first place. Structuralists claim to be studying deliberation when they study discourse with structural conditions allegedly favourable for deliberation. This claim is problematic because individuals can exercise their agency and choose to reason under unfavourable conditions, or conversely, choose not to reason under favourable conditions. Functionalists claim to be studying deliberation when they study discourse with functional outcomes associated with deliberation. This claim is problematic because individuals can exercise their agency and choose to reason or respond to reasons in ways that defy functionalist expectations in a deliberation.

Modernists cannot easily respond to this problem by claiming that deliberation is not about reasoning, especially if their empirical research seeks to address challenges against deliberative claims of legitimacy. After all, reasoning plays a central role in both intrinsic and instrumental theories of deliberative legitimacy. For intrinsic theorists, political legitimacy arises from the fact that participants justify their political action with reasons that others can accept. Empirical research arises to answer questions such as how the mass can be motivated to engage in such a process. Modernist research misses the point when it merely shows that the mass can be motivated to participate in a discursive process without showing that the participants reason in that process. For instrumental theorists, political legitimacy arises from the fact that the reasoning of participants gives others the reason to presume that the resulting outcomes are reasonable. Empirical research arises to show that this is indeed the case. Modernist research misses the point when it merely shows that the resulting outcomes from a discursive process have been epistemologically enhanced without showing that the participants reason in that process.

Modernists might reply that because empirical scholars are undertaking research agendas significantly different from those of normative theorists, they should be free to operationalize deliberation differently and employ different approaches. As Neblo (2015: 33) explains with the analogy of courage, while it is reasonable to have a normative definition of courageous behaviour, it would be a mistake to limit empirical research to only courageous behaviour, because that may risk censoring important causes and consequences that distinguish courageous and cowardly behaviour. Can a similar claim not be made about deliberation? While we may want to have a normative definition of deliberative behaviour, would it be useful to expand our concept of deliberation so that we can investigate the causes and consequences associated with different communicative behaviour?

While we recognize that some research agendas may compel researchers to look at more encompassing forms of communication, we do not see why we have to extend the label of deliberation to all of them. Researchers may want to study the relationship between strategic bargaining and deliberating, but it is not clear what is gained by labelling both as deliberating. To use Neblo’s analogy, it may be interesting to see why some soldiers acted courageously while other soldiers in the same platoon acted cowardly, but that does not require us to stretch the concept of courageous behaviour to include both as courageous. The systemic turn is instructive here: we do not need to understand every part of the system as deliberative for us to study the relationships between these different parts. For researchers who do not think that deliberation needs to be legitimated, they may conceptualize deliberation radically differently. If so, it should also be obvious that these researchers mean by deliberation something different from what deliberative democrats usually understand as deliberation. To the extent that researchers seek to either indirectly or directly prove the legitimating power of democratic deliberation, they cannot undertake a modernist approach without undermining their enterprise.

Normative Concepts vs Modernist Approaches

We have argued that when researchers employ modernist approaches, their works tend to suppress or ignore the intentional, and so normative, dimensions that distinguish deliberative theories of democracy. Conversely, when researchers want to retain or recover these dimensions, they have to abandon modernist approaches in favour of interpretive ones. Although few deliberative democrats recognize that they face such a dilemma, many of them have, over the past two decades, come to realize that they are struggling to differentiate deliberation from other discursive practices in their empirical research. It might seem that they could respond to the dilemma in one of two ways: either they could tweak their methods while clinging to a modernist approach or they could adopt an alternative interpretive approach. In practice, as we shall show in this section, empirical researchers often confuse the issue. First, some researchers claim to rely on a modernist approach while implicitly shifting towards an interpretive one. Second, other researchers claim to shift towards an interpretive approach while implicitly continuing to rely on modernism. We will discuss each side of this confusion to show that they can fully resolve the tension between normative concepts and empirical research only by adopting an interpretive approach to social science. Only by so doing can they properly respect deliberative ideas of reasoning and so intentionality.

Let us begin by looking at empirical researchers who claim to rely on a modernist approach while implicitly shifting towards an interpretive one. This implicit shift towards interpretive social science appears in the history of attempts to distinguish deliberation from other discursive practices. Deliberative democrats often wonder how they can determine whether participants are engaged in sincere attempts to persuade or merely in strategic and rhetorical communication (Hanrieder, 2011: 399–404; Jörke, 2013: 356; Saretzki, 2009: 168–172). At first, empirical researchers responded to this problem by explicitly endorsing a structuralist approach. They tried to bypass the beliefs and intentions of participants and ‘focus on the social and institutional context in which arguing takes place’ (Risse, 2004: 302). Over time, however, they have in practice sometimes turned to different interpretive approaches.

Consider, for example, a study by Risse and Kleine (2010) contrasting the deliberation in the European Convention with the bargaining stalemate in the intergovernmental conferences (IGCs) that preceded the convention. Risse and Kleine argue for a structuralist approach to operationalizing deliberation. They suggest that because it is impossible to ascertain actors’ true motivations, empirical researchers should concentrate on the institutional scope conditions conducive to deliberation (Risse and Kleine, 2010: 711). They use this structuralist approach to establish their causal claims, arguing that the institutional differences between the two sites enabled deliberation to happen at the European Convention but not at the IGCs. Both their operationalization and their causal claims thus appear to bypass individual intentionality and agency.

Despite their modernist proclamations, however, Risse and Kleine turn to an interpretive approach when they try to separate deliberation from other discursive practices. Crucially, they recognize that we need to have confidence that the processes we are studying entail elements of deliberation (2010: 712). Consequently, they offer a set of indicators for these elements. One of their indicators is that actors offer the same reasons for their agreement since that would suggest some actors have persuaded others. Finally, these indicators prompt them to understand the reasoning and intentionality of actors by interpreting the actors’ speeches and decisions.

Risse and Kleine (2010: 714) also oscillate between modernism and interpretivism when they establish their causal claims. For example, they explicitly claim that leadership becomes more conducive to deliberation when institutional norms and procedures require neutral chairs of negotiations. This claim focuses on structure, bypassing individual agency. However, in their discussion, they track the contingent decisions made by the chair of the European Convention working group, showing how he facilitated the deliberative process by framing the question as legal rather than political (Risse and Kleine, 2010: 718). A reasonable reading of the discussion shows that the results of the European Convention at least in part depend on the contingent decisions made by the chair and not simply on the institutional structure.

We would suggest that Risse and Kleine drift towards interpretivism in practice because only by doing so can they retain the normative dimensions of deliberation. It is true that Risse and Kleine (2010: 712) explain their oscillation by claiming that they have to use these temporary measures only because our understanding of institutional scope conditions remains unsatisfactory. Surely, however, this claim is rather misleading. The problem lies not with the state of our knowledge but rather with the modernist approach to which they cling. As long as researchers try to operationalize deliberation through institutions, they have to suppress the agency of individuals crucial to our understanding of deliberation. As Risse and Kleine try to recover that agency, they inevitably drift away from modernism and towards an interpretive social science.

Let us turn now to researchers who claim to shift towards an interpretive approach while implicitly continuing to rely on modernism. This implicit attachment to modernism appears in the history of attempts to measure deliberation. At first, empirical researchers typically drew on a functionalist approach, measuring the quality of deliberation through its functional outcomes. Over time, however, some researchers became concerned that functionalist attributes do not travel, so a functionalist approach falsely homogenizes deliberation across different contexts. Unfortunately, although they proposed turning to interpretive approaches, measuring the quality of deliberation through the understanding of the participants, they have often struggled to break with modernism.

Consider, for example, the development of the Deliberative Quality Index (DQI). The authors of this index make sincere efforts to avoid a modernist approach (Bächtiger et al., 2007; Bächtiger and Steiner, 2005; Steiner et al., 2004). The deliberative quality of each speech is evaluated according to the following coding categories: participation, level of justification, content of justifications, respect and the constructiveness of politics. Steiner and his colleagues acknowledge that ‘[b]oth the qualitative and our approach require extensive contextualized interpretation’. They even stress that the DQI cannot be used mechanically, because ‘assessing the quality of discourse requires interpretation’ and ‘one needs to know the culture of the political institution, the context of the debate, and the nature of the issue under debate, to get a true understanding of how actors in the institution use and interpret language’ (Steiner et al., 2004: 60).

Despite their interpretive commitments, when they try to scale up the index, they often relapse into modernism. For instance, when Steiner et al. (2004: 56) attempt to measure the quality of justifications, a speech gets the lowest possible value merely for mentioning ‘one or more groups or constituencies’, which is a rather blunt and indirect way of assessing justificatory quality. Bächtiger et al. (2010) acknowledge the limitations imposed by the formal coding process that is used to produce the DQI. Formally coding the properties of ideal deliberation poses the risk of creating a rigid framework that is insensitive to some of the relevant phenomena, as well as exacerbating the concerns of critics who contend that deliberation excludes forms of speech that fail to fit a relatively constrained model. The formal properties of discourse are at best an approximation to the actual phenomena under scrutiny. The authors note that many of the important subtleties of actual discourse are lost, and that the quality of arguments often depends heavily on a context that is stripped away by formal coding. These problems, among others, have led Bächtiger and Parkinson (2019: 138) to declare that ‘the time of searching for a grand, unified index of deliberative quality are [is] over’.

Employing functional outcomes to measure the quality of deliberation is problematic because these measurements ignore the intentionality of participants. Because participants can reason and respond to reasons in creative ways, the quality of deliberation is not necessarily tied to some predestined functional outcomes. The DQI avoids this problem by using coding categories that prompt the researcher to study the intentionality of participants and the context of the deliberation instead of focusing on outcomes. However, researchers who employ the DQI succumb to modernist influences when they fail to realize that these categories are merely heuristics. For example, what counts as ‘respect’, one of the coding categories, can differ drastically between different cultural contexts.

In sum, the empirical research to date shows that the conceptual tension between deliberative democracy and modernist approaches is manifested in the evolution of the methods used to study deliberation. Faced with the difficulties of operationalizing deliberation, scholars have sometimes searched for strategies within modernist approaches and at other times drifted towards interpretive approaches. We have argued that strategies within modernist approaches cannot lead to progress in studying the deliberative process itself. Gaining empirical traction on the deliberative process while maintaining the normative substance of deliberative theory requires turning to interpretive approaches to uncover the relevant phenomena.

The Interpretive Alternative

We have illustrated the tensions between modernist approaches and deliberative ideals. When empirical researchers lean towards modernist approaches, they undercut deliberative ideals. When they try to retain those ideals, they drift from modernism to an interpretive social science. Recent years have thus seen considerable growth in the number of interpretive studies of deliberation (Ercan et al., 2017). Although we cannot discuss all these interpretive studies, we want to emphasize the importance of treating interpretivism less as a matter of qualitative methods than as an approach that focuses on beliefs, reasoning and agency.

Sometimes an interpretive approach is equated with qualitative methods in contrast to a modernism or positivism that is equated with quantitative methods (Curato et al., 2017: 34). So, for example, Bächtiger and Parkinson (2019: 140) recognize that ‘future research on deliberation and deliberativeness must grapple with the meaning of deliberative and other communicative acts in various venues of democratic systems’, but they then suggest research should not restrict itself to the qualitative methods they seem to equate with an interpretive approach. As this example suggests, the conflation of an interpretive approach with qualitative methods can make researchers reluctant to embrace it. We therefore want to emphasize that this conflation is mistaken (Bevir and Blakely, 2018: 88–89). Approaches and methods are different. Approaches embed philosophical assumptions about human reason, action and practice, and so about the nature of valid social explanations. Methods are instrumental tools for generating data and finding patterns among data. Because different approaches favour different explanations, they can lead researchers to privilege different kinds of data and so methods. Nonetheless, there is no reason why proponents of any approach should reject the data generated by any method. They might question its relevance. They might question the way in which proponents of another approach conceive of it. But they have no reason to reject it.

Although there are different varieties of interpretivism, they generally build on philosophical assumptions about the intentionality of reasoning and the holistic nature of reasons and beliefs. Habermas (1990: 26) argues, for example, that when a social scientist tries to understand people, he has to ‘grapple with the problem of the context dependency of his interpretation’ and understand first the ‘background assumptions and practices’ of the research subjects. The social scientist explains people’s beliefs, and so actions, in large part by relating them to their background assumptions, or, in other words, to their whole web of beliefs. Although interpretivists believe that people are agents capable of reasoning, they need not thereby deny any role to social background. On the contrary, many interpretivists are committed to some form of situated agency, seeing people as capable of modifying or creatively altering beliefs and actions that they acquire in large part from their social background.

These philosophical assumptions commit interpretivists to certain forms of explanation. Because people can respond to their social backgrounds in creative ways, the interpretivist cannot explain behaviour solely by reference to those backgrounds. Instead, the interpretivist has to relate people’s behaviour to their reasons and webs of belief, and only then relate those reasons and webs of belief to the social background. Clearly, this mode of explanation fits neatly with deliberative ideals that emphasize the giving and responding to reasons. Interpretivism just does favour a focus on people’s intentionality – their reasons and reasoning. What is more, there is no need for interpretivists to take a particular view of what counts as a reason or reasoning. They are interested in the ways in which people reason, not in imposing a priori categories of correct reasoning on people. Interpretivists thus agree with Bächtiger and Parkinson (2019: 23–24) that reasons need not take the traditional form of ‘because’ statements but instead can be expressed in many different ways, including testimonies and storytelling.

Having distinguished the interpretive approach from qualitative methods, we can reconsider the recent growth in interpretive studies of deliberation. On one hand, deliberative democrats do not adopt interpretivism simply because they use qualitative methods. We have to look instead to their philosophical assumptions as these appear in their forms of explanation. Consider, for instance, Curato’s (2019) ethnographic work on the relationship between populism and deliberation in the Philippines. Curato relies on an interpretive approach for much of her study. She studies the speeches and conversations of the supporters of Duterte to show that they engage in deliberative practices. She shows both that they have good reasons for their choice and that they can be critical of some of Duterte’s pronouncements. And she thus argues persuasively that populism is not simply the antithesis of deliberation. Curato relies on an interpretive approach here not because she uses ethnographic methods, but because she focuses on the people’s intentionality. Indeed, despite her use of ethnographic methods, Curato seems clearly to depart from an interpretive approach when she tries to generalize her conclusions. She concludes that populism should be understood in terms of its functions and dysfunctions for deliberative democracy. She argues that populism broadens the range of discourse, creates room for discursive pluralism, but fails to prompt reflection (Curato, 2019: 141). Surely, however, this functionalist understanding of populism is at odds with her contextualized approach to Duterte supporters. If Duterte supporters can defy conventional expectations when they exercise their situated agency, surely other populists might act in other creative ways that might defy Curato’s functionalist expectations. As Curato generalizes her conclusions, so she takes people’s intentionality out of her explanation thereby falling into modernist tropes.

On the other hand, deliberative democrats can adopt interpretivism without limiting themselves to qualitative methods. As long as they attend to their philosophical assumptions and forms of explanation, quantitative and computational methods can be part of their research. Consider, for instance, a study of the deliberations surrounding the Scottish Independence Referendum by Parkinson et al. (2020). These authors pay careful attention to people’s intentionality as they operationalize their research. In developing their methods of data analysis, they explicitly avoid deductively categorizing texts in advance and instead inductively cluster texts together as bags of words. They also recognize that these clusters can still obscure the contexts of the deliberation, and advise researchers to drill down to the original source data to recover the contextual differences. Finally, they recognize that in running structural topic modelling, they have to, as researchers, make a judgement about the number of topics in advance. They validate their model by checking these topics with the issues that arose in the preliminary interviews and workshops they have done (Parkinson et al., 2020: 8–9). In every instance, they ensure that they operationalize their research in ways that do not bypass people’s intentionality and context. Their research yields many fruitful conclusions. For example, because their data shows that the single largest set of topics in the discussions surrounding the referendum concerns the quality of the debate itself, they can conclude they have here ‘direct evidence of citizens holding each other to deliberative standards of respect, evidence and argument in mass democracy’ (Parkinson et al., 2020: 19). The success of this research shows how an interpretive approach might use quantitative methods.