Gendered Institutions and Where to Find Them: A Critical Realist Approach

Abstract

The gender and politics literature offers diverse views on the causes of gendered practices and the best methodologies for studying them. This article advances efforts to take stock of and systematize this diversity by grounding the feminist institutionalist perspective in critical realism. The article posits that gendered institutions are real entities with independent powers, while also emphasizing the crucial role that human ideas play in upholding and contesting gendered practices. To faithfully capture gendered institutions and their relationship with human agency, the article promotes the use of the abductive-retroductive research design. This approach allows feminist institutionalist scholars to construct and test multiple competing theories about gendered institutions, drawing from various empirical manifestations of institutional power. These expressions range from observable actions to codified rules, socially shared norms, and other subtle discourses. By shedding light on the principles at the heart of realist-oriented feminist research, this work paves the way for a more standardized and transparent approach to feminist inquiries.

Introduction

As the gender and politics literature grows in volume, there is a need to systematize and standardize the metaphysical and methodological diversity within it (Bacchi and Rönnblom 2014; Childs and Krook 2006; Schmidt 2008; Tripp and Hughes 2018). Such an endeavor will help facilitate a more efficient exchange of views and make the literature more accessible to external review. Existing studies have taken stock of the gender and politics literature and identified a shift toward a greater analytical and methodological pluralism (Childs and Krook 2006; Krook and Squires 2006; Stauffer and O’Brien 2018, 2019; Tripp and Hughes 2018). This article joins the ongoing audit of the field by providing a critical realist foundation for feminist institutionalism, a popular perspective among feminist scholars.

Feminist institutionalist (FI) approaches have gained recognition for their capacity to shed light on how institutionalized prescriptions—or institutions—about identity and behavior shape human actions (Bacchi and Rönnblom 2014; Bjarnegård 2013; Driscoll and Krook 2012; Kenny 2014; Kulawik 2009; Lowndes 2020; Mackay, Kenny, and Chappell 2010; Waylen 2014). These institutions have a gendering effect on social practices when they prescribe distinct actions for individuals with different gender identities in comparable social situations. However, FI scholars disagree about the nature of gendered institutions and their relationship with human agency. Realist-oriented FI scholars view institutions as real and stable entities with autonomous powers to constrain and enable human agency (Gains and Lowndes 2014; Kenny 2014; Lowndes 2020; Waylen 2014). Their constructivist counterparts perceive institutions as ideational constructs and highlight the role that these constructs play in shaping gendered practices (Bacchi and Rönnblom 2014; Kulawik 2009; Schmidt 2008).

This metaphysical diversity is FI’s strength, allowing FI scholars to generate rich and varied knowledge about the roots of gendered practices. However, such diversity also poses challenges. One significant challenge is the risk of blending contradictory understandings of gendered institutions within the same research, a criticism that some contemporary FI studies have faced (Bogaards 2022). To bolster the clarity and internal consistency of FI research, we need to expose the metaphysical views within the perspective and propose empirical strategies consistent with these views (Krook and Squires 2006). Within the constructivist strand of FI, such efforts are already well underway (Bacchi and Rönnblom 2014; Kulawik 2009; Miller 2021b; Schmidt 2008). This article further enriches the discourse by grounding FI in critical realism (CR) (Archer 2003; Bhaskar 2008; Decoteau 2017; Fletcher 2017). CR enables feminist scholars to devise inquiries that incorporate the realness of institutions and the role of human ideas in resisting and transforming these entities (Archer 2003; Archer and Elder-Vass 2012; Bell 2011, 2012; Lowndes 2020).

CR is a theory of science that integrates realist ontology with epistemic relativism, positioning itself as a middle ground between positivism and interpretivism (Bhaskar 2008; Danermark, Ekström, and Karlsson 2019; Fletcher 2017; Hoddy 2019). CR holds significant value because of its clearly defined methodological guidelines that align its metaphysical standpoints with empirical research (Zachariadis, Scott, and Barrett 2013). Within the CR framework, institutions are conceptualized as autonomous entities that exist beyond human consciousness (Fleetwood 2008). These entities exercise their powers over human agency by enabling certain behaviors and constraining others. Yet humans are not merely passive vessels for institutional powers (Archer 2003). On the contrary, the power of institutions relies on human agency—our capacity to think and act—for activization and realization. This dynamic relationship means that humans can resist and redefine institutional powers (Rees and Gatenby 2014; Sharpe 2018). A key strength of CR is its acknowledgment of both structure and agency as having independent ontological statuses, thereby avoiding the analytical pitfall of subsuming one into the other (Bertilsson 2004; Fletcher 2017, 66).

However, a challenge arises when studying institutions. Institutions are not material objects that can be directly accessed or observed (Danermark, Ekström, and Karlsson 2019, 19; Fletcher 2017, 182). Furthermore, institutional powers are partially local, inherently complex, and, because of their dependence on human interpretation and action, ever changing. Owing to the complex nature of institutional powers, multiple explanations are often consistent with their observable manifestations. In order to produce valid theoretical explanations of gendered practices, feminist scholars should ideally construct and test multiple theories about gendered institutions and consult multiple expressions of institutional powers, ranging from observable actions to subtle discourses (Fletcher 2017, 189). The adoption of an abductive-retroductive research design enables feminist scholars to do just that. Through repeated rounds of empirical testing and theory refinement, feminist scholars are able to produce valid and empirically corroborated explanations of studied gendered practices and offer recommendations for how to de-gender them (Kenny 2014).

By thoroughly discussing the nature of real gendered institutions and the relationship with human agency, this article offers an internally coherent ontological justification for studying institutions and ideas about institutions as separate yet equally essential components of gendered practices. It also provides an ontological argument for conceptually distinguishing between institutions as real and elusive entities and their empirical manifestations, including actions, rituals, norms, rules, laws, conventions, and so on. These discussions highlight that understanding institutions necessitates moving away from pure deduction as a way of evaluating the validity of institutional truth claims in favor of the more complex abductive-retroductive research design, combined with a multimethod approach to data generation.

For those unfamiliar with the field, these methodologies—which already enjoy popularity among feminist scholars—might appear to lack parsimony, internal consistency, or even analytical rigor (Bogaards 2022). The arguments presented in this article demonstrate that these methodological approaches are necessary for generating valid and empirically corroborated truth claims about entities that are elusive, contingent, and changing. The article also provides conceptual tools for feminist scholars who share its metaphysical principles, enabling them to organize and present their empirical inquiries in a coherent and transparent manner. This may facilitate further development of realist-oriented feminist research designs and their integration into mainstream political science. The significance of this article goes beyond the FI perspective alone. As the gender and politics literature gains prominence in political science, reflecting on the nature of the abstract entities at the core of our investigations becomes critical. Engaging in such reflections can enhance the precision, transparency, and internal validity of existing methodological approaches.

What Are Gendered Institutions? A Tapestry of Views

The FI perspective enjoys popularity among students of gender and politics for its ability to illuminate the effects of gendered institutions on social practices (Bacchi and Rönnblom 2014; Driscoll and Krook 2012; Gains and Lowndes 2014; Kenny 2014; Kulawik 2009; Lowndes 2020; Mackay, Monro, and Waylen 2009; Schmidt 2008). Drawing upon earlier strands of new institutionalism (NI), FI sees institutions as socially devised prescriptions that constrain and enable human agency (Annesley, Beckwith, and Franceschet 2019; Chappell and Waylen 2013; Kenny 2014; Waylen 2014). These prescriptions have varying degrees of applicability, ranging from context specific to general (Kenny 2014). Institutions are gendering if they ascribe different strategies to individuals with different gender identities in equivalent social situations (Bjarnegård 2018; Gains and Lowndes 2014; Lowndes 2020). We learn about institutions by observing how others act in various situations and/or becoming familiar with discourses/ideas about proper (and improper) behavior (Peters 2019, 40). There is a general consensus among FI scholars about the significance of human discourses about institutions in shaping and gendering human practices (Bjarnegård 2013; Kenny 2013, 2014; Mackay, Kenny, and Chappell 2010; Mackay, Monro, and Waylen 2009). These discourses can take various forms, ranging from formally codified directives to socially shared norms and customs. However, FI scholars disagree on whether human ideas about institutions are institutions (Bacchi and Rönnblom 2014; Kulawik 2009; Schmidt 2008) or whether they are traces of more tangible entities that are external to our minds (Bell 2011, 2012).

Discussions of ontology are relatively rare within FI scholarship (Bacchi and Rönnblom 2014; Kulawik 2009). While the terms like “institutions,” “gendered prescriptions,” or “institutional change” feature extensively in the literature, their meanings are ambiguous (Kenny 2014; Waylen 2017a, 2). What are (gendered) institutions? Are they external to us and knowable? And how should we understand the relationship between institutions and human agency and the nature of institutional change? Taking a position on questions that pertain to the nature of gendered institutions is a prerequisite for our ability to design feminist inquiries that are philosophically coherent and empirically plausible (Archer and Elder-Vass 2012, 94; Stanley 2012). On this front, Fleetwood (2008, 243) notes that

stating the social ontology that forms the basis of [the] analysis … allows the reader to identify fundamental points of agreement or disagreement, without having to guess [the author’s] position; obviates the need to repeat arguments that, if not widely known, are readily available; and highlights those issues and arguments that, whilst important, are beyond the scope of [the] article.

FI draws upon both the ontologically realist strands of NI, including rational choice institutionalism (RI), sociological institutionalism (SI), and historical institutionalism (HI) (Hall and Taylor 1996; cf. Hay and Wincott 1998), as well as the ontologically constructivist discursive institutionalism (DI) (Bacchi and Rönnblom 2014; Schmidt 2008). Consequently, these two broad ontological perspectives have been integrated into the FI perspective.

Most realist-oriented FI scholars perceive gendered institutions as autonomous entities, endowed with the power to constrain and enable human agency (Bjarnegård 2013; Chappell 2006; Driscoll and Krook 2012; Gains and Lowndes 2014; Kenny 2014; Mackay, Monro, and Waylen 2009; Waylen 2014). However, there are disagreements among realist FI scholars about the origins and knowability of these constraints. In line with the core principles of RI, some scholars view institutions as socially devised tools—encompassing both material and discursive elements—that social actors put in place to safeguard their interests (Annesley, Beckwith, and Franceschet 2019, 27; Bjarnegård 2018, 8; Lowndes 2020; Zetterberg 2008, 443). This view assumes a certain degree of intention behind the creation of institutions, as well as social actors’ awareness of their existence (Bjarnegård 2018; Lowndes 2020). Lowndes (2020, 548), for instance, argues that “actors in a political setting should be able to recognize shared rules (formal and informal), even if they disagree with them or (at times) ignore or adapt them.”

Assuming that institutions are knowable has clear implications for the design of scholarly inquiry (Lowndes 2020). As a result, RI-oriented FI scholars will often combine the analysis of written and spoken accounts of institutions with observations of agential actions in their efforts to construct theoretical models of institutions (Bjarnegård 2013; Gains and Lowndes 2014), some even opting for a deductive research design (Zetterberg 2008). Other realist FI scholars, drawing upon the tenets of HI and SI, emphasize the subtle and often subconscious nature of real institutional powers (Mackay, Kenny, and Chappell 2010, 583; Waylen 2014, 213). This view sees institutions as more organic and less intentional products of myriad past human actions that, through the processes of routinization and institutionalization, become real prescriptions about proper conduct (Hall and Taylor 1996). These prescriptions sit deep “at the back of our minds” and guide our actions even if we are unaware of them (Chappell 2006; Kenny 2014, 679). This is because institutions create logics of appropriateness and historical path dependencies that shape human action in almost an automatic fashion. Chappell and Waylen (2013, 600) argue that institutions are the “unquestioned ways of operating seen as natural and immutable, if participants are even aware of them.” The process of learning about institutions that, despite being real, are not entirely knowable is more complicated, often involving complex multimethod research designs (Childs and Krook 2006, 23; Kenny 2013, 51; Kenny 2014).

Various forms of constructivist or anti-foundationalist ontology can also be identified within the FI literature, owing to influences from DI (Bacchi and Rönnblom 2014; Kulawik 2009; Schmidt 2008). According to this perspective, institutions are sedimented ideas about appropriate being and behaving that are shared among social actors (Freidenvall and Krook 2011; Kulawik 2009; Miller 2021a; Schmidt 2008). Schmidt (2008, 314), for instance, rejects the view that institutions are “external to actors,” as well as “the rule-following logic [of the three older new institutionalisms], whether an interest-based logic of calculation, a norm-based logic of appropriateness, or a history-based logic of path dependence.” The heart of Schmidt’s critique is the perceived failure of RI, SI, and HI to account for human agency and institutional change (Bell 2012). If institutions are solid frameworks of reference that constrain and enable human agency in a regular and predictable way, then why do dictatorships fall, laws change, and customs disappear? The answer, according to Schmidt, lies in social actors (agency) who think about, interpret, enforce, resist, and challenge different ideas within institutions (Erikson 2017). Institutions, in this view, are contingent on, and thus internal to, actors’ thoughts, words, and actions. There is seldom consensus about the meanings behind institutions, and shifting coalitions of actors can empower some meanings at the expense of others. If institutions are shared meanings, then the process of learning about them involves identifying these meanings. Constructivist scholars strive to map out the dominant meanings that constitute institutions and study the power struggles over meaning (Erikson 2017; Kantola and Lombardo 2017). Poststructuralist scholars, conversely, critique constructivist attempts to delineate and thus risk solidifying and perpetuating the dominant institutional meanings (Miller 2021b). Instead, poststructuralist FI scholars focus on mapping out dissenting minority voices overshadowed by dominant institutional narratives within a specific context (Bacchi and Rönnblom 2014; Miller 2021a).

A common approach taken by realist feminist scholars in response to the missing agency critique is to incorporate human agency into their analytical frameworks, often by drawing on various strands of NI (Annesley, Beckwith, and Franceschet 2019; Kenny 2014; Lowndes 2020; Waylen 2014). In their cross-country study of cabinet appointments, for instance, Annesley, Beckwith, and Franceschet (2019, 26–27) cite all strands of NI and claim that they “follow a ‘border-crossing’ approach that draws from all variants of new institutionalism that we believe are useful for explaining how the process of cabinet appointment [is gendered].” Furthermore, Kenny (2013, 51–58) combines HI and DI in her study of candidate selection. However, without an explicit ontological justification for this practice, FI scholars risk introducing metaphysical inconsistencies into their inquiries. In a review of eight new books in the FI literature, Bogaards (2022) identifies multiple such inconsistencies. Assessing a recent book by Freidenvall (2021), Bogaards (2022, 422) notes that

Although Freidenvall aims to make a contribution to feminist discursive institutionalism, other variants of new institutionalism are never far away. Historical institutionalism is frequently mentioned, with the incremental process of ideational and institutional change described as “path-dependent, leading to a ‘politics of no return’”… . Sociological institutionalism is present implicitly in the chapter with case studies of how local parties selected their candidates and how the parties’ formal rules were implemented on the ground.

Schmidt (2008, 314) herself argues that “DI can be seen as complementary to the other three institutionalisms.” The issue with taking this position is that one might, unwittingly, simultaneously subscribe to the view that institutions are real entities, external to social actors, and to the view that institutions are sedimented meanings, internal to social actors. By doing so, one imports two mutually exclusive ontological perspectives into one’s inquiry. Fortunately, there is no need for such inconsistencies, because there are viable ontological solutions. The ontological perspective of critical realism allows scholars to reconcile the views that institutions are real constraints, independent of human consciousness, and that they are dependent on human agency for reproduction (Archer 2020; Bhaskar 2008; Fleetwood 2008; Fletcher 2017). It also provides scholars with concrete ontological tools to make sense of institutional change without having to “sacrifice” their commitment to the “realness” and relative stability of institutions. What is more, the perspective offers concrete methodological tools for those who believe that agential knowledge of institutions is always partial (Chappell 2006; Chappell and Waylen 2013; Kenny 2014; Waylen 2014) and yet wish to generate valid and empirically corroborated knowledge about these entities.

The Critical Realist Ontology of Gendered Institutions and Human Agency

Although the FI literature presents various forms of realist ontology, the bulk of these definitions align closely with the ontology of critical realism (Annesley, Beckwith, and Franceschet 2019; Chappell 2006; Chappell and Galea 2017; Kenny 2013; Lowndes 2014; Waylen 2014). CR views institutions as humanly devised prescriptions about proper being and behaving that shape human action in all social situations (Danermark, Ekström, and Karlsson 2019; Decoteau 2017, 71; Hall and Taylor 1996, 939). In every social context, we encounter a multitude of institutional prescriptions, ranging from how to stand and where to look to what to wear and when to speak (Rees and Gatenby 2014; Sharpe 2018, 384). These prescriptions represent established practices that, through socialization and routinization processes, have evolved into real entities with their own causal powers (Bhaskar 2008; Danermark, Ekström, and Karlsson 2019; Decoteau 2017; Hoddy 2019). These powers, to some extent, operate independently of our (accurate) knowledge of institutions (Danermark, Ekström, and Karlsson 2019, 21; Fletcher 2017, 183). When we take action, we rely on our personal knowledge of institutions that we accumulate over time (Bell 2011, 891; Danermark, Ekström, and Karlsson 2019, 19). This knowledge is acquired by observing how others act in different social situations and becoming familiar with human accounts of rules and norms (Bell 2011; Danermark, Ekström, and Karlsson 2019, 37–38).

However, this knowledge is never complete. In fact, we often act based on incomplete or even erroneous notions of what is expected of us in different social situations (Decoteau, 2017, 58–59). This is because institutions exist outside our sensory experience and the powers through which they affect our action are often complex and intertwined. What is more, institutions, though independent of human consciousness, are dependent on human abilities to think and act for the materialization and reproduction of their powers (Bell 2011, 892). Though changing an institutional status quo is not easy, humans have the ability to resist the institutional pressures they are subjected to and, through concerted action, change the ways things are done in their environments (Bell 2012, 715; Peters 2019; Steinmo, Thelen, and Longstreth 1992). A comprehensive model of gendered institutions must therefore find ways of capturing the essence of these elusive entities (Alina-Pisano 2009; Danermark, Ekström, and Karlsson 2019; Fletcher 2017), while accounting for the role of human agency in their reproduction and transformation (Decoteau 2017, 69).

Elusiveness of Institutions

Although institutions and the powers associated with them are real, they are not material beings that can be directly observed (Bhaskar 2008; Danermark, Ekström, and Karlsson 2019; Decoteau 2017; Fletcher 2017). We cannot simply go out and capture images of real institutions. Instead, we can perceive the effects institutions have on the social world (Archer 2003; Danermark, Ekström, and Karlsson 2019). For instance, we might witness someone removing their hat upon entering a place of worship, overhear a discussion condemning theft as immoral behavior, or read a new directive regarding appropriate work attire. These observable expressions of institutions and their powers are like puzzle pieces that our minds transform into personal knowledge about institutions. Over time, as we immerse ourselves within specific contexts, our understanding of how things are done in these contexts becomes more nuanced (Fletcher 2017). However, the concept-laden nature of human cognition limits our ability to fully register and accurately interpret all relevant manifestations of institutional power (Danermark, Ekström, and Karlsson 2019, 37; Fletcher 2017; Hoddy 2019, 113). The words and concepts available to us act as filters through which we interpret our observations and auditory experiences (Decoteau 2017, 69). Consequently, there is always a risk of misinterpreting what we encounter and drawing incorrect conclusions about institutions (Fletcher 2017, 189–90). Furthermore, our structural position and access to various social arenas within a particular context also influence how we perceive and understand institutions (Archer 1995, 195–98; Danermark, Ekström, and Karlsson 2019, 86). New employees, for instance, often require time to comprehend the routines and practices they encounter in their new workplace. In certain organizations, significant events may transpire behind closed doors. Individuals lacking access to these environments are unlikely to have a good knowledge of all the prescriptions governing the organization (Archer 2003; Danermark, Ekström, and Karlsson 2019, 86–87).

Complexity, Partial Idiosyncrasy, and Contingency of Institutional Powers

Another key ontological feature of institutions concerns the contingency, complexity, and partial idiosyncrasy of their causal powers—or mechanisms through which institutions affect human behavior (Bhaskar 2008; Danermark, Ekström, and Karlsson 2019). First, institutional powers are contingent because they are not always activated in seemingly equivalent social situations (Decoteau 2017, 64; Hoddy 2019, 118). While gendered institutions have the power to influence a particular course of action, their effectiveness relies on human agency to actualize that power (Archer 1995, 2020; Danermark, Ekström, and Karlsson 2019, 93; Fleetwood 2008). Humans do not always do what institutions require of them, which creates a contingency in the exercise of institutional power.

Second, institutional powers are complex because of their ability to interact with one another. These interactions might sometimes engender surprising courses of action (Hoddy 2019, 119; Rees and Gatenby 2014, 139). For instance, human behavior is often profoundly shaped by prescriptions that are specific to our group identities, such as being a woman, a young person, or a member of a particular social class. Identity-based prescriptions tend to interact with other prescriptions, engendering identity-based behavioral disparities in contexts where no such disparities are expected (Fleetwood 2008). For instance, norms in the political sphere about what defines a successful politician may clash with party-specific rules aimed at boosting women’s descriptive representation (Bjarnegård 2018; Gains and Lowndes 2014; Kenny 2014; Waylen 2017b). Individuals whose actions are conditioned by the norm may be unaware of its influence, attributing reform failures to a perceived lack of suitable candidates (Bjarnegård 2013).

Finally, institutional powers are partially idiosyncratic because they often contain elements that are specific to a particular social context (Rees and Gatenby 2014; Sharpe 2018). Rees and Gatenby (2014), for instance, demonstrate how different managerial styles engender company-specific patterns of behavior. These features of institutional powers complicates the process of learning about the entities that produce these powers (Danermark, Ekström, and Karlsson 2019; Decoteau 2017). One of the consequences of the elusive and complex nature of institutions and their powers is that multiple plausible institutional explanations are often consistent with their observable manifestations. This is something we need to be aware of when designing inquiries about the roots of gendered practices.

Structure, Agency, and Institutional Change

An integral element of CR ontology is the relationship between structure and agency, each of which is attributed their own ontological reality (Archer 2020; Danermark, Ekström, and Karlsson 2019; Fleetwood 2008). This means that institutions cannot be simply reduced to human discourses and behaviors, and vice versa (Fleetwood 2008, 243). The dynamic between these two fundamental components of social reality is marked by mutual codependence (Archer 2020). On one hand, institutions influence and shape specific courses of action (Rees and Gatenby 2014, 138). On the other hand, institutions rely entirely on the human capacity to interpret societal expectations and subsequently act upon those interpretations (Sharpe 2018, 385). Thus, the materialization and continuity of institutional powers depend on human actions. When we cease to comply with institutional requirements, they gradually lose their influence and significance (Danermark, Ekström, and Karlsson 2019; Decoteau 2017; Fletcher 2017). To conduct a comprehensive institutionalist inquiry, it is essential to consider both structure and agency. Accounting for institutions is vital to understanding practices and behavioral regularities. However, equal attention must be paid to conflicts over ideas and power relations that envelop these situations, as they play a critical role in explaining how and when institutions undergo change (Archer and Elder-Vass 2012; Bell 2011; Chappell and Waylen 2013)

Because institutions rely on human agency for the realization of their powers, they are subject to change (Hoddy 2019). Institutions have the tendency to create unequal playing fields by allocating social actors to different structural positions, some of which are more enabling, while others are more constraining (Archer 1995; Danermark, Ekström, and Karlsson 2019, 89). For instance, consider the employer-employee relationship in a low-skill industry during a period of high unemployment. Under such circumstances, the employer is relatively unrestricted in defining the terms of the workers’ employment. Given the limited options available, workers often have little choice but to accept these conditions. This may entail agreeing to mass layoffs, irregular pay schedules, or even salary decreases in order to protect their job security. The unequal power dynamics between employers and employees in certain situations can leave workers feeling compelled to make difficult decisions, sacrificing certain benefits or rights in exchange for continued employment. However, in certain cases, persistent mistreatment of workers can lead to collective action and collaboration in response to their shared grievances. Humans possess the capacity to respond in novel and surprising ways to their structural conditioning, using this ability to advance their own interests and enhance their structural positioning (Danermark, Ekström, and Karlsson 2019; Rees and Gatenby 2014). This adaptive behavior allows individuals and groups to navigate within institutional constraints and find innovative solutions that can potentially lead to better outcomes for themselves (Archer 2003; Danermark, Ekström, and Karlsson 2019). Human action that defies our structural conditioning might be the beginning of an institutional transformation. In our example, the formation of a labor union can emerge as a potential response, leading to a significant and lasting transformation of the relationship between the owner and the workers.

Institutional transformation often arises from a deliberate battle over meaning in which coalitions of actors successfully alter specific rules or practices, often at the expense of those vested in preserving the old ways of doing (Danermark, Ekström, and Karlsson 2019). However, institutional change can also be triggered by a series of unconscious responses to the structural conditioning experienced by social actors. Because of the interdependent relationship between structure and agency, institutions remain in a perpetual state of flux (Archer 1995; Fleetwood 2008). Consequently, our knowledge of institutions requires regular reassessment (Decoteau 2017, 78).

Formal versus Informal “Institutions”

Though invisible in their true form, institutions leave plenty of footprints in the social world around us (Peters 2019). On one hand, we observe traces of institutional powers in the practices, rituals, or actions that are conditioned by institutions. On the other hand, institutional prescriptions are reflected in human discourses of various kinds, such as laws, directives, rules of conduct, norms, conventions (Fleetwood 2008). Some of these expressions are formally codified, while others are socially shared. Some even take a material form, like statues, art pieces, or memorials. Many FI scholars find it useful to distinguish between formal and informal discursive manifestations of institutional powers (Azari and Smith 2012; Bjarnegård 2013; Freidenberg and Levitsky 2006; van Dijk 2023; Waylen 2017b). However, it is rarer that FI scholars make an explicit conceptual distinction between institutions, on the one hand, and their empirical manifestations, on the other (Archer 2003; Fleetwood 2008). On the contrary, institutions, rules, norms, conventions, practices, and rituals are often treated synonymously by FI scholars, as well as other institutionalists (Fleetwood 2008). This lack of conceptual nuance can negatively affect analytical precision of feminist research. The next section explores how different expressions of institutional power can be used to build and test valid theories about these elusive entities.

Constructing and Testing Theories about Gendered Institutions

With an understanding of what gendered institutions are, we can now delve into the question of what can be known about these entities. For scholars who adhere to some form of realist ontology, the objective typically revolves around generating knowledge that, as accurately as possible, captures the true essence of the studied object (Danermark, Ekström, and Karlsson 2019; Gschwend and Schimmelfennig 2007; Lowndes 2020; Ostrom 2007). For FI scholars, the challenge lies in generating knowledge about entities that are hidden from view, constantly evolving, and idiosyncratic, all while recognizing the inherent limitations of human cognition. This process is fraught with challenges. One such challenge is the unescapable fact that human practices can have multiple plausible institutional explanations, some context specific and others more general. Another issue is that human accounts of institutions are almost always incomplete and sometimes outdated. These factors limit the effectiveness of deduction—the approach favored by positivism/empiricism—as a fruitful means of generating knowledge about gendered institutions and their powers (Decoteau 2017; Gorski 2004). This approach is associated with an unacceptably high risk of corroborating invalid theories about institutions.

The epistemology underpinning CR, termed epistemic realism, represents a middle ground between positivism/empiricism and judgmental relativism associated with some strands of poststructuralist research (Decoteau 2017; Fletcher 2017, 182; Zachariadis, Scott, and Barrett 2013). While rejecting deduction as a fruitful means of generating knowledge about institutions, the perspective also rejects judgmental relativism, which posits that the concept-laden nature of human cognition makes it impossible to adjudicate between competing truth claims about invisible social entities and forces (Decoteau 2017, 61). Epistemic relativism concedes that all scientific knowledge is partial and fallible because of the concept-laden nature of human cognition (Bhaskar 2008; Danermark, Ekström, and Karlsson 2019, 21; Decoteau 2017, 68; Elder-Vass 2008). However, the perspective also posits that it is possible to adjudicate between competing truth claims about institutions and their powers through the means of empirical testing (Alina-Pisano 2009; Fletcher 2017, 182).

To produce scientific knowledge about gendered institutions, we must approximate the process of how humans learn about these entities in our everyday lives. We do so by observing how people act in various social situations and getting acquainted with the discourses surrounding this behavior. By studying various empirical manifestations of gendered institutions, we may be able to construct valid theoretical models of these entities. To this end, we need a design that allows for a gradual refinement of the conceptual lenses through which these expressions are studied to capture the essential local elements of institutional powers. Abduction and retroduction are suitable analytical approaches for facilitating this process (Decoteau 2017; Fletcher 2017; Zachariadis, Scott, and Barrett 2013).

Abduction allows researchers to go through multiple iterations of data generation and theory testing within a single investigation (Alina-Pisano 2009; Decoteau 2017; Zachariadis, Scott, and Barrett 2013). This method facilitates the gathering of data on various tangible manifestations of institutional powers, including actions, practices, rituals, and varied discursive interpretations of institutions. Retroduction, on the other hand, is the process of analyzing these expressions and turning them into theories about the forces that could have produced them (Fletcher 2017; Zachariadis, Scott, and Barrett 2013). To construct plausible explanations of studied practices, multiple rounds of data generation and retroduction are often necessary. This iterative process enables the researcher to gradually falsify some of these explanations while refining those that remain consistent with the generated material. The process is ideally repeated until only one explanation remains in place (Decoteau 2017). Such an explanation captures both the local and broader components of institutional powers and accounts for the role social actors play in reproducing them. Armed with this knowledge, the researcher is well positioned to propose informed remedies for addressing studied gendered practices (Kenny 2014).

Abduction and Retroduction in Feminist Institutionalist Research: How to Go about It?

Many FI inquiries aim to explain gendered practices that manifest themselves in contexts in which the formal accounts of institutions are seemingly gender-neutral (Bjarnegård and Kenny 2016; Bjarnegård and Zetterberg 2019; Chappell and Waylen 2013; Kenny 2014). For instance, women are less likely to run for office even if most political parties describe their recruitment practices in gender-neutral terms (Bjarnegård and Kenny 2016; Culhane 2017; van Dijk 2023). Women have also been shown to be less likely than men to hold parliamentary speeches in contexts in which no formal rules prevent them from taking to the floor (Bäck, Debus, and Müller 2014; Yildirim, Kocapınar, and Ecevit 2021). This is usually the case because the existing formal rules do not faithfully—or exhaustively—capture all the real institutions that govern these practices (Bjarnegård and Kenny 2016; Bjarnegård and Zetterberg 2019; van Dijk 2023). How, then, can we uncover these institutions?

The first step in explaining a gendered practice is to ascertain whether the practice exists. This can be achieved either by consulting existing evidence or by undertaking an empirical study. To efficiently navigate such an investigation, the researcher can refer to the formal accounts of rules specific to the arena under study. These serve as a conceptual framework for gathering initial behavioral and discursive data (Kenny 2013; van Dijk 2023). For example, some political parties have instituted formal rules regulating the activities of their members of parliament (MPs), occasionally specifying a minimum number of speeches, bill proposals, or op-eds each legislator is expected to produce. Identifying whether there are systematic gender differences between those who fulfill these expectations and those who do not is a prerequisite for identifying gendered practices (Smrek 2022).

Once a gendered practice is identified, attention shifts to unearthing the discourses enveloping this practice. Various materials—from interview transcripts and internal communications to posters and artwork—can be used in this process (Evans and Kenny 2020). For instance, to explain gendered disparities in legislative speechmaking, interviews with women and men MPs—but also their party superiors—could prove invaluable. With a sufficient amount of empirical material in hand, the stage is set for the initial round of retroduction. The analysis of the generated material is used in the process of constructing theoretical explanations that are congruent with this material (Decoteau 2017, 71). Danermark, Ekström, and Karlsson (2019, 122) offer an excellent description of what this process entails:

Counterfactual thinking is fundamental for all retroduction. We ask questions like: How would this be if not …? Could one imagine X without …? Could one imagine X including this, without X then becoming something fundamentally different? In counterfactual thinking, we use our experiences and knowledge of social reality, as well as our ability to abstract and to think about what is not, but what might be.

Initially, multiple plausible explanations might align with the material, requiring the collection of additional material (Alina-Pisano 2009). The theoretical explanations from the first round of retroduction serve as new conceptual frameworks through which new data can be generated (Decoteau 2017; Rees and Gatenby 2014). This process allows the researcher to relatively quickly adjudicate between competing explanations of the studied practice. The conceptual lenses through which data is generated become more precise with each iteration (Rees and Gatenby 2014, 140; Zachariadis, Scott, and Barrett 2013). By becoming more attentive to behaviors or discourses that might have been overlooked, the researcher can pose more targeted questions to make sense of the observed phenomena, resulting in more valid data that captures local expressions of institutional power (Alina-Pisano 2009; Danermark, Ekström, and Karlsson 2019, 110). The researcher is gradually able to discern the structural positions occupied by social actors within the context and identify the roles they play in perpetuating the studied practice (Lowndes 2020). For instance, throughout the research process, it may become evident that a party-specific mechanism exists, permitting party superiors to allocate floor time among MPs in a manner that disproportionately favors men MPs. Further investigation may reveal that this skewed allocation is conditioned by deeply entrenched gendered norms about political competence held by the elites who allocate the floor time.

This iterative process of data generation and retroduction is ideally repeated until only one theoretical model remains, accounting for the entire corpus of generated material (Danermark, Ekström, and Karlsson 2019, 109–17). While the model may not perfectly depict the true relationship between gendered institutions and studied gendered practices, it outperforms other plausible alternative explanations of the practice under examination. Furthermore, the model’s validity has been ascertained through not one but multiple rounds of empirical testing.

A Multimethod Approach to Learning about Gendered Institutions

This article advocates for the use of the abductive-retroductive research design as an effective approach for generating valid knowledge about elusive and imperfectly knowable gendered institutions, as well as the role of actors in sustaining and challenging them. The FI literature, like the broader gender and politics literature, encompasses a wide range of methods and methodological approaches that are compatible with the demands of abduction and retroduction (Krook and Squires 2006; Tripp and Hughes 2018). This section argues for adopting a multimethod approach as a way of maximizing the analytical potential of the abductive-retroductive research design (Zachariadis, Scott, and Barrett 2013). Institutions leave diverse marks in the empirical realm, encompassing actions and discourses. To map out these empirical expressions, multiple tools are needed. Behavioral data is essential in ascertaining whether seemingly gendered actions indeed constitute gendered practices (Chappell and Galea 2017; Fletcher 2017, 186–87), while discursive data helps make sense of these practices (Alina-Pisano 2009; Decoteau 2017, 76). The remainder of this section outlines the methods that can be used in the process of learning about gendered institutions.

Actions Shaped by Institutions: Behavioral Data

In order to establish whether a particular gendered action constitutes a gendered practice, relevant behavioral data is often needed. This material can be sourced either from publicly available sources (public registers, parliament and/or party websites) or through direct observation. Various statistical analyses can subsequently be used to analyze whether there are systematic gender differences in the studied practice. For instance, Verge and Claveria (2018) utilize data from 12 European democracies to analyze gendered patterns in access to viable political candidacies and ministerial office. Similarly, Smrek (2020) uses detailed data on Czech MPs spanning three decades to examine whether access to the incumbency advantage is gendered. Htun and Weldon (2012) use a comprehensive data set of social movements in 70 countries to identify a correlation between feminist social movement activity and the development of policies on violence against women. The statistics used to determine whether a particular behavioral pattern is gendered do not need to be complex. Many FI scholars use simple descriptive statistics and/or time series plots to identify and visualize gendered practices over time (Gains and Lowndes 2014; Johnson 2016; Zetterberg 2008).

Observations of human interactions also serve as conduits to understanding intricate power relations between social actors (Gains and Lowndes 2014). Chappell and Galea (2017) use a rapid ethnographic approach to identify instances of norm enforcement, such as nods, laughter, frowns, or scolding. By studying who enforces these norms, they are able to discern whether men or women are more likely to occupy structurally privileged positions within the studied contexts. Similarly, Smrek (2022) observes exchanges between men and women legislators and their administrative staff. He finds that women are more likely to challenge the existing party rules, while men are more inclined to enforce them. Gains and Lowndes (2014) use data on the gender composition of the staff working for the policy and crime commissioners in the United Kingdom to identify the conditions under which women commissioners are able to adopt a women-friendly policy agenda.

Agential Accounts of Institutions

Institutions imprint themselves on human discourses. However, given the constantly evolving nature of institutions and the concept-laden nature of human cognition, these discourses only offer partial and sometimes inaccurate or outdated reflections of institutions (Archer 1995; Fletcher 2017). Consequently, the process of constructing valid models of real institutions and their powers should ideally involve a range of discursive material (Evans and Kenny 2020; Kenny 2014). To address this, FI scholars employ a variety of material generating approaches, including archival work and document analysis (Colley and van Acker 2021; van Dijk 2023; Morrison and Gibbs 2023), interviews (including focus groups, in-depth interviews, and life histories), process tracing (Kenny 2013), and political ethnography (Gains 2011; Galea et al. 2020; Smrek 2022).

Political ethnography, for example, combines the methodological advantages of directly observing human actions with the analysis of discursive material (Alina-Pisano 2009). By spending time in the field, ethnographers build trust with local actors. This trust grants them access to actions and discourses that are otherwise unavailable to outsiders (Chappell and Galea 2017; Galea et al. 2020; Smrek 2022). Smrek (2022) generates rich discursive material about instances in which internal rules are evoked or enforced by talking to several actors who witnessed these events. This type of material enables him to identify those who are privileged by the existing institutional arrangements and to adjudicate between competing agential explanations of the same event. On the other hand, diverse interview-based methods might allow for a more rapid generation of agential accounts about institutions. Unlike ethnography, interviews can be conducted outside the immediate research setting. Under certain favorable conditions, such as when discussing nonsensitive practices or when anonymity can be granted, this approach can enhance participant comfort and result in rich and multifaceted material (Culhane 2017, 50–51; Josefsson 2020; Zetterberg 2008). Culhane (2017) opts for in-depth interviews in her study of candidate selection in Ireland and finds that ensuring anonymity has a positive effect on participant comfort and candor. To maximize the validity of her material, Culhane engages a broad group of relevant stakeholders and compares the interview data to other discursive accounts of institutions, including party rulebooks, constitutions, and internal party documents. Analyzing a range of written documents—from various formal accounts of rules to correspondence, newspaper articles, novels, and imagery—can serve either for triangulation (Culhane 2017; Gains and Lowndes 2014; Kenny 2013) or as a primary stand-alone method (Colley and van Acker 2021; van Dijk 2023). Colley and van Acker (2021, 192–93) draw solely on policy analysis and analysis of formal party rules and election reports in their bid to offer an institutional account of the rise in women’s descriptive representation in Australia. In a similar vein, Morrison and Gibbs (2023) use autobiographical and biographical material to study the development of women’s political leadership style in pre- and post-devolution Scotland.

Discursive data does not necessarily need to be generated and/or analyzed qualitatively (Bjarnegård and Zetterberg 2019). A relatively new approach to generating data on people’s views and attitudes that shape their actions are survey experiments. Devroe and Wauters (2018), for instance, use a survey experiment to study how Belgian voters perceive women and men candidates (see also Håkansson 2023). Another relatively recent innovation in the FI literature is the use of quantitative text analysis as a way of analyzing large corpuses of textual material. Hargrave and Blumenau (2022) use U.K. parliamentary debates between 1997 and 2019 to analyze changes in women MPs’ debating styles, revealing a gradual shift toward more masculine styles.

The More, the Merrier: A Multimethod Approach

This section encourages realist-oriented FI scholars to use both behavioral and discursive material in their bids to reveal the powers that hide behind gendered social practices. Such multimethod approaches allow feminist scholars to fully capitalize on the abductive-retroductive mode of inquiry by identifying multiple expressions of institutional power (Alina-Pisano 2009; Krook and Squires 2006; Zachariadis, Scott, and Barrett 2013). The analysis of these expressions is used to construct and test multiple institutional explanations of studied gendered practices, accounting for both real institutional constraints and the role social actors play in maintaining them.

There are numerous examples of realist-oriented FI inquiries that rely on multimethod approaches to studying gendered institutions. Many of these studies—some of which are summarized in Table 1 ¹—successfully combine the analysis of behavioral and discursive data in their efforts to construct institutional explanations of studied gendered practices (Bjarnegård 2013; Franceschet and Piscopo 2008; Gains and Lowndes 2014; Johnson 2016; Josefsson 2020; Kenny 2013; Zetterberg 2008). Franceschet and Piscopo (2008) complement a temporal analysis of bill initiation patterns in the Argentine Parliament with 54 semistructured interviews. Their multimethod approach reveals the constraining effect of gender quotas—and the gendered expectations they embody—on women MPs’ legislative activity and the resistance from men party elites to these quotas. Kenny (2013) combines an analysis of women’s descriptive representation in the Scottish parliament with an analysis of official party documents, candidate campaign materials, party meeting minutes, media coverage, and 15 in-depth interviews with social actors involved in the candidate selection process. Using this material, Kenny shows how institutions that gender political selections in the broader U.K. context make their way into the new—and seemingly more gender-neutral—institutional setup of post-devolution Scotland. In a similar vein, Bjarnegård (2013) combines a large-N analysis of the correlation between clientelism and the political representation of men spanning two decades with an analysis of 145 interviews with a multitude of actors involved in the Thai candidate selection process. She identifies multiple cases when local men officials enforce and reproduce the existing gendered institutions that place women political hopefuls and candidates in a structurally underprivileged position vis-à-vis their men counterparts.

Table 1. Realist-oriented multimethod FI research on political selections and recruitment

While the existing realist-oriented FI studies are characterized by many strengths, there are aspects that could be improved. First, not all of the reviewed inquiries make a conceptual distinction between institutions and their empirical expressions. On the contrary, practices, rituals, rules, and norms are often treated as synonymously with institutions, which might create confusion regarding what is being studied (Fleetwood 2008). Second, a conceptual distinction between actors who maintain and resist institutions and institutions is not always made (Gains and Lowndes 2014; Lowndes 2020). This is an unnecessary self-limitation. Distinguishing between these two core components of social action allows feminist scholars to produce more nuanced and transparent analyses of studied practices. Third, while the bulk of the reviewed work are—at least implicitly—abductive, the abductive process is seldom described in detail. As a result, it is difficult to see which alternative explanations of the studied practices have been considered—and eliminated—by the authors. This lack of transparency makes it more difficult for the reader to assess the internal validity of the findings. While word-count-related constraints often prevent scholars from listing all the tentative explanations they considered (and tested), a good practice in this regard could be to include these in an online appendix. In light of these criticisms, the contribution of this article lies in demonstrating why it is ontologically desirable and epistemologically necessary to distinguish between institutions and their empirical manifestations, on the one hand, and between institutions and agency, on the other. If institutions are defined as real entities that are elusive, complex, partially idiosyncratic and changing, then studying their multiple empirical expressions in an iterative fashion is the most direct approach to getting as close as possible to true essence of these entities. Using the practices in this article can standardize realist-oriented feminist inquiries and enhance awareness about the core components of feminist research design.

The good practices presented in this article should not be interpreted as a zero-sum game. Many excellent FI inquiries reviewed here do not employ a multimethod approach, and yet they provide nuanced analyses of gendered institutions and their powers. Depending on the topic and/or funding, it is not always feasible to conduct multiple iterations of data generation and/or focus on both behavioral and discursive expressions of institutional powers. Sometimes, the existence of a particular gendered practice is so well documented that there is no need to generate original behavioral data. Other times, discursive material might be the only empirical material that is available. This does not necessarily make such inquiries inferior to the ideal presented in this article. Indeed, living up to the good practices outlined here can, at times, prove to be too costly or impractical. In light of this, it is important to recall that science is a cumulative enterprise and there is no reason for why the abductive-retroductive ideal presented here should be confined to a single FI study. A good practice in this regard is to describe which plausible institutional explanations of the studied practice have been eliminated in the course of the study and which ones remain consistent with the available material. This will help future scholars know where to start when contributing to illuminating the roots of a particular gendered practice.

Conclusion

The wide range of metaphysical views and methodological approaches within the gender and politics literature facilitates the production of varied knowledge about the role gender plays in political processes (Childs and Krook 2006; Krook and Squires 2006; Stauffer and O’Brien 2018; Tripp and Hughes 2018). The feminist institutionalist perspective, which highlights the role of gendered prescriptions in shaping social practices, is no exception to this trend (Bacchi and Rönnblom 2014; Driscoll and Krook 2012; Kulawik 2009; Mackay, Monro, and Waylen 2009; Waylen 2014). Clarifying the main focus of our inquiry and what we seek to understand is particularly important in fields that are metaphysically diverse (Archer and Elder-Vass 2012; Fleetwood 2008; Stanley 2012). The growing diversification of FI literature calls for an audit of the analytical perspectives and methodological approaches within it.

The close coexistence of realist and constructivist strands within FI has created tensions, making the perspective susceptible to internal misunderstandings and external criticism (Bogaards 2022; Waylen 2017a). A recent external review of eight books in the FI literature highlights the tendency among FI scholars to draw upon mutually exclusive views about the relationship between gendered institutions and human agency (Bogaards 2022). While the constructivist-oriented FI scholars are engaged in an ongoing debate about the nature of gendered institutions and the role of social actors in their reproduction (Bacchi and Rönnblom 2014; Kantola and Lombardo 2017; Kulawik 2009; Miller 2021b; Schmidt 2008), their realist counterparts appear to be less active on this front. A clearer understanding of the principles guiding the production of realist-oriented feminist knowledge can facilitate the standardization of this literature, making it more transparent and accessible to a wider audience. This article attempted to facilitate this process by grounding FI in critical realism.

This article offers a critical realist view of gendered institutions and their relationship with human agency (Archer 2020; Archer and Elder-Vass 2012; Bhaskar 2008). It defines gendered institutions as real entities that constrain and enable human action and are external to human minds. In the same vein, the article acknowledges the role social actors—and their ideas about institutions—play in reproducing and transforming existing institutions (Archer 2003). This approach, which grants independent ontology to institutions and agency, allows feminist scholars to analyze both the core components of gendered practices without reducing one to the other. To be realized, institutional powers depend on human capacity to think and act. This makes them contingent, partially idiosyncratic, and complex. To generate knowledge about these entities, we need to find ways of constructing and testing multiple plausible explanations of studied gendered practices (Danermark, Ekström, and Karlsson 2019; Decoteau 2017; Fletcher 2017). To facilitate this process, this article made a case for the abductive-retroductive research design. This approach allows feminist scholars to consult multiple expressions of institutional power, including actions, rituals, norms, rules, laws or conventions, and use these to gradually construct, refute, and refine institutional explanations of studies practices until only one remains. This explanation, validated through successive rounds of empirical testing, encompasses both the core local and broader facets of institutional powers. It further elucidates the roles various social actors undertake in perpetuating these powers. By employing the knowledge-generating process outlined in this article, realist-oriented FI scholars enhance the internal validity of their explanations, enabling them to offer well-informed policy proposals geared toward promoting gender equality.

Many existing FI inquiries come close to the good practices put forward in this article, but there is room for improvement. First and foremost, this article lays a solid ontological foundation for distinguishing between institutions and human agency as distinct entities. This differentiation allows both to be given their due analytical emphasis. Previous feminist studies rooted in realism have faced criticism for either downplaying the pivotal role that agency plays in shaping gendered practices (Schmidt 2008) or integrating actors into the analysis in ways that lead to metaphysical inconsistencies (Bogaards 2022). This article offers guidelines for how to include both institutions and human agency in the analysis while avoiding the trap of metaphysical stretching.

Second, the conceptual distinction between institutions and their empirical manifestations—such as actions, rituals, laws, rules, norms, or conventions—provides feminist scholars with a potent analytical tool justifying why it is essential to study a wide array of observable implications of institutional power when constructing theories about institutions. Often, the two are conflated in FI literature (Fleetwood 2008). Keeping them apart will enhance the analytical precision and rigor of realist-oriented FI inquiries.

Third, feminist research designs are frequently of a complex nature, involving a multitude of empirical material and favoring abductive approaches over deductive ones (Bjarnegård 2013; Gains and Lowndes 2014; Zetterberg 2008). To an independent observer, these approaches might appear as lacking in parsimony or analytical rigor. This article offers feminist scholars a clear epistemological rationale for choosing more complex research designs over deduction, ensuring the production of valid and empirically corroborated institutional explanations.

Lastly, by introducing the abductive-retroductive research design, this paper aims to contribute to the standardization of realist-oriented FI inquiries. This design underscores the necessity of constructing and testing multiple plausible explanations when seeking to understand a specific gendered social practice. While providing a detailed account of every iteration might contradict the principles of parsimony, listing all the tentative explanations considered—and ultimately refuted— throughout the study can enhance the internal validity of feminist theoretical models and ward off potential misunderstandings and criticism.

While the primary focus of this article is on the FI perspective, its significance reaches beyond the confines of this perspective. It is widely acknowledged that the rich methodological diversity within the gender and politics literature stems from the complex and multifaceted nature of the concept of gender (Stauffer and O’Brien 2018; Tripp and Hughes 2018). As this body of literature continues its rapid expansion, there is a pressing need to periodically reassess and refine its foundational metaphysical assumptions. Such a practice will enhance the internal coherence, transparency, and integration of gender and politics literature into the mainstream of political science research.