Replication or Innovation? Structuration in Policy Implementation

Abstract

Why does a program, policy, or management approach implemented with success in one jurisdiction or organization fail to achieve similar results in another context? There is a large body of literature in public affairs and related fields that wrestle with this question. Scholars place varying emphasis on the constraints of the institutional system relative to humans’ agency in bringing about successful outcomes, and there is a tendency to generate lists of factors that enable or impede successful implementation. In this article, we present an alternative theoretical approach grounded in structuration processes. We turn to recent empirical scholarship and theory to re-examine what is known about structural elements that influence the implementation process: rules, routines, culture, and resources. This literature emphasizes that the work of these mechanisms is fundamentally shaped by endogenous factors within a system, fueled by the agency of actors within the setting. This is a more robust way to understand how microdynamics shape meso conditions in organizations and networks. Rather than understanding the implementation puzzle as how to replicate effective ideas, this frame suggests more attention to how to support innovation and learning is warranted.

Introduction

For decades, scholars of public affairs have wrestled with a perennial puzzle of implementation—policies or programs that are successful in one context fail to be replicated in another context. In public policy circles, we frequently hear discussions about the challenges of scaling up; how to implement a program with fidelity across implementation settings (List 2011; Weiss et al. 2014). In public management circles, we hear of challenges to effectively reproduce process and performance management reforms; how to lead and manage the adoption of new ideas in governance systems to enable the most effective outcomes (Gerrish 2016; Heinrich et al. 2009). Recent claims by behavioral and implementation science promote behavioral nudges and fidelity to proven program models as answers to these questions. Yet these solutions are rarely sufficient to produce the change desired.

Part of the challenge lies in the implicit theory underpinning conventional understandings of how implementation of public programs is accomplished. Many approaches see the problem as one of controlling ‘replication’ and reducing confounding factors, and this assumption is reflected in most studies of policy diffusion (Haider-Markel 2001; Shipan and Volden 2008), behavioral science (Madrian 2014; Richburg-Hayes et al. 2017), and implementation science (Aarons et al. 2011; Meyers et al. 2012). In these traditions, the core implementation problems are conceptualized as technical. The emphasis is on “knowing what”—what variables can explain readiness for change or predict successful implementation? Yet this overlooks what scholars have uncovered about how change is actually accomplished in complex social and economic systems (Colander and Kupers 2014; Fligstein and McAdam 2012; Innes and Booher 2018; Rogers 2003).

Rather than focus on means of prediction and control, this alternative approach emphasizes human agency—the way individuals facilitate change and learn within organizations and networks. Rather than conceiving of agency as vested in exceptional “leaders,” our notion sees change as potentially coming from people occupying various roles in a social context. It is developed from a close read of sociological and organizational theory, as well as trends in public management research. This analytical turn is significant, as we move from consideration of “knowing what” to “knowing how,” privileging an understanding of context and social dynamics within meso-level institutions. This allows us to re-situate the classic implementation puzzle of how to replicate programs within a broader understanding of how social systems operate when encountering what is perceived to be an innovation. It also is consistent with pragmatic philosophy reflected a century ago in the work of Mary Parker Follett and John Dewey (Ansell 2009, 2011; Cohen 2009; Graham 1995), which point to ways of integrating theory and practice, idealism and empiricism. It allows us to develop a theoretical approach that is both descriptive (allowing us to explain what happens) and normative (allowing implementers to be more intentional about specific ends).

To help illustrate these theoretical ideas, we describe a particular case of policy change. Much has been written about the inconsistent implementation of cash assistance welfare policy in the United States (Allard 2007; Brodkin 2006; Hill 2005; Riccucci et al. 2004; Sandfort et al. 2019; Soss et al. 2011). In the last 20 years since welfare reform, absolute grant levels have decreased and caseloads have plummeted—not because there is less need, but because of a fundamental shift in the federal policy. In the current program configuration, administration of the Temporary Assistance for Needy Families (TANF) program focuses on tracking program participants’ activities rather than supporting them to they find sustainable or adequate employment (Government Accountability Office 2010; Hahn et al. 2012). This exacerbates the administrative burden experienced by families living with low-incomes who turn to the program for temporary financial assistance (Herd et al. 2013) and reinforces social constructions of the poor as deviants who need to be closely monitored (Fording et al. 2007; Hasenfeld 2010; Schneider and Ingram 1993). Overall, the implementation of this policy has institutionalized a desensitized response to citizens’ experiences of living in poverty (Brodkin 2006; Soss et al. 2011; Watlkins-Hayes 2013).

It is tempting to see the rules, routines, culture, and resources of the TANF program as given, rather than a product of the particular institutional configuration created during implementation. When one does, there seem to be few avenues available to improve implementation. Yet, we are arguing here for a theoretical account that has a more robust understanding of the interplay between institutional structures and human agency. There are variations observed at the local level of implementation—such as variation in interpretation of performance measures, rigidity of application processes, or cultural understandings of the poor in the TANF program—which are not simply explanatory variables that predict differing outcomes such as caseload levels (Sandfort et al. 2019). These structures are malleable and can be leveraged by human ingenuity to shift implementation practices and improve outcomes.

We begin to lay the foundation for this alternative theory by first establishing the historical and analytical roots within social theory and pragmatic philosophy. We then drill down on the core theoretical constructs underlying the agency-structure interactions around program implementation. Rather than understanding this process as “replication,” this theoretical foundation posits that this process is more akin to diffusion of “innovation,” with an institutional twist.

Intellectual Aims

In the legacy of pragmatic philosophy (Cohen 2009; Dewey 1910; Farjoun et al. 2015; Graham 1995; Gross 2009; Houser and Tienne 1998), we are interested in overcoming conventional dualities that separate theory from practice, environment from action, means from ends. By studying policy implementation in practice, we can explore how each are mutually constituted. Behavior can be recognized as purposive and the product of choice. But, unlike rational choice theories, people’s aims do not have to be separated from their conditions (Whitford 2002). We can explore the meanings people make of their actions, privileging the human experience of sensemaking about the ends-in-view as significant in determining what they do, and the actions they take. We can acknowledge that in attempting to solve problems, people both rely on habits but also exhibit creativity if these habits fail to bring about desired results (Gross 2009). As Chris Ansell (2007) has noted, such a pragmatic approach enables analytical holism; it is focused both on understanding a particular problem—in this case, the puzzle of program replication across settings—and in synthesizing theoretical explanations of such a problem.

There is a large body of literature in public affairs and related fields that wrestle with issues of public policy implementation and program replication (Sandfort and Moulton 2015). Some theories and frameworks emphasize institutional structures and resources—it is the variation in the regulatory environment, the capacity and resources, or the bureaucratic processes and red tape—that set the “rules of the game” that lead to differences in outcomes. Other theories and frameworks emphasize the people—the politicians, managers, and entrepreneurs—who can lead change and motivate innovation towards desired outcomes. Different approaches also place varying emphasis on action occurring at a particular level of the governance system.

In the last 40 years, considerable implementation research has generated long lists of variables and matrices of factors that attempt to predict what is necessary to “implement with fidelity” a policy or program. One widely applied framework in health sciences identifies 37 distinct constructs relevant to predicting faithful implementation (Damschroder et al. 2009). Yet even when replication is controlled, rarely is this care sufficient to create uniform results. For example, in a meta-analysis of the results of an energy conservation initiative replicated in more than 100 different settings, Allcott (2015) found that the early adopters had significantly larger impact estimates than those sites included in later impact evaluations.

Part of the challenges of replication with fidelity in the public arena is that public programs and policies are implemented through multi-level systems of governance (Heinrich et al. 2009; Hill and Hupe 2014; Lynn et al. 2001; Moulton and Sandfort 2017). Rather than a single public agency tasked with the implementation of a public service, most public goods and services are carried out by a networked array of actors with varying degrees of direct and indirect control (Milward and Provan 2000). Institutional structures at various levels of the system constrain and enable action that shape and lead to the ultimate policy and program outcomes. These include regulatory mechanisms generated from the policy field level of the system (e.g., formal laws or rules that accompany financing) and managerial processes at the organizational level (e.g., operational routines or staff cultures), all that potentially constrain and enable frontline service delivery structures and processes. The aggregation of these forces ultimately shape the outputs and outcomes of policy and programs.

While the multi-level logic of governance offers a comprehensive framework to describe the context within which implementation takes place, empirical research tends to emphasize either the macro-structural dynamics of the system that lead to observed outcomes, or the micro-behavioral actions of individuals within the system (Barzelay and Gallego 2006; Moynihan 2018; Robichau and Lynn 2009). In the policy implementation literature, this has been expressed as a tension between top–down and bottom–up approaches to understand (and control) implementation (e.g., Goggin et al. 1990). Yet there is a significant missing piece. There are important meso-level processes—interactions between agency and structure around the core program technology—that occur at the organizational and service delivery levels of the system (Hill 2003; Knox et al. 2018; Sandfort 2010). To understand these, we must depart from conventional assumptions that essentialize either structure or human agency. Both are important and cannot be understood in isolation. Structures shape how action is directed. But structures attain their significance because of how actors invest in them. And actors, themselves, have agency in the valuation of institutional mechanisms and communication about them. This is the essence of structuration. And it is the recursive dance between institutions and agents that is central to enable system change during policy and program implementation.

The Structuration of Implementation

Giddens’ theory of structuration recognizes the dual nature of social structures (Giddens 1981, 1984). In this theory, social structures are viewed as part of a process rather than a constant state. They are central to both constraining and enabling human agency within social systems, providing an explanation for how and why patterns of interaction are reproduced. As Sewell notes, “agents are empowered with and against others by structures” (Sewell 1992, 20). Thus, understanding the potential for human agency within social systems requires an understanding of the structures at play. But neither agency nor structural forces are inherently deterministic. This idea is echoed by other important social theorists of the last 40 years, such as Bourdieu (1977), Emirbayer and Mische (1998), Feldman (2000), Latour (2005), Orlikowski (1992), and Fligstein and McAdam (2012).

For example, Emirbayer and Mische (1998) elaborated on the importance of time. Human agency in changing structures is simultaneously informed by the past (observing what was viable), oriented toward the present (understanding current conditions), and anticipating the future (imagining alternative possibilities). In this way, they point towards an explanation of how an individual might alter structures in relation to time, with actors seeking to create interventions for the future, in relation to the past, responding to outcomes developed in the present. Fligstein and McAdam (2012) take another approach, articulating a theory of human action they call “strategic action fields” that provides a sociological alternative to rational choice. Rather than assuming self-interested individuals, Fligstein and McAdam (2012) posit the sociability of human beings as the central driver in the quest for resources and the maintenance and alterations of institutions.

These perspectives add nuance to the ways in which we understand how social structures are institutionalized and altered by the actions and beliefs of individuals and groups. Across these works, distinctions are often made between structural elements that are intangible and reflect virtual understandings, akin to schema or habits, and those that are tangible, often with physical attributes that operate more overtly as sources of power (Giddens 1984; Gross 2009; Sewell 1992). Here, we are interested in the structuration of implementation, or how actors use and shift both tangible and intangible elements to bring about new program or policy change. Tangible structuring elements include rules and resources—attributes that are more easily observed, and that have been commonly applied in top–down approaches to studying policy implementation. Policy makers and public administrators believe these elements can control and direct attention. And, in some ways they do. However, the studies we discuss here also highlight the significance of more intangible structuring elements such as routines and cultures. These are more commonly explored in bottom-up studies of implementation.

The structuration of policy and program implementation occurs at the intersection of macro-political and micro-behavioral levels, at the meso-level through the ways individuals and groups work with these structuring elements. Figure 1 provides a simple depiction of this relationship.

The structuration process involving the strategic deployment of rules, resources, routines, and culture is a critical—but often taken for granted—aspect of policy and program implementation. In their day to day actions, public servants at all levels of the implementation system evoke a “logic of appropriateness” (March and Olsen 1998) grounded in their understanding of their particular social structure. Said another way, socially skilled actors know what is appropriate and legitimate in their particular context (Fligstein 1997, 2001). Sometimes they deploy structuring elements to promote harmony and shared understanding; other times, they introduce discord. Both shared understanding and discord are used to create changes in organizations, networks, and staff practices during the policy and program implementation process. Recasting the implementation puzzle through the lens of structuration not only offers richer theoretical insights, but it offers a pragmatic hook to describe how agents affect implementation. It also provides a more robust explanation of variation in implementation processes and outcomes across settings.

The four structural elements described here are not new to research on implementation, nor are they an exhaustive list. Certainly, there are many factors that contribute to heterogeneous implementation outcomes, some related to characteristics of the context and technology, some related to the qualities and abilities of individual actors. Our purpose is not to offer a recitation or framework for these factors— that has been done elsewhere (Goggin et al. 1990; Greenhalgh et al. 2004; Mazmanian and Sabatier 1989; Meyers et al. 2012). Rather, our purpose is to develop a better explanation for “know how”—how do actors involved in policy implementation facilitate institutional learning and change in systems that are apparently constrained by a high degree of institutionalized rules, resources, routines, and cultures?

Most prior research treats tangible and intangible structures as relatively immutable, exogenous factors that are correlated with variation in implementation outcomes across settings. However, a structuration lens calls attention to the mutability of structures. Table 1 summarizes the differences between seeing these structural elements as fixed or changeable. There is significant empirical evidence that these elements are endogenous, shaped by people within the contexts in which they work. Yet implementation research tends to ignore this reality, treating them as exogenous. In the text that follows, we elaborate on how rules, resources, routines, and culture are shaped by agency and human ingenuity. If we adopt a structuration approach to implementation, we will more clearly see how people take purposeful actions that affect implementation outcomes. These actions may improve—or hinder—the creation of public value throughout the implementation process.

Rules

Rules or schema are often considered the basic building blocks of social structures (Giddens 1984). Elinor Ostrom (2011) defines rules as the “shared understandings among those involved that refer to enforced prescriptions about what actions (or states of the world) are required, prohibited, or permitted (17).” Some rules are superficial and tangible, such as written legislation or program guidelines, whereas others are deeply embedded in day to day practices, so much so that they can be difficult to even identify, let alone change (Sewell 1992). Here, we conceptualize rules as being of the first more readily observed type, allowing deeper rules or schema to be part of our exploration of routines and culture.

In implementation settings, rules include formal legislation passed by a central government, but they also include laws enacted at lower levels of government as well as regulatory restrictions and informal requirements imposed by governmental and non-governmental actors. There is a large body of literature in policy studies that examines the evolution of rules and formal policies, and the role of human agency in shaping them. Laws are passed and rules are written, with coalitions of actors vying for their positions to be reflected in formal policies. Research on agenda-setting, advocacy coalitions, and institutional design has advanced understanding beyond top–down, “command and control” perspective of policy processes to consider the role of agents—and in particular, coalitions of agents—who shape the policies and rules considered to be legitimate within policy fields (Ostrom 2011; Weible et al. 2009, 2012).

Individuals attempting to bring about change in macro policy structures are often referred to as policy brokers, or more generally, policy entrepreneurs. Kingdon (1984) first introduced the role of the policy entrepreneur as an individual within the policy field—from a policymaker to a member of an interest group or an implementing organization—who invests their time, resources, and reputation to bring about policy change; these individuals watch for and take advantage of windows of opportunity to push forth new ideas.

The concept of the policy entrepreneur can be combined with an understanding of the structuration of rules and policies over time. For example, a recent sociological study of U.S. child labor law reform in the first half of the nineteenth century documents that variations in labor laws between states can best be understood as policy entrepreneurs’ actions to shift collective understandings and create changes in the macro-structural factors over time (Anderson 2018). While this line of research offers promise for understanding institutional change in formal policies over a considerable period of time, it does not inform how actors maneuver and create change in institutional settings around implementation on a day to day basis. Much of this type of action takes place at the organizational and service delivery levels of the system.

At the organizational level, the focus shifts from formal legislation to agency rules and rulemaking processes. Public programs are notorious for having bureaucratic rules that are often in conflict and that may not be oriented towards the most efficient—or effective or equitable—way to bring about policy and program results. Red-tape is defined as the “rules, regulations and procedures that remain in force and entail a compliance burden for the organization but have no efficacy for the rules’ functional object” (Bozeman 1993, 283). Excessive rules and bureaucratic processes can reduce employee motivation and job satisfaction and can lead to worse agency and program performance (Bozeman and Feeney 2014; Brewer and Walker 2009; DeHart-Davis and Pandey 2005; Moynihan and Pandey 2007).

While bureaucratic rules can and certainly do impact implementation outcomes, prior research that treats rules as merely exogenous predictors of outcomes misses a fundamental element of rules—they are not fixed or complete, but are shaped and interpreted by agents within the system (Blau 1955; March 1997; March et al. 2000). For example, state agencies administering social welfare programs make intentional and unintentional decisions regarding the onerousness of intake processes that may discourage client follow-through with the process and thus receipt of program benefits (Heinrich 2018; Herd et al. 2013). The use of rules to shift costs from the public agency to individuals creates administrative burden that imposes learning, psychological, and compliance costs on citizens (Herd and Moynihan 2018). Additionally, the many interventions using behavior nudges (Thaler and Sunstein 2008) presumes that small alterations in rules and administrative procedures can improve efficiency or effectiveness in program implementation. Human agency and ingenuity are always at work shaping the rules guiding implementation.

Resources

A common way to analyze organizational phenomena, particularly in open system approaches, is to consider the resources on which an organization or program is dependent for survival. According to resource-dependence theory (Pfeffer and Salancik 1978), organizations are dependent on their external environment for money, human talent, and other resources. Most fundamentally, resources are sources of authority. Strategic management, then, focuses on how to reduce environmental uncertainty and ensure stable and diverse resources that allow for maximum organizational independence and resilience. For example, in nonprofit organizations, managers are encouraged to develop heterogeneous funding portfolios because of the different levels of predictability and control associated with various sources of funds (Gronbjerg 1993).

The value of financial resources is often assumed to be fixed, tied to its objective value. In structuration theory, resources are recognized as unequally distributed within social systems and can be used by agents to enhance or maintain power (Sewell 1992). Said another way, resources are not exogenous to the system, but are themselves shaped by actors operating in the system.

Of course, there are objective variations in the amount and type of resources in organizations or communities. But the same resource endowments are frequently used in different ways based on agents’ understanding of their significance and utility (Cyert and March 1963; Gross 2009). For example, there are endless cases of how organizations with comparable financial resources, such as those operating within the same grant program or under the same government contract, have very different programmatic outcomes. Human resources are similar. Teams made up of people who lack credentials on paper might create more value for the organization than those with the best training or expertise. Resources are significant when they are used to alter relationships and results. In fact, their relative worth is dependent upon how they are configured and used within the field more than any objective valuation (Feldman 2004; Sandfort and Quick 2017).

This conception is also relevant for how other material resources are used. While scholars have long stressed material artifacts help bring sense and meaning to organizational life (e.g., Schein 1985), recent scholarship shows the critical role artifacts can play in activating others and crossing divides (Hill 2003). For example, Carlile (2002) documents how material objects function to bridge professional differences. Best practice reports, visual analytics, or performance measures are regularly used to bridge the syntactic, semantic, and pragmatic boundaries that separate departments or organizations. In fact, he terms them “boundary objects” that offer a way to represent information to others, learn other alternatives, and transform understanding. Artifacts operate as resources that can both be adapted to local needs and constraints, and maintain commonalities across sites (Star and Griesemer 1989). When deployed through inclusive management approaches (Feldman and Khademian 2007; Quick and Feldman 2014; Sandfort and Quick 2017), they enable a means for people to do work together and potentially change understandings or motivate new actions in the process.

These studies show that rather than being fixed, resources are malleable. When people apply their knowledge about a particular context and bring those resources into use, they become sources of power. In fact, it seems more accurate to understand the process of using resources as an action verb, rather than focusing on resources as a noun. Resourcing is the way of taking various assets—money, people, material artifacts, knowledge—and using them to bring about change or stability in the social setting (Feldman 2004; Sandfort and Quick 2017).

Routines

The topic of routines shifts our focus from tangible to intangible elements that include the interpretation of formal rules (Reynaud 2005) and the development of habits (Gross 2009). Routines are repetitive patterns of action that involve coordinating the activities of multiple actors within an organization, such as hiring, budgeting, program design, or performance assessment. They are so interesting to organizational scientists because they provide connection among organizational activities and, in a sense, make them collective.

While conventionally understood to be formal mechanisms that stabilize organizations, studies of how routines are actually carried out reveal there is more indeterminacy (Feldman 2000; Feldman and Rafaeli 2002; Feldman and Pentland 2003; Pentland and Feldman 2005). In fact, from a number of field studies, Feldman and colleagues document frequent changes are not uncommon. Rather than merely stabilizing the organization, they are regularly adjusted by actors trying to improve what results. As Feldman (2000:620) writes,

“[Routine changes]… are related to different kinds of outcomes. One reason is that sometimes actions do not produce the intended outcomes. Another is that sometimes actions produce outcomes that create new problems that need to be solved. A third reason is that rather than producing problems, actions can result in outcomes that produce new resources, and therefore enable new opportunities (Feldman 2004). A fourth possibility is that the outcome produced is intended but that participants still see improvements that could be made.” (620).

This diversity leads them to posit that routines in practice actually involve a “repertoire” of possible actions. When used, routines reflect “collective repertoires for thinking and acting vis a vis a set of problems” (Gross 2009). This performative characteristic of routines illuminates how actors shape what routines actually look like in specific times and places. People assess the relative success and limitations of a routine process. Influenced by their cares and commitments, they make decisions about what can be reasonably altered, and apply themselves to making improvements. Understood in this way, routines are the building blocks of both learning and institutionalization. For example, there may be a collective routine for “hiring,” but the specific practices involved in carrying it out are adjusted and refined depending on the particular people involved, the timing of the market, and the candidates interviewed (Feldman and Pentland 2003).

In studies of policy and program implementation, there has been much focus on the routines of workers at the frontlines of service delivery (Hill 2005; Sandfort 2000). Street-level workers interpret and establish routines that allow them to accomplish their tasks and adjust in response to institutional pressures. Yet, there is varying emphasis in this stream of literature on the extent to which front-line workers have agency to push back against and shape routines. Often front-line worker agency is understood as a coping mechanism rather than strategic action to shape the direction of policy or programmatic outcomes (Lipsky 1980). However, there is increasing recognition that a dynamic tension exists between micro-level sense-making and organizational factors (Meyers and Vorsanger 2003; Oberfeld 2009; Riccucci et al. 2004; Soss et al. 2011; Zacka 2017). Empirical studies highlight how frontline staff either develop routines or challenge the ones imposed by managers in light of their own sense of how their actions are related to outcomes (Maynard-Moody and Musheno 2000; Sandfort 2000; Walkins-Hayes 2013).

Culture

The cultural aspects of social structures are even less easy to observe and often are more deeply embedded in the taken-for-granted interactions of day to day life. If routines are assumed to stabilize the structure of organizations, a wide variety of scholars recognize that culture operates more informally, developed from the beliefs and values of organizational members to shape implementation processes (Miller 1992; Morrill 2008; Weber and Dacin 2011). In recent years, research about organizational culture proliferated and scholars note five prominent conceptions developed to understand the construct (Giorgi et al. 2015): values, stories, frames, toolkits, and categories. For example, values were a topic of early scholarly exploration that emphasized how shared meaning guides and constrains individual and groups’ thoughts and behavior within organizations. Frames were conceptualized as filters that limit cognitive attention, defining a situation within its social context. While this research initially conceptualized culture as a fairly inflexible constraint, operating through collective values or frames that were difficult to challenge, this has not held up empirically (Martin 1992). Rather, much like studies of routines, researchers document that organizational actors actually use culture to their advantage, to accomplish some purpose such as solving software glitches (Perlow and Weeks 2002), improving product innovation (Seidel and O’Mahony 2014), or enhancing operational effectiveness (Meyerson 2003; Morrill et al. 2003).

Swidler’s (1986) classic articulation named this idea the “cultural tool kit.” She used this terminology to evoke the way she observed people using various configurations of symbols, stories, rituals, and worldviews when confronting problems in order to develop action strategies. In this conception, the cultural tool kit is a “grab bag” of factors that offer ways actors might mobilize others to solve problems. When applied to organizational studies, this idea has spawned a vast empirical literature (Giorgi et al. 2015); because individuals and organizations draw from larger cultural repertoires to pursue their desired ends, studies consider how people with social skill or salient identities acquire these tools and deploy them within a particular context (Weber and Dacin 2011).

Anne Khademian’s (2002) work in public organizations stresses how culture is produced from group efforts to accomplish a programmatic purpose, with particular resources in a particular environment. While a group develops shared commitments, managers decide whether or not to and how to develop connections between those values and the programmatic work which needs to be done by that group. In these decisions, they either reinforce or challenge the existing culture. In this regard, public organizational culture is rooted in the interplay of program responsibilities, environment, and resources. Managers can draw upon the cultural tool kit that is created by this interplay and use their knowledge of it to fuel change. Like the research on rules and routines, culture is not merely something that creates stability. Practices—what people do—develop and alter organizational culture. Thus although organizations have cultures, “culture is also something that people do” (Giorgi et al. 2015: 30).

One of the ways that culture is enacted within an organization is through the process of legitimizing a new practice. When adopting new practices in policy implementation, actors must consider the value and understandability of the new approach. If the new practice is something that seems aligned with current practices of the setting, its value often will not be questioned. This requires managers to think carefully about how they introduce new practices. If they are framed in relation to current approaches, values, or aspirations shared by people in that context, they are more likely to be integrated into the setting. When managers use terminology that resonates with shared values—“person-centered planning,” “human-centered design,” “green development,”—changes are more easily embraced. In the popular management literature, Johnson (2010) has talked about this idea as working in the “adjacent possible,” emphasizing that when using the cultural toolkit, managers should make purposive choices in how they frame and name the change they desire.

Structuration in TANF Implementation Processes

To illustrate how structuration during implementation unfolds, let’s consider the case of a web-application developed to improve the implementation of the TANF policy in Minnesota. The federal policy provides minimal cash support to low-income families and requires that in exchange they participate in approved activities. The 2005 reauthorization of the law both strengthened mandates for states to report program participants’ involvement in required activities and imposed sanctions if they fail to comply (Allard 2007; Government Accountability Office 2010). In the politically charged arena of welfare reform (Brodkin 2006; Fording et al. 2007), the desired outcome of the policy does not explicitly emphasize increasing wages or providing training for better jobs, although rigorous research and program evaluations of welfare-to-work programs point to how to accomplish these outcomes (Autor and Houseman 2010; Hamilton 2002).

Because of the disconnect between federal law and empirical evidence, some states try to broaden the implementation of cash support programs. Some secure alternative financing to fund activities more likely to advance economic security. Others design state-funded efforts for distinct populations such as young parents or develop innovative “work arounds” (Hahn and Loprest 2011; Schott and Pavetti 2013). While each state has its unique story of how these things come to be, it is worth considering—How does this variation in policy implementation occur?

The state of Minnesota created alternative performance measures to incentive counties to increase TANF participants’ incomes (Sandfort et al. 2019). They also developed separate state-funded programs for hard-to-serve populations and supported administrative innovations such as a web-application to improve service effectiveness. Program managers at the state and local levels led this activity because they recognized their own abilities to create new rules, marshal necessary resources, influence program processing routines, and frame the policy area as one in which innovation could occur. Institutionalizing this alternative system took many years and is far from perfect. But it does provide a simple illustration of the structuration of implementation.

Although federal rules that define performance are clear—track hours of work experience, on the job training, and employment for more than 30 hours a week—leaders in Minnesota challenged this rule. They developed an alternative performance management system that directed local governments to report on their substantive outcomes—how many families found full-time employment, and how many left the cash assistance rolls. They passed state law that provides financial benefits to local governments with the best performance and included the new measures in a management dashboard for state government (Sandfort et al. 2018).

Even though Minnesota developed an alternative performance system and rewarded counties that achieved superior performance, local service agencies still needed to report client activities, such as job search activities, to state officials. To ease client experience in reporting their activities and to free up time for local organizations to engage more substantively with families, the state developed a new program technology– a mobile application (web app) for tracking TANF client activities. In developing the tool, they engaged clients and local service agencies in a series of design labs that shaped the functionality and feel of the product, and a general process for integrating it into local office contexts. To actually bring the app into existence, state managers needed a means for convening the various stakeholders, securing the web-developer, and recruiting local sites for the field test. State managers secured fairly flexible federal grant dollars to contract for these services. They were used to build necessary implementation tools—the first iteration of the product, a web-site to explain it, and video and brochure to motivate local managers to participate.¹ Administrative tools were subsequently created to allow frontline workers to have a dash-board summarizing their interactions with clients. Training materials were built, including a “sand-box” for local sites to play with the app's functionality before taking it “live.” The field tests conducted by three local offices helped to identify both technical enhancements and improvements in process routines. As Feldman (2000) notes, the tests produced outcomes, generated new resources, and enabled new opportunities for improvement.

Standard routines within TANF service organizations typically focus on processing program applicants and ensuring their eligibility for cash assistance. To actually implement the web-application in local offices, training sessions were held in each local government office about the core technology of the tool. However, managers in each local office determined how to best embed the web-app in their particular routines. There were many details that needed to be decided, and specific ways that existing activities needed to be altered or aligned to enable frontline staff to deploy the application. In some local organizations, managers were particularly concerned with ensuring clients could “opt in,” and steps were added into the routine to maximize client choice; in other offices, good relationships with nonprofit employment service providers simplified the workflow; in still other offices, relationships with nonprofit partners were tense, and better relationships needed to be facilitated to ease implementation. Each adjustment in the routine was led by a program manager with social skill, who knew the context, the people, and the ways to get things done in that setting.

Finally, understanding the various manifestations of culture was a critical element necessary to develop and implement the technology. The team who facilitated the design and implementation process worked for over a year with the state agency before the idea of the web-application crystalized as a possibility. They knew state program managers were motivated to reduce the administrative burden caused by federal rules on program participants. They also navigated the competing values of accountability, privacy, and efficiency articulated during initial design sessions and crafted a product to reflect those commitments. Staff also needed to ascertain the most effective way to frame understanding of the technology. Depending on the dominant values in each context, the innovation was framed as a method to improve customer satisfaction, a tool for improved efficiency, or a modernization of existing program delivery. The external consultants also depended on managers, often those who had agreed to champion the app, to inform them about the salient values and landmines of that context.

This simple case illustrates how the TANF system in Minnesota evolved to enable successful implementation outcomes—despite rigid legislation and burdensome rules passed down from federal policymakers. Scholars studying variation in TANF implementation across states might document different outcomes for Minnesota TANF recipients, but would be hard-pressed to identify how these outcomes came to be different. Empirical analyses could analyze agency rules, survey employees to measure agency culture, or observe employee routines; yet these analyses would miss the dynamic interplay between human agency and the range of structural elements that shape implementation in practice.

Implications for Research and Practice

Our reading of the research and our practice experiences like that in Minnesota suggests a new and, we believe, more robust way for public management scholars to understand how meso-level organizations are stabilized or changed through micro-level behaviors of individuals and groups during implementation. This theoretical move complements new understandings of organizational innovation and change more broadly (Barzelay and Gallego 2006).

With the rise of interest in public administration on behavioral science (Moynihan 2018), it is important to note that our approach is complimentary but distinct. Whereas behavioral science draws insight from psychology and cognitive sciences, this approach is informed by sociology. The structuration of implementation is grounded in a fundamental ontological assumption—individuals are not strictly self-regarding but are motivated to make meaning with others and they have a fundamental need for social acceptance (Fligstein and McAdam 2012). The collective provides “existential refuge” in modern life and collective action is motivated and sustained by people’s efforts to make meaning. Rather than attending to isolated factors that predict individuals’ behavior, this approach privileges exploring how social meaning and membership shapes that behavior.

This understanding also provides a significant challenge to current public management research that includes time-invariant models with only meso-level variables (e.g., organizational rules, routines, culture, or resources). These cross-sectional analyses often assume (for empirical simplicity) that meso-level structures are given, ignoring how micro-processes shape the meso-level factors. We must take seriously the developments in organizational theory (Feldman 2000; Giorgi et al. 2015; May and Finch 2009; Orlikowski 1992; Romme 2003; Yoo et al. 2006) that show how individual agency operating within the constraints of institutions shapes meso-level realities. As Sewell (1992) writes, “Structure is dynamic, not static; it is the continually evolving outcome and matrix of a process of social interaction. Even the more or less perfect reproduction of structures is a profoundly temporal process that requires resources and innovative human conduct. But the same resourceful agency that sustains the reproduction of structures also makes possible their transformation” (27). Said another way, there is a dynamic interplay between stability and change, constraints, and autonomy, at the meso-level of institutional life (Feldman and Orlikowski 2011).

Finally, this conceptualization moves beyond seeing organizational, meso-level factors as exogenous to the social system, to be controlled or imported in attempts to ensure fidelity in replication of program technologies. Instead, we posit that the important meso-level structural elements are endogenous to the particular field. They are both shaped and legitimated by human agency within that context. If an “evidence-based” program or management practice is not legitimated by those within the institution, then symbolic or mimetic adoption is the result. Implementation is aborted or subverted rather than seized as an opportunity for actual change in the system (Chi Lin 2000).

This is why, in our mind, it is important to move from thinking about the task of implementation as something technical, focused on preserving internal validity and fidelity to a designer's intent. Instead, the interactive, social process that surrounds implementation is more akin to innovation.

Innovation is a new product or administrative process introduced into a particular context (Lam 2005; Pavitt 2006; Torfing 2016; Van de Ven et al. 2008). Unlike invention, which is the development of something entirely new, innovation is something perceived as new by actors within a particular setting. Thus the concept of innovation is intrinsically tied to the social context. Certainly, while innovations differ in the extent to which they represent a significant departure from what is already being done, they inevitably involve disruption of “established practices and routines” (Ansell and Torfing 2014; Bessant (2005); Torfing 2016; Van de Ven et al. 2008). Research about the diffusion of innovation draws attention to communication and learning (Rogers 2003). Diffusion is a social, adaptive, and collaborative process. It requires skillful attention to power, cultural meanings, and developing shared understanding to motivate collective action (Fligstein and McAdam 2012). In fact, many times, as innovations are taking hold, they actually alter what is understood to be viable within the implementation setting (Van de Ven et al. 2008). Reframing implementation as innovation enables us to see that this type of change requires organizations and the people within them to learn new ways to understand problems and act to resolve them. And, seeing innovation as a process of structuration provides additional purchase for the theoretical ideas described here, and how they relate to other aspects of public management.

The practical implications of this orientation are not that difficult to glean. The process of taking a good idea or intervention from one context and bringing it to another is not one that can overlook the understanding, ingenuity, and skill of people. Frustrated efforts to control the replication of evidence-based programs (Aarons et al. 2012) have led some to try to articulate the “core components” of an intervention. Purveyors—entities focused upon promoting evidence-based programs—try to focus attention on these core components (Fixsen et al. 2005, 2009). Yet it is often difficult to say definitively which components are the most essential. Rather than focusing on technical control, the structuration of implementation suggests more purposive attention to foster learning when we want to bring new policy into practice (Knox et al. 2018; May and Finch 2009).

Learning should not be focused only within organizational boundaries. The implementation system is comprised of larger policy fields (Moulton and Sandfort 2017). Public managers should create collective forums, virtual and face-to-face, where field actors can discuss the merits and limitations of program ideas, explore options, and collectively debate what might be an appropriate action. Methods like “Breakthrough Series Collaboratives” operate in this regard, helping practitioners to consider how research-based program ideas can be integrated into particular contexts (Institute for Health Care Improvement 2003). Public managers can create implementation teams tapping people from diverse perspectives who meet to understand the change being proposed, develop trials, and identify feedback tools (Langley et al. 2009). They can sponsor design processes that use short-term events to bring people together to design or redesign services or products (Bason 2017).

While engagement processes create buy-in to change, more fundamentally they provide opportunities where people together can influence the structural elements that shape implementation in ways that “make sense” in their particular context. For example, this theoretical lens brings into focus a more comprehensive way for purveyor organizations to operate to support the use of evidence-based programs. Rather than promoting a few elements of core practice, their key role is creating and sharing implementation resources—recruitment materials, manuals, and virtual trainings. These resources stimulate leaders in other social settings to learn more, understand more, and help diffuse the innovation (Hill 2003). In the hands of informed, strategic actors, such resources can help convince others to change the rules, routines, or culture necessary to implement policy or program ideals.

Conclusion

In previous work (Moulton and Sandfort 2017; Sandfort and Moulton 2015), we developed the strategic action field framework for public policy and program implementation. This framework applies the social theory developed by Fligstein and McAdam (2012) to provide a new explanation for persistent questions about policy and program implementation. It builds upon multi-level governance (Hill and Hupe 2014; Lynn et al. 2001) to reflect that any analysis of policy implementation is nested within a larger system. Such analysis must focus on scale, because distinct implementation actions occur at different levels. Important to what we are arguing here, there is a nested theory of structuration in which social dynamics and human agency are critical in determining variation in implementation and its outcomes at different levels in a complex system (Perlow et al. 2004).

In this article, we further develop these ideas in relation to an important puzzle of program replication across settings. We argue that human agency is an essential ingredient for understanding how the structural elements of rules, resources, routines, and culture are actually enacted in practice; the structuration process explains how ideas about program technology—both those which have been subjected to rigorous social science testing and those which have not—are integrated into different organizational contexts.

This theoretical approach allows us to both describe what happens and provide more clarity for the practices of those implementing public programs. We do not mean to suggest that it is not important for practitioners to consult social science and evaluation evidence about the programs they are interested in implementing. Rather we want to encourage them to feel comfortable in letting down the elusive goal of implementing evidence-based programs with fidelity. This theory privileges what practitioners know about their context, appreciating their understanding of what is possible within that context, and focusing on the learning that needs to occur to achieve desired results. Rather than emphasizing implementers attempt to control their context, this theory stresses that the actions of implementers shape their contexts. As we further develop that awareness, we are more likely to be able to offer strategic insights about how managers and leaders can affect changes to rules, routines, culture or resources to support the integration of innovation in those settings. In this way, we recognize and engage others in practices of innovation, helping them to shift the constraints that get in the way of improving policy and program outcomes.

References

We would like to thank the participants at the Workshop on Organizational Theory held at the University of Arizona, particularly Mary Feeney, Andy Whitford, Brint Milward, and Martha Feldman who provided specific feedback on the ideas in this article.

Footnotes