We can extend traditional ideas about the execution of case studies (
Eisenhardt, 1989;
Yin, 1994) by applying the blueprint of complexity science. This will lead to new research strategies for fruitfully using case studies in health care settings. In this section, our purpose is to identify several of the potential extensions of case study design. By no means do we claim to have exhausted these potentials. Nor do we wish to suggest that conventional understandings of case study research need to be discarded. Rather, we present the case study as a research approach uniquely suited to carrying out a study designed from a blueprint of complexity theory. The case study strategy with these extensions becomes a powerful tool for increasing our understandings of health care. These extensions are as follows:
• Understand interdependencies
Through complexity theory we recognize that systems do have elements but it is the interdependencies and interactions among the elements that create the whole. Thus complexity theory suggests that studying the interdependencies and interactions among the elements, as well as the unity of the system itself (
McDaniel, 2004;
Price, 1997), will provide critical insights for understanding an organization and its system properties. Identification of these interdependencies requires prolonged engagement with the system. Actions are interdependent with actions. Ideas are interdependent with ideas. And, importantly, actions are interdependent with ideas. Our tendency in case studies is to isolate actions and ideas, that is, we describe them independent of each other. To understand the system, however, requires that we understand these interdependencies (
Capra, 1996;
Lee, 1997). Thus, when we see either a discrepancy or a consistency between ideas and actions, this is a cue to search for and describe the underlying interdependencies. For example, the first author and colleagues collected in-depth case study data over a six-month period from a nursing home revealing that nursing assistants held child care/rearing as a guiding mental model of a patient’s behavior and thus interpreted a patient’s crying, not eating, and taking to the bed as a temper tantrum. The nurse aides acted accordingly by giving her a “timeout.” Understanding the nurse aides’ mental model (ideas) shed meaning on the action; it makes sense to give a time out for a temper tantrum, a standard child-rearing practice. However, in isolation, the action appears thoughtless and cruel. The case study method with the blue print of complexity science, revealed this interdependency through direct observation combined with interview methods that explore the participants explanations, and analysis that paid attention to the interdependencies between thought and action.
Further, because of the co-evolutionary nature of the system, we must pay more attention to the interdependencies across the boundaries of systems. Traditionally case studies bound the case and then study phenomenon within the boundary. Complexity science suggests that important insights can be gleaned by studying the behavior that occurs at and across the boundaries that define the case. For example, in another nursing home in the nursing home study mentioned previously, interdependencies were identified between external regulators (surveyors), the nursing home, and the resultant relationships between managers and staff. A history of multiple survey deficiencies coupled with frequent surprise visits from surveyors caused the nursing home managers to believe that the surveyors held a bias against the facility because of past poor performance and that they were citing them for things that would be overlooked on a nursing home with a better history. In other words, the regulators were co-evolving through interaction with the facility over time. In turn, the managers constantly monitored nursing home staff for rule violations with the strategies of correcting behavior. As a result, staff described the nature of their interactions with managers as “scolding” and “chewing out.” Morale was low and turnover was high. The managers had difficulty seeing beyond the regulatory issues to other important aspects of managing the nursing home. Here we thus suspect that the interdependencies across external boundaries were co-evolving with the relationships within the facility and knowing the system at this level, enabled better explanation of internal behaviors. These findings were revealed through direct observation and interviews with multiple agents at multiple levels in the system as well as review of survey reports. In addition, the analysis allowed for synthesis such that the patterns were revealed.
• Be sensitive to dimensions of relationships
There are several dimensions of relationships you want to be sensitive to and you should decide ahead of time which may be important for your research questions while also remaining open to the unexpected. Example dimensions are mindfulness (
Weick, Sutcliffe, & Obstfeld, 1999), heedfulness (
Weick & Roberts, 1993), loose/tight (
Granovetter, 1973;
Papa, 1990), quantity (
Kauffman, 1995;
McKelvey, 1999), and quality of connections (
Daft, 1989;
Thompson, 1967). When we use complexity science, we need to have richer understandings of relationships in our case studies. Traditionally, we have looked for rich understandings of the elements in the case. We also must pay attention to the ways in which elements are similar to or different from each other. This means that we must pay attention to system diversity on a wide variety of dimensions (not just race and gender) and try to understand how that diversity might help the organization and how it might hurt the organization (
McDaniel & Walls, 1997).
For example, building on the nursing home case example above about the crying patient, the registered nurse (RN), holding a clinical mental model of the patient’s behavior (crying, not eating, taking to the bed) would likely have considered it a symptom of depression. Thus, had the RN been aware of the patient’s behavior, she would have investigated to see if it was possible to rule-out depression as the primary cause of the observed behavior. However, there were several barriers to the RN detecting this issue. First, sparse interaction occurred between the RN and nurse aides and hence the RN was not likely to just stumble onto the relevant information. The nurse aides, while they would report certain things such as an elevated temperature, did not report this behavior because it was clear to them that it was a behavioral issue that they could manage without bothering the nurse—a concern expressed by the nurse aides that caused them to censor their interaction with the nurse. Finally, the RN does not recognize interdependency between her role and that of the nurse aides and thus did not actively seek out what the nurse aide “knew” about the patient. In this study, the researchers observed that the two types of workers (nurse and nurse aide) held very different mental models of the patient’s behavior. Because of the nature of the relationships in the nursing home, the diverse views were never explored together leading to a potentially poor outcome for the patient. In this case study design, the researchers stated the goal of understanding the nature and quality of connections among agents, however when the analysis revealed aspects of mindfulness (connection between thought and action), this dimension was added. Relationship patterns were assessed through direct observation of multiple processes (e.g., shift change reports, care planning meetings, direct care routines), shadowing the nurse aide and the RN while they worked, and through depth-interviews where explanations were obtained from the agents about their actions and thought processes.
• Focus on nonlinearities
It is difficult to detect nonlinearities. Therefore, try to look for instances where small events have led to large outcomes. For example in one of the nursing home case studies, a patient’s daughter had a habit of leaving post-it notes stuck all over the patient’s room with instructions to the nurse aides about things such as laundry, placement of personal items, and meal preferences. Rather than seeing the notes as useful information for the patient’s care, the nurse aides were highly insulted and viewed the action as the daughter trying to be “the boss” of the nurse aides. Significant staff time (multiple levels of managers as well as nurse aides) was invested in talking about the issue and meeting with the daughter to try to get the daughter to stop posting notes. The issue became so disruptive that it was suggested that the patient find another nursing home. Thus, the daughter’s seemingly “small” act of leaving notes to the staff resulted in a disproportionately “large” outcome of the daughter being asked to move her mother to another nursing home.
In contrast, examine nonlinearities by looking for instances where large events have led to small outcomes. A one nursing home in the nursing home case study, for example, turnover of the nursing home administrator was a seemingly large event (i.e., it occurred three times in just over a year) but seemed to have a disproportionately small impact on the nursing staff working on the patient units. The staff’s explanation was that they could and would outlast any administrator, and thus had developed a resistance to change efforts of each new administrator. Why bother doing what he wants when he’ll be gone soon?
Because nonlinearities are keys to understanding the system the researcher must be paying attention in ways that they will be noticed. The case study method allows such nonlinearities to be explored.
• Look for the unexpected
We must ask ourselves what potentially useful behaviors, processes, and outcomes are we missing because we were only looking for outcomes we had predicted? Heisenberg’s Uncertainty Principle demonstrates in experiments that when we measure one aspect of matter, other aspects are less observable. “Matter’s total identity (known as a wave packet) includes potentialities for [two] forms—particles and waves….We can measure position, and thus get a fix on the particle aspect; or we can study momentum, and observe the wave. But we can never measure both simultaneously” (
Wheatley, 1992, p. 35). This suggests that research intended to understand how health care organizations evolve successfully will need to use multiple lenses (methods) to observe it from more than one position and time period. The case study method lends itself to multiple lenses across time. For example, multiple lenses can be used by observing and interviewing people at all levels of the organizations (e.g., patients, nurse aides all the way to the top administrator) and across disciplines (e.g. nursing, food service, social services, housekeeping) asking about the same phenomena. The case method is particularly useful in identifying the unexpected because the researcher is in the field and can ask the agents what about the system has surprised them or caught them off-guard, providing new targets for understanding the system dynamics. Traditionally research has focused on average behavior and thus other events (including unexpected events) are considered anomalous and outliers to be ignored. Complexity theory, however, suggests that it may be fruitful to pay greater attention to outliers because they may be a source of new structural arrangements and patterns of behavior. Thus, in choosing cases for comparison, it is often useful to look to the extremes—comparing the very best with the very worst (
Anderson, Hsieh, & Su, 1998).
• Examine unexpected events
Deeper understanding of the organization can be gained by a search for actions taken in the organization that deviated from the “plan.” Successful organizations are often those in which people are attentive enough to improvise—that is deviate from plans or routines—when events suggest that some new or different behavior is needed (
Eisenhardt & Tabrizi, 1995). One of the ways that people treat the unexpected is to normalize it (
McDaniel et al., 2003). The case study researcher therefore must be careful not to accept explanations that normalized something that initially was unexpected. For example, engineers called the failure of the o-ring on the Challenger space shuttle the normal way that the o-ring behaves rather than a potential source of disaster (
Vaughan, 1996). The case study researcher must see disruptions in the state of the systems as an opportunity rather than a distraction or barrier to the research. Be sure to try to detect the nature of the organization’s response to uncertainty. In particular, to what extent do they try to control uncertainty versus leverage uncertainty and what strategies do they use? Look for examples of creativity (
Guastello, 1995;
Jones, 1997;
Stacey, 1992), improvisation (
Brown & Eisenhardt, 1997;
Crossan, 1998) and bricolage (
Weick, 1993a), as well as rules, policies and procedures. Complexity science suggests that rules have less relevance than we traditionally thought while creativity has more relevance than traditionally thought.
• Focus on processes as well as events
Case studies traditionally search for decision points as major events for revealing the nature of the organization. Complexity science suggests that you should look instead for sense making properties as revealing the nature of the organization (
Weick, 1995). Pay attention to sense making as a process not just decision making as an event. Complexity sciences ask that we focus on processes. In the example above, if the researchers had not explored the sense making process of the nursing aides, links between thought (child care/rearing guiding mental model) and action (timeout) would not have been revealed and a potential conclusion for the event might have been that the nurse aides were thoughtless and cruel. Instead, much richer patterns were revealed with better potentials for intervention. Researchers usually try to understand what an organization knows, but from a complexity viewpoint, we are more interested in how an organization learns. For example, how are errors treated (
Edmondson, 1996)? How are samples of one turned into learning opportunities (
March, Sproull, & Tamus, 1991)? What is the balance between exploration and exploitation (
Levinthal & March, 1993)? Treat conflict in the organization as part of the routine ebb and flow rather than as a disruptive event (
March, 1958).
• Recognize dynamics
Self-organization and emergence are ongoing dynamic properties of organizations. You must not let the formal organizational documents and policies mask the nature of the organization, which is defined by the informal organization. The organization, thought of as a verb rather than a noun (
Weick 1993b), is not something that is; it is something that is becoming. Applied to health care organizations, the concept of emergence will draw the researcher’s attention to such things as the “informal” organization. The informal organization is emergent because it is defined as “spontaneously occurring organizational events, structures, processes, groups, and leadership that occur outside of officially sanctioned channels” (
Goldstein, 1999, p. 65). Complexity theory is a guide to learning about the ways in which the informal organization evolves and the adaptive functions (or destructive functions) it performs for an organization. Other emergent phenomena in health care organizations might include leaders that emerge in work groups and the unexpected configurations of health care networks that have emerged through mergers and/or acquisitions. The case study method is well suited to recognizing dynamics because the method facilitates exploring the informal organization. In particular, using strategies of participant observation of agents’ interactions and processes, the dynamics of the informal organization will quickly emerge.
Within the case study, the use of social network methods (
Borgatti, Everett, & Freeman, 1999) is a strategy for measuring actual communication flows that occur, whether they result from formal or informal mechanisms (
Morrissey et al., 1994). Thus, these measures may assist in describing relationship patterns. Relationships represent the ways in which work is carried out and are the conduits for understanding what is to be accomplished. Also, network analysis methods can characterize these patterns for each person in an organization and for the organization as a whole. For example, network analysis can assess: 1) the nature of new information flows through an organization; 2) the density and intensity of those flows (
Bovasso, 1996); 3) how monopolized or centralized those flows are (
Rowley, 1997); and 4) the extent to which a small number of groups comprising cliques of individuals can keep information from diffusing through the organization, creating fragmentation (
Cott, 1997). These measures are just a few examples of how organizations and individuals can be characterized in terms of actual social processes.
• Describe patterns as well as events
Research observations that target patterns of relationships, interactions, and processes, over time, are keys to understanding a system (
Capra, 1996;
Lee, 1997). A search for patterns implies attention to the flow of behavior within organizations rather than merely describing static behavior (
Camazine et al., 2001;
Goldstein, 1999). As an example, when one enters a particular nursing home, invariably it is apparent that it belongs to the class of organizations called nursing homes and not to the class of organizations called family practices. Nursing homes have regularities in their characteristics that make them recognizable as nursing homes. Despite such macro-level regularities, however, internal processes differ significantly from organization to organization (
Tallia et al., 2003). Particularly important patterns are likely to be found in the relationships among people in the organization and the ways they interact (
Watts, 2003). In the nursing home case study example above, describing the pattern of relationship between the nurse aides and the RN provided important information for understanding the event (i.e., the timeout). By using the case study methods with attention to relationship patterns, results were richer and provide more avenues for potential intervention.
• See patterns across levels
Complexity theory suggests that a health care organization is best understood as a system and that a system is best understood as nested within of a larger network of systems (
Watts, 2003). The same holds true for individual people or units within a health care organization. There is likely to be a fractal (
Liebovitch, 1998) or self similar set of relationships between phenomena at different levels of the organization. The example above in which the surveyors were making surprise visits to the nursing home, finding fault and making citations is a macro-level pattern that is similar to the pattern at the subsystem level in which the managers were making frequent rounds, finding fault with staff behaviors and making corrections. Case studies can be designed to look for this self-similarity in analyzing patterns.
• Understand that patterns change
Traditional case study research design seeks to identify trends and trajectories. Case study designs using a complexity science blueprint will also seek to discern patterns in the behaviors and would recognize that the patterns themselves may well change over time. For example when doing a case study to help understand nurse behaviors and the pattern of pain medication, it is useful to examine patterns of use across patients rather than the individual use by a patient. We might find that patients who have advocates might have a different pattern of pain medications use than those who don’t have advocates.
• Recognize that in any given situation different patterns may be successful
Because the nature of a complex adaptive system emerges through self-organization and has the property of equifinality (
Knight & McDaniel, 1979), when more than one case is studied, more than one successful configuration is likely to be found. In health care, much value is placed on identifying and disseminating “best” practices. Complexity theory suggests, however, that there may be more than one way for organizations to be successful. In research, if we seek that one best answer, we will probably find it. Research that is open to more than one way of looking at situations however will lead to more useful knowledge. There is likely to be more than one successful process, structure, or configuration of processes and structures (i.e., patterns of organization) within any complex adaptive system. Because case studies are designed first to describe the uniqueness of each case, it is a method that is suited to finding multiple successful patterns.
• Shift foreground and background
Creating new views of organizations is a key to a better understanding of them. Using a model with boxes and arrows as a metaphor for shifting foreground and background,
Lissack (1999) describes the organizational chart as a model of boxes with lines between them. He suggests that traditionally we put most import on the boxes, which define roles and formal organizational position. Shifting, however, and placing most import on the lines between the boxes will bring to life the “relations, flows and exchanges” (
Lissack, 1999, p. 120) represented. The case method can facilitate shifting foreground and background multiple times during a research study. For example, examining the system with the patient at the center will reveal certain issues and then shifting and examining it with the physician at the center will reveal other issues that are most likely linked to the patient issues through system processes. Shifting foreground and background is another way to change the lens used to study the same phenomena.
• Redefine observer roles
Treat the case study researcher as an intruder who is providing an opportunity to observe how the system dynamic unfolds as it adapts to that intruder. This idea goes beyond the idea of research rigor in which reflexivity and relationality are addressed through “attention to making the effects of interactions of investigators and participants more transparent during data collection and analysis” (
Hall & Callery, 2001, p. 270). It suggests that responses to the researcher or research process can provide considerable information about the nature of the system itself. For example, in a nursing home case study introduced above, one of the team members interviewed the medical director about practice guidelines used in the nursing home. He indicated that they were not currently using them and had not previously considered using them. Soon after, when the researcher interviewed the nurse about practice guidelines, the nurse indicated that the medical director had just suggested that using practice guidelines might be a good idea. Thus this system responded to ideas introduced by the investigators, which could be a distinguishing factor for this nursing home if others in the study do not demonstrate uptake of ideas in this way.
Recognize also that because of the coevolutionary nature of complex adaptive systems the role of the observer changes over time as a result of the fact that the system changes and the system changes as a result of the observer’s presence. Observing these coevolutionary changes is a rich opportunity for gaining insights into system dynamics.
• Learn the system’s history
What the health care organization is today is in large part due to what it was yesterday. In complexity theory, this phenomenon is referred to as interdependency of present and past. Thus learning how the system has evolved over time will provide insight into its present patterns of behaviors. Take for example the case study (described above) in which the nursing home is playing out patterns that are linked to its history of very poor survey results. In describing the system’s history, significant events are important. But true understanding of the system will come from describing its configuration of relationships over time (
Capra, 2002;
Stacey, Griffin, & Shaw, 2000). Using the case study method, this suggests studying how managers and staff have historically related to each other within the organization and to people outside the organization. Additionally, it suggests exploring what types of relationships have been most intense, relied on in crisis, or relied on when thinking about what to do next.
The publisher's final edited version of this article is available at 