Process-Based Research

Let us go back to the metaphor of video. If content-based research is com­parable to hitting the pause button, then the film is rolling again in process- based research. The time dimension is placed at the heart of the managerial issues studied. That which researchers intend to describe and understand (for example, decision-making or strategic changes in organizations) is opera­tionalized in the form of a variable whose evolution, transformation or change is being studied. The dynamic and temporal aspects are primordial here.

Beyond this general characteristic in relation to evolution, process-based studies do not form a homogenous body of research. On the contrary, they are extremely diverse from both a theoretical and a methodological point of view (Pettigrew, 1992).

1. Why Choose Process-based Research?

1.1. The goals of process-based research

Process-based research describes and analyzes the way in which a variable evolves over time (Van de Ven, 1992). For example, a researcher’s goal may be to analyze how a strategic decision is taken within an organization, to under­stand how an idea takes shape and turns into strategic innovation or to grasp how the firm acquires knowledge.

In order to study ‘how’, researchers may need to reveal the evolutionary path of the variables over time (Monge, 1990). In this way, they can measure a variable’s time span (how long it is present for), periodicity (is the observed variable’s behavior stable over time or not?) or else its evolutionary trend (is it decreasing or increasing over time?).

But studying a process should go beyond that. Reconstituting the evolution of a variable should lead to revealing the different ‘time intervals’ that make up its process and articulate its evolution over time (Pettigrew, 1992). The process then appears as the sequence of change of an organizational variable (Miller and Friesen, 1982). Thus process-based research enables researchers to identify and articulate intervals such as sequences, cycles or phases that describe a vari­able’s behavior over time (refer to Chapter 15 in this book for definitions of the terms sequence, cycle and phase). One delicate task consists of naming these intervals, in order to explain the process under study in as precise and helpful a way as possible. For example, Miller and Friesen (1982) suggested a momentum- revolution model to explain change in organizations. The terms ‘momentum’ and ‘revolution’ are explicit. The first describes a long period of steady, contin­uous evolution within an organization, while the second applies to an (often brief) period of radical upheaval.

Finally, process-based research may aim to describe or explain the studied object’s evolution over time.

1.2. Descriptive and explanatory process-based research

Descriptive process-based research Describing a process leads a researcher to pay particular attention to the elements the process is composed of, as well as their order and sequence over time. Observing its constitutive variables is here central to a processual analysis with a descriptive purpose.

Three main (and complementary) goals can explain why researchers carry out descriptive process-based research.

The first goal is to develop an in-depth description over time of the object under study. The value of this description is based on the value of the collected data and the identification of dimensions or subvariables that will be useful in understanding the process. A researcher can thus demonstrate regularity – patterns or configurations – in the process, and identify and name the sequences and phases it is composed of.

Describing the process itself, as studies about organizational change (Pettigrew, 1992) encourage us to do, can be another goal of process-based research of a descriptive nature. Historical studies of the structural and strate­gic development of organizations (Chandler, 1962) also serve this purpose. The environment is taken into account not to explain how a phenomenon came about but rather to situate it in the context of the collected information.

Finally, researchers may wish to compare two or more observed processes. In this way, in his work on decision-making processes, Nutt (1984) compared 78 processes to find patterns and identify different temporal developments within the decision-making process. The work of Mintzberg et al. (1976) on non­structured decision-making processes corresponds to the same research objective.

Example: How is decision-making structured over time?

(Mintzberg et al., 1976)

Using a study of 25 decision-making processes, the authors break the processes’ temporal development down according to the activities it is composed of. They pre­sent a very general model combining three phases: identifying, developing and selecting. This model can be articulated seven different ways, depending on the nature of the solution adopted.

Explanatory process-based research The goal of analyzing a process can be to explain the phenomenon observed – to explain how a variable (the object under study) evolves over time, relative to the evolution of other variables. In this case, researchers attempt to answer the following question: ‘Would an evolution or modification of variable X be related to or necessarily imply a modification of variable Y?’

Example: How did Intel decide to abandon the computer memory sector? (Burgelman, 1994)

Burgelman wanted to understand the decision-making process whereby, in the mid-1980s, the Intel firm stopped researching and producing computer memory in order to concentrate only on microprocessors. The author began by investigating the evolution and dynamics of the variable ‘the process of deciding to leave an industrial sector’. He wanted to describe this variable, and to understand its evolu­tion particularly in terms of the skills that distinguish Intel from the competition. Strategic and industrial analysis calls for researchers to understand the distinctive competencies a firm has that may provide sources of advantage over the competi­tion in the firm’s economic sector. Burgelman followed the evolution of these competencies in the field of electronic components (microprocessors and memory chips) over the period studied. The author tried to understand how a sector’s evo­lution can influence a firm’s decision to leave it. Burgelman’s work is based on a logic of co-evolution. An organization’s strategy (the decision to abandon an indus­trial sector) appears as the result of a co-evolution between an external, sector vari­able (a sector’s competitive advantages) and an internal, organizational variable (the firm’s unique skills). Only process-based research allows for perception of the dynamic of this co-evolution.

2. Conducting Process-based Research

Let us look at the main stages of process-based research through the following example. While this example illustrates a possible modus operandi for researchers wishing to describe or explain the phenomena they are studying, it should not be taken as the model to follow. Researchers may have very good reasons for adopting a different research design. This example does provide, however, an interesting illustration of what a process-based study can be. By setting forth the main steps in the research process, it enables us to enumerate the main problems researchers are likely to encounter.

2.1. How is process-based research conducted?

The example, based on the work of Van de Ven and his team of researchers (Van de Ven et al., 1989), illustrates descriptive process research.

Example: How do innovations appear and develop within an organization? (Van de Ven et al., 1989; Van de Ven and Poole, 1990) – a descriptive process-based study

Van de Ven and his team wished to describe very concretely ‘the temporal order and the sequential steps involved when innovative ideas are transformed and implemented into concrete reality’ (Van de Ven and Poole, 1990: 313). A major research program was launched at three different sites. Data collection and analysis were articulated around the four main steps described below.

The first step in the research process was to specify the study’s process variable (the innovative process, or the birth, transformation and implementation of new ideas into reality).

The second step was to clearly define the period of observation and the observa­tion sample.

The third step consisted in defining the core concepts, or subvariables, which would permit the researchers to observe the evolution of the ‘innovation’ variable. Five core concepts were defined – actors, ideas, transactions, context and results. These core concepts demonstrated the manner in which the authors defined the innovative process in these organizations. Through them, the researchers were able to follow and characterize the ‘innovation’ variable over time. The authors broke down the observed history of an innovation into its critical incidents, and described and studied each incident from the perspective of the five core concepts. Each inci­dent was the subject of a binary analysis. The five core concepts were coded 0 or 1, depending on whether the actors, ideas, transactions, context or results of the inno­vation changed over time or not. This breakdown and encoding of the innovation’s history over time was based on the principles of sequential process analysis.

The fourth, and final, step consisted of classifying critical incidents and deter­mining phases of the processes, which enabled researchers to follow the develop­ment of these innovation processes over time.

Through their research, the authors were able to describe how an innovation process unfolds within an organization, by breaking the longitudinal history down into phases and describing each phase in terms of the evolution of the variables ‘ideas’, ‘actors’, ‘transactions’, ‘context’ and ‘results’.

The type of research described above results in reconstructing history over time and permits us to ‘describe’ what happened.

2.2. The main stages of process-based research

The study presented above demonstrates the following principal stages:

  1. The researcher breaks down the process variable he or she is studying into core concepts (or subvariables). This first decomposition stage permits the researcher to become familiar with the process under study, and to follow its evolution through its constitutive elements. The researcher is at this point confronted with the problem of how to decompose the process variable.
  2. Once the process variable has been decomposed, the researcher will study and describe the research subject over time. The researcher will follow its evolution through the different dimensions of the concepts that make up its process. During this essential stage, researchers may have some difficulty in delimiting the process under study. This delimitation is, above all, tem­poral. The researcher is confronted with the problem of deciding when the phenomenon being studied actually begins and ends. The question of delimitation must then be seen in terms of the subject of the research. Researchers who wish to observe, for example, the decision-making process within organizations, will soon realize that the decision whose develop­ment they intended to track turns out to be inextricably bound up with other, concomitant decisions (Langley et al., 1995). The researcher has to decide how to isolate the single phenomenon under observation from a myriad of other phenomena, as the firm is functioning and changing throughout the observation period. The question of delimitation must therefore be considered in terms of both the internal and external contexts in which the process is taking place. Researchers are faced with the ticklish problem of needing to take several contexts into account (on several levels of analysis: personal, organizational and environmental), and of having to absorb a tremendous amount of data relative to actors, the organization and its external environment.
  3. The researcher will need to identify the critical incidents, and analyze and classify them in order to reveal the temporal intervals affecting the process’s development. The researcher will then be confronted with another problem – that of having to articulate the identified ‘intervals’ over time, relative to each other. These intervals will often appear to overlap to the point that it can be extremely difficult to separate them from one another, or they may follow one after another in exceedingly variable ways, depend­ing on the organizations studied. The studied process may follow an evolu­tion that may be, to a greater or a lesser extent, anarchic, non-linear or complex.

3. Troubleshooting Process-based Research

Three principal problems confronted in process studies are in the areas of:

(a) recognizing, and then decomposing the process variable to be studied;

(b) delimiting the process under study, and (c) arranging temporal intervals over time (that is, reconstituting the studied chronology).

3.1. Decomposing the process variable

The process variable will remain abstract if it is not broken down into the ele­ments (or subvariables) that participate in its development over time. The essential problem is that of deciding which elements should be included. Within the framework of purely inductive reasoning (Glaser and Strauss, 1967), researchers use empirical data to reveal ‘sensitive concepts’ – concepts that give meaning to (or make sense of) the information that has been collected – and the different dimensions they may take. With other methodologies (par­tially inductive or deductive), the researcher will draw on existing work as well as empirical data to construct a conceptual framework that will integrate all the constitutive subvariables of the studied variable.

Most process studies are based on generally accepted codification systems for longitudinal studies. These codification systems are not themselves depen­dent on the contents of the study – they define the chief domains within which the codes should be established empirically. Miles and Huberman (1984a) propose several codings.

A researcher may not wish to decompose the process variable in such a detailed manner, preferring to adopt a more general coding. For instance, in their study of the decision-making process, Mintzberg et al. (1976) decided to decompose the 23 processes they studied solely in terms of different activities (which they called routines). By this coding they identify seven different modes of decision-making.

Researchers generally articulate their codification of the processes they study around the following three generic concepts: the actors involved, the activities taking place and contextual elements. Using this as a broad framework, researchers may choose to focus particularly on one of the three categories, depending on their subject’s particularities or the objective of their study.

3.2. Delimiting a process in time

The question of delimitation is twofold; it is posed both in terms of time and in terms of the subject and context of the study.

Process-based research aims to describe and analyze the evolution of a variable across time – to describe what happens between a given point in time (T) and a later point (T + n). But it is not always easy for researchers to establish the starting point and the end point of a phenomenon. Organizations are constantly in flux: making decisions, moving, hesitating,
deciding, going forward then questioning the idea, etc. The decision to proceed with a structural change may be preceded by a long maturation period, the beginning of which is often hard to identify. Key actors may begin to discuss the idea of ‘change’ informally among themselves, before addressing the firm’s executive management in an equally informal manner, and before a single word has ever been written about such an important issue. Can ideas discussed among staff members at the beginning of the process of structural change be taken into account, or are we dealing with nothing more than the usual exchange of ideas and opinions which we all express at our place of work? This question of temporal delimitation is important for two reasons. First, it obliges researchers to decide when to begin collecting data. Second, the way in which the beginning of the process is defined can influence our interpretation of the process itself.

Example: How the way a process is delimited in time can influence its analysis

In studying a change in a firm’s strategic vision, seemingly initiated by a new actor, defining the beginning of change before or after that actor’s arrival at the firm may lead to different analyses of the process. In the first case (situating the beginning of the change of vision before the actor’s arrival), explaining the process depends less on an actor-based logic and more on a systems-based logic, related to the organiza­tion itself (the organization is viewed as a system, itself made up of different sys­tems). Whereas, in the second case (situating the beginning of the change of vision after the actor’s arrival), the process is explained on the basis of the actor’s capacity to induce the emergence of new representations and a new vision within the organi­zation (Gioia and Chittipeddi, 1991).

To address this issue, Hickson et al. (1986) recommend following a process ‘from the first deliberate action that initiates movement towards a decision (when, for example the subject is discussed at a meeting or a report is requested) until it has been approved (when the decision and its application are made offi­cial)’ (Hickson et al., 1986: 100). Researchers may also form their own opinion based on interviews with organizational actors in which a reconstruction of the process may be solicited. Finally, we recommend that researchers do not hesi­tate to go as far back as possible into the organization’s past and to collect even seemingly out-of-date data. A good knowledge of the firm’s past can allow researchers to judge whether a simple, informal announcement might indeed indicate the beginning of a process, or if it is no more than a common remark for the firm in question.

Delimiting a process in time poses one other major problem. Time is a relative concept. An individual’s time frame may be quite different from that of an organi­zation. The more closely we study daily events the more likely we are to notice change. Conversely, the more we study an event as a whole, going back to its origins, the more likely we are to perceive continuity (Van de Ven and Poole, 1990: 316). There are no hard and fast rules defining the ‘right’ level to observe a process from. To compensate for this, it is often recommended to adopt an observation perspective including multiple levels of analysis. In this way, an organization’s evolution can be studied alongside the actions taken by key actors within it.

3.3. Arranging temporal intervals

The time intervals that make up the process of a variable will usually emerge through studying the variable’s evolution.

Before arranging these intervals logically, researchers may run into difficulty in deciding how many intervals are applicable to their processual model. Hannan and Freeman’s (1977) ecological model, for example, is based on three phases (variation, selection, and retention) while Miller and Friesen’s (1980) model of organizational change is constructed around just two phases (momentum-revolution). Other models are based on a more detailed evolution of the process studied. Pounds (1969), for instance, constructed his processual model of the emergence and resolution of strategic problems within an organi­zation around eight phases:

  • Choosing a resolution model.
  • Comparing it to the actual situation.
  • Identifying the differences.
  • Selecting a particular difference.
  • Considering alternative operators (solvers).
  • Evaluating the consequences of the choice of operators (solvers).
  • Choosing a particular operator (solver).
  • Putting the chosen operator (solver) into action in order to resolve the strategic problem.

Studies seem to waver between a low number of intervals, which make it easier to comprehend the pace of a process, and a larger number of intervals, enabling a more detailed explanation. The question of how many intervals should be used in building a processual model remains largely up to the researchers’ judgement. It will depend on the level of detail the researcher intends to give in describing the temporal arrangement of the studied process.

Once the time intervals have been identified, the researcher is confronted with the problem of their articulation over time. The researcher must consider whether these time intervals follow one another or whether they overlap, with a second appearing before the first is really over. Different models of processual development have been presented in organizational literature.

Models of management processes abound in management literature. Such work provides researchers with representations of different ‘types’ of evolution that the process they intend studying may follow over time. Researchers may decide to adopt a particular model before starting their research, or they may attempt to use their data to establish a model. This choice will be dependent on whether the researcher adopts an inductive or deductive position in conduct­ing the research (see Chapter 3).

Source: Thietart Raymond-Alain et al. (2001), Doing Management Research: A Comprehensive Guide, SAGE Publications Ltd; 1 edition.

Leave a Reply

Your email address will not be published. Required fields are marked *