Organizational Change and Innovation Processes: Theory and Methods for Research – Book Review
Jan Faber
M.S. Poole, A. H. Van de Ven, K. Dooley and M. E. Holmes: Organizational Change and Innovation Processes: Theory and Methods for Research
2000, New York: Oxford University Press. 406 pages.
This book provides a unique contribution to research into organizational change and innovation processes. It presents a very structured discussion of the methodology of studying these subjects. After an introduction to the book, two major aspects of this research are addressed, namely theory formation (Part I) and research methods (Part II). Part I (Theory) comprises three chapters (88 pages) and Part II (Methods) encompasses six chapters (253 pages). The last chapter of the book summarizes the main results of the study and discusses future directions for process research.
Part I starts with a comparison of three types of research to be applied in the study of organizational change: cross-sectional survey research, panel data research and process research. The first two research designs are criticized by the authors for not providing insight into the causal dynamics of organizational change. Cross-sectional survey research rests upon data on organizational change measured at just one moment in time. Consequently, the ordering of the independent and dependent variables in time remains unnoticed in survey data. The conceptual model of the causation of organizational change tested on these data is therefore an interpretation made by the researcher in trying to explain the variance in organizational change and its consequences within a (large) set of entities.
A solution to this shortcoming of cross-sectional survey research is panel-data research, in which the temporal ordering of causes and effects is preserved in the research design by repeated measurements of relevant independent and dependent variables associated with the organizational change of various entities, at successive discrete moments in time. The validity of the causal effects of independent on dependent variables can now be assessed more reliably. However, these causal effects still constitute a causal black box of change, because they provide no insight into the causal processes at work during the period of time between two moments of measurement. For this reason, the authors advocate process research.
Process research focuses on the identification of events of organizational change in continuous time wherein organizational dynamics occurs. Each event is explained by a narrative story based on theoretical notions. Additionally, those stories are compared to derive generic theoretical characteristics of organizational change. In this respect, process research is complementary to panel-data research and cross-sectional survey research; the generic theoretical characteristics of organizational change can be mapped into the concepts used in panel-data research and cross-sectional survey research.
In fact, two well-known methodological principles are recalled in this study of research into organizational change. First — by introducing a research design capable of tracing time dynamics in the unit of analysis — the introduction of the time dependency of causal effects. This implies moving from a static model to a time-dependent model. Second — by introducing a research design capable of tracing variable time dynamics in the unit of analysis — the introduction of continuous, rather than discrete time in causal models of organizational change.
Part I continues with a comparison of variance- and process-oriented explanations of organizational change. Variance explanations are criticized by the authors for utilizing: (1) fixed units of analysis; (2) both necessary and sufficient conditions for causal effects; (3) reliance on only one form of causality, i.e. efficient causality; (4) context independency; (5) ignorance of the temporal order wherein independent variables exert influences on dependent variables; (6) ignorance of indirect causal effects; and (7) ignorance of varying meanings of the same variable over time. It is argued that process explanations address these issues more properly. Process explanations should therefore be based on narrative, theory-based, stories in which probabilistic laws are applied concerning formal and final causation derived from sequences of human actions.
In fact, the well-known discussion about structuralism versus functionalism in sociology and system theory is repeated here for the study of organizational change. The result is that the latter approach is favoured.
Part I is concluded by the identification of ‘motors of change’, which are argued to be generic mechanisms derived via induction from the narrative stories of events of organizational change. Four motors of change are identified: life-cycle theory, teleological theory, dialectical theory, and evolutionary theory. These four motors of change are thought of as ideal types resulting from the combination of two independent underlying dimensions, namely ‘the mode of change’ (prescriptive or constructive) and ‘the level of change’ (one unit or multiple units). Combinations of 1, 2, 3, and 4 motors of change result in 16 categories, which are related by the authors to established models of organizational change. Introducing the nesting of motors, the differential timing of motors and the complementarities of motors, together with a balance between prescriptive and constructive modes, increase the 16 categories of theories of organizational change even more. These categories of theories of organizational change allow for the explanation of equilibrium, oscillation, bifurcation and catastrophes, and chaos in organizational change. In order to analyze these possible (dynamic) states of organizational change, the following stages in process research are prescribed: (1) identification of organizational-change events; (2) characterization of event sequences and their attributes; (3) testing the time dependency of events in identified sequences; (4) evaluating hypotheses concerning causal relations; and (5) identifying the ‘motors of change’ at work. Methods to be applied during each of these stages are presented in Part II.
However, the four motors of change cannot be conceived as Weberian theoretical-ideal types derived from the four combinations of the dimensions ‘mode of change’ and ‘level of change’. The reason for this is that the dimension ‘level of change’ is not a content-related or theoretical dimension like ‘mode of change’, but a level-of-aggregation-related or methodological dimension. This implies that less aggregated ‘motors of change’ (i.e. the teleological and dialectical motors of change) are complementary to or a part of more aggregated ‘motors of change (i.e. the lifecycle and evolutionary motors); or, in other words, the latter ‘motors of change’ are not theoretically distinct from the first ones. Consequently, the rules of classification developed by Max Weber before 1920 and methodologically grounded by Johan Galtung are not consistently applied.
Part II discusses various methodologies to be applied in process research. First, attention is given to which methodologies should be applied at every stage of the process research mentioned above. Additionally, the representation of (characteristics of) events in a process dataset built on incident measurements is outlined. This is illustrated by the cochlear implant program (CIP) and the therapeutic apherisis technology (TAP) studies undertaken within the Minnesota Innovation Research Program (MIRP; 1977-1989). In these programmes, emphasis is put on three important conceptual categories in the underlying learning model of organizational change: (1) the course of actions performed in each event, (2) the outcomes of events of organizational change, and (3) context events.
Secondly, selected methodological issues involved in the design of process research are discussed. These issues concentrate on three important concerns in process research: (1) formulating the research plan; (2) establishing and validating the observational systems and instruments applied; and (3) the transformation of observational data into useful forms for analysis. Illustrations are again derived from the aforementioned CIP study. In fact, the methodological issues discussed are common issues in the methodology of empirical research to be found in standard textbooks.
The next four chapters of Part II discuss relevant categories of quantitative methods to be applied in process research; i.e. stochastic modelling, phasic analysis, event time series regression analysis, and event time series nonlinear dynamical analysis. All categories of quantitative methods are well-explained at an introductory level and illustrated by clear examples and applications to the CIP data. As this review focuses on the methodology of research into organizational change and innovation processes, the results obtained from the various analyses of the CIP data are left to the reader to explore.
Stochastic modelling is applied to identify and characterize event sequences, and to identify temporal dependencies in event sequences. Two types of stochastic models and their analysis are discussed. First, discrete state/discrete time stochastic models (i.e. homogeneous Markov processes and lag sequential analysis) for nominal level variables representing the occurrence of (different types of) events are described. Secondly, discrete state/continuous time stochastic models (i.e. semi-Markov processes and event history analysis) for ratio variables representing the duration of (different types of) events are described. However, every stochastic modelling technique discussed requires a fairly large amount of data.
Phasic analysis is applied to identify in which phase of the developmental process of organizational change a particular event of organizational change occurs. As individual events are the elementary units in a set of events representing a sequence of events, a phase is a distinguishable sub-set of successive events in this sequence of events. In this respect, phases in the development process of organizational change are macro-level characteristics of micro-level realizations of that process. Depending on whether events are measured by just their occurrence or their duration, different methods, e.g. Markov processes and time series analysis, should be applied to identify phases. Phasic analysis is helpful for identifying different types of events. Additionally, information may be derived about layers of phases and ‘the motors of change’ at work.
Event time series regression analysis (ETSRA) can be applied on frequency counts of events during fixed time periods. However, because of its reliance on statistical methods, the data requirements are quite large. Furthermore, when predictions of successive frequencies of events are found to be dependent on prediction errors, rather complicated estimation methods need to be applied. The strengths of ETSRA are that it produces insight into the dynamic structure, stationarity, predictability and causality of the developmental process of organizational change occurring within a (set of) event sequences experienced by an organization(s). ETSRA applies standard time series analysis methods.
The last category of quantitative methods discussed is event time series nonlinear dynamical analysis (ETSNDA). ETSNDA allows the researcher to analyze whether a time series of frequencies of events, which was found to be a random series by ETSRA, was generated by chaos, coloured noise or white noise. Methods to identify these processes are discussed and illustrated. To help the reader, a flow diagram is included of the sequence of analyses to be performed (Fig. 9.16: 339). Furthermore, the ‘motors of change’ are associated with four dynamical patterns in a time series of frequencies of events. These patterns are: static equilibrium (life-cycle theory), periodicity (life-cycle theory and/or teleological theory), coloured noise (life-cycle theory, dialectical theory and/or evolutionary theory, and chaos (life-cycle theory and/or dialectical theory). However, the results obtained by the ETSNDA methods do not uniquely discriminate one ‘motor of change’ from another, thereby leaving room for doubt and choice.
There are two critical issues to be discussed in this book. First, there is the problem with the application of nonlinear analysis to event time series. Secondly, there is a problem with the unit of analysis chosen, i.e. an event of organizational change.
The discussion of nonlinear analysis in this book assumes that the event time series are measured validly and reliably. However, as is known from the social and organizational sciences, such measurements almost always, for various reasons, contain measurement errors. Then it becomes seemingly impossible to identify the underlying generating mechanism without a theoretical model of nonlinear behaviour, It Would be useless to explore a seemingly random time series of events. The only thing that might be fruitful, would be to test a theoretical model of nonlinear organizational change on such a time series using the methods discussed in the book (Faber and Koppelaar 1994).
The unit of analysis chosen in the book, i.e. an event of organizational change, is not beyond doubt. As argued by the authors, such an event is the outcome of sequences of human actions and contextual conditions. These latter conditions influence the sequences of human actions undertaken. A narrative story of these sequences of human actions under particular contextual conditions is, then, not just an explanation of the event of organizational change. Organizational dynamics over time, wherein the event of organizational change is just a particular state of these dynamics, due to human decision making and related (goal-oriented) behaviour within a social network, delimited by the organization or of which the organization is a part, provides a better foundation for explanation. The explanation of individual human decision-making and related behaviour in social networks appoints the human individual as the unit of analysis to be chosen in the study of organizational change (Faber and Scheper 1997).
Despite the critical comments made, this book discusses the most fundamental aspects of research into organizational change and innovation processes in a very structured way. Put differently, it is because the book is methodologically so well organized that these comments can be made and various ways of improving research into these subjects can be indicated and elaborated further. Therefore, I can recommend reading this book to every researcher in this field of interest.
References
Faber, J. and H. Koppelaar
1994 ‘Chaos theory and social science: A methodological analysis’. Quality & Quantity 28: 421-433.
Faber J., and W.J. Scheper
1997 ‘Interdisciplinary social science: A methodological analysis’. Quality & Quantity 31: 37-56.
COPYRIGHT 2002 Walter de Gruyter und Co.
COPYRIGHT 2003 Gale Group
