Multi-Level Issues in Strategy and Methods: Volume 4

1. INNOVATION IN ORGANIZATIONS: A MULTI-LEVEL PERSPECTIVE ON CREATIVITY

Fred Dansereau is a Professor of Organization and Human Resources in the School of Management at the State University of New York at Buffalo. He received his Ph.D. from the Labor and Industrial Relations Institute at the University of Illinois with a specialization in Organizational Behavior. Dr. Dansereau has extensive research experience in the areas of leadership and managing at the individual, dyad, group, and collective levels of analysis. Along with others, he has developed a theoretical and empirical approach to theorizing and testing at multiple levels of analysis. He has served on the editorial review boards of the Academy of Management Review, Group and Organization Management, and Leadership Quarterly. Dr. Dansereau is a Fellow of the American Psychological Association and the American Psychological Society. He has authored eight books and over 70 articles and is a consultant to numerous organizations, including the Bank of Chicago, Occidental, St. Joe Corp., Sears, TRW, the United States Army and Navy, Worthington Industries, and various educational institutions.

“Multi-Level Issues in Strategy and Methods” is Volume 4 of Research in Multi-Level Issues, an annual series that provides an outlet for the discussion of multi-level problems and solutions across a variety of fields of study. Using a scientific debate format of a key scholarly essay followed by two commentaries and a rebuttal, we present, in this series, theoretical work, significant empirical studies, methodological developments, analytical techniques, and philosophical treatments to advance the field of multi-level studies, regardless of disciplinary perspective.

Recognizing the impact of innovation on organizational performance, scholars from a number of disciplines have sought to identify the conditions that make innovation possible. Although these studies have served to identify a number of key variables, the relationship between these variables and innovation is complex. In this chapter, we argue that the apparent complexity of these relationships may be attributed to cross-level differences in the requirements for innovation and the existence of complex interactions among the phenomena operating at a given level of analysis. The implications of this multi-level perspective for understanding how innovation occurs in organizational settings are discussed.

This chapter on Mumford and Hunter's chapter “Innovation in Organizations: A Multi-Level Perspective on Creativity” (this volume) describes both its contributions and limitations to the development of a cross-level theory of innovation. To resolve some of the cross-level paradoxes highlighted by Mumford and Hunter, we propose five variables that operate at multiple levels including trust, social identity, mental models, networks, and time, and formulate some new multi-level propositions. Future directions for innovation theory development and research are also discussed.

This chapter complements that of Mumford and Hunter (this volume) by pointing out ways in which creativity can be more or less rewarded, depending on its type. Whereas Mumford and Hunter (this volume) discuss levels of organization, I discuss here different kinds of creative contributions. These contributions can either accept existing paradigms, propose new ones, or integrate new ones with old ones.

In their articles on “Innovation in Organizations: A Multi-Level Perspective on Creativity,” Robert Sternberg, along with Jane Howell and Kathleen Boies, broach a critical question bearing on the nature of creativity in organizational settings. Why is creativity in organizations so difficult even though organizations say they want creativity? In the present chapter, we examine some likely sources of this paradox and the ways one might go about resolving this paradox. Subsequently, we discuss directions for future research.

Most research issues in strategic management are essentially problem focused. To one extent or another, these problems often span levels of analysis, may align with different performance metrics, and likely hold different implications from various theoretical perspectives. Despite these variations, research has generally approached questions by taking a single perspective or by contrasting one perspective with a single alternative rather than exploring integrative implications. As such, very few efforts have sought to consider the performance implications of using combined, integrated, or multi-level perspectives. Given this reality, what actually constitutes “good” performance, how performance is effectively measured, and how performance measures align with different perspectives remain thorny problems in strategic management research. This paper discusses potential extensions by which strategic management research and theory might begin to address these conflicts. We first consider the multi-level nature of strategic management phenomena, focusing in particular on competitive advantage and value creation as core concepts. We next present three approaches in which strategic management theories tend to link levels of analysis (transaction, management, and atmosphere). We then examine the implications arising from these multi-level approaches and conclude with suggestions for future research.

This chapter applies arguments advanced by Drnevich and Shanley (this volume) to the strategic leadership literature – an area of work where such multi-level analyses seem likely to be particularly appropriate. In an analysis of the relationship between managerial capabilities and firm performance, this chapter breaks from tradition in the strategic leadership literature by examining the interaction between three levels of analysis. In doing so, this chapter identifies the conditions under which leadership can be a source of competitive advantage for a firm, when labor markets will allocate managerial talent imperfectly across competing firms, and when managers will and will not be able to appropriate the rents their specific managerial talents might generate.

Multi-level issues (Klein, Dansereau, & Hall (1994). Academy of Management Review, 195–229) are critical to both strategic management research and practice, and yet, we have few approaches for dealing with them both systemically and systematically. In this chapter, I take a resource-based approach to exploring multi-level linkages, suggesting that such an approach has wide applicability. A resource-based view (RBV) of competitive advantage and value creation illustrates the multi-level nature of these concepts and shows how the RBV is itself linked to the external market environment. The RBV also provides a way to link a variety of types of levels of analysis. These include different organizational levels of analysis, content and process linkages, and linkages across time.

In this reply to the articles about “Multi-level Issues for Strategic Management Research: Implications for Creating Value and Competitive Advantage” (Drnevich and Shanley, this volume), we consider and applaud the applications of our ideas presented by Mackey and Barney (this volume) and Peteraf (this volume). We also note three general issues that arise from considering the two articles together: (1) the complexity of multi-level constructs of management; (2) the importance of strategic decision processes (and process approaches to strategy) in a multi-level approach; and (3) the need for dynamic approaches to bridge temporal periods in crafting strategic theories.

The upper-echelons model of Hambrick and Mason ((1984). Academy of Management Review, 9, 193–206) launched a new area of research and provided the first overall theoretical framework for use in understanding how the experiences, backgrounds, and values of senior executives in organizations can influence the decisions that they make. The model is typically assumed to be what Rousseau ((1985). In: B. M. Staw, & L. L. Cumming (Eds), Research in organizational behavior (Vol. 7, pp. 1–37). Greenwich, CT: JAI Press) calls “multi-level,” as it describes how both individuals and top management teams (TMTs) make decisions in line with their preferences, biases, and values; the same model is applicable to both individuals and groups. However, the levels issues in the model have never been subjected to rigorous analysis. This chapter juxtaposes levels concepts and theories on the upper-echelons model, in an effort to highlight its strengths as well as its weaknesses. While the majority of researchers use the model to describe team-level decision making, the analysis presented here reveals that the model is inherently individual-level in focus, and several important limitations must be overcome before the model will provide a full explanation of team-level decision making.

This chapter examines Cannella and Holcomb's (this volume) multi-level analysis of Hambrick and Mason's ((1984). Upper echelons: The organization as a reflection of its top managers. Academy of Management Review, 9, 193–206) original upper echelons perspective, and the strategy formulation studies that have consequently employed that perspective. I highlight five key contributions made by Cannella and Holcomb (this volume), and suggest three supplementary avenues for inquiry. I close by arguing that if we are aiming to encourage researchers to move to a truly multi-level upper echelons model, then such a multi-level emphasis must encompass both strategy formulation and strategy implementation given their implicit interdependence.

Cannella and Holcomb ((this volume). In: F. Dansereau & F. J. Yammarino (Eds), Research in multi-level issues (Vol. 4). Oxford, UK: Elsevier Science) are unconvinced that top management teams (TMTs) are the appropriate level of analysis for upper echelons research and are, accordingly, unenthusiastic about the promise of multi-level analysis for research of this type. We agree and discuss (1) the fragility of agency theory as it pertains to TMT research, (2) various issues pertaining to TMT turnover (or lack thereof), (3) paradoxes in practice and theory regarding TMT homogeneity/heterogeneity, (4) the absence of boards of directors in the upper echelons perspective, and (5) the implications of these issues on the theory/conceptualization of TMTs and of the research dedicated to them. We question whether the variables, as currently configured, relied on in this literature are sufficiently developed to adequately test an upper echelons perspective, or to sensibly warrant a multi-level analytical approach.

We thank Carpenter and Dalton and Dalton for their insights on our earlier chapter, and on the promise (and perils) of upper-echelons research in general. We set out to closely examine the levels issues in Hambrick and Mason's ((1984). Academy of Management Review, 9, 193–206.) original upper-echelons model, and the research initiatives that have applied this theoretical framework. We are encouraged by the initial reception that we have received from these authors. We continue to believe that top management teams (TMTs) are an important level of analysis for strategic leadership research, though the original upper-echelons model proposed by Hambrick and Mason cannot be directly applied at the team level. Our reply highlights several joint and individual concerns raised by the articles. We close by reiterating our call for continued analysis of the upper-echelons model.

Over the last decade, latent growth modeling (LGM) utilizing hierarchical linear models or structural equation models has become a widely applied approach in the analysis of change. By analyzing two or more variables simultaneously, the current method provides a straightforward generalization of this idea. From a theory of change perspective, this chapter demonstrates ways to prescreen the covariance matrix in repeated measurement, which allows for the identification of major trends in the data prior to running the multivariate LGM. A three-step approach is suggested and explained using an empirical study published in the Journal of Applied Psychology.

Multivariate latent growth modeling (multivariate LGM) provides a flexible data analytic framework for representing and assessing cross-domain (i.e., between-constructs) relationships in intraindividual changes over time, which also allows incorporation of multiple levels of analysis. Using the chapter by Cortina, Pant, and Smith-Darden (this volume) as a point of departure, this chapter discusses important preliminary data analysis and interpretation issues prior to performing multivariate LGM analyses.

Every “structural model” is defined by the set of covariance and mean expectations. These expectations are the source of parameter estimates, fit statistics, and substantive interpretation. The recent chapter by Cortina, Pant, and Smith-Darden ((this volume). In: F. Dansereau & F. J. Yammarino (Eds), Research in multi-level issues (vol. 4). Oxford, England: Elsevier) shows how a formal investigation of the data covariance matrix of longitudinal data can lead to an improved understanding of the estimates of covariance terms among linear growth models. The investigations presented by Cortina et al. (this volume) are reasonable and potentially informative for researchers using linear change growth models. However, it is quite common for behavioral researchers to consider more complex models, in which case a variety of more complex techniques for the calculation of expectations will be needed. In this chapter we demonstrate how available computer programs, such as Maple, can be used to automatically create algebraic expectations for the means and the covariances of every structural model. The examples presented here can be used for a latent growth model of any complexity, including linear and nonlinear processes, and any number of longitudinal measurements.

In response to the Three-Step-Approach (TSA) that Cortina, Pant, and Smith-Darden (this volume) have suggested, Chan (this volume) expressed his reservations regarding the usefulness of a procedure that explicitly ignores measurement considerations and does not include mean scores. In this reply, we argue that the purpose of TSA is heuristic in nature and does not involve statistical testing of assumptions. In this spirit, the software, illustrated by Grimm and McArdle (this volume), rounds out our more conceptual considerations.

Most multi-level studies are cross-sectional and focus on a certain point in time, though various changes within levels may occur over time. This chapter presents a statistical method for assessing whether the degree of interdependency within a group has changed over time, using the intraclass correlation coefficient (ICC) as an indicator of the degree of homogeneity within the groups. It then shows how to apply this method using the SAS MIXED procedure. The problem was motivated by a study in which 120 subjects were divided into 40 groups of three. In a portion of the study, collective efficacy was the dependent variable measured for each subject under four different conditions (two levels of task interdependence at two points in time). ICC was used as a measure of group homogeneity with respect to collective efficacy, and the problem was how to compare the dependent ICCs associated with the different conditions.

In this chapter, we discuss Cohen and Doveh's (this volume) proposed protocol for testing differences in intra-class correlation coefficients (ICCs). We believe that there are many research questions that can be addressed by this procedure. We provide several potential examples of using this procedure at the individual, group, and organizational/society levels of analysis. We do, however, raise concerns about interpreting the ICC as an index of within-group homogeneity.

In this commentary, we discuss the tutorial by Cohen and Doveh (this volume) on using multi-level models to estimate and test hypotheses about (dependent) intra-class correlations (ICCs). The goal in this context is to find subsets of homogeneous ICCs across time points and/or experimental tasks. We suggest that one should approach this model selection problem by (a) specifying the set of all possible models, (b) using a systematic top-down selection strategy that avoids possible inconsistent and intransitive patterns in the results, and (c) using the entire dataset available when fitting each model. Our goal is to discuss the larger framework in which the specific models considered by Cohen and Doveh in their analysis are embedded, which should be useful in guiding researchers faced with a similar analysis task.

This article is a response to the two articles about our chapter (Cohen & Doveh, this volume). The first article was written by Viechtbauer and Budescu and the second written by Hanges and Lyon (both in this volume). The main contribution in the first article relates to the statistical methodology, while in the second article the authors introduce further applications to our method and discuss the interpretability of intra-class correlation coefficients (ICC). We concur with most of the ideas expressed in these articles and elaborate on some of the points raised in them.

Jay Barney is a Professor of Management and holds the Bank One Chair for Excellence in Corporate Strategy at the Max M. Fisher College of Business, The Ohio State University. He received his undergraduate degree from Brigham Young University, and his master's and doctorate from Yale University. He taught at the Anderson Graduate School of Management at UCLA and Texas A&M University before joining the faculty at Ohio State in 1994, where Professor Barney teaches organizational strategy and policy to MBA and Ph.D. students.