Click here to return to Research in Multi-level Issues page
The Many Faces of Multi-Level Issues
Francis J. Yammarino
In numerous fields of study, dealing with the problems that arise from a consideration of levels of analysis (for example, individuals, groups, and organizations) is becoming necessary not just to understand a topic but also to avoid making serious mistakes in theorizing, conceptualizing, and analyzing data. Such problems and their solutions are arising in various fields of study such as organizational behavior, human resources management, organization theory, business policy and strategy, psychology, sociology, education, and political science as well as leadership, decision-making, communication and information processing, and stress.
The purpose of the Research in Multi-Level Issues series, of which this book is Volume 1, is to provide an outlet for discussion of such multi-level issues (problems and solutions) across a variety of fields of study. In these annual volumes, we will present timely, scholarly work on multiple levels of analysis, especially multi-level theory, research, and methods. The focus is on "critical essays" (i.e., critical literature reviews and new model development), purely theoretical work, significant empirical studies, methodological developments, analytical techniques, and philosophical treatments in the field of multi-level studies. While we have been associated with a particular approach to multi-level issues, the varient approach and WABA (e.g., Dansereau, Alutto, & Yammarino, 1984; Dansereau & Yammarino, 1998a, 1998b, 2000; Dansereau, Yammarino, & Kohles, 1999), we strongly believe that various useful approaches to multi-level issues arise within different fields of study but often are not communicated across these fields. This series is an attempt to fill that void and to help each area avoid reinvention of the same wheel. The goals here are to provide high-quality scholarly work and to foster a constructive scientific debate for the efficient advancement of knowledge in the realm of multi-level issues. Thus, in addition to key scholarly essays that focus on a topic from the perspective of multiple levels of analysis, there are two commentaries and a rebuttal for each major essay.
This volume contains five major essays. The essays and associated commentaries are similar in that they highlight the reason why a consideration of multi-level issues is becoming essential across a wide variety of fields. But, at the same time, the essays differ in their approaches to multi-level issues. We describe the similarities and differences of these approaches very briefly in this overview.
The following sampling of some of the commentaries about the five key essays in the book help show how important multi-level issues are becoming. In essence, the point is that ignoring multi-level issues can result in serious misconceptions of theories/concepts and misinterpretations of data.
In discussing the Mumford, Schultz, and Osborn’s essay on planning, Sydney Finkelstein from Dartmouth says this paper:
tells us that cross-level research reminds us that we may not only be missing something by retaining a single-level focus but also misspecifying that which we wish to study.
In her commentary on the Ployhart and Schneider essay about multi-level selection, Cheri Ostroff from Columbia University says their work suggests that
the traditional selection model should be leveled (both figuratively and literally). An organization could carefully follow the traditional selection model, develop very valid tests and adhere to technically superior practices, but if they do not translate into performance and effectiveness at higher levels of analysis, then ultimately the organization’s overall performance will suffer and it may not survive.
In discussing the essay by Ferris, Adams, Kolodinsky, Hochwarter, and Ammeter about perceptions of politics, Robert Dipboye and Jessica Foster from Rice University point out:
While each individual’s perceptions of politics may be unique in some respects, it is likely that a significant portion of the variance in perceptions is attributable not to the individual, but to higher-level influences within the organization. A comprehensive understanding of these complexities will require a more aggressive multilevel approach than seems to have characterized research on POPS thus far.
In discussing the essay by Earley and Mosakowski about culture, Joseph Alutto from Ohio State suggests:
It is all too common to assume that “cultural differences” really do make a difference but [there is] little understanding of how level effects can confound understanding of critical relationships. This occurs in research and in the teaching activities of business schools.
In discussing the essay about multi-level simulation by Seitz and Hulin, Robert Bernstein of the United States Agency for International Development says:
choosing to study one causal level (of analysis) over another may be a convenience to researchers but it produces incorrect results.
He also states if Seitz and Hulin are correct with the addition of process decompositioning “then the paper raises the haunting spector that much of the social and behavioral research over the past several decades may be fundamentally flawed.”
The similarity of the concerns of the authors and commentators across the topics makes it clear that new fundamental approaches to multi-level issues are essential to perform successful research and to avoid flaws in traditional research.
Although the essays are similar in that they demonstrate that a consideration of multiple levels of analysis is critical, they differ in terms of the way they conceptualize and consider levels of analysis and focus on variables. To illustrate this point, each major essay and its associated commentaries are considered briefly below.
In the essay by Mumford et al., the focus is on the planning process and successful planning at multiple levels of analysis. Here the planning process is viewed as distinct at the individual, group, and organization levels. This is a particularly interesting way to think about planning because the variables at one level of analysis are not viewed as reducible to lower levels or extendable to higher levels of analysis. The following table illustrates the point. In the table, “yes” means a level is considered and “no” means a level is not considered as a part of the formulation about the variable.
Individual Group Organization
Planning Planning Planning
Variable Variable Variable
Individual level Yes No No
Group level No Yes No
Organization level No No Yes
Thus, we believe the variables and relationships are viewed as specific to each level but that planning manifests itself and includes different variables at multiple levels.
In his commentary about this essay, Bluedorn raises the possibility that planning may be the same at all levels of analysis and that this approach should be tested in addition to the multiple-level model described by Mumford et al. Likewise, Finkelstein points out the importance of understanding planning from a broader perspective. Mumford responds by extending the model to focus on cross-level and within-level interactions. Mumford also emphasizes the need for a multi-level focus on planning from an applied practitioner perspective.
Ployhart and Schneider focus on predictor and criterion variables at each of several levels of analysis with an emphasis on the selection of individuals for jobs. Similar to Mumford et al., they also focus on different variables at each of several levels of analysis. In addition, Ployhart and Schneider seem to imply for each variable that it too can be extended to higher and lower levels so that one can also examine direct effects from a higher level of analysis on a lower level of analysis and vice versa. If we recognize that different variables apply at different levels, Ployhart and Schneider focus on variables at each level as well as across levels as shown in the following table.
Individual Level Group Level Organization Level
Predictor-Criterion Predictor-Criterion Predictor-Criterion
Individual Level Yes Yes Yes
Group Level Yes Yes Yes
Organization Level Yes Yes Yes
In this formulation, the variables and relationships apply at each of several levels of analysis. This approach, for example, allows an examination of group-level variables with individual-level variables at the individual and group levels.
In her commentary, Ostroff offers various suggestions to enhance the multi-level selection model described by Ployhart and Schneider. Schmitt, in his commentary, raises various questions about aggregation problems. Ployhart and Schneider respond to these issues and point out that through careful multilevel explications and model testing, considerable progress is likely to occur in personal selection practices as well as in theory development.
Perceptions of politics
The essay by Ferris et al. suggests a very different view of multiple levels of analysis. Here the interest is in the perceptions of individuals as they are influenced by variables at multiple levels of analysis. In a sense, all of the variables involve individual-level perceptions of variables at different levels of analysis. At first glance, this might seem to be an approach that is anchored only at the individual level. In one sense this is correct. But, at the same time, individuals’ characteristics are assumed to interact with other variables that are situational in nature. This approach assumes that multiple levels of analysis are a given, and then assesses variables at different levels by means of individuals’ perceptions of these variables at an assumed level.
In commenting on the Ferris et al. paper, Dipboye and Foster suggest an approach more similar to that described by Ployhart and Schneider where the focus is not only on the individual level but also on the group and organizational levels. Fedor and Maslyn offer additional propositions to those suggested by Ferris et al. Adams et al., in response, suggest that indeed additional attention needs to be paid to multiple levels of analysis in perceptions of politics.
The essay by Earley and Mosakowski presents yet another view of levels of analysis, similar to that of Mumford et al. as well as Ployhart and Schneider, but adds a slightly different angle. Here, variables at higher levels are thought to manifest themselves in different (not the same) variables at lower levels of analysis. Each level is unique, as was the case with the work of Mumford et al., but the higher-level variables may manifest themselves via different variables at lower levels of analysis. Moreover, these variables at different levels are apparently viewed as casually related. For example, organizational-level culture produces lower-level (group-level) roles that in turn moderate other effects at the (lower) individual level of analysis. To begin to explain such casual linkages across levels, we would need to examine how changes at the higher levels relate to variables at a later time point at a lower level which may in turn moderate variables at the lowest level at an even later time point. This “cascading over time” approach is clearly different from the Ferris et al. approach where the individual level is considered the most important. In Earley and Mosakowski’s approach, it is higher levels of analysis that associate with culture that are of importance.
Alutto suggests the possibility that Earley and Mosakowski may have gone too far in implying that empirical tests for alternative levels of effects are unnecessary if they assert effects at a particular level. It is clear that testing this model will require a quite complex methodology that goes well beyond, for example, HLM. We believe Alutto is stressing the importance of fully testing all theoretical assumptions and assertions and the use of WABA as one approach for doing so. This allows for different degrees of agreement under different conditions or cultures. Earley and Mosakowski respond by raising questions about how this should be done. Morris and Young, in their commentary, argue strongly in favor of the type of reconceptualization offered by Earley and Mosakowski. Indeed, they say that traditional views of culture as stemming from values have run dry. In their reply, Earley and Mosakowski generally agree.
Seitz and Hulin, in their essay about multi-level simulation, examine multiple levels by transforming an individual-level variable to a group-level variable. They then use contextual type analysis equations with simulations to focus on how context variables and their interactions with individual-level variables combine to influence the individual-level dependent variable—rate of transmission of HIV. This is somewhat similar to the Ferris et al. view of levels of analysis but with a difference: the level of analysis for all empirical estimators is groups. Nevertheless, similar to Ployhart and Schneider, the focus for Seitz and Hulin is on different variables at different levels. But, estimations of effects across levels are made by focusing at the group level. Seitz and Hulin show how this type of multi-level simulation applies to work organizations as well.
Bernstein provides two clear illustrations of how the very sophisticated modeling by Seitz and Hulin can be viewed from a very straightforward applied perspective and draws out the implications of this work for much of the social and behavioral sciences. Markham, in his commentary, contrasts Seitz and Hulin’s approaches to hierarchical linear modeling (HLM) and within and between analysis (WABA). In this way, Markham provides a perspective on various empirical methods for considering multiple levels of analysis. In their reply, using the commentaries of Bernstein and Markham, Seitz and Hulin elaborate the implications of their work for the social and behavioral sciences.
The essays, commentaries, and rebuttals in this book reflect the diversity and rich ways in which a consideration of multiple levels of analysis can help researchers avoid mistakes and better understand virtually any concept of interest. In this way, this book provides a look at the “many faces of multi-level issues.” At the same time, the essays and commentaries offer a real challenge to researchers, theorists, and methodologists who want to develop their ideas with an eye toward multiple-level thinking. In future volumes of the series, we hope to be able to bring forward additional responses to this real challenge. We believe it will be a continuing process requiring significant effort that will result in the development of various valid ways to focus on levels of analysis. It is our intention that, over time, this series will describe valid and integrated ways to deal with these issues. If you would like to contribute your thinking to this process, please read our “Call for Papers” for the series at www.levelsofanalysis.com/rmli1.html or just contact us directly.
Dansereau, F., Alutto, J.A., & Yammarino, F.J. (1984). Theory testing in organizational
behavior: The varient approach. Englewood Cliffs, NJ: Prentice‑Hall.
Dansereau, F., & Yammarino, F.J. (Eds.) (1998a). Leadership: The multiple-level approaches
(Part A: Classical and new wave) (Vol. 24 of Monographs in Organizational Behavior
and Industrial Relations). Stamford, CT: JAI Press.
Dansereau, F., & Yammarino, F.J. (Eds.) (1998b). Leadership: The multiple-level approaches
(Part B: Contemporary and alternative) (Vol. 24 of Monographs in Organizational
Behavior and Industrial Relations). Stamford, CT: JAI Press.
Dansereau, F., & Yammarino, F.J. (2000). Within and between analysis: The varient paradigm
as an underlying approach to theory building and testing. In K.J. Klein & S.W.J.
Kozlowski (Eds.), Multilevel theory, research, and methods in organizations:
Foundations, extensions, and new directions (pp. 425-466). San Francisco, CA: Jossey-
Dansereau, F., Yammarino, F.J., & Kohles, J. (1999). Multiple levels of analysis from a
longitudinal perspective: Some implications for theory building. Academy of
Management Review, 24, 346-357.
We would like to thank Tom Clark, Deborah Raven, and Diane Cogan at Elsevier Science for their help in publishing this book series. We also want to thank our respective Schools of Management for providing the time and resources necessary to initiate and continue this series, and to thank Cheryl Tubisz, Robyn Washousky, and Marie Iobst for their help in the preparation of this book for publication. Finally, we owe a special debt to the authors of the essays, commentaries, and rebuttals in this book who were willing to expend the significant amount of time and effort necessary to begin to unravel the challenges that arise from considering multiple levels of analysis.
State University of New York at Buffalo
Francis J. Yammarino
State University of New York at Binghamton
Click here to return to Research in Multi-level Issues page
<Order DETECT or Books> <Home Page> <Info. Page> <Contact us: Info@levelsofanalysis.com>