The many faces of multi-level issues: Volume 1

In complex, dynamic environments where people must coordinate their activities, planning represents a key influence on performance. Accordingly, one might expect planning to affect performance in organizational settings. In this article, we define planning as the mental simulation of future actions. The processing operations that make the generation of viable mental simulations possible are then described along with requisite information requirements. The variables that shape the need for, and effectiveness of, planning activities are then considered with respect to the individual, group, and organizational levels of analysis. Based on these observations, we argue that planning represents a cross-level performance phenomenon of critical importance in understanding organizational behavior. Directions for future research are discussed.

Three points are addressed concerning, “Planning in Organizations.” The first deals with what has become a fundamental image of planning; the second, with problems in conceptualizing the quality of planning; and the third, with the possibility of developing some of the propositions further by combining them with other theory.

This commentary suggests what I consider to be the best points of the article by Mumford et al. and offers some possible improvements.

Allen C. Bluedorn and Sydney Finkelstein have provided an unusually insightful set of commentaries on “Planning in organizations: Performance as a multi-level phenomenon”. In this article, I respond to these commentaries beginning with an examination of the assumptions made about planning over the years. Subsequently, I examine Bluedorn's and Finkelstein's critiques of this article with respect to three key issues: Complexity, performance, and history. With regard to history, I argue that the use of case-based reasoning in planning insures that planning performance is as much a function of the past as forecasting of the future. With regard to performance, it is argued that planning performance must be framed in a multi-level model. The issue of complexity is addressed by examining the role of cross-level and within-level interactions in shaping planning performance. The implications of these observations for theoretical integration and research needs are discussed.

Personnel selection has a long and successful record for effectiveness in applied psychology. We propose that this record for effectiveness has been narrowly focused on the individual level of analysis, resulting in a lack of suitability for addressing conceptual and applied phenomena at unit (group, organizational) levels of analysis. The chapter integrates the traditional personnel selection focus on individuals with recent thinking on multiple levels of analysis and we show how this alternative has implications for selection system design, assessment procedures, and validation research. Specifically, we first review and critique the individual selection model from a multi-level orientation and then explicate how multi-level selection procedures may be enacted and evaluated. We then compare the development and validation of selection practices in two fictional organizations, one using the traditional focus on individuals and one using our revised multi-level methodology, to-illustrate the benefits of the new approach. We conclude with several recommendations for future research and practice.

The multi-level perspective on selection offered by Ployhart and Schneider provides an important framework for enhancing the utility of our selection procedures. By examining relationships between predictors and criteria across multiple levels of analysis, new insights into how the selection process contributes to group and organizational performance and effectiveness can be gleaned. Further, tradeoffs between using sets of predictors at different levels of analysis become evident, again highlighting the importance of moving beyond consideration of the individual level of analysis in selection. Indeed, it is argued that the traditional model should be replaced by a multi-level expanded model. The strength of this framework is evaluated and suggestions to enhance the multi-level model are offered.

The article by Ployhart and Schneider provides an excellent specification and explication of a multilevel model of personnel selection hypotheses. However, questions concerning the nature of group- or organization-level tasks and KSAs, how the appropriateness of aggregating individual level data is determined, the practicality of conducting meaningful multi-level research, the practical utility of research results, and the long term reliance on individual difference paradigms represent obstacles or challenges to the development of multilevel research in personnel selection. Applications of multi-levels research in personnel selection are very much needed.

We cluster the issues raised by Ostroff and Schmitt and respond to these clusters rather than to the more detailed ways in which each raised them. These issues concerned: (1) the collection of job analysis information at higher units of analysis (e.g. teams), (2) the analysis of such job analysis information, (3) the use of such job analysis information as a basis for the selection of people to be team members, especially when different selection procedures might be appropriate for different team tasks, (4) the nature of the linkage of criteria internal to organizations and those external to organizations, and (5) how decision makers might weight different criteria of effectiveness as guides to hiring decisions when the criteria exist at different levels of analysis. We note that these are all important issues but spend the most time on the first three having to do with team/group issues in personnel selection psychology. We conclude with a call for researchers to identify the “architecture of organizations” through the use of multi-level computational models. These models would require detailed specification of critical variables at different levels of analysis to permit preliminary exploration of hypothesized relationships. Through such careful explication and model testing, we envision considerable future progress in cross-levels personnel selection practice.

Organizational politics has intrigued academicians and practitioners for decades. Yet, serious scholarship on politics in organizations has emerged as a viable body of scientific inquiry just within the past twenty years. In general, theory and research on organizational politics has been sorted into the two categories of political behavior and its effects, and the nature of organizational politics perceptions. With few exceptions, these areas of inquiry have been treated as largely independent of one another. Whereas it is useful periodically to take stock of the current status of our knowledge base in particular areas of inquiry, we are often remiss in not engaging in such activity frequently enough. The present paper seeks to address this void by determining the status of theory and research on perceptions of organizational politics. First, we report on a comprehensive review of the literature designed to convey the current state of the field with respect to theory development, testing, and validation, as well as methodological considerations, including levels of analysis issues. Then, we propose future challenges with respect to construct expansion and validation, theory refinement, multi-level considerations, and integration with other constructs in the organizational sciences. Other issues that need to be addressed in future work are also examined in an effort to propose a revised model of politics perceptions to guide future research.

In this commentary, we review some of the major conclusions reached by Ferris and colleagues in their review of research on perceptions of organizational politics. We suggest that the Ferris et al. model and previous research have not gone far enough in considering the phenomenology of the individual's perception of politics and the multi-level linkages with perceived politics at both the organization-wide and subunit/group levels of analysis.

In this paper, we explore a number of politics-related issues that are raised in the extensive review by Ferris, Adams, Kolodinsky, Hockwarter, and Ammeter (this volume). In some cases, we have attempted to add important details (e.g. providing possible scale items to tap positive politics across multiple levels of the organization) and, in other cases, we have applied different viewpoints to identify alternative possibilities (e.g. the role of performance as a predictor of political perceptions). In all cases, we offer propositions to foster future politics-focused research.

In this response, we address three central themes of the Fedor and Maslyn and Dipboye and Foster commentaries. In doing so, we attempt to integrate their perspectives by presenting possible extensions to the current research stream. We suggest that these research extensions will generate a broader understanding of the perceptions of politics construct both across levels and between organizations.

To guide scholars interested in incorporating culture into research on behavior in organizations, this chapter discusses cross-level approaches to the study of culture that go beyond simplistic comparative analyses. We focus on the major issues confronting a cross-cultural management researcher. We consider diverse theoretical, research design, and analytical approaches that allow the researcher to link culture to organizational behavior. A central theme in our discussion is the paramount importance of a model that specifies mechanisms that link culture to lower levels of analyses, such as organizations, teams, and individuals. Our recommendations for empirical research revolve around the informed use of cross-level theoretical models to guide research design and analytical choices. We conclude with general recommendations for future research on culture and behavior in organizations.

The Earley and Mosakowski article effectively demonstrates the importance of culture as a predictor and moderating variable in the understanding of human and organizational behavior. However, the importance of four issues is highlighted in these comments. First, a definition of culture that recognizes variability in the uniformity of common norms and values is particularly relevant for understanding organizational behavior. Second, levels of effects are particularly important in cultural research and should not be simply assumed to occur due to theoretical preferences. Third, one area that has not received sufficient attention in organizational research is that of cultural transition. That is to say, what dynamics are particularly critical at what levels of analysis when there are significant cultural changes, and how does the time available for adaptation impact these processes? Finally, while Earley and Mosakowski assert that there should be constant movement from theory to data and data to theory as a guide for determining which level effects one should assess, it is not at all clear that they have justified a position that empirical tests for alternate level effects are unnecessary if theory asserts effects at a particular level (e.g. individual, group or collective).

In this commentary, we reiterate and build upon Early and Masokowski's call for cultural researchers to investigate underlying cognitive structures through which culture influences behavior, looking beyond the models of value-orientation that have dominated previous research. We assess evidence that tapping specific, knowledge structures — as opposed to focusing on value dimensions — has more successfully provided proof of mediating and moderating cultural effects on behavior. Finally, we explore conceptual challenges to this approach of seeking proximal knowledge structures — namely, tapping knowledge that is culturally implicit as well as explicit, further exploring conceptions of agentic groups, and examining other types of agency.

In this response to the commentaries of Alutto, Morris and Young, we explore several important issues raised by the authors. The points raised in the commentaries highlight several important directions for cross-level and cross-cultural organizational research to follow including a strong cognitive orientation and an emphasis on contextualization.

Multi-level causation generates serious methodological issues that are not always appreciated in contemporary social research. This chapter uses the dynamics of HIV/AIDS to illustrate three such issues. First, the failure to take both individual-level and group- or context-level forces into account leads to systematic bias in the statistical analysis of observed data. We decompose a fully specified cross-level regression model into separate individual- and group-level components to illustrate the resulting biases in data analysis. Second, we look at the application of fully specified cross-level regression models to processes that are not in equilibrium. Static cross-level regression models cannot properly estimate multi-level cause-and-effect when there are non-linear feedback effects among independent and dependent variables over time. Finally, we explore how computational modeling can be used to study these feedback dynamics in multi-level causal processes. We illustrate two computational methods that help researchers unravel such complex causal environments: counterfactuals and process decomposition.

The use of logic models and simulation modeling are among the techniques recommended by the Centers for Disease Control and Prevention (CDC) for gathering and analyzing evidence for the purposes of planning and evaluating public health policies, intervention strategies, and programs. However, in most countries, logic models are much more commonly used than simulation modeling, and the capacity for computational modeling is a rare luxury. Most public health professionals recognize that static models, regardless of how good the logic or how detailed the mathematics of the simulation, may nevertheless be severely constrained or weakened if they can not provide information about the impact of dynamic changes and secular trends in important parameters. The chapter by Seitz and Hulin is a well-documented argument in favor of increased exploration, and greater use, of computational modeling in social and behavioral research, including program evaluation. The authors make their argument by providing an extended case analysis of HIV transmission, because this global health problem has both multi-level causal components and their feedback over time changes the transmission dynamics. Using computational modeling methods, including “counterfactuals” and “process decomposition”, to exploit alternative scenarios for prevention and control of HIV/AIDS, Seitz and Hulin make an important contribution by focusing the attention of public health experts and others on the strengths and limits of current methods of evaluation research and analysis. They have succeeded in drawing our critical attention to computational modeling as an emergent research paradigm.

The Seitz and Hulin article applying multi-level simulation to the analysis of the AIDS pandemic fills a much needed void in the social sciences literature. In light of its pioneering aspects, four topics will be highlighted by way of review: (1) methodological contributions, (2) theoretical contributions, (3) need for visualization techniques, and (4) possible applications in management above and beyond health issues. Once its initial assumptions are recognized, this article represents a positive addition to the multi-level modeling and simulation literature.

Computational modeling brings unique and critical contributions to behavioral and social research. Computational modeling can transform logic models into dynamic models and helps formalize complex theory construction. Computational modeling opens new vistas in data designs and data analysis. Computational models allow us to explore systems not in dynamic equilibrium, to understand the implications of different initialization conditions, to examine complex system synergies through process decomposition, and to provide policy-related tools such as counterfactual simulations.