Multilevel Organizational Analysis: Benefits and Research Challenges

Introduction: Organizational Complexity and Multilevel Approaches

Given today's increasingly complex organizational structures, equally nuanced levels of organizational research are required to ensure a full and comprehensive portrait of the environment. Different approaches have been created to deal with the need for organizational complexity. Two dominant approaches are that of multilevel and comparative approaches, both of which "present rather distinct traditions in organization studies, each with its own epistemological assumptions and associated methods" (Lacey & Fiss 2009: 3).

Organizations can be compared with other organizations as a whole, or they can be compared in terms of their different internal levels. According to the multilevel approach, organizations are made up of a series of interconnected individuals, dyads, groups, organizations, industries, markets, and other components based upon the belief that "to examine organizational phenomena is thus to encounter levels issues" (Schnake & Dumler 2003). Multilevel organizational analysis breaks down the organization, implicitly arguing that the study of the parts is just as valuable as the study of the whole. As Rousseau (1985) notes, levels are by definition hierarchical; thus the comparisons take place between dissimilar rather than similar levels of the organization (Rousseau 1985:4).

Three Multilevel Analysis Strategies

Several different multilevel approaches exist within these theoretical constructs: composition, cross-level, and full multilevel analysis. "Composition theories are essentially conceptual theories of aggregation from one level to the next and ask how, or if, concepts located at level 1 relate to the same concept at level 2 and level 3. In contrast, cross-level theories aim to test whether a variable at level 3 or 2 affects behavior at the lower level 1" (Lacey & Fiss 2009: 4).

For example, an analysis of cross-functional teams might attempt to understand how perceptions of organizational mission at the individual level relate to behaviors in dyads and teams. Sometimes certain behaviors such as learning are generalizable from the individual to the collective level; other behaviors such as absenteeism and turnover may take on a different character when compared as to how they function in the behavior patterns of the individual versus as aggregate data (Rousseau 1985:13).

Cross-level theories, in contrast, do not view different levels as enclosed entities and examine how behaviors at different levels affect one another, specifying a causal effect between different behaviors (Rousseau 1985:14). For example, the intransigent behavior of management and refusal to promote organizational learning has measurable effects upon the behavior of work teams and individuals. A failure of guidance creates a particular kind of organizational fallout.

Methodological Challenges in Multilevel Research

Finally, full multilevel theories attempt to explain whether the relationships of the parts—such as relationships between workers on an individual level—are true of other types of levels in other areas of the organization. For example, these theories ask if cultural conflicts on a team can be analogized to cultural conflicts between different branches of a larger organization (Lacey & Fiss 2009: 5). An organization that does not promote diversity at the macro level may allow intolerance to fester at the local level. Decisions create an organizational culture with holistic implications for the survival of the organization, and no policy can be analyzed in isolation from its potentially far-reaching consequences. Another way to classify these levels of analysis is in terms of whether they relate to individual (compositional), group (cross-level), or mixed level (multiple-level actors) (Schnake & Dumler 2003).

Obviously, there are many different assumptions embedded into the construction of these levels of analysis in terms of the dialogue that takes place between organizational levels. Regarding its standards of proof, multilevel research presents a number of challenges to researchers.

The first challenge is misspecification, or attributing a behavior to another level other than where it occurs within a specific unit (Rousseau 1985: 5). For example, the behavior of a specific employee might be assumed to be reflective of the behavior of upper-level management, or vice versa. However, prejudice on a lower level or incompetence does not necessarily mean that such values are reflective of higher levels of the organization.

Another problem is aggregation bias, in which data about the aggregate or whole organization is assumed to be generalizable to all of the organization's parts (Rousseau 1985:6). This type of thinking is often seen when consumers express outrage about a defective product and take their feelings out at lower-level employees. For example, all Toyota employees are assumed to be incompetent because of the fact that so many of the car company's products have been defective, or Starbucks workers are held accountable for the high prices set by the home office.

Advantages and Practical Benefits of Multilevel Analysis

Cross-level fallacies suggest the opposite of aggregation bias—bad behavior by an individual employee, such as a rude McDonald's employee, is assumed to be true of the whole organization (Rousseau 1985:8). A still more complex problem with multilevel analysis is that of cross-level fallacies, in which the different levels of the organization are anthropomorphized to such a degree that the analysis lacks logical substantiation (Rousseau 1985:7). For example, assuming that an organizational level is stressed because of an increased workload, similar to how a person feels stressed when overworked. Notions of organizational conflict and organizational learning can also fall prey to this fallacy. Because of the rhetoric commonly used in the business press as well as existing corporate law, it is very common to take very literally the assertion that corporations are persons when they are not.

Finally, contextual fallacies are fallacies that do not take into consideration the context of a behavior when making a generalization about a particular phenomenon (Rousseau 1985: 10). Context shapes how behaviors manifest at different organizational levels and cannot be ignored without invalidating conclusions.

Conclusion: Balancing Rigor with Comprehensive Understanding

However, despite these many disadvantages to using multilevel organizational models, there are still many advantages to doing so. First of all, the existence of multiple levels is a reality, and the alternative is postulating a black box organizational actor. This means it is very difficult to determine how internal power structures can impact organizational decision-making in a meaningful way, leading the observer to fall prey to a rational actor analysis of the organization, assuming that it always makes the best organizational decision, regardless of the behavior of individual members. It could be argued that some of the fallacies inherent to multilevel organizational research can be guarded against by using different forms of analysis. And taking an overly simplistic approach is itself fallacious and can fall prey to making hasty generalizations based upon a relatively limited array of observations of organizational conditions.