Multilevel theory (MLT) details how organizational constructs and processes operate and interact within and across levels. MLT focuses on two different inter-level relationships: bottom-up emergence and top-down effects. Emergence is when individuals’ thoughts, feelings, and/or behaviors are shaped by interactions and come to manifest themselves as collective, higher-level phenomena. The resulting higher-level phenomena can be either common, shared states across all individuals (i.e., compositional emergence) or stable, unique, patterned individual-level states (i.e., compilational emergence). Top-down effects are those representing influences from higher levels on the thoughts, feelings, and/or behaviors of individuals or other lower-level units. To date, most theoretical and empirical research has studied the top-down effects of either contextual variables or compositional emerged states. Using predominantly self-report survey methodologies collected at a single time point, this research commonly aggregates lower-level responses to form higher-level representations of variables. Then, a regression-based technique (e.g., random coefficient modeling, structural equation modeling) is used to statistically evaluate the direction and magnitude of the hypothesized effects. The current state of the literature as well as the traditional statistical and methodological approaches used to study MLT create three important knowledge gaps: a lack of understanding of the process of emergence; how top-down and bottom-up relationships change over time; and how inter-individual relationships within collectives form, dissolve, and change. These gaps make designing interventions to fix or improve the functioning of organizational systems incredibly difficult. As such, it is necessary to broaden the theoretical, methodological, and statistical approaches used to study multilevel phenomena in organizations. For example, computational modeling can be used to generate precise, dynamic theory to better understand the short- and long-term implications of multilevel relationships. Behavioral trace data, wearable sensor data, and other novel data collection techniques can be leveraged to capture constructs and processes over time without the drawbacks of survey fatigue or researcher interference. These data can then be analyzed using cutting-edge social network and longitudinal analyses to capture phenomena not readily apparent in hierarchically nested cross-sectional research.
Investigating multilevel pathways of developmental consequences of maltreatment