Nested Data and Multilevel Models: Hierarchical Linear Modeling

Abstract

Most of the people and cases that are subject to research in business and information systems are nested within hierarchies. A hierarchy attaches roles to certain levels and typically makes higher-level roles responsible for lower-level roles. At all levels of the organizational hierarchy, this approach translates into small clusters of managers and larger clusters of team members (which may include managers of lower-level teams). Such a hierarchy could range from a CEO and her team of executives to a line manager and his team of operators, but the hierarchy even continues beyond the organization, as organizations are nested within industries, industries within countries, and so on. Sometimes we want to study effects that cross these hierarchical layers. For example, we may be interested in the effect of managers' behavior on their team members' behavior, or the effect of remuneration policies at the level of the organization on individual performance and individual turnover intentions. In other words, we may want to study the effect of a variable that varies at the group level (i.e., between groups) on another variable that differs for every individual (i.e., it varies within groups). This kind of investigation calls for the use of hierarchical linear models.