Multilevel Item Response Theory: Modeling Changing Measurement Structures in Longitudinal Studies

• Activity Number: 626
• Type: Topic Contributed
• Date/Time: Thursday, August 8, 2013 : 8:30 AM to 10:20 AM
• Sponsor: Biometrics Section
• Author(s): Robert Wirth*+
• Companies: Vector Psychometric Group, LLC
• Keywords: Item Response Theory ; Psychometrics ; Longitudinal ; Measurement Invariance ; Differential Item Functioning ; Multilevel

Abstract

Many constructs studied by social, behavioral, and medical researchers cannot be directly observed. In such cases, scales are created that reflect the construct of interest. Observed behaviors, often responses to a set of items, are taken as manifestations of an unobserved common cause. Understanding the measurement structure of a construct is especially important when trying to understand individual or group changes. However, the manifestation of a construct can change over time thereby changing its measurement structure. For example, a set of items may function differently (e.g., become more or less related to the construct) over age. Historically workarounds for modeling the nested structure of data has been common place in the item response theory (IRT) framework. This presentation will address recent advancements in multilevel IRT that allow for such nesting to be appropriately modeled when examining changing measurement structures. Two hypothetical examples (changing item sets and changing dimensionality) are presented using traditional and multilevel IRT methods. The results as they apply to over-time/cross-study analyses are compared and interpretations discussed.