Efficient Assessment of Multidimensional Processes in Intensive Longitudinal Designs Using Adaptive Ecological Momentary Assessment

Intensive longitudinal designs, such as ecological momentary assessment, rely on frequent and repetitive measurement to capture psychological processes in naturalistic settings over micro-longitudinal timescales (e.g., days or weeks). However, these designs can impose substantial response burden on respondents which impacts data quality and study compliance. This forces researchers to limit designs to few variables measured by ultra-brief static measures resulting in a burden-fidelity tradeoff. Such limitations often preclude effective assessment of complex multidimensional processes (e.g., suicidal thoughts). The present work proposed and evaluated multidimensional computerized adaptive testing (MCAT) methods as a means improve the burden-fidelity tradeoff in intensive longitudinal measurement. Two studies examined the performance and efficiency of multidimensional adaptive testing for high dimensional tests with realistic intensive longitudinal burden constraints. The first study focused on contrasting MCAT item selection rules across test dimensionality and test structure. The second study focuses on a subset of test dimensionalities across intensive longitudinal data structures. Multiple test initialization strategies are also examined. Results generally supported MCAT as a viable assessment method for intensive longitudinal data. Test dimensionality, item selection rules, and underlying change dynamics are critical, however. Future directions and practical considerations for adaptive measurement in intensive longitudinal designs are discussed.