Intensive Longitudinal Data Collection Methods in Health Research: Tutorial for Selecting Measures and Designing a Sampling Protocol

Vollständiger Abstract

Worum geht es in dieser Arbeit?

Abstract Intensive longitudinal data (ILD) include frequent and dense repeated measures captured over acute timescales (eg, every second, hour, or day) that are used to investigate within-person processes both within and across days. ILD are collected via wearable sensor data, ecological momentary assessments, or daily diaries and provide unique insights into within-person processes under ecologically valid conditions that can strengthen understanding of temporal relationships among variables and potential causal processes, while also informing the development of just-in-time adaptive interventions. Advancements in mobile and sensor technology have facilitated an explosion of ILD studies that have outpaced formal training in ILD study design. When designing ILD studies, researchers need to make careful decisions about the frequency (ie, how often) and timing (ie, when) of measurements. Decisions about the frequency and timing of measurement are influenced by issues such as variability across individuals and constructs, the purpose of the assessment, concerns about recall biases, saliency, or missing information, and participant needs. The interpretation of study results, causal inferences, and the predictive value of ILD are also impacted by decisions related to the timing of assessments, temporal lags between measures, how a “day” is defined, and data aggregation choices. Due to the increased interest in and adoption of ILD studies, and a lack of formal training, researchers can benefit from guidance on how to design ILD studies. Therefore, this paper aims to provide practical guidance to researchers on how to plan ILD studies using a step-by-step decision-making tutorial with applied health behavior research examples examining phenomena (eg, physical activity and alcohol use) that vary over acute time scales.