Statistical Issues in Intensive Longitudinal Data Analysis

Abstract Innovation in scientific data collection involving new kinds of protocols has proceeded at a breakneck pace over the past 10 years. Starting from a tradition of diary studies (Almeida, 2005; Bolger, Rafaeli, & Davis, 2003), researchers have increasingly used emerging “smart” electronic devices to collect data reflecting experience over time in real-life contexts. Many scientific studies have emerged using these devices; for example, there are studies using beepers for diary data collection (Csikszentmihalyi & Larson, 1987), computerized ambulatory recording devices (e.g., Wilhelm, Roth, & Sackner, 2003), automated behavioral observational tools (e.g., Borrie,Jonsson, & Magnusson, 2002; Hirschhauser, Frigerio, Grammer, & Magnusson, 2002), and sensing devices in physical and architectural environments (Intille, Chapter 16, this volume). Consistently, in many areas of psychology and in behavioral medicine, studies using Ecological Momentary Assessment (EMA) techniques have become commonplace (Stone & Shiffman, 1994). At the same time, other technological advances, including developments in GPS and GIS technology, fMRI and PET imaging, road network analysis, Web interfaces for transactions, and many other technology-based data collection mechanisms have equipped researchers with tools to collect much more detailed and complex time-graded data than ever before.