John S. Searles

Comparison of drinking patterns measured by daily reports and timeline follow back.

The major objective of this study was to compare near real-time daily alcohol consumption data over the course of 366 consecutive days with retrospective reports by means of the timeline follow-back (TLFB). Participants (N = 33) responded for 366 days on an interactive voice response (IVR) system by entering alcohol consumption data daily using the touch-tone pads of their telephones. In-person interviews were conducted every 13 weeks during which participants were administered the TLFB. The correlations between the IVR and TLFB for amount consumed, drinking days, and heavy drinking days were modest. There was a wide variability across participants in their individual correlations for these variables. Participants who were diagnosable with a lifetime DSM-IV alcohol disorder at baseline significantly underreported their drinking compared with participants who were not diagnosable. The authors were unable to ascertain variables that influenced accurate reporting on the TLFB.

USING INTERACTIVE VOICE RESPONSE TECHNOLOGY AND TIMELINE FOLLOW-BACK METHODOLOGY IN STUDYING BINGE EATING AND DRINKING BEHAVIOR: DIFFERENT ANSWERS TO DIFFERENT FORMS OF THE SAME QUESTION?

As part of a study of the relationship of binge eating, alcohol use, mood, and stressors, we compared the results of two forms of reporting on binge eating and drinking behavior. Forty-three first-year college women participated in an interactive voice response (IVR) study for 12 weeks. Participants answered computer-administered questions daily via IVR technology on number of eating binges and number of alcoholic drinks consumed. After 12 weeks, participants completed a Timeline Follow-back (TLFB) interview retrospectively for number of binges and drinks in the past 12 weeks. Results of this distally retrospective methodology (commonly used in drinking research and applied here also to binge eating) were compared to the results of daily IVR reporting. There was convergence across measures for drinking behavior, but divergence between IVR and TLFB for binge eating reports. TLFB reports underrepresented actual binge eating frequency, which calls into question the validity of applying this methodology to the assessment of binge eating.

AN APPLICATION OF INTERACTIVE VOICE RESPONSE (IVR) TECHNOLOGY TO LONGITUDINAL STUDIES OF DAILY BEHAVIOR

A relatively new technology-an interactive voice response (IVR) system-was used to collect selfreport data from 51 male subjects (21–56 years old) concerning daily tobacco and alcohol consumption over a 112-day period. After resolution of initial technical problems, the Touch-Tone data-entry methodology of the IVR was shown to provide notable advantages over traditional longitudinal self-report methodologies. The high compliance rate (93%) for this study was attributed to (1) implementation of important study design characteristics; (2) an innovative compensation scheme that reinforced consistent daily reporting and incorporated feedback to the subjects on a daily basis; (3) a simple, unobtrusive Touch-Tone data-entry interface with the interview and data-collection system that provided subjects with flexibility and time efficiency; and (4) direct coupling of the automated data-collection system and a relational database-management structure, providing immediate access and analysis of collected data.

Psychological performance assessment via interactive voice response systems

Two experiments examined the feasibility of psychological assessment using interactive voice response (IVR) technology and the potential sensitivity of such assessments to alcohol and fatigue effects. In Experiment 1, 10 subjects performed a 12-min battery of six IVR-administered tasks, Monday through Friday, over 2 weeks. Minimal learning effects were evident during training. Repeated administrations indicated high test-retest reliabilities. In Experiment 2 (double-blind, alcohol/placebo crossover design), 7 subjects were tested every 2 h over a 24-h period during two experimental sessions (peak blood alcohol concentrations =80 mg/dL). Several IVR-administered tasks were sensitive to alcohol impairment, but not as sensitive as laboratory-based measures specifically designed to assess alcohol impairment. Little evidence for fatigue-related impairment was obtained. The results support optimism for the potential to assess psychomotor and cognitive functioning distally via telephony; however, further refinement and validation of the methods are needed.

Conducting longitudinal studies of behavior using interactive voice response technology

Interactive voice response (IVR) systems represent a convergence of automated computerized interview techniques with telephone survey research. This convergence affords unparalleled opportunities for conducting longitudinal behavioral research. Our experiences, after several years of conducting multiple IVR studies investigating the circumstances and consequences of daily alcohol use patterns, continues to generate excitement regarding the research potential of this methodology. This article discusses design and implementation considerations, and potential pitfalls, for others interested in conducting this type of research. The discussion is structured around: (1) IVR script development, (2) programming implementation issues, (3) research administration of longitudinal IVR studies, and (4) database management and hardware configuration issues: The future of IVR research and further integration of developed and emerging technologies are also discussed.