Explaining the Attenuation Effect in Epidemiological Studies

点击翻译按钮获取中文摘要

DESCRIPTION (provided by applicant): Although epidemiologists and clinical researchers depend on longitudinal assessments to study process and change, for decades it has been known that survey reports of health problems show a bias where they decline inexplicably in frequency and severity from first to subsequent interviews. This decline, called the attenuation effect, is better documented in psychiatric than in alcohol use disorders, but this difference could be due to the paucity of research on the latter. The presence of attenuation biases in surveys of alcohol abuse and psychiatric disorders can lead to erroneous conclusions about the course of these diseases and of their precursors. Several explanations for the attenuation effect have been proposed, including a therapeutic benefit of repeated assessments, participants' changing interpretations of those assessments, or a process of learning/optimization of participant responses, but these explanations have not been tested. Moreover, the explanations have not been jointly considered in contrasting contexts of psychiatric disorders where the attenuation effect is well established, and alcohol disorders, where it is not yet established. We propose two specific aims. 1) To carefully delineate the attenuation effect in two intensive longitudinal studies of college students, by estimating its magnitude in a) surveys of alcohol problems and mental health assessed repeatedly over a four month period, and b) daily diary studies of alcohol use and mental state among students facing an acutely stressful event. 2) To reproduce the effect in experimental studies and to test hypotheses that it is due in part to a) a therapeutic effect of the repeated assessments, b) participants' increased understanding of the assessment instrument, or c) a tendency for participants use conversation norms to judge what they should report and what they think investigators expect to hear. These aims will be approached with four studies, three of which are randomized experiments and one a psychophysiology lab study. PUBLIC HEALTH RELEVANCE: This study will help researchers gather more accurate data on alcohol and psychiatric disorders assessed in longitudinal surveys. This, in turn, will facilitate understanding of causes and courses of these disorders, enabling prevention and intervention programs to be designed.