AMERICAN JOURNAL OF INDUSTRIAL MEDICINE 53:1019–1031 (2010) The Influence of Sociodemographic Characteristics on Agreement Between Self-Reports and Expert Exposure Assessments Grace Sembajwe, ScD, 1,2,3 Margaret Quinn, ScD, 1 David Kriebel, ScD, 1 Anne Stoddard, ScD, 4 Nancy Krieger, PhD, 2 and Elizabeth Barbeau, ScD 2,3 Background Often in exposure assessment forepidemiology, there are no highly accurate exposure data and different measurement methods are considered. The objective of this study was to use various statistical techniques to explore agreement between individual reports and expert ratings of workplace exposures in several industries and investigate the sociodemographic influences on this agreement. Methods A cohort of 1,282 employees at 4 industries/14 worksites answered questions on workplace physical, chemical, and psychosocial exposures over the past 12 months. Occupational hygienists constructed job exposure matrices (JEMs) based on worksite walkthrough exposure evaluations. Worker self-reports were compared with the JEMs using multivariable analyses to explore discord. Results There was poor agreement between the self-reported and expert exposure assessments, but there was evidence that agreement was modified by sociodemographic characteristics. Several characteristics including gender, age, race/ethnicity, hourly wage and nativity strongly affected the degree of discord between self-reports and expert raters across a wide array of different exposures. Conclusions Agreement between exposure assessment tools may be affected by socio- demographic characteristics. This study is cross-sectional and therefore, a snapshot of potential exposures in the workplace. Nevertheless, future studies should take into account the social contexts within which workplace exposures occur. Am. J. Ind. Med. 53:1019– 1031, 2010. ß 2010 Wiley-Liss, Inc. KEY WORDS: exposure assessment; occupational epidemiology; job exposure matrix; agreement; discord; higher estimation; lower estimation INTRODUCTION In exposure assessment for occupational epidemiology, questionnaires are commonly used to collect exposure data, largely because they are simple and cost-efficient [Nieu- wenhuijsen, 2005; Perry et al., 2006]. Although other qualitative and quantitative (e.g., biomonitoring) methods for data gathering exist, questionnaire surveys are a major ß 2010 Wiley-Liss, Inc. Contract grant sponsor: NIOSH; Contract grant numbers: R01OH07366-01, R01 OH07366-01S. *Correspondenceto:Dr.GraceSembajwe,ResearchFellow,DepartmentofSociety,Human DevelopmentandHealth,HarvardSchoolofPublicHealth,Boston,MA02115. E-mail:grace___sembajwe@dfci.harvard.edu Accepted9 January 2010 DOI10.1002/ajim.20821.Publishedonline19March2010inWileyOnlineLibrary (wileyonlinelibrary.com). 1 Department of Work Environment, University of Massachusetts Lowell, Lowell, Massachusetts 2 Department of Society, Human Development and Health, Harvard School of Public Health,Boston,Massachusetts 3 Center for Community-Based Research, Dana-Farber Cancer Institute, Boston, Massachusetts 4 Department of Statistical Analysis and Research, New England Research Institute, Watertown,Massachusetts Noneoftheauthorshaveanaffiliationwithanorganizationthat,totheirknowledge,hasa directinterest,particularlyafinancialinterest,inthesubjectmatterormaterialsdescribed.