Addictive Behaviors, Vol. 25, No. 5, pp. 693–704, 2000 Copyright © 2000 Elsevier Science Ltd. Printed in the USA. All rights reserved 0306-4603/00/$–see front matter PII S0306-4603(00)00076-9 693 Pergamon SMOKING BEHAVIOR AMONG CASINO EMPLOYEES: SELF-REPORT VALIDATION USING PLASMA COTININE HOWARD J. SHAFFER, GABRIEL B. EBER, MATTHEW N. HALL, and JONI VANDER BILT Harvard Medical School Abstract — The veracity of behavioral self-reports is often challenged, particularly when the motivation to avoid stigma and win social approval holds potential to introduce bias into the data collected. This study employed plasma cotinine tests to validate the self-reports of to- bacco use collected from 3,841 casino employees as part of a comprehensive health survey. Rates of discordance were calculated by comparing employee self-reports with results from plasma cotinine tests. This study provides evidence that casino employees can provide valid self-report data. Further, discordance rates of self-reported tobacco use vary according to op- erational definitions of tobacco use. These findings highlight the methodological importance of recognizing the inherent heterogeneity of smoking behavior. © 2000 Elsevier Science Ltd. This article reports the results of a validation analysis performed on the self-reports of tobacco use obtained from a large sample of casino employees. Rarely studied, these workers represent a unique population with higher-than-typical health risks (Shaffer, Vander Bilt, & Hall, 1999) who labor in an environment that is often based on distrust (Frey, 1986; Frey & Carns, 1988). The validity of self-reports has been a recurrent is- sue across a variety of scientific disciplines, occasionally calling into question the util- ity and value of even the most highly esteemed research (e.g., Ericksen, 1998; Free- man, 1983; Weinhardt, Forsyth, Carey, Jaworski, & Durant, 1998). Concerns about the integrity of self-reported data escalate when the population from which the sample is drawn is popularly considered less than honest and forth- right. Population credibility deficits, real or imagined, can lead to serious doubts about the validity of data. For example, the veracity of self-reports from insurance and car salesmen, two of the least trusted occupations, likely will be held to closer scrutiny than those of pharmacists and clergymen, who rank among the most trusted profes- sions (McAneny & Saad, 1999). However, determining the validity of data by judging the moral character of any sample risks the fundamental attribution error (Ross, 1977). That is, explaining deceptive behavior with a model that focuses on an individ- ual’s dispositional and personality-related traits ignores the possibility that external, situational factors also may be at work. Thus, when considering the veracity of self- reported data, it is important to consider the social context in which each individual self-report is elicited. Babor, Brown, and Del Boca (1990) propose that the question- answering process occurs in a social context that consists of four factors: characteris- Casino, Inc. provided funding for this project. This project received additional funding support through a grant from the National Center for Responsible Gaming (NCRG). The National Center for Responsible Gaming is the first national organization devoted exclusively to funding independent research on problem and underage gambling. The authors wish to thank the management and employees of Casino, Inc. and Stu Gurnee and Brenda Broberg of Informed Technologies for their important role in this project. Requests for reprints should be sent to Howard J. Shaffer, Harvard Medical School, Division on Addictions, 220 Longwood Avenue, Goldenson Building, Room 523, Boston, MA 02115-5729; E-mail: howard_shaffer@hms.harvard.edu