Effects of Survey Mode on Self-Reports of Adult Alcohol Consumption: A Comparison of Mail, Web and Telephone Approaches
Journal article by Michael W. Link, Ali H. Mokdad; Journal of Studies on Alcohol, Vol. 66, 2005
Journal Article Excerpt
Effects of Survey Mode on Self-Reports of Adult Alcohol Consumption: A Comparison of Mail, Web and Telephone Approaches *.by Michael W. Link , Ali H. Mokdad ALCOHOL CONSUMPTION results in approximately 85,000 deaths each year and is the third leading preventable cause of death in the United States (Mokdad et al., 2004). Heavy episodic drinking is the most common pattern of alcohol misuse and accounts for many of these alcohol-related deaths (Naimi et al., 2003a; Wechsler and Nelson, 2001). Heavy alcohol consumption is also associated with many other health and social problems, such as unintended pregnancy, sexually transmitted infections, violence, crime, child abuse and lost productivity (Naimi et al., 2003b). Valid and reliable public health surveillance is essential to develop a more comprehensive public health response to alcohol misuse. However, participation in general population surveys, one of the primary tools for public health surveillance, is declining, particularly for telephone-based surveys (Groves and Couper, 1998; Steeh et al., 2001). Response rates for one of the world's largest telephone-based health surveys, the Behavioral Risk Factor Surveillance System (BRFSS), declined from 63% to 51% between 1996 and 2001 (Link, 2003). Men and those ages 18-24 years are among the groups most likely to be underrepresented in random-digit dialed (RDD) surveys, thereby potentially leading to an underestimate of alcohol consumption and related behaviors. Furthermore, survey data in general and telephone surveys in particular tend to underestimate alcohol consumption by the general public because of a combination of sampling noncoverage and underreporting of alcohol use (Feunekes et al., 1999; Gfroerer and Hughes, 1991; Lemmens et al., 1992; Nelson et al., 2001). As a result, researchers need to consider alternatives or complements to traditional telephone surveys to ensure the validity and reliability of the data collected.Paper-and-pencil surveys conducted by mail and computer-assisted self-interviews (CASIs) conducted over the Internet could potentially be used in place of or in conjunction with telephone surveys. Mail surveys have an established history in epidemiology, whereas the Web survey is a relatively new tool for epidemiologic research but one that is being used across a number of disciplines, including medicine, dentistry, psychology and sociology (Eysenbach and Diepgen, 1998; Hilsden et al., 1999; Lakeman, 1997; Schleyer et al., 1998). Both mail and Web surveys are self-administered questionnaires, which allow potential respondents privacy and flexibility in completing a survey (Dillman, 2000). When compared with surveys conducted by interviewers, SAQs have been shown to result in higher levels of reporting of sensitive or stigmatized behaviors, including the use and misuse of alcohol (Couper, 2000; Dillman, 2000; O'Reilly, 1994).Web surveys vary considerably in their design and whether or not they employ probabilistic sampling methods (Couper, 2000). Because Internet surveys are a form of CASI they provide a number of benefits over pencil-and-paper approaches. CASI modes have been shown to increase data quality by reducing the level of missing data and "out-of-range" responses and ensuring that instructions for skipping questions based on previous responses are followed consistently (Couper, 2000; Dillman, 2000). CASI approaches also have been shown to produce higher estimates for sensitive behaviors than non-CASI approaches (Tourangeau and Smith, 1996; Turner et al., 1998; Wright et al., 1998). Web surveys do, however, differ from most other CASI modes in that most CASI surveys are completed with an interviewer or field agent nearby. During a Web survey, if assistance is available, it usually must be accessed by calling a toll-flee number or sending an email to a help-desk agent.A number of studies have examined how survey mode can affect the responses obtained to alcohol-related questions, and some have specifically compared mail and Web surveys with other survey modes (Bongers and Van Oers, 1998; McCabe et al., 2002; Miller et al., 2002; Nelson et al., 2003). These studies found few substantive differences in estimates of alcohol use. Little is known, however, about how responses to mail and Web surveys compare with responses to telephone surveys. Moreover, most Web-based surveys of alcohol consumption have focused on younger, typically college and university populations; few have used the Web to survey the general public on this topic (Bason, 2000; McCabe et al., 2002). This is understandable given that Internet availability and penetration is more limited among the general public (Spooner, 2003). In this study, we provide one of the first assessments of how responses to questions about alcohol consumption by adults in the general public compared across mail, Web and telephone modes of data collection. The Web design tested here is a variation of Couper's (2000) "mixed ... End of free preview...
|
To continue reading this publication, you must have a Questia Subscription.
