The effects of interview mode on 18to 34-year-olds' self-reported tobacco, alcohol, marijuana, and cocaine use were investigated. RDD telephone and personal drug use surveys of the state of New Jersey were conducted in 1986-87. In the personal interview, respondents recorded their drug use on selfadministered answer sheets. Compared to the area probability sample in the personal interview, RDD yielded a sample of blacks higher in income and education, and more likely to be married and currently employed; white SES was also higher in the telephone sample, but to a lesser degree than for blacks. Controlling for demographic characteristics and RDD's exclusion of nontelephone households, the telephone survey yielded significantly lower estimates of blacks' alcohol consumption, and lifetime and recent marijuana use. Whites' alcohol consumption was slightly lower by telephone; otherwise, estimates of whites' use of the four substances were nearly identical in the two modes. Sample coverage, respondent demographic characteristics, and racial matching of interviewer and respondent did not account for the significant mode differences. Characteristics of the interview situation itself, such as provision of privacy in the self-administered format, may have influenced tendencies toward socially desirable responding to a threatening topic such as drug use. Telephone interviewing has increased in popularity relative to personal interviewing due to cost savings, efficiency, and the availability of comWILLIAM S. AQUILINO iS a Research Scientist with the Center for Demography and Ecology, Department of Sociology, University of Wisconsin-Madison. LEONARD A. LO SCIUTO is Director, Institute for Survey Research, Temple University. The study was supported by grant No. RO1-DA04280 from the National Institute on Drug Abuse, Bethesda, Maryland, and by PHS National Research Service Award #AG00129-03 from the National Institute on Aging. Surveys were conducted by the Institute for Survey Research at Temple University. The personal interview data were collected by the New Jersey Household Survey on Drug Abuse, funded by the Department of Public Health, State of New Jersey. The authors wish to thank Linda Thompson for helpful comments on an earlier version of this paper. Public Opinion Quarterly Volume 54:362-395 ? 1990 by the American Association for Public Opinion Research Published by The University of Chicago Press / 0033-362X/90/0054-03/$2 50 This content downloaded from 157.55.39.248 on Thu, 28 Jul 2016 05:21:29 UTC All use subject to http://about.jstor.org/terms Interview Mode Effects in Drug Use Surveys 363 puter-assisted methods to speed data collection and processing (Groves, 1983; Klecka and Tuchfarber, 1978). Compared to face-to-face interviewing, telephone surveys are cheaper, have shorter data collection periods, and lend themselves to closer monitoring of interviewer performance, thus enhancing quality control during data collection. The utility of the telephone survey has been demonstrated in marketing, political polling, health, and numerous forms of attitudinal research. Herzog and Rodgers (1988), for example, found almost no differences in response distributions obtained from telephone and personal interviews in a survey of older Americans' political attitudes, health, well-being, and life satisfaction. Telephone approaches have been shown to produce reliable and valid results for relatively nonthreatening interview topics. Much less is known, however, about the efficacy of telephone surveys for highly threatening interview topics, such as sexual behavior, criminality, and the use of illicit drugs. Surveys on these topics have relied almost exclusively on costly and timeconsuming face-to-face interviewing. This research addressed the question: To what extent does the validity of sensitive surveys differ by mode of interview? Specifically, is there a difference between faceto-face surveys using self-administered questionnaires (SAQs) and telephone surveys in their estimates of lifetime and current use of tobacco, alcohol, marijuana, and cocaine? Our research questions centered on comparing telephone and personal survey modes in terms of nonresponse, demographic characteristics of the respondents, and validity of drug use estimates: (1) what differences in sample demographic characteristics are produced by multistage area probability sampling versus Random Digit Dial (RDD) methods; and (2) do personal surveys using SAQs and telephone surveys provide different estimates of drug use for the same population? If so, to what extent are mode differences attributable to sample coverage, nonresponse bias, or to the interview situation itself?