Interview mode effects on estimates of need for alcohol and drug treatment among welfare recipients: evidence from a quasi-experiment

Employing a quasi-experimental design, this study compared responses from face-to-face and telephone surveys of welfare recipients to assess the effects of survey modality on prevalence estimates of alcohol and drug abuse and need for treatment. Samples were drawn from adult welfare recipients in Oklahoma in 2001. We generated 1-year and 30-day prevalence estimates of alcohol and drug use, employed several common measures of substance-related behavioral and physical problems to estimate treatment need, and used odds ratios to compare prevalence estimates. Results indicated that telephone surveys provided estimates that were (1) statistically no different from those of face-to-face administration in the categories used to estimate treatment need and (2) significantly higher than those of face-to-face surveys for admission of ever having a substance problem and driving after consuming alcohol or drugs. We conclude that, for welfare recipients, telephone surveys provide similar data quality about treatment need as do more costly face-to-face interviews, a finding that is important and of practical concern to both practitioners and methodologists. More broadly, we argue that further survey methodology research is required for specific populations and subgroups that are often the focus of criminal justice and public health research and response.