Web-Enabled Coding Procedures in the National Survey on Drug Use and Health
Handley, W., Kroutil, L. A., Smarrella, D. J., & Chien, C. C. (2005, May). Web-Enabled Coding Procedures in the National Survey on Drug Use and Health. Presented at International Field Directors and Technologies Conference, Miami Beach, FL.
The National Survey on Drug Use and Health (NSDUH) is anongoing cross-sectional face-to-face household survey. Data are collected quarterly and yield a final sample of approximately 67,500 persons each year. The survey is conducted by RTI International for the Substance Abuse and Mental Health Services Administration (SAMHSA) and is the primary source of information on the prevalence, patterns, and consequences of alcohol, tobacco, and illegal drug use and misuse ofprescription medications in the general U.S. civiliannon-institutionalized population, aged 12 or older. Since 1999, the survey has used a computer-assisted interview mode for data collection. The NSDUH uses open-ended questions to collect information on the use of "other" drugs, other tobacco brands, and other behaviors of interest (e.g., arrests, mental health service utilization) that are not specifically covered in the interview. Responses to these open-ended questions need to be coded numerically for further data processing and analysis. The potential exists for a high volume of responses that require coding each quarter. Consequently, we needed to develop asystem that would (1) reduce coder burden, (2) standardize coding procedures, and (3) reduce the turnaround time from receipt of raw, uncoded data to delivery of coded variables. We have developed anautomated system featuring the SAS statistical software package and a project intranet Web site to facilitate and standardize the coding of data online. The system allows for multiple analysts with appropriate access permissions to code or review data. An additional strength of the system is that it utilizes responses that have been entered in previous years to assign standardized codes and reduce coder burden. Further, the system allows coding and updating of the drug and tobacco brand data to proceed on a flow basis throughout the year, thereby reducing the turnaround time for coding and processing of these critical variables. This presentation will outline the major processes involved in our Web-enabled coding system, including the important steps involved at each stage of the coding and processing. We will illustrate these procedures with dummy data and will provide data to demonstrate the system's impact on coder burden and turnaround time.