Republic of Georgia estimates for prevalence of drug use: Randomized response techniques suggest under-estimation
Abstract
Background: Validity of responses in surveys is an important research concern, especially in emerging market economies where surveys in the general population are a novelty, and the level of social control is traditionally higher. The Randomized Response Technique (RRT) can be used as a check on response validity when the study aim is to estimate population prevalence of drug experiences and other socially sensitive and/or illegal behaviors.
Aim: To apply RRT and to study potential under-reporting of drug use in a nation-scale, population-based general population survey of alcohol and other drug use. Methods: For this first-ever household survey on addictive substances for the Country of Georgia, we used the multi-stage probability sampling of 18-to-64-year-old household residents of 111 urban and 49 rural areas.
During the interviewer-administered assessments, RRT involved pairing of sensitive and non-sensitive questions about drug experiences. Results: Based upon the standard household self-report survey estimate, an estimated 17.3% [95% confidence interval, CI: 15.5%, 19.1%] of Georgian household residents have tried cannabis.
The corresponding RRT estimate was 29.9% [95% CI: 24.9%, 34.9%]. The RRT estimates for other drugs such as heroin also were larger than the standard self-report estimates.
Discussion and conclusions: We remain unsure about what is the "true" value for prevalence of using illegal psychotropic drugs in the Republic of Georgia study population. Our RRT results suggest that standard non-RRT approaches might produce 'under-estimates' or at best, highly conservative, lower-end estimates.