675 – Practical Aspects of Survey Design and Analysis
First Reproducible Nationwide Survey on Substance Use in Brazil: Survey Design and Weighting
Pedro Luis do Nascimento Silva
IBGE-ENCE
Mauricio Teixeira Leite de Vasconcellos
IBGE-ENCE
Francisco Inacio Pinkusfeld Monteiro Bastos
FIOCRUZ
Raquel B. De Boni
FIOCRUZ
Neilane Bertoni dos Reis
Inst
Carolina Coutinho
FIOCRUZ
Jurema Mota
FIOCRUZ
The paper describes the study design adopted in the III Brazilian Household Survey on Substance Use (BHSU-3). This is the first truly reproducible nationwide survey to investigate substance use and related issues carried out in Brazil in 2015. The study adopted a stratified multi-stage probability sampling design and interviewed 16,273 individuals in their households, using face-to-face paper and pencil interviewing. The study population included all residents of private households aged 12-65 at the time of the survey. Calibration weighting was used to compensate for differential age-sex nonresponse. R statistical software was used throughout weighting and tabulation, thus enabling fully reproducible results by analysts having access to the survey microdata. Innovative combination of proven survey methods and statistical leadership of study design team were crucial elements for successful planning, realization and conclusion of the study.