Estimation of small area proportions under a bivariate logistic mixed model

Abstract A variety of data is geographic interest but not available at a small area level from large-scale national sample surveys. Small estimation can be used to estimate parameters target variables detailed geographical scales based on relationships between the and relevant auxiliary information. proportions topic great in many fields study, where binary are diffused, such as labour force, business, social exclusion The univariate generalised mixed model with logit link function widely adopted this context. literature has shown that multivariate estimators, correlations among response taken into account, provide more efficient estimates than traditional approaches. However, problem been studied yet. In article, we propose bivariate estimator function. simulation study an application presented evaluate good properties compared its setting. We found extent improved efficiency over approach associated degree correlation area-specific random effects intraclass correlation, whereas it strongly related size.