Efficiency of Standard Regression Model-Based Ratio-Synthetic Estimators in Sample Surveys Combining Time Series and Cross-Sectional Data

• Activity Number: 541
• Type: Contributed
• Date/Time: Wednesday, August 12, 2015 : 10:30 AM to 12:20 PM
• Sponsor: Survey Research Methods Section
• Author(s): Prabhakar Ghangurde*
• Keywords: efficiency ; ratio-synthetic ; model-based ; time-series ; cross-sectional ; data

Abstract

Efficiency of ratio-synthetic estimator compared to BLUP estimator, both based on cross-sectional data, depends on ratios of harmonic means of auxiliary variable values of sample units to sample sizes in respective domains, domain of interest Ui and complementary domain Uc. Assuming equal unit error variances in model-based domain estimation BLUP estimator of domain mean or total is less efficient when compared to ratio-synthetic estimator(Ghangurde,P.D.; JSM(2014)).In this paper efficiencies of ratio-synthetic estimators based on time series data as compared to the same based on cross-sectional data are analyzed for assumed values of correlation coefficient in AR(1)model for sampling errors in the regression model assuming 5 and 10 time-points or occasions with diagonal covariance matrix for sampling errors. These efficiencies are not affected by small area domain effects in the mixed model assumed to derive BLUP estimator (pages 159-62; Rao,J.N.K.(2003)). Empirical results can be similarly obtained for ratio-synthetc estimators in household surveys with rotation sample designs.