A Note on the Efficiency of Geographic Stratification

Abstract

Often times, administrative convenience or the need for estimates for the domain of study dictates the use of Stratified Sampling using Geographic Stratification (GS). This method of strata boundary determination is ill adapted in practice due to its less amenability to mathematical approach. Despite its poor performance in terms of Precision, empirical investigation using four sets of real-life data with varying degrees of skewness shows that GS yields minimum Mean Square Error (MSE) value when compared with popular strata construction methods like cumulative square root of frequency method Dalenius and Hodges, DHR (1959) and Geometric Stratification of Gunning and Horgan, GMS, (2004) using optimum allocation.