Design and analysis of two-phase sample survey: Statistical application software

Abstract

The purpose of this project was to develop computer application software to estimate parameters in double sampling design. Double sampling or two-phase sampling is used in a wide variety of surveys which includes situations of non-response, missing data, populations that are rare, remote or hard to access, sampling at multiple occasions and in quality control studies. In making inference about any population, we usually estimate parameters like the means, variances and standard errors by using sample data obtained. However, the formulae involved in these calculations are complex and quite hard to memorize and hence time consuming. The application software developed randomly selects a sample, estimates parameters involving double sampling for stratification, optimum allocation technique, double sampling for ratio and regression estimation and Double sampling for analytical comparisons. Previous work done in the area of double sampling has failed to address the specific need for a quicker way to estimate parameters in the double sampling applications. With an application in place where by the user just inputs the sample data, then estimates are easily obtained under double sampling. Application software was developed using C++ programming language whereby the formulae used in these different applications were in-built in the system to enable easy computation of parameters for the different well-known applications like double sampling for stratification, double sampling for ratio estimation, double sampling for regression estimation, optimum allocation and repeated sampling of the same population. It is expected that this application will be used as a tool by researchers and scholars for easy computations while performing analysis in the area of double sampling. The application will ease the computation of the parameters under the situations of non-response, missing data, populations that are rare, remote or hard to reach and in quality control studies since double sampling is better used in these scenarios. Recommendations based on this study and areas of further research related to this field of study have been presented and these include to develop a graphical user interface for the system and to consider designing a statistical package that handles non-nested double sampling.