Data Analysis and Related Applications, Volume 2

7The Implementation of Hierarchical Classifications and Cochran’s Rule in the Analysis of Social Data

In social sample survey research and the census, international organizations have developed classifications for the measurement of background variables such as the level of educational attainment (ISCED), economic activities (ISIC) and occupations (ISCO) to ensure the cross-national and overtime comparability of measurement. In this context, Eurostat developed the Nomenclature des Unités territoriales statistiques – Nomenclature of Territorial Units for Statistics (NUTS) and the European socio-economic classification (ESeC). All of these classifications – with the exception of ISCED – are categorical schemas, defined hierarchically to allow for different levels of detail in the analyses. However, when performing bivariate analyses with these classifications, Cochran’s well-known rule of thumb of expected values larger than five in order to use the chi-square test should also be taken into account in presenting the results. In this chapter, we investigate the implementation of ESeC to regions (NUTS) to demonstrate how to statistically decide on the appropriate level of classification to be used in the analysis of social data. The analysis is based on the 2016 European Social Survey (ESS) datasets for five European countries: Austria, Belgium, France, Ireland and Italy.

7.1. Introduction

In social sample survey research and the census, as mentioned in previous work (Yfanti and ...