To Measure is to Know? A Comparative Analysis of Gender Indices

Abstract

In this paper, I present a comparative analysis of five cross-country composite gender indices. Although there is a relatively high correlation between the indices, the overlap of underlying indicators is low. Country rankings both at the top and at the bottom have parallels but are quite distinct. The differences are explained in two ways: methodologically and theoretically. The methodological differences concern in particular weights, capping, and aggregation. The Capability Approach helps to explain the different focus of each index by distinguishing between four stages of human development, which include distinct types of indicators. The substantial differences that exist between the gender indices require a cautious selection between these for research and policy analysis. This is shown in a few examples with policy variables. Finally, I present a set of three decision trees, which enables an informed choice between the indices.

Keywords:

• gender indices
• women
• inequality measurement
• index
• policy research

Notes

For the calculation of the average correlation, the auto-correlations have been ignored.

For the calculation of the average overlap, the 100% overlap between the same indices has been ignored.

The rationale for SIGI states it thus: “In many countries of the world, social norms lock women in traditional roles, for example activities as housewives, responsible for taking care of the children and preparing food. SIGI variables try to capture the social institutions that manifest such stereotypes, for example by measuring the percentage of girls; who get married at very young ages, and indication of forced or arranged marriages” (http://genderindex.org/content/rationale-social-institutions-and-gender-index).

Combining indicators does not eliminate measurement error, but if one assumes that errors are uncorrelated between data sources and that the size if the error is constant across items, then the combination of multiple sources will progressively reduce error as the number of indicators increases.

ILO Convention 111 concerns discrimination in employment and occupation.

Except for the first question, which asks about dimensions.