SMABS 2004 Jena University
SMABS 2004 Home Organization About Jena Sponsors Links Imprint / Contact SMABS Home

European Association of Methodology

Department of methodology and evaluation research

Jena University

Contributions: Abstract

Assessing the quality and comparability of data between countries in cross-national studies using Multiple Correspondence Analysis

Jörg Blasius
University of Bonn

Meaningful international comparisons based on social surveys presuppose uniformly high quality of data for each country and common understanding of the items used for the comparisons. If one cannot guarantee that the quality of data and the understanding of the questions is high in all countries, the results in international comparison are biased.

To assess the quality of data in an international survey, we utilize the 1994 ISSP data set, which focuses on family and gender roles. This data contains a large number of opinion statements in the areas of women and work, marriage, and children using five-point Likert response formats, with a neutral "neither agree nor disagree". It also includes a "can't choose" response as a second possible non-substantive response.

With multiple correspondence analysis we discuss a technique for assessing the quality of data within a set of Likert-type items within 24 countries participating in the ISSP 1994. The analysis shows that the response structure of the gender role items varies substantially by country. In some countries, such as Norway and The Netherlands, a clearly interpretable underlying structure of support for single versus dual household earner emerges. In other countries, including Japan and some of the East European countries, the underlying dimensions are less clear and appear to have high proportions of measurement error. The analysis highlights some of the difficulties in quantitative cross-cultural comparisons.