Quantitative tales of ethnic differentiation: measuring and using Roma/Gypsy ethnicity in statistical analyses

Abstract

This paper analyses the use of ethnicity in quantitative models, focusing on Roma/Gypsy ethnic affiliation. Three research models are identified, together with characteristic measurement and model specification issues. A path model is estimated as a methodological example in order to explore the association of Roma/Gypsy ethnic affiliation with church attendance in Romania, using data from two surveys with national and Roma samples. Direct, mediated and moderated relationships involving ethnic affiliation are highlighted and discussed in relation to the processes of ethnic differentiation.

Keywords:

• Ethnicity
• measurement
• causality
• survey
• Roma
• gypsy

Acknowledgements

I am grateful for the generous advice, comments and moral support of Dumitru Sandu, Mălina Voicu, Bogdan Voicu and Puiu Lătea, and for the helpful observations and suggestions of the two anonymous reviewers.

Notes

1. Details about the survey are available on the project website: http://www.see-educoop.net/aeiq/outputs.htm (accessed 30 April 2009).

2. Part of this work was carried out at the European Data Laboratory for Comparative Social Research (EUROLAB) – GESIS. Access to the EUROLAB was supported by the European Community under the ‘Structuring the European Research Area’ specific programme, Research Infrastructures Action in the 6th Framework Programme.

3. Estimates of Roma people who self-identify as such in census and surveys range from 2.5 per cent in the 2002 census to 2 per cent or less in nationally representative samples, such as the biannual Public Opinion Barometer of the Soros Foundation Romania. Therefore, I have computed a weighted variable, so that the self-identified Roma respondents represent 2.5 per cent of the total sample, while keeping the total number of cases constant.