Conferences

This paper discusses methodological challenges researchers face in collecting and analyzing quantitative data on the Roma minority in Romania. Specifically, we are interested in two issues: (1) the operational definition the Roma minority a priori in the sample frame and post hoc in the actual date sets, and (2) the sensitivity of questionnaire items for this minority in comparison with the majority, or the general population. Estimates of the Roma minority in Romania range from 2.5% of the total population (2002 Census) to up to 7.9% (the World Bank). This discrepancy is primarily driven by the definitions used in identifying the target population: self-identification gives much lower estimates than hetero-identification, as Roma are reluctant to express their identity in a society that strongly discriminates against them. Using various survey data we analyze the consequences of four forms of delineating the Roma minority: (1) self-identification, (2) interviewer’s identification, (3) institutional identification, and (4) community identification. In particular, we demonstrate that demanding on the definition of Roma, the basic demographic characteristics of this minority – such as age of marriage, number of children, or household composition – have different distributional properties. In addition, we analyze whether persons from the Roma minority tend to respond to different questions through the same type of answers. Specifically, we are interested in the extent to which response-set poses a measurement problem. We assess this through the proportion of ‘don’t know’ and ‘yes’ answers across different questions for self-identified Roma in relation to non-Roma and hetero-identified Roma.