Obtaining accurate survey data about ethnic minorities is not easy. Ethnic minorities are usually underrepresented in surveys. Furthermore, it is not certain that people that do take part in surveys are representative of the group one is interested in. Do, for instance, only people with high education take part? To ensure that everybody can participate, one needs a tailor-made approach, such as translating the questionnaires and using interviewers with the same ethnic background as the intended respondents. This tailor-made approach can lead to higher response, but does a translated questionnaire still measure the same things as the original? In order to be able to trust the results, survey data needs to be of good quality, different groups need to be readily comparable - in spite of linguistic differences - and the data should also, preferably, be available quickly. How quickly one wishes to obtain the results and the overall budget also determine the upper limit of the data quality: to get a high response, for instance, one needs a longer fieldwork period.
The relation between the survey design and the quality of the survey data is the main focus of this dissertation. With respect to data quality, the focus is on two aspects of data quality that seem the most relevant in this context: accuracy and comparability. For accuracy, we study mainly 1) representation, or how well the population under study is represented by the respondents of a survey, and 2) measurement, meaning how the manner in which data has been collected among respondents may affect the answers they provide. With respect to comparability, the focus is on the comparability of data collected from different minority groups. The data come from several large-scale surveys conducted among non-Western minorities in the Netherlands. Attention is also paid to the costs of surveys among non-Western minorities.