The purpose of this paper is to analyse the character of response bias in a Danish survey among native Danes and immigrants from Turkey, Iran and Pakistan and thus enable the tailoring of future surveys to particular immigrants groups. We find that both contact rates and cooperation rates are lower for immigrants, leading to a significantly lower overall response rate. We also find important differences between groups - immigrants from Pakistan are especially difficult to reach, while refusals are particularly high among those from Turkey. Language is likewise important as a very large share of women could not be interviewed in Danish. We analyse not only the determinants of the probability of contact and the probability of cooperation but also the determinants of overall nonresponse, by looking at individual characteristics and observable interviewer characteristics. We find that the characteristics of the sample persons are important for both contact and cooperation rates, with different factors affecting each. Yet none of the observable interviewer characteristics appear to affect the response rate. Furthermore, after controlling for all the other variables, we find that the lower probability of response among immigrants compared to native Danes persists. The analysis clearly points to the need for
tailoring surveys directed to immigrant groups to avoid response bias.