Introduction

In Denmark, as in other European countries, it has turned out to be very difficult to achieve a satisfactorily high participation of different immigrant groups in surveys. Both in general population surveys and in surveys specifically targeted to immigrant populations, non-response rates are typically relatively high, and the basic lesson from these surveys is that interviewing immigrants requires considerations other than those applied to interviewing the majority population. Nonetheless, few studies have focused on non-response among immigrants (see, e.g. Feskens, Hox, Lensvelt-Mulders & Schmeets 2007; Dale & Haraldsen 2000; Van den Brakel, Vis-Visschers & Schmeets 2006).

The main reason for concern about high non-response rates is that nonresponse may generate bias problems. In particular, non-response poses a problem if it is correlated with the variables of interest. Previous research has shown that while non-response bias occurs, the non-response rate of a survey alone is not a good predictor of the magnitude of the bias (Groves 2006). Blind efforts to reduce non-response may increase the bias problems. Instead, efforts at reducing non-response should be guided by knowledge about the character of the non-response bias and about the ways in which groups are affected by efforts to reduce it (Groves 2006).

A survey on education and labour market affiliation among persons in the 18-45 age group from three different immigrant groups (with origins in Iran, Turkey and Pakistan) and native Danes, carried out in 2006, provided the opportunity to look further into the characteristics of non-respondents in the different groups of the population. Although great effort was made in the survey design and data-collection process, the survey more than fulfilled the expectation that non-response among the immigrant groups would be high – on average the response rate among the three immigrant groups was about 20 percentage points lower than for the Danes.

However, as the sample for the survey was drawn from the Danish population register, we were able to compare the groups of respondents and non-respondents within each of the four population groups on the basis of administrative register data available at Statistics Denmark. Furthermore, we had access to some information on the interviewers from the survey organisation and to rather detailed data on the number of contacts and information on reasons for non-response.