Introduction
Most Western societies have seen massive international migration inflows during the past decades. Unlike the typical migrant of the 1960s, however, today's migrants can hardly be standardised in terms of skills, gender and country of origin. This results in a high heterogeneity in terms of educational levels, social positions and gender (Koser & Lutz 1998; Kofman, Phizaclea, Raghuram & Sales 2000).
Many authors (see Wimmer & Schiller 2002; Chernillo 2006; Wimmer & Min 2006) have addressed the issue of whether social surveys are (still) able to represent foreigners both with respect to heterogeneity of their cultural background, educational level and position in the hosting country's social structure. There is evidence that foreign minorities in general (especially the more marginal and vulnerable individuals) are under-represented among respondents (Vandecasteele & Debels 2007). There are several reasons for this: under-coverage in sampling frames (Lipps & Kissau 2012), higher non-response rates in cross-sectional surveys (Deding, Fridberg & Jakobsen 2008; Feskens, Hox, Lensvelt-Mulders & Schmeets 2006, 2007; Camarota & Capizzano 2004; Eisner & Ribeau 2007; Jakobsen 2004; Nielsen & Pedersen 2000) and higher attrition rates in longitudinal surveys (Lipps 2007; Peracchi & Depalo 2006). To analyse under-representation, the literature has either focused on selected minorities (Deding et al. 2008) or used a rough distinction between Western and non-Western minorities (Feskens et al. 2007) that is not likely to adequately capture the increasing heterogeneity.
In this chapter we would like to shed light on the question of whether there is a bias in Swiss social surveys regarding the representation of foreign minorities. If so, are there sub-categories according to social class or education within different nationality groups that make them especially prone to being under-represented? A related question refers to the extent to which this possible bias changes depending on the type of survey: does the under-representation found in cross-section surveys increase in panel surveys? We use data from three Swiss social surveys: the Swiss Household Panel (SHP), the Swiss National Labour Force Survey (LFS) and the Swiss sample of the European Social Survey (ESS). We use three different surveys to check if possible patterns of under-represented national minorities are independent of survey design features. In addition, using panel data like the SHP and the rotating LFS panel allows analysing attrition as a second dimension of under-representation.