Introduction

Carrying out surveys among ethnic minority (EM) groups raises a number of problems above and beyond those to do with surveys of the general population. These difficulties arise because EM groups may appear with a low frequency in the population, may be geographically unclustered, and may be difficult to access.

This chapter examines some of the key issues to do with designing rigorous, high-quality surveys of ethnic minorities, whether as a ‘boost’ to increase their numbers in a general survey or as a targeted survey among particular EM groups. By ‘high quality’, the survey needs not only to provide accurate data, but it also must relate to the needs of the users (e.g. it should be relevant, timely and accessible). While having indicators of survey quality is important – for example, to evaluate the usefulness of the data provided and to differentiate between ‘good’ and ‘bad’ data – this chapter is not concerned with issues of general survey design or quality (for a general discussion of survey methodology, see Groves, Fowler, Couper, Lepkowski, Singer & Tourangeau 2009; Fowler 2008; Lyberg, Biemer, De Leeuw, Dippo, Schwarz & Trewin 1997).

The tension between cost and quality that is found in all surveys is even more acute when carrying out surveys of ethnic minorities, since these specialist surveys tend to be more expensive in terms of cost per participant. Therefore, an EM survey is likely to cost more than a survey among the general population of a similar size and quality. This will need to be reflected in the survey budget (unless reductions in sample size or quality are acceptable).

The following sections in this chapter cover approaches to categorising EM groups and the classification commonly used in surveys in the United Kingdom; the practicalities influencing decisions about which EM groups to include in surveys; methods for sampling EM groups, including an example of how a probability sample can be designed cost-effectively using stratification and clustering; and data-collection issues including response rates, the translation process and ethnic matching of fieldworker and respondent.