Sampling Establishments for Social Survey Research

Abstract

Over the last 10 years the Census of Employment has been increasingly used as a sampling frame for government-funded, large-scale national interview surveys. Based on experience of the three Workplace Industrial Relations Surveys, the paper discusses a number of issues surrounding the sampling of organisational units for survey research. These include the definition and choice of organisational units appropriate to a given survey design, and particularly the problems of establishment definition; specific problems of using census data units to provide samples of establishments; issues arising from the age of the frame and an attempt to overcome this by supplementary sampling of new workplaces; the use of role-holders as informants about the workplace; and finally some of the problems of access and their resolution. accommodate this. Two basic adaptations are possible. The characteristics of the social units may be obtained by aggregating or summarising the scores of individuals within each unit. For example, different schools can be compared by averaging individual scores on tests for each pupil, computing the proportion in each gender, using a head-count of pupils for school size and so on to give a data set where every variable is a school-level variable, of which many if not all, are derived from observations on individuals. Alternatively, surveys of social units may involve selecting prominent individuals or role-holders and using their observations about the unit as the survey responses. A survey of school head-teachers would be an example. This paper addresses some of the issues surrounding this latter approach, where the social unit of interest is the whole or part of a work organisation. Many work organisations and virtually all large ones-have a complex structure which any survey design must take account of. Descriptions of the complex structure may vary with the subject-matter of the survey, but in most instances it is useful to recognise two fundamental organisational units: the whole organisation under common ownership or control; and the workplace or establishment-an individual site or address at which part of an organisation's employment-related activities are carried out. In surveys of industrial or commercial organisations, the larger unit is often termed an 'enterprise', while the individual workplace may well be a plant or office. Complex work organisations may contain a variety of subunits which could form the focus of survey research. Large enterprises often consist of a number of companies associated in a complex hierarchical structure; as legal units, the companies themselves may be the appropriate unit for some surveys (Knight, 1979; Smith, 1986). In economic research, the theoretical notion of the 'firm' is sometimes operationalised as a 'product line