David Rose, David Pevalin and Karen O’Reilly, March 2001
The support of ESRC, the Office for National Statistics and the University of Essex is gratefully acknowledged. The work reported here is part of the scientific programme of the Institute for Social and Economic Research (ISER) conducted under ESRC Grants H542255001 and H501265031.
Introduction
Until 2001, the UK had two official socio-economic classifications (SECs), Social Class based on Occupations (SC) and Socio-economic Groups (SEG). Each was based on different principles and conceptions and there was no straightforward mapping between them. Here, we discuss the work undertaken to replace these SECs, thereby producing a new social classification to be used from the time of the 2001 Population Census.
SC was first introduced in 1913, but was revised into something like its present form in 1921. SEG was produced in 1951 and substantially revised in 1961. SC is the most widely used in analysis, but has also been subject to some severe criticisms. SEG is less criticised, although it shares the same fundamental problem as SC: what are these SECs measuring?
Given this background, the Census Division of the former Office of Population Censuses and Surveys (OPCS, but now Office for National Statistics, ONS) requested the Economic and Social Research Council to undertake a zero-based review of SC and SEG. Were SECs still necessary? If so, might it be that two occupationally-based SECs could be one too many?
The Review of Social Classifications
The first report on the Review (Rose, 1995) demonstrated a continuing need for government SECs. It also argued for a single, revised, occupationally-based socio-economic classification which united the most important features and advantages of both SEG and SC.
The second phase of the Review (Rose and O’Reilly, 1997) involved the task of producing, validating and testing an interim version of the revised SEC, using data from the ONS Omnibus and other surveys such as GHS). After the second phase, an edited book, Constructing Classes, was produced. The third phase of testing and validation work, using especially collected data from the LFS was completed in 1998 (Rose and O’Reilly 1998).
The new SEC was then given the title of the National Statistics Socio-economic Classification (NS-SEC). In the fourth phase of the Review, the NS-SEC was re-based on SOC 2000 . A User Manual for this new version of the NS-SEC is available on the ONS website.
Conceptual basis for the NS-SEC: in terms of its conceptual basis, the NS-SEC follows a well-defined sociological position that employment relations and conditions are central to delineating the structure of socio-economic positions in modern societies. SEG, designed by a social scientist, David Glass, has a greater affinity to sociological conceptions of social class than the current SC and was therefore already amenable to this conception. Moreover, the proposed conceptual basis, as operationalized in the widely-used Goldthorpe class schema, has already been subjected to a good deal of impressive criterion and construct validation.
Why do we need an SEC conceived and constructed in this manner? One of the strengths of the approach we have taken in the Review, indeed its underlying principle, is that the NS-SEC offers not necessarily improved statistical associations over the old SECs, but that it lends itself to the possibility of explaining the associations we find. Because we know what the NS-SEC is measuring – employment relations and conditions, i.e. aspects of the work situation and the labour contract – we can construct causal narratives which specify how the NS-SEC links to a range of outcomes via a variety of intervening variables. This is decidedly not the case for the old government SECs, since it is not clear what they measure. They may show statistical associations with dependent variables of interest, but they do not lend themselves to causal explanations.
Properties of the NS-SEC: the NS-SEC is nested, with an operational version which can be collapsed into a smaller number of categories for analytic purposes. Because of the close relationship between SEG and SC this strategy will yield high continuity between the revised SEC and both of the current classifications.
Population coverage: government, academic and private users of SECs have stressed the importance of including within the NS-SEC as many as possible of those adults not in paid employment. Recent research by sociologists demonstrates that classifying individuals not currently in paid employment by their last main job is a satisfactory procedure, even for those who have been out of the workforce for many years. For this reason the Review committee has recommended that both the Census and government surveys collect information on last main job for all those not in paid employment.
Registration and other administrative data: it is recognised that the level of detail about occupation and employment status available for deriving SECs from registration and other administrative data may not be as extensive as it is for Census and survey data. Hence reduced and simplified forms of NS-SEC are available.
Validation
The NS-SEC has been subjected to both criterion and construct validity tests. Criterion validation has involved assessing the relationship between the measure itself and some of the criterion variables for which it is a proxy. Data collected on the ONS Omnibus and Labour Force Surveys have demonstrated that the NS-SEC successfully discriminates between positions in terms of varying employment relations.
However, it has been argued that criterion validity has limited usefulness in the social sciences and that, where possible, one should also assess the construct validity of a measure. This involves assessing how the measure relates to other variables in ways predicted by theory, and in substantive areas of importance to government researchers and social scientists. We have therefore examined the construct validity of the NS-SEC using a variety of relevant data sets (GHS, 1% Longitudinal Study, SARS, BHPS). These analyses provide positive evidence for the construct validity of the NS-SEC by distinguishing, for example, groups with different mortality risks, and by showing that the NS-SEC produces comparable health gradients to those of the old SECs in terms of both limiting long-standing illness and self-assessed health.
The full operationalization of the NS-SEC required the collection of primary data on employment relations and conditions for the Occupational Unit Groups of the Standard Occupational Classification (SOC). Each OUG by employment status combination was examined for its position within the NS-SEC, beginning with the fifty largest OUGs which between them account for over half of all employees. To help us to achieve this, we had a set of seven variables on employment relations (dealing with matters such as mode of payment, job security and autonomy) which were derived from questions carried on a recent quarter of the Labour Force Survey. This type of exercise could be repeated inter-censally for NS-SEC validation and revision purposes.