Developments in Sociology

Robert Burgess, Anne Murcott

• Publisher: Routledge
• Year: 2014
• eBook ISBN: 9781317875925
• Topic: Social Sciences
• Subtopic: Sociology
• Index: Social Sciences

Appropriate as a supplemental text to courses in Sociology. Providing an overview grounded in research. Developments in Sociology focuses on the major areas of theoretical, methodological and substantive developments in sociology. Each author takes a field of study in which they are an acknowledged expert and highlights the way in which the subject has developed over the last fifty years.

part 1

Theory, methodology and methods

chapter 1

Sociologists and the survey: potential and pitfalls

Sara Arber

This chapter is dedicated to Cathie Marsh who presented a paper on developments in quantitative methodology to the President’s Symposium at the British Sociological Association annual conference in 1991. It was a great loss to British sociology that Cathie died of cancer on New Year’s Day 1993.

Introduction

British sociology has never whole-heartedly embraced the survey, unlike our North American cousins. During the postwar period, the dominance of surveys in American sociology led to numerous critiques, which had a profound impact on a generation of British sociologists. Students still quote C. Wright Mills’ (1959) characterization of social surveys as representing ‘abstracted empiricism’ which involves no sociological imagination, and Cicourel’s (1964) argument that surveys impose a ‘deterministic grid’ on respondents which artificially generates attitudes and obscures respondents’ meanings. Cathie Marsh played a key role in re-establishing the value of surveys within British sociology. Her book The Survey Method (1982), provided a trenchant critique of Cicourel (1964) and the ‘anti-positivists’, and argued convincingly that social surveys can measure respondents’ meanings.

British sociologists, and particularly graduate students, are usually encouraged to collect their own data. This emphasis, together with constraints of time and money, has led to a preponderance of smaller-scale qualitative research in British sociology, as well as many poor-quality surveys. Although an increasing amount of high-quality national survey data is available, it is only analyzed by a small number of sociologists and few graduate students. Government surveys are rich sources of data for secondary analysis, since many are underanalyzed, both from a statistical and a theoretical viewpoint.

This chapter demonstrates that survey researchers require sociological imagination, that surveys have contributed much to sociological understanding, and that the secondary analysis of large, primarily government, surveys has untapped potential as a source of sociological insights. Although such data are ‘official statistics’, government surveys must be distinguished from official statistics based on administrative sources, such as crime, suicide, and unemployment statistics. The problems associated with administrative statistics (Kitsuse and Cicourel 1963; Hindess 1973; Scott 1990; Levitas and Guy 1996; Dorling and Simpson 1999) are distinct from those associated with government surveys.

Developments in surveys

Surveys comprise the collection of standardized data from a sample of individuals or aggregate units, such as households or small businesses, using structured interviews or self-completion questionnaires. The defining characteristic of a survey is that questions are standardized, allowing the comparison of responses, and surveys aim to be representative of a specified population, so that inferences can be made from the survey results to the wider population from which the sample was drawn.

The methodology of surveys has developed in numerous ways since the 1950s. Telephone surveys are more common and in some areas of market research have largely replaced face-to-face interviews because of the cost advantages (Lavrakas 1986; Frey 1989). Telephone surveys developed hand-in-hand with the development of computer-assisted telephone interviewing (CATI), which alters the process of coding, data entry and editing, as well as the relationship between the interviewer and respondent. The fall in the cost of laptop computers in the 1990s prompted the development of computer-assisted personal interviewing (CAPI), which is now used on nearly all government surveys and surveys conducted by large research organizations (Sainsbury et al. 1993; Couper et al. 1998). For example, the British government Labour Force Survey (LFS) uses CAPI to collect data at the first household interview, with subsequent interviews using CATI, if the respondent has a telephone and is agreeable (OPCS 1993a).

Many parts of the survey process are increasingly being controlled through computer programs, a development known as CASIC (computer-assisted survey information collection) (Couper et al. 1998; Westlake et al. 1998). This trend is transforming the survey data collection process and is likely to continue. The required substantial investment in hardware and software means that CASIC can be embraced by government and large survey organizations, but is largely outside the reach of sociologists wishing to conduct their own one-off survey.

Over recent years developments in sampling have led to modifications in sample designs to reduce sampling errors, e.g. by improving the precision of stratification factors (Bruce 1993). The representativeness of samples has been enhanced by using sampling frames with more complete coverage. For example, government surveys have changed from using the electoral register to the Postcode Address File (PAF) as a sampling frame of the population, because of its improved coverage (Butcher and Dodd 1983; Dodd 1987; Butcher 1988; Foster 1993).

Technical advances in data storage, processing and handling, together with the reduction in the cost of computing, have encouraged the growth in the number and size of surveys. In the 1960s, all computing was undertaken on mainframe computers with limited capacity to hold and analyze large datasets; overnight turnaround was the norm and statistical analysis packages, such as SPSS, were only beginning to be developed. By the early 1980s, personal computers had become more common, but at first only had two floppy disk drives and insufficient disk capacity to hold most survey datasets, let alone the statistical packages to analyze them. By the late 1990s, personal computers had sufficient disk space to run the largest statistical analysis packages and hold and analyze surveys containing upwards of 20,000 cases. In addition, statistical packages have become more user-friendly, capable of handling more complex survey data structures, and produce high-quality graphical output.

The growth of surveys has also been stimulated by social and political changes. The information needs of government and other organizations have increased alongside the speed of change in society. In the early 1990s, Conservative political ideology, encapsulated in the Citizen’s Charter and the Patient’s Charter (DoH 1992), emphasized that services should respond to the needs and priorities of the individual consumer. Thus, surveys were designed to identify the customer’s priorities and attitudes towards the services provided. The Labour government has continued this emphasis: for example, in 1998, the Cabinet Office launched a people’s panel of 5000 adults to monitor change over time in views about a wide range of public services (Page 1998; Cabinet Office 1999).

The proliferation of survey research makes it all the more important to evaluate the adequacy of survey data, and for sociologists to analyze only data which are considered to meet criteria of reliability, validity and representativeness. A further issue relates to the conceptual assumptions embedded in surveys. These are often implicit, reflecting the underlying conceptual framework of the organization or individuals conducting the research. Inevitably surveys, like sociologists, are prisoners of specific values and theoretical assumptions, which need to be subjected to critical scrutiny. Irvine et al. (1979) and Dorling and Simpson (1999) demonstrate that the collection of official survey statistics is not a purely technical matter, showing how conceptual assumptions influence both the production and presentation of official statistics. Only certain types of data are collected by the government. Data of interest to the sociologist may be collected in government surveys, but are presented in a way that reflects contested conceptual assumptions: for example, sexist assumptions about the identity of the head of household or that the husband’s occupation is an adequate measure of the social class of married women (Oakley and Oakley 1979). However, secondary analysts may be able to apply alternative conceptual assumptions in their own analyses, for example by classifying women by their own rather than their husband’s occupation or characterizing the household by the characteristics of the highest income earner rather than the head of household (Arber 1997).

The purposes of surveys

Surveys of value to sociologists are conducted for a wide variety of purposes. Bulmer (1978) provides a framework which distinguishes four types of purpose. ‘Intelligence and monitoring’ is an essentially descriptive enterprise, whereby the survey originator is interested in ‘fact-finding’, relating to the collection of demographic and socio-economic data, or information about attitudes and knowledge. Primary research undertaken by sociologists is rarely of this type, but this is the major type of survey research conducted by central government, local authorities and health authorities.

‘Specific problem-orientated research’ is geared towards providing solutions and recommendations to a specific problem, with the problem defined by a particular ‘customer’, such as a government department or health authority. Problem-orientated research may be conducted by sociologists working on government research contracts or for a specified ‘paymaster’.

Bulmer characterizes ‘strategic social science’ as theoretical, but orientated towards providing a better understanding of a social problem, with the aim of producing policy-relevant results, referred to by the Economic and Social Research Council (ESRC) as ‘applied strategic’ research, whereas Bulmer’s fourth type, ‘basic social science’, aims to advance knowledge through theory building and is not designed to have any direct practical implications, referred to by ESRC as ‘basic research’. Both these types are orientated to advancing knowledge and theoretical understanding, and are the main province of the academic sociologist. The ESRC and charitable trusts are the main funders of both ‘basic’ and ‘strategic’ social science, but such funders increasingly emphasize the latter, with a concern that research addresses the needs of ‘users and beneficiaries’.

Surveys conducted by sociologists

This section focuses on surveys designed by sociologists, and the next section outlines surveys conducted by government and other organizations which may be used to address sociological questions.

The restricted amount of funding available from the ESRC for surveys has been one reason why the number of major British sociological surveys barely reaches double figures. Conforming to the canons of conventional sociological enquiry, most of these surveys have explicitly aimed to test specific hypotheses. A classic is The Affluent Worker study in the early 1960s which tested ‘the embourgeoisement thesis’ among affluent male workers in Luton (Goldthorpe et al. 1969). Social mobility and class analysis have been repeatedly investigated by surveys. Glass (1954) examined inter-generational social mobility in the early postwar period and the Oxford Mobility Study surveyed a national sample of men in 1972 (Goldthorpe and Hope 1974; Goldthorpe 1980). Roberts et al. (1977) surveyed economically active men to examine class imagery. Male manual workers were interviewed to examine the degree of constraint and choice in their working lives (Blackburn and Mann 1979). The Cambridge Social Stratification Group later conducted a survey of friendship patterns among men which formed the basis of the Cambridge Social Class scale (Stewart et al. 1980).

The tradition of surveys on social class culminated in the Essex Social Class survey (Marshall et al. 1988), which formed part of a cross-national project researching class structure and class consciousness, initiated by Erik Olin Wright in the USA. For the first time, women as well as men were interviewed, but the survey questions were firmly locked in a male-stream concern with paid employment. Surveys have sometimes been used in community studies, such as the second study of Banbury, which interviewed 1500 men and women, alongside participant observation, examination of documentary sources and neighbourhood studies (Stacey et al. 1975).

Structural changes within the British economy, and particularly the growth of male unemployment, led to surveys to examine the social consequences of economic restructuring. For example, Harris (1987) conducted a survey among redundant South Wales steel workers funded by the ESRC. Pahl’s interest in the ‘informal’ (or black) economy developed into the Isle of Sheppey study (Pahl 1984). He examined unpaid as well as paid work, taking the household as the unit of survey analysis, thereby examining women’s as well as men’s contribution to the household economy. This survey represents a milestone in its conceptualization of work, and its inclusion of young people’s employment within a household context (Wallace 1987).

Sociological surveys relating to class and labour markets reached their zenith with the ambitious ESRC-funded Social Change and Economic Life Initiative (SCELI). In six local labour markets, surveys were conducted of men and women under 60, followed by a household survey which included interviews with the respondent and their partner, and a telephone survey of some 200 employers in each locality (Gallie 1988; Gallie et al. 1994). A great deal of attention went into the design of complex survey instruments, including the collection of full work, employment, family and migration histories. The detailed work and family histories produced data which captured the complexity of individuals’ working and everyday lives, but sacrificed ease of analysis, although the development of easily used software for analyzing event history data has tended to lag behind methods to collect such data.

Surveys funded by ESRC include Finch and Mason (1993), who used the innovative approach of vignettes to assess the normative obligations of caring for older relatives according to the history of inter-family relationships (Finch 1987). Some sociological surveys involve rela-tively small samples, e.g. a team at the University of Keele (Bernard et al. 2000) received ESRC funding to resurvey the family lives of older people in three communities previously studied over forty years earlier. This study only interviewed 200 people in each community, making it difficult to undertake more detailed statistical analysis.

It seems likely that the ‘academic-survey gap’ in Britain (McKennell et al. 1987) will remain substantial or increase, as the expertise of professional survey experts in government and large organizations becomes increasingly divorced from that of academic sociologists. The high costs of surveys and lack of technical survey infrastructure within university sociology departments means that, in future, sociologists are less likely to seek funding to...

Table of contents

• Cover
• Half Title
• Title Page
• Copyright Page
• Table of Contents
• Tables
• Notes on contributors
• Series preface
• Preface
• Introduction
• Part One: Theory, methodology and methods
• Part Two: Substantive areas
• Part Three: Policy and problems
• Index