Orientation Observation In-depth interviews Document analysis and semiology Conversation and discourse analysis Secondary Data Surveys Experiments Ethics Research outcomes



Social Research Glossary

About Researching the Real World



© Lee Harvey 2012–2017

Page updated 29 May, 2017

Citation reference: Harvey, L., 2012–2017, Researching the Real World, available at
All rights belong to author.


A Guide to Methodology

7. Secondary data

7.1 Introduction to secondary analysis
7.2 Extent of re-analysis of secondary data

7.2.1 Minimal re-analysis
7.2.2 Significant re-analysis
7.2.3 Meta-analysis

7.3 Nature of the data
7.4 Data sources

7.5 Examining data sources
7.6 Methodological approaches

7.7 Summary and conclusion

7.2.2 Significant re-analysis
An example of significant reananlysis of survey data is that undertaken by Lisa Strohschein and Alvinelle Matthew (2015). They used the 2006 Canadian International Youth Survey, which asked 3101 adolescents in Toronto, aged between 12 and 15, about their problem behaviour. The analysis involved Poisson regression models that revealed parental monitoring, school performance, peer approval of illegal activities and neighborhood social disorder were the main factors associated with
adolescent problem behaviours.

In some cases, a set of data may be re-examined and re-analysed from scratch asking very different questions to that for which the data was first collected.

The study by John Goodwin and Henrietta O’Conner (2005) is an example of significant reanalysis of secondary data. The authors found old questionnaires from a project called Adjustment of Young Workers to Work Situations and Adult Roles, carried out in Leicester between 1962 and 1964. The original research team had completed 882 interviews from the 1150 individuals, the bulk of the interview data has never been fully analysed or published. In the mid-1970s the interview schedules were archived ‘and have remained untouched until recently, when 851 of the original interview schedules were rediscovered’ (Goodwin and O’Connor, 2005, p. 205)

Goodwin and O’Connor claimed that earlier research on youth transitions understated the level of complexity that characterised transitions into work in the early 1960s and 1970s and focused on ‘macro’ or more structural issues such as class and gender. In reanalysing the material from the Adjustment of Young Workers to Work project, Goodwin and O’Connor challenged the extent to which transitions during this time were complex, lengthy, non-linear and single-step. Instead they used the questionnaires to question the assumed linearity and uncomplicated nature of school to work transitions in the 1960s.

In Leicester, the 1960s was a period of excellent employment prospects for young unqualified workers with the wealth of low-skilled jobs available and a low rate of unemployment. However, the orthodox view ‘ignores the fact that many local labour markets were characterized by large fluctuations in their buoyancy and prosperity’. (Goodwin and O’Conner, 2005, p. 203). They concluded:

Despite assertions that past school to work transitions were single step, simple and homogenous, data from the Adjustment of Young Workers to Work Situations and Adult Roles project provides clear evidence of frequent job moves (often to very different job roles and different industries), with many young people having four or more jobs within the first year or so of employment. The findings also question the assumed seamless transition from school to work as it appears that transitional experiences during this period were not straightforward. The data reveals that many of the young workers already felt disillusioned with work and were anxious about their future prospects or concerned about their lack of training. Some of the young people had experienced periods of unemployment either before entering work or between jobs. The young people interviewed also clearly felt increased levels of insecurity (and risk) brought on by being out of work, or when threatened by unemployment. Likewise, the view that young people in the past made homogenized transitions to work, sharing the same experiences with friends, neighbours and relatives is also problematic. For example, over half of the sample interviewed here clearly had individualized experiences entering different firms at different times during the first years of their working life. The data also suggests that many of the young people in this survey had not made the step of disengaging with the family of origin. They consequently remained very dependent on their parents and family for housing, money and decision-making long after they had made the transition from school to work. (Goodwin and O’Conner, 2005, pp. 216–7)

The research does not just provide historical detail but also casts doubt on the prevailing presumption that transitions have moved from smooth and uncomplicated in the 1960s to complex and problematic in the new millennium. They suggest that the fact that the complexities were not highlighted in the 1960s and 1970s may be explained by the changing nature of the theoretical and methodological approaches to researching youth transitions: from exploring the impact of social structure to the current individualised approaches. Thus, those currently involved in trying to understand transitions have become concerned with different phenomena and have different academic preoccupations to those of past scholars who were not looking for the individualised, subjective, complex transitional experience.

A related conclusion, therefore, must be that a secondary analysis of old sociological data and the re-reading of classic studies is both worthwhile and insightful. Being able to interrogate historical data with contemporary ideas and concepts has obvious value and can change (or contribute to) previous understandings of the social world…. (Goodwin and O’Conner, 2005, pp. 217–18)

Mike Savage (2005) undertook secondary analysis of the fieldnotes collected by John Goldthorpe, David Lockwood, Frank Bechhofer and Jennifer Platt as part of their studies of affluent workers in Luton in the early 1960s. On the basis of his reanalysis, Savage argued that:

the ideal type distinction between power, prestige and pecuniary images of society, elaborated by Lockwood (1966), fails to recognize that money, power and status were often fused in the statements and attitudes of the workers they interviewed.

Savage claimed that most respondents did have ‘a keen sense of dominant social classes’ and that the ‘hesitations evident in the fieldnotes when respondents were asked about class were not due to defensiveness so much as fundamental differences in the way that the researchers and the workers thought about class’.

Savage argued that ‘the single most important influence of the Affluent Worker Study was its repositioning of class as a structural, rather than a cultural, concept’. Goldthorpe et al. (1969) sought to develop a holistic understanding of the meaning of workers’ class identity. Their 229 male respondents, drawn from three workplaces in Luton, taken as exemplars for affluent manual workers, were interviewed both at their workplace and at their home, when their wives were also present. These interviews were comprehensive and ‘the questions on class identity, asked towards the end of the home interview, were remarkably full’ (Savage, 2005, p. 931). This generated a large amount of material (about a quarter of a million words in total), much of which was discarded because it failed to show unambiguously that the embourgeoised worker had a clear conception of class: at least a conception that reflected the mapping of class devised by the research team.

Through the work of Lockwood (1966) they had proposed three models of class identification: power, prestige and pecuniary. Goldthorpe et al. advocated a deductive approach to sociology and tended to discard what they considered inconsistent or ambiguous material, which meant most of the qualitative data went unreported. ‘Rather than inductively seeking to make sense of this complexity, they treated Lockwood’s three working-class images of society as ideal types, and considered how far the interview material could be read to indicate the predominance of any one of them’ (Savage, 2005, p. 933).

Savage (2005, p. 933–4) explained what he did:

I have examined the data on class identity in the 227 questionnaires, now held as part of the Qualidata Archive at the University of Essex. So far as I am aware this is the first attempt to re-analyse this data. I conducted some counts of easily quantifiable issues, namely the number of classes reported, the order in which they were reported, and the class that respondents themselves reported. More significantly, I recorded full quotations that bore on issues of money, power and status, and under themes raised in recent research: ordinariness, hesitancy and individuality. In what follows I seek to make sense of the respondents’ accounts of class using direct quotation and reporting general patterns emerging out of the interviews.

Savage (2005, p. 943–4) argued, then, that

Goldthorpe et al.’s findings about the nature of class identities and the dominance of money models of society only stand up in the context of the analytical categories they deployed, namely the distinction between money, power and status models of society. Although the data can be interpreted in these terms, this unhelpfully obscures the close association between these axes in the minds of the respondents. …However, today we might instead look more critically at the limitations of that tradition of sociological inquiry itself, and return to the data with a different set of questions and interests. Here, we can argue that people’s conceptions of class need to be understood as anchored in people’s understandings of what it is to be an ‘ordinary’ individual, with ‘natural’ attributes. My excavation of the views of 1960s affluent workers show how the idea of class is a necessary, though also shadowy, concomitant of people’s individualism….

My interpretation of affluent worker’s images of society, if true, has some telling implications, and suggests certain constants and continuities in popular identities. It suggests that the Luton affluent workers might not have been so distinct from other groups within the working class which were researched at the same time and where there is evidence that respondents recognized power divisions (see Cousins and Brown, 1975, for instance). The way that the Luton workers identified an affluent upper class as the most visible class in Britain, and their identification of themselves as having relatively little money compared to such groups, clearly indicates a sense of relative deprivation. It also helps explain why working-class identities have persisted even in the contemporary period of de-industrialization and the decline of manual employment.

The secondary analysis of the archive raises questions not just about the perceptions of the Luton workers but about methodological approaches, the value of all types of evidence and, crucially, sociological theorising. In similar vein to Goodwin and O’Conner (2005), Savage (2005, p. 944) stated:

My final point is methodological. There is a genuine difficulty in knowing whether the kind of claims made about the extent of changing popular identities, such as those of individualization, post materialism and the like reflect the different orientations and perspectives of sociologists and other scholars, rather than any shift in measurable popular attitudes and values themselves. I hope I have shown that the secondary analysis of archived qualitative data can be used to shed new insights on this issue, in ways which complicate and qualify any simple accounts of epochal change.

One might argue that the reanalysis by Savage and by Goodwin and O’Connor are such radical reanalyses that their use of secondary data is close to primary data analysis, the only difference being that they did not collect the empirical data themselves.


Next 7.2.3 Meta-analysis