Qualitative longitudinal research seeks to uncover and understand processes of change over time. This focus on change, both on how people change and on how people respond to change, is very relevant in the current policy context in which individual behaviour change is seen as key to achieving desired policy goals. Across a range of policy areas – including for example healthy living, education and lifelong learning, welfare to work, drug addiction, anti-social behaviour – policy-makers are seeking to understand what factors are important in determining choices and behaviour. There is also a growing theoretical interest in people as active agents, constructing their own biographies and lifecourse, in the context of particular social and economic constraints. Having people look back over time can provide insight into how they perceive and explain their actions, given the opportunity to discuss and reflect. Following people forward over time provides an opportunity to explore how and why people make the individual choices that add up to particular cumulative trajectories, and more specifically to understand the ways in which people respond to and use social and welfare services.
The articles in this section discuss the value and contribution of qualitative longitudinal methods for social policy with reference to examples of recent or ongoing research projects. All are based on repeat interview methodology, which reflects the fact that in-depth interviews tend to be the most common method for qualitative data collection in social policy research. Two are part of multi-method evaluations of welfare to work pilot programmes. Jane Lewis discusses the longitudinal qualitative element of the evaluation of the ‘Job Retention and Rehabilitation Pilot’, which started in 2003 and ran for two years. The policy aim was to help people avoid long spells out of work by providing early intervention and support. The ‘Pathways to Work Pilot’, discussed by Anne Corden and Katharine Nice, was introduced in phases in 2003 and 2004. New claimants to incapacity benefits were required to attend a series of work-focused interviews and offered a range of services and support. Again the aim was to promote early return to work and to reduce the risk of long-term benefit receipt. In both studies, there was a comprehensive programme of evaluation, providing information on outcomes (in particular on duration of benefit receipt and job entry rates) as well as on the processes of engagement with the services, which was the focus of the qualitative longitudinal research. Jane Millar also reports on research which is engaged with the welfare to work policy agenda, in this case in relation to lone mothers, but is not an evaluation of a specific policy intervention. The research is exploring issues of well-being and employment sustainability, and includes interviews with both the mothers and the children, starting from when the mothers left income support and started working and receiving tax credits. Rachel Thomson's research is rather different in that the ‘Inventing Adulthoods’ study is a much longer-term project following a group of young people for ten years, as they leave school for work or higher education. This research is not directly policy-focused but addresses many issues that have implications for policy in various ways. The fifth article in the section is a review article which provides some additional examples from other recent longitudinal qualitative studies.
By definition longitudinal qualitative studies take place over time, and time is an explicit element in the definition of longitudinal qualitative research. We discuss this more fully in our review article, which is the final article in this themed section. All longitudinal qualitative studies involve data collection at more than one point in time, but the overall time period for the research can differ widely, as can the number of points of data collection. The articles here reflect that. As just noted, Thomson's research has covered a ten year period, starting with a sample of 100 young people, and including up to six interviews over that period. The study which Millar discusses started with 50 lone mothers and 61 children, and so far has involved two interviews about 18–20 months apart. Corden and Nice have a sample of 105 people, in three cohorts, each person being interviewed three times over a period of around nine months. Jane Lewis has a sample of 36 people, in three cohorts, and interviewed six times at roughly monthly intervals. For the individuals in these studies, data collection ranges from ten years to six months. However the research study itself may cover a longer time period. The cohort design in the two evaluation studies meant that data collection covered a longer time span, 21 months for Corden and Nice and 23 months for Lewis. For Millar, the first interviews took place about 10–12 months after the women left income support, and the initial interview included retrospective biographical material covering a much longer time period. As each of these authors explains, decisions about the number of interviews, and the time gap between these, are made in the context of the overall aims and purposes of the research. The relatively short time gap between interviews for Lewis is intended to pick up contemporaneous reactions to the services offered and used. The longer time gap between interviews for Thomson is appropriate to the main aim of that research, which is to explore the construction of gender identity for young people.
The total amount of data generated by repeat interviews is substantial, even for the minimal longitudinal design of two rounds of interviews at two points in time. How to approach the management and analysis of such complex data sets is therefore an important issue for anyone contemplating such a study. The articles by Lewis and by Thomson provide the most extensive discussion of this issue, and outline contrasting approaches to analysis. Lewis sets out seven ways of reading the data (cross-sectional, repeat cross-sectional, individual case narrative, thematic, linked thematic, between-case comparison, between-group comparison) and describes the way in which material from each case needs to be extracted and summarised schematically within and across interviews in order to facilitate these different types of analysis. Thomson focuses on the individual ‘case history’ and describes how these can be analysed across different domains (for example, family, education, play), using concepts of biographical fields (for example, self, relationship to others) and biographical motifs (for example, loyalty, escape, determination), to explore how people find solutions to dilemmas or problems. Putting these individual case histories ‘in conversation’ with each other allows exploration of both individual and temporal change and of social and spatial change.
Analysing change over time is the specific contribution of longitudinal methodology. This is true of both quantitative and qualitative research and Millar makes an explicit link between these, discussing three key concepts that have been used in the analysis of the dynamics of poverty and employment – transitions, adaptations and trajectories – and arguing that a longitudinal qualitative approach can develop and extend our understanding of concepts such as these. She illustrates this by using these analytical categories to explore her data from interviews with lone mothers and argues that the qualitative approach provides useful insights into the value and limits of these concepts as ways of summarising change. Lewis discusses the different domains or levels at which change takes place – at the individual level (in health condition for example), at the level of the service (with different approaches to delivery over time), as well as at the policy level and at the broader structural level. She discusses the different types of questions that need to be addressed at each level. She notes that events, experiences and feelings can be interpreted differently by people at different standpoints in time, and this is true for both participants and researchers.
This important point about how interpretations can themselves change over time is addressed in all four articles. On the one hand, changing interpretations over time provide an opportunity to explore, and to seek to explain, the gaps between what people say and what they do. Asking people explicitly to do this, to apply hindsight to their narratives, is one of the tools of longitudinal qualitative research and can produce important insights. On the other hand, such changing interpretations present a challenge to drawing clear conclusions about the meaning and significance of the research findings. Corden and Nice discuss this in relation to the production of ‘emerging findings’ and the presentation of these to policy makers. As they note, the picture that is drawn from the first interview may change as the interviews progress and people reassess their earlier views, with perhaps differing degrees of awareness of how their views have apparently changed. Corden and Nice argue that working closely with the policy customer or wider policy audience is important so that the progress of the interpretation of the findings is part of a shared and iterative process.
Other important themes or issues also emerge from these articles. All longitudinal research of necessity has a starting point and a finishing point, but this window of observation may not be able to pick up all the relevant change. As Millar points out, some knowledge about the previous employment histories of the lone mothers was needed in order to be able to contextualise this particular move into work. Lewis acknowledges that the longer term impact of being part of an employment project may only emerge when the research is over. There are various statistical techniques for addressing this ‘left and right censorship’ in the analysis of quantitative survey data. In the analysis of qualitative research it is important to start with an explicit awareness of the type and nature of change that is the focus of attention and to be clear about what is the ‘baseline’, the point from which change is being explored. This is not necessarily the point when the sample was drawn but, in the case of a service evaluation for example, may be the point of first contact. Another key point concerns the potential importance of no change, and of non-linear change, in the data. Concepts of ‘progress’ or ‘advancement’ may not be appropriate to understanding the complex twists and turns and changes of direction over time.
As we noted at the start, there is an increasing interest in longitudinal qualitative methods for social policy research. The literature is growing rapidly and more studies are getting underway in a range of policy areas. We hope that this themed section will encourage others to contribute to the further development of this methodology. There are many important issues yet to be addressed in respect of the design, analysis, reporting, and archiving of longitudinal qualitative data.
