Qualitative Longitudinal Research: The Timescapes Project PDF

66 Research designs Research in focus 3.12 Qualitative longitudinal research: the Timescapes project Qualitative longitudinal research (often abbreviated to QLL) that involves repeat qualitative interviews with...

66 Research designs Research in focus 3.12 Qualitative longitudinal research: the Timescapes project Qualitative longitudinal research (often abbreviated to QLL) that involves repeat qualitative interviews with research participants has become more common since the turn of the century. This is particularly apparent with the Timescapes project, which is a major project that began life in February 2007. The aim is to interview and re-interview people on several occasions to capture social changes and shifts in peoples life course and thoughts and feelings. It comprises seven relatively independent projects. Through these projects the researchers aim to track the lives of around 400 people. One of the projects is entitled Maculinities, identities and risk: transition in the lives of men as fathers and aims to get a sense of how masculine identities change in the wake of rst-time fatherhood. This particular study builds on research that originally began in Norfolk in 1999, well before the Timescapes project began. Thirty fathers were interviewed in 2000–1 both before and after the birth of their rst child. Each man was interviewed three times (two interviews were scheduled after the childs birth). This group of men was then followed up in 2008. A further set of interviews was conducted with eighteen men from south Wales in 2008–9 with the same pattern of one interview before and two interviews after birth. In the course of the interviews use was made of photographs of families and men with their children to stimulate reection on fatherhood. The use of photographs in interviews is explored in Chapters 19 and 20. The materials will eventually be made available for secondary analysis (see the section on Secondary analysis of qualitative data in Chapter 24). Sources: Guardian, 20 Oct. 2009: www.guardian.co.uk/education/2009/oct/20/timescapes-leeds-research-memories?INTCMP=SRCH Project website: www.timescapes.leeds.ac.uk For information on the masculinities project, see: www.timescapes.leeds.ac.uk/research-projects/projects/masculinities-fatherhood-risk.php For some methodological reections on the Timescapes project, see: www.timescapes.leeds.ac.uk/methods-ethics All the above websites were accessed 18 May 2011. unplanned elements. This is also an interesting illustration East End of Boston, M. Staceys (1960) research on of a mixed methods study, in that it combines quantita- Banbury, and OReillys (2000) research on a commun- tive and qualitative research. ity of Britons living on the Costa del Sol in Spain. Increasingly, social researchers are becoming inter- ested in the study of online communities (see Chap- Case study design ter 28 and Research in focus 28.4 in particular); The basic case study entails the detailed and intensive a single school, such as studies by Ball (1981) and analysis of a single case. As Stake (1995) observes, case by Burgess (1983) on Beachside Comprehensive and study research is concerned with the complexity and Bishop McGregor respectively; particular nature of the case in question. Some of the best-known studies in sociology are based on this kind of a single family, like O. Lewiss (1961) study of the Sánchez family or Brannen and Nilsens (2006) design. They include research on: investigation of a family of low-skilled British men, a single community, such as Whytes (1955) study which contained four generations in order to uncover of Cornerville in Boston, Ganss (1962) study of the changes in fathering over time;  Research designs 67 Research in focus 3.13 A planned and unplanned longitudinal design In the 1940s Sheldon and Eleanor Glueck of the Harvard Law School conducted a study concerned with how criminal careers begin and are maintained. The study entailed a comparison of 500 delinquents and 500 non-delinquents in Massachusetts. The two samples were matched in terms of several characteristics, such as age, ethnicity, and the socio-economic status of the neighbourhoods from which they were drawn. The sample was aged around 14 at the time and was followed up at ages 25 and 33. The data were collected by various means: interviews with the 1,000 participants, their families, and various key gures in their lives (for example, social workers and school teachers); observations of the home; and records of various agencies that had any connection with the participants and their families. Obviously, data concerning criminal activity were collected for each individual by examining records relating to court appearances and parole. While all these sources of data produced quantitative information, qualitative data were also collected through answers to open questions in the interviews. Around the mid-1990s Laub and Sampson (2003, 2004) began to follow up the 500 men who had been in the delinquent sample. By this time, they would have been aged 70. Records of death and criminal activity were searched for all 500 men, so that patterns of ongoing criminal activity could be gleaned. Further, they managed to nd and then interview fty-two of the original delinquent sample. These cases were selected on the basis of their patterns of offending over the years, as indicated by the criminal records. The interviews were life history interviews to uncover key turning points in their lives and to nd out about their experiences. This is an extremely unusual example of a longitudinal study that contains planned elements (the original wave of data collection, followed by the ones eleven and eighteen years later) and an unplanned element conducted by Laub and Sampson many years later. Research in focus 3.14 A case study Holdaway (1982, 1983) was a police ofcer who was also conducting doctoral research on his own police service, which was located in a city. His main research method was ethnography, whereby he was a participant observer who observed interaction, listened to conversations, examined documents, and wrote up his impressions and experiences in eld notes. Holdaways superiors did not know that he was conducting research on his own force, so that he was a covert researcher. This is a controversial method on ethical grounds (see Chapters 6 and 19). Holdaways research provides insights into the nature of police work and the occupational culture with which ofcers surround themselves. a single organization, such as studies of a factory a single event, such as the Cuban Missile Crisis (Allison by writers such as Burawoy (1979), and Cavendish 1971), the events surrounding the media reporting (1982), or of pilferage in a single location like a bakery of a specic issue area (Deacon, Fenton, and Bryman (Ditton 1977), of a single police service (Holdaway 1999), the Balinese cockght (Geertz 1973b), and the 1982, 1983; see Research in focus 3.14), or of a single study of a disaster incident (Vaughan 1996, 2004). call centre (Callaghan and Thompson 2002; Nyberg 2009); What is a case? a person, like the famous study of Stanley, the jack- The most common use of the term case associates the roller (Shaw 1930); such studies are often character- case study with a location, such as a community or organ- ized as using the life history or biographical approach ization. The emphasis tends to be upon an intensive (see the section on Life history and oral history inter- examination of the setting. There is a tendency to associ- viewing in Chapter 20); and ate case studies with qualitative research, but such an 68 Research designs identication is not appropriate. It is certainly true of qualitative research within a cross-sectional design that exponents of the case study design often favour framework (see Research in focus 3.9). However, it has qualitative methods, such as participant observation and been described as providing case-study evidence by unstructured interviewing, because these methods are Davies et al. (1994: 157), presumably on the grounds viewed as particularly helpful in the generation of an that the eldwork was undertaken in a single location. intensive, detailed examination of a case. However, case I would prefer to reserve the term case study for those studies are frequently sites for the employment of both instances where the case is the focus of interest in its quantitative and qualitative research, an approach that own right. The study in Thinking deeply 3.3 is no more a will receive attention in Chapter 27. Indeed, in some in- case study of Kidderminster than Beardsworth and Keils stances, when an investigation is based exclusively upon (1992) research is based on a case study of the East quantitative research, it can be difcult to determine Midlands. McKee and Bells (1985) research is concerned whether it is better described as a case study or as a cross- with the experience of unemployment among the forty- sectional research design. The same point can often be ve couples whom they interviewed. It is not concerned made about case studies based upon qualitative research. with Kidderminster as such. The town provides a kind of As an illustration of the difculties of writing about backdrop to the ndings rather than a focus of interest in case studies, consider the study described in Thinking its own right. The crucial point is that Kidderminster is deeply 3.3. Ostensibly, it is similar to Beardsworth and not the unit of analysis; rather it is the sample that is the Keils (1992) study of vegetarians, in that it is a piece unit of analysis. Thinking deeply 3.3 What is the unit of analysis? McKee and Bell (1985: 387) examined forty-ve couples in a single location (Kidderminster in the West Midlands) in order to examine the impact of male unemployment on family and marital relations. They describe their research instrument as an unstructured, conversational interview style. In most cases, husbands and wives were interviewed jointly. The interviews were very non-directive, allowing the couples considerable freedom to answer in their own terms and time. Their research focused on the range of problems faced by unemployed families, the processes by which they cope, and the variations in their experiences. Thus the focus was very much on the experience of unemployment from the perspective of the couples. The authors show, for example, that the impact of husbands unemployment on their wives is often far greater than is usually appreciated, since research frequently takes the unemployed person as the main hub of the enquiry. Couples often reported changes to the domestic division of labour, which in turn raised questions for them about images of masculinity and identity. Is this study a case study of unemployment in Kidderminster or is it better thought of as a cross-sectional design study of unemployed men and their wives? As I suggest in the text, it is not terribly helpful to think of it as a case study, because Kidderminster is not the unit of analysis. It is about the responses to unemployment among a sample of individuals; the fact that the interviewees were located in Kidderminster is not signicant to the research ndings. However, it is not always easy to distinguish whether an investigation is of one kind rather than another. As these reections imply, it is important to be clear in your own mind what your unit of analysis is. Similarly, Powell and Buttereld (1997) present a experience—which in turn affected promotion. The im- quantitative analysis of promotion decisions in a US gov- pact of race on these two variables was such that people ernment department. They were concerned to investigate of colour were disadvantaged with respect to promotion. how far race had an impact on promotions within the Once again, we see here a study that has the hallmarks of department. The researchers found that race did not both a cross-sectional design and a case study, but this have a direct effect on promotion, but it did have an indir- time the research strategy was a quantitative one. As ect effect. This occurred because race had an impact on with the McKee and Bell (1985) research, it seems better two variables—whether the applicant was employed in to describe it as employing a cross-sectional design rather the hiring department and the number of years of work than a case study, because the case itself is not the  Research designs 69 apparent object of interest: it is little more than a location on a national, random sample of the population of Great that forms a backdrop to the ndings. Britain would have to be considered a case study of Great Similarly, I would tend to argue that the study of Britain! However, it also needs to be appreciated that, redundant steelworkers by Westergaard et al. (1989) when specic research illustrations are examined, they is a case study of the effects of redundancy in which a can exhibit features of more than one research design. quantitative research strategy was employed with clear What distinguishes a case study is that the researcher is indications of a cross-sectional design. With a case study, usually concerned to elucidate the unique features of the the case is an object of interest in its own right, and the case. This is known as an idiographic approach. Research researcher aims to provide an in-depth elucidation of it. designs like the cross-sectional design are known as Unless a distinction of this or some other kind is drawn, nomothetic, in that they are concerned with generating it becomes impossible to distinguish the case study as statements that apply regardless of time and place. a special research design, because almost any kind of However, an investigation may have elements of both research can be construed as a case study: research based (see Research in focus 3.15). Research in focus 3.15 A cross-sectional design with case study elements Sometimes, an investigation may have both cross-sectional and case study elements. For example, Leonard (2004) was interested in the utility of the notion of social capital for research into neighbourhood formation. As such, she was interested in similar issues to the study in Research in focus 2.2. She conducted her study in a Catholic housing estate in West Belfast, where she carried out semi-structured interviews with 246 individuals living in 150 households. Her ndings relate to the relevance of the concept of social capital, so that the research design looks like a cross-sectional one. However, on certain occasions she draws attention to the uniqueness of Belfast with its history in recent times of conict and the search for political solutions to the problems there. At one point she writes: In West Belfast, as the peace process develops, political leaders are charged with connecting informal community networks to more formal institutional networks (Leonard 2004: 939). As this comment implies, it is more or less impossible in a study like this to generate ndings concerning community formation without reference to the special characteristics of Belfast and its troubled history. With experimental and cross-sectional designs, the in this chapter—measurement validity, internal validity, typical orientation to the relationship between theory external validity, ecological validity, reliability, and repli- and research is a deductive one. The research design and cability—depends in large part on how far the researcher the collection of data are guided by specic research feels that these are appropriate for the evaluation of case questions that derive from theoretical concerns. However, study research. Some writers on case study research, like when a qualitative research strategy is employed within Yin (2009), consider that they are appropriate criteria a cross-sectional design, as in Beardsworth and Keils and suggest ways in which case study research can be (1992) research, the approach tends to be inductive. In developed to enhance its ability to meet the criteria; for other words, whether a cross-sectional design is induc- others, like Stake (1995), they are barely mentioned, if at tive or deductive tends to be affected by whether a quan- all. Writers on case study research whose point of orien- titative or a qualitative research strategy is employed. tation lies primarily with a qualitative research strategy The same point can be made of case study research. When tend to play down or ignore the salience of these factors, the predominant research strategy is qualitative, a case whereas those writers who have been strongly inuenced study tends to take an inductive approach to the relation- by the quantitative research strategy tend to depict them ship between theory and research; if a predominantly as more signicant. quantitative strategy is taken, it tends to be deductive. However, one question on which a great deal of discus- sion has centred concerns the external validity or general- Reliability, replicability, and validity izability of case study research. How can a single case The question of how well the case study fares in the possibly be representative so that it might yield ndings context of the research design criteria cited early on that can be applied more generally to other cases? For 70 Research designs example, how could the ndings from Holdaways (1982, commonplace situation (Yin 2009: 48). Thus a case 1983) research, referred to in Research in focus 3.14, be may be chosen because it exemplies a broader cat- generalizable to all police services in Great Britain? The egory of which it is a member. The notion of exempli- answer, of course, is that they cannot. It is important to cation implies that cases are often chosen not because appreciate that case study researchers do not delude they are extreme or unusual in some way but because themselves that it is possible to identify typical cases that either they epitomize a broader category of cases can be used to represent a certain class of objects, or they will provide a suitable context for certain whether it is factories, mass-media reporting, police research questions to be answered. An illustration of services, or communities. In other words, they do not the rst kind of situation is Lynd and Lynds (1929, think that a case study is a sample of one. 1937) classic community study of Muncie, Indiana, in the USA, which they dubbed Middletown precisely Types of case because it seemed to typify American life at the time. The second rationale for selecting exemplifying cases Following on from the issue of external validity, it is is that they allow the researcher to examine key social useful to consider a distinction between different types processes. For example, a researcher may seek access of case that is sometimes made by writers. Yin (2009) to an organization because it is known to have imple- distinguishes ve types. mented a new technology and he or she wants to know what the impact of that new technology has The critical case. Here the researcher has a well- developed theory, and a case is chosen on the grounds been. The researcher may have been inuenced by that it will allow a better understanding of the cir- various theories about the relationship between tech- cumstances in which the hypothesis will and will nology and work and by the considerable research not hold. The study by Festinger et al. (1956) of literature on the topic, and as a result seeks to exam- a religious cult whose members believed that the end ine the implications of some of these theoretical and of the world was about to happen is an example. The empirical deliberations in a particular research site. fact that the event did not happen by the appointed The case merely provides an apt context for the day allowed the researchers to test the authors pro- working-through of these research questions. To take positions about how people respond to thwarted a concrete example, Russell and Tylers (2002) study expectations. of one store in the Girl Heaven UK chain of retail stores for 3–13-year-old girls does not appear to have The extreme or unique case. The unique or extreme been motivated by the store being critical, unique, or case is, as Yin observes, a common focus in clinical by it providing a context that had never before been studies. Margaret Meads (1928) well-known study of studied, but was to do with the capacity of the re- growing up in Samoa seems to have been motivated search site to illuminate the links between gender and by her belief that the country represented a unique consumption and the commodication of childhood case. She argued that, unlike most other societies, in modern society. Samoan youth do not suffer a period of anxiety and stress in adolescence. The factors associated with this The revelatory case. The basis for the revelatory relatively trouble-free period in their lives were of case exists when an investigator has an opportunity interest to her, since they might contain lessons for to observe and analyse a phenomenon previously Western youth. Fielding (1982) conducted research inaccessible to scientic investigation (Yin 2009: on the extreme right-wing organization the National 48). As examples, Yin cites Whytes (1955) study of Front. While the National Front was not unique on the Cornerville, and Liebows (1967) research on unem- British political scene, it was extremely prominent at ployed blacks. the time of his research and was beginning to become an electoral force. As such, it held an intrinsic interest The longitudinal case. Yin suggests that a case may be chosen because it affords the opportunity to be inves- that made it essentially unique. tigated at two or more junctures. However, many case The representative or typical case. I prefer to call this an studies comprise a longitudinal element, so that it is exemplifying case, because notions of representative- more likely that a case will be chosen both because it ness and typicality can sometimes lead to confusion. is appropriate to the research questions on one of the With this kind of case, the objective is to capture other four grounds and also because it can be studied the circumstances and conditions of an everyday or over time.  Research designs 71 Any case study can involve a combination of these that the evidence they present is limited because it has elements, which can best be viewed as rationales for restricted external validity by arguing that it is not the choosing particular cases. For example, Margaret Meads purpose of this research design to generalize to other (1928) classic study of growing up in Samoa has been cases or to populations beyond the case. This position is depicted above as an extreme case, but it also has ele- very different from that taken by practitioners of survey ments of a critical case because she felt that it had research. Survey researchers are invariably concerned to the potential to demonstrate that young peoples re- be able to generalize their ndings to larger populations sponses to entering their teenage years are not deter- and frequently use random sampling to enhance the mined by nature alone. Instead, she used growing up in representativeness of the samples on which they conduct Samoa as a critical case to demonstrate that culture their investigations and therefore the external validity of has an important role in the development of humans, their ndings. Case study researchers argue strenuously thus enabling her to cast doubt on notions of biological that this is not the purpose of their craft. determinism. It may be that it is only at a very late stage that the Case study as intensive analysis singularity and signicance of the case becomes appar- Instead, case study researchers tend to argue that they ent (Radley and Chamberlain 2001). Flyvbjerg (2003) aim to generate an intensive examination of a single case, provides an example of this. He shows how he undertook in relation to which they then engage in a theoretical a study of urban politics and planning in Aalborg in analysis. The central issue of concern is the quality of the Denmark, thinking it was a critical case. After conducting theoretical reasoning in which the case study researcher his eldwork for a while, he found that it was in fact an engages. How well do the data support the theoretical extreme case. He writes as follows: arguments that are generated? Is the theoretical analysis incisive? For example, does it demonstrate connections between different conceptual ideas that are developed Initially, I conceived of Aalborg as a most likely critical out of the data? The crucial question is not whether the case in the following manner: if rationality and urban ndings can be generalized to a wider universe but how planning were weak in the face of power in Aalborg, well the researcher generates theory out of the ndings. then, most likely, they would be weak anywhere, at This view of generalization is called analytic generaliza- least in Denmark, because in Aalborg the rational tion by Yin (2009) and theoretical generalization by paradigm of planning stood stronger than anywhere J. C. Mitchell (1983). Such a view places case study re- else. Eventually, I realized that this logic was awed, search rmly in the inductive tradition of the relation- because my research [on] local relations of power ship between theory and research. However, a case study showed that one of the most inuential faces of power design is not necessarily associated with an inductive in Aalborg, the Chamber of Industry and Commerce, approach, as can be seen in the research by Adler and was substantially stronger than their equivalents Adler (1985), which was referred to in Chapter 2. Thus, elsewhere. Therefore, instead of a critical case, case studies can be associated with both theory genera- unwittingly I ended up with an extreme case in the tion and theory testing. Further, as M. Williams (2000) sense that both rationality and power were unusually strong in Aalborg, and my case study became a study has argued, case study researchers are often in a position of what happens when strong rationality meets strong to generalize by drawing on ndings from comparable power in the area of urban politics and planning. But cases investigated by others. This issue will be returned this selection of Aalborg as an extreme case happened to in Chapter 18. to me, I did not deliberately choose it. (Flyvbjerg 2003: 426) Longitudinal research and the case study Case study research frequently includes a longitudinal element. The researcher is often a participant of an organ- ization or member of a community for many months or Thus, we may not always appreciate the nature and sig- years. Alternatively, he or she may conduct interviews nicance of a case until we have subjected it to detailed with individuals over a lengthy period. Moreover, the re- scrutiny. searcher may be able to inject an additional longitudinal One of the standard criticisms of the case study is element by analysing archival information and by retro- that ndings deriving from it cannot be generalized. spective interviewing. Research in focus 3.16 provides an Exponents of case study research counter suggestions illustration of such research. 72 Research designs Research in focus 3.16 A case study of ICI Pettigrew (1985) conducted research into the use of organizational development expertise at Imperial Chemical Industries (ICI). The eldwork was conducted between 1975 and 1983. He carried out long semistructured interviews in 1975–7 and again in 1980–3. During the period of the eldwork he also had fairly regular contact with members of the organization. He writes: The continuous real-time data collection was enriched by retrospective interviewing and archival analysis... (Pettigrew 1985: 40). Another way in which a longitudinal element occurs is when a case that has been studied is returned to at a later Figure 3.5 stage. A particularly interesting instance of this is the A comparative design Middletown study that was mentioned previously. The town was originally studied by Lynd and Lynd in 1924 –5 (Lynd and Lynd 1929) and was restudied to discern T1 trends and changes in 1935 (Lynd and Lynd 1937). In 1977 Obs1 the community was restudied yet again (Bahr et al. 1983), Obs2 using the same research instruments but with minor Obs3 changes. Burgess (1987) was similarly concerned with Case 1 Obs4 continuity and change at the comprehensive school he had Obs5 studied in the early 1970s (Burgess 1983) when he re-... turned to study it ten years later. However, as he observes, Obsn it is difcult for the researcher to establish how far change is the result of real differences over the two time periods Obs1 or of other factors, such as different people at the school, Obs2 different educational issues between the two time peri- Obs3 ods, and the possible inuence of the initial study itself. Obs4 Case n Obs5... Comparative design Obsn It is worth distinguishing one further kind of design: comparative design. Put simply, this design entails study- ing two contrasting cases using more or less identical methods. It embodies the logic of comparison, in that it implies that we can understand social phenomena better when they are compared in relation to two or more meaningfully contrasting cases or situations. The com- when individuals or teams set out to examine particular parative design may be realized in the context of either issues or phenomena in two or more countries with the quantitative or qualitative research. Within the former, express intention of comparing their manifestations in the data-collection strategy will take the form outlined in different socio-cultural settings (institutions, customs, Figure 3.5. This gure implies that there are at least two traditions, value systems, life styles, language, thought cases (which may be organizations, nations, communities, patterns), using the same research instruments either police forces, etc.) and that data are collected from each, to carry out secondary analysis of national data or to usually within a cross-sectional design format. conduct new empirical work. The aim may be to seek One of the more obvious forms of such research is in explanations for similarities and differences or to gain a cross-cultural or cross-national research. In a useful de- greater awareness and a deeper understanding of social nition, Hantrais (1996) has suggested that such research reality in different national contexts. occurs