THE RESEARCH METHODOLOGY
4.4 DATA COLLECTION METHODS
As with research strategy, the decision over which data collection methods to use in a given study arguably lends itself to the nature of the research question being asked, within the confines of the research paradigm being assumed (Miles and Huberman, 1994). However, for the purpose of the present study there were a variety of potential data collection methods that could have been used, each with their own benefits and drawbacks. Clearly from the research strategy, and indeed the critical realist perspective, there was a natural tendency towards the adoption of in-depth qualitative methods, in order to capture some of the complexity and context of the experiences investigated. And in the end, the decision was made to collect most of the data through semi-structured in-depth interviews.
Yet, this was not a totally straight forward decision, and other options were considered as well (and in some instances conducted, although not formally included in the thesis23).
4.4.1 SEMI-STRUCTURED IN-DEPTH INTERVIEWS
The decision to use in-depth interviews for the present study mainly centred on the relative advantages the method embraced in relation to helping answer the research objectives and questions of the thesis.
According to numerous literatures (Warren, 2002; Hesse-Biber and Leavy, 2011), in-depth interviews
23 A survey was conducted for the present research but disregarded because of methodological problems in the data generation.
Key:
80 allow researchers to ‘tap’ or access ‘rich’ descriptions about an individual’s experience of events and activities. As noted by Carson (2001, p 77),
“In-depth interviews can provide an open, flexible, experiential and illuminating way to study complex, dynamic interactive situations”.
In addition to rich data, the attractions of using semi-structured in-depth interviews also related to (i) the face to face contact that interviews provide, (ii) the opportunity to clearly explain the topic being investigated and ensure participants understood what was being asked, and (iii) the level of flexibility available when collecting the data (Mahoney, 1997; Saunders et al., 2011). Indeed, the matter of flexibility was particularly pertinent for the study as the semi-structured interviews enabled the capacity to change the format, order and wording of questions when necessary and probe and follow-up on issues of interest and relevance as well. As the primary data collection was informed by existing research, and at times drawing on ‘a priori’ notions (e.g. elements of the content of social exchange such as aspects of pay, training, development, involvement, support etc), the use of unstructured interview would have been incongruous to the conceptual framework being applied.
But having the semi-structured interviews seemed to allow for a healthy degree of focus and flexibility: allowing existing notions to be tapped, while facilitating the space for new insights to
‘emerge’ inductively. On the back of this, although like all qualitative methods (and to a less extent quantitative methods) there remains the potential criticism for ‘bias’ in the process of collecting data in this way (Kaar, 2009), the argument here is that by collecting data in a fairly systematic and reflexive way, these biases are able to be reduced and/or openly acknowledged (see Chapter Ten for reflexivity).
Compared to other data collection methods, semi-structured interviews also seemed to encapsulate fewer limitations for the present research. For example, while questionnaires may have garnered a more ‘objective’ means of generating data, they would potentially only gather superficial snapshots of outsourcing experiences – even if a host of variables were included and a longitudinal design adopted.
Certainly they would not offer much in the way of explanation or context for the critical realist perspective. Furthermore, there is also the practical problem of collecting meaningful ‘before’ and
81
‘after’ data on the specific phenomena being investigated in this thesis. That is, it would have been practically very difficult to get accurate ‘before’ snapshots of worker attitudes of the employment relationship in the outsourcing setting because the survey questionnaires will almost always be conducted with workers after they have knowledge that the outsourcing transition is being considered or actually happening (by which time, as some participants told me during the field work, their perceptions of the employment relationship would be altered).
Focus groups were also discounted owing to the lack of privacy they provide participants. Indeed, although focus groups would have facilitated the benefit of inciting and eliciting ideas and issues which may not have arisen in one-to-one interviews (Hesse-Biber and Leavy, 2011), sensitivty was seen as a significant problem owing to the nature of the topic being researched. And, in relation to participant observation, although this method would have no doubt provided some interesting and contextualised findings on the experience of outsourcing, the problem here related to access, as well as the lack of benefits of such research. In terms of access, participant observation is acknowledged to be notoriously difficult owing to its intrusive nature, particularly in sensitive situations (Bryman, 2012).
The lack of benefits related to the argument that many social exchange interactions are difficult to
‘observe’ in the direct sense (and over such a long period of time), and so would inevitably require interviews to extract perceptions of any changes, and the meaning attributed them for the individual workers.
Consequently, on the back of this, the primary form of data collection used in the present research was semi-structured in-depth interviews – and 85 interviews were conducted in total. Most of these were conducted with workers for the main data analysis into changes to the employment relationship.
However, interviews were also conducted with senior managers of the Leisure Trusts and, on occasion, Local Councils, in order to capture broader contextual insights and inform the explanations that emerged from the wider study.
On the back of this, it is worth pointing out two further points regarding data collection though.
The first point to mention is that seven interviews were conducted with senior managers from other
82 local authorities, leisure organisations and trade unions prior to the case study research – i.e. as a sort of ‘preliminary study’ (as well as discussions with academics in the field). These interviews were not part of the official data analysis, though they did help me get a better grasp on the nature of the study during the initial year as a PhD student, and offered some useful direction as to what to investigate.
The second point is that when ‘conducting’ the research at the case study sites there were many opportunities for ‘conversations’ with workers outside of the formal interviewing processes.
For example, I quite regularly spoke to employees who did not have ‘formal’ interviews, and I often held record’ chats to those who did. To recognise the significance of such conversations, the ‘off-record’ chats sometimes lasted up to 20 minutes and often included discussions relating to the topics of the study. Silverman (2011) discusses the importance of ‘naturally occurring talk’ and ‘naturally occurring data’ for qualitative research, and argues that qualitative researchers should take seriously the data found in ‘natural’ settings, as well as the ‘formal’ settings, as this can be a gateway for deeper insights. Moreover, Richie and Lewis (2004) note that being aware of the insights garnered from natural settings may provide a more complete picture of the phenomena and context being explored – and in case study research such a position is likely to be particularly valuable, especially if the phenomena in question is quite sensitive. Hence, it seems important to recognise that such conversations took place in order to outline the extent of research carried out during the field work and to note that this was in fact quite useful for gaining a deeper impression of the culture and context of the case studies being investigated.
4.4.2 APPLICATION
In terms of practicalities for the formal interviews, the details below offer some insight into how the interviews were organised, how they were managed, and the way in which the interview questions were developed.
83 Organising the Interviews
The process of gaining access to each of the case studies typically consisted of emailing and/or telephoning senior managers at the leisure centres or local authorities to explore the possibilities of conducting the research. Subsequently, visits were then made to the respective organisations to deliver a presentation to the senior managers about the nature of the study, what participation would involve, and what the study was being used for. After they granted access, I then pursued possibilities to see if there was a supervisor who might be kind enough to act as a ‘research co-ordinator’ – i.e. to organise and recruit participants. In each case there was, and the research co-ordinators were very helpful in this respect, and I remain thankful for each of their support.
At each data collection wave, the research co-ordinator placed a letter/advert on the staff notice about the study that asked employees if they were interested in participating. Employees then registered interest with the research co-ordinator who would organise a time for them to have an interview.
During a wave of data collection, a whole working week would typically be spent at the leisure centre, whereby I would be open to have interviews with employees at most times. This also allowed me to get time to chat to other workers who were not formally having interviews, and get a better ‘feel’ for the organisations and their cultures.
Interview Questions
In terms of planning and conducting good qualitative interviews, Mason (2002) elucidates that it requires “hard, creative, active work” (p. 67) to get it right. She consequently shares that the significance of planning and preparing for the interviews cannot be overstated. For the present study, and in line with the semi-structured approach, developing a relevant guiding set of questions to ask participants was deemed important in order to be adequately prepared to tap the topics and themes under investigation. In general, these were largely developed from the objectives of the study (p. 12), and the reviews of literature that had taken place before the time of data collection. However, as the interviews and waves were conducted, new topics and themes emerged as the study evolved, in line
84 with the critical realist paradigm (Ackroyd, 2009). Power and identity, for example, were not initially intended to be studied but emerged in the first wave of data collection as important notions to consider. At the same time, many of the original themes remained steadfast throughout the entire data collection process. For the list interview questions please see Appendix 1.
Managing the Interviews
In terms of managing the interviews, Miller and Crabtree (1992) note the importance of this for building a climate of trust and open communication. It could be argued there is also significance in letting individuals ‘tell their own story’ at the beginning of interviews in order to get a better understanding of individuals and their perceptions. Thus, in terms of the actual interviews themselves, each interview always opened with a light informal discussion first, in order to build rapport and make the participants feel at ease – particularly given the interviews were being recorded via electronic Dictaphone. Reassurance was also given around confidentiality and anonymity. Some of the initial questions about the participants’ experience were intentionally broad, allowing participants to give their own accounts of events that they perceived to have happened for themselves. This proved quite useful because it allowed participants to talk about what was important to themselves first, whilst also allowing new insights to emerge before I asked participants about specific topics. After the broad questions, participants were then asked about a range of specific topics relating to the employment relationship – e.g. how and why terms and conditions, training and development, job designs, job security, support, trust, commitment etc changed during the course of the transition (see Appendix 1).
When it came to closing the interviews, the aim was to always end on a positive note and offer a sincere thank you to the interviewee24.
Other points to mention about the management of the interviews relate to (i) the environment, and (ii) demeanour of the interviewer. In relation to the environment, the interviews were typically carried out in an office at each of the leisure centres I was based at. Occasionally, they were conducted in more
24 I remain immensely grateful to all the participants who took part, as without them this research project would simply not have been possible.
85 open spaces such as leisure centre cafés, but this tended to be rare. The office locations appeared to work well, as they seemed to give a professional feel to the interviews, but also provide a level of privacy so participants felt they could freely discuss their thoughts and feelings.
Mason (2002) offers some advice about the interviewer demeanour, and suggests a few factors that can enhance the willingness of participants to be more open. These include: (i) being meaningful to the interviewees; (ii) being able to relate to the interviewees circumstances and experiences based on what one already knows about them and their situation; (iii) being able to ensure the flow of interview interaction – ‘the conversation with purpose’; being sensitive to the needs and rights of the interviewees in accordance within one’s own ethical position (and that of the research establishment);
and being able to empathise with interviewees when discussing sensitive and emotionally charged topics. All of these things I sought to do in the interviews. In doing so, I presented myself as a
‘researcher’ to participants, explaining that I was collecting data for a PhD project (participants were also given full information about the study – see p. 94 and Appendix 2). Giving a warm sense of professionalism also felt important for building trust with participants, and was something I always tried to do.
4.5 SAMPLING
Although the significance of sampling in quantitative research lies in the ability to define a finite population and thereafter select representative case units through probability sampling, the purpose of sampling is rather different in qualitative research. For, as a starting point, the goal of qualitative research is typically not based on statistical generalisibility to broader populations (Garson, 2002). As Cresswell (1994) notes, the intention of qualitative research is about the transmission of ‘trustworthy’
findings that offer contextualised insights about social phenomena. Similarly, Lincoln and Guba (1985) suggest that researchers should be more pre-occupied with ensuring credibility, trustworthiness and dependability in qualitative research – i.e. making sure the study gains adequate data from the field from those being investigated, gaining rich and contextualised forms of data, and being open and reflexive about how the data was collected. To this end, the logic of sampling in qualitative research
86 is conspicuously different from quantitative studies (Patton, 2005), and usually involves non-probability techniques such as purposive, snow-ball and quota sampling.
4.5.1 PURPOSIVE SAMPLING
Morrow (2005) argues that by nature qualitative research typically has a purposive dimension, as participants are deliberately selected to serve a particular purpose. Barbour (2001) elucidates that purposive research can be particularly useful as it provides researchers with a clear direction as to whom to recruit data from, whilst simultaneously allowing space for researchers to seeks out ‘outliers’
when necessary. Miles and Huberman (1994) further explain that purposive sampling is particularly useful when researching ‘typical cases’ (the norm of those being studied), ‘deviant cases’ (unique participants who have a specialist account of the phenomena for certain reasons), or ‘negative cases’
(participants who ‘buck the trend’). For the present study, therefore, purposive sampling was applied to both the case organisations and the individuals within the case organisations. This was to enable the sampling choices to be pragmatic and deliberate about which organisations and participants to include in the research, in order to enable the research objectives to be most appropriately addressed (Devers and Frankel, 2000).
4.5.1 APPLICATION
In terms of application, pre-defined ‘inclusion’ criteria were developed for both the organisations and the participants in order to enhance the plausibility and trustworthiness of the data and thus the conclusions being drawn (Le Compte and Preissle, 1993). The criteria employed are described below, along with details of the final samples.
Outsourced Leisure Trusts
The criteria used for sampling the case study organisations in this sector were based on four overarching factors. Firstly, the organisations had to be located in the UK public leisure sector (prior
87 to outsourcing). Secondly, the organisations had to be planning to outsource or recently outsourced from a local authority. Thirdly, the ‘acquirer’ of the public leisure organisation had to be a non-profit Leisure Trust. Fourthly, the organisations needed to be able to provide research access to workers in line with the timeframes outlined earlier in the chapter (i.e. at approximately three months and twelve months after the transfer).
The characteristics of the final sample of case study organisations are given below in Table 5. Further details of each of the case studies can be found in Chapter Five. More specific details have been withheld owing to anonymity promises to the managers of the leisure organisations.
Table 5: Characteristics of Case Study Organisations
Type Leisure Centres Leisure Centre Leisure Centres
No. One One Five
Workers
No. of Staff Transfers 108 65 300
Leisure Workers
The guiding criteria for the employee participants related to the following factors. Firstly, the participants would have to have experienced working for the leisure service under the Local Authority and the Leisure Trust - and experienced the transition of outsourcing25. Secondly, it was the endeavour to recruit participants from all levels of the hierarchy (e.g. team members, supervisors, managers) and a diversity of departments (e.g. the leisure, administration, reception, food and beverage, and housekeeping departments typically found in leisure centres).
25 On this point, while this typically was meant to be directed at those experiencing the ‘legal transfer’ from the local authority to the Leisure Trust, it is worth noting that six participants from across the case studies had interviews in the second wave of data collection, although they were recruited and employed by the Leisure Trust after the transfer had taken place. These could be enlisted as ‘unique’
(i.e. deviant’) cases.
88
The process of bringing order, structure and interpretation to qualitative data is not as linear and neat as many theoretical descriptions of data analysis might suggest. But this does not mean it should be without organisation and rigor (Marshall and Rossman, 2010). As Gibbs (2007) notes, qualitative data analysis involves some form of ‘transformation’, and this means taking a mass of data and analytically processing it into intelligent and understandable pieces of work. However, there are different forms of qualitative data analysis, and how one goes about this process depends on the nature of the research and the preference of the researchers involved. Some forms of qualitative data analysis emphasise the importance of scientific-like rigor in the sorting and indexing and ordering of data systematically (e.g.
Miles and Hubermann, 1994; Richie and Lewis, 2003). Others focus on the significance of narrative deconstruction to convey meaning from the data (e.g. Denzin, 1997; De Fina and Georgakopoulou, 2012). This study sought to embrace rigour in the process of analysing themes and trends in the data.
Yet, it also tried to capture something of the participants’ stories as well, in order to try to ensure the
‘voice’ of participants was accounted for.