Chapter 5: The method
5.4 The methods: semi-structured interview
The previous chapter established that this research has adopted the ethnographic case study, a qualitative approach, as the preferred method for generating data. The choice of research methods is determined by the research questions, (see Introduction) yet it is also constrained by cost, time, and the strengths of the researcher (L. Cohen, et al., 2000, p. 74). Since the 1980s, the mixed methods approach has become more common (Alise & Teddlie, 2010, p. 120; Bryman, 2004, p. 463). The paradigm wars, which contrasted ‘deep rich observational’ data with ‘hard generalisable’ data, have been superseded by debate of the value of the mixed methods research as a third way (R. B. Johnson & Onwuegbuzie, 2004, p. 15). Its proponents argue that it gives research greater rigour and validity (Bryman, 2004, p. 448), but such an approach cannot be an add-on, it must fit within the ontological and epistemological framework of the research and should not be employed on the basis of the more data generated the more valid the research findings. Therefore this research remains wholly qualitative, rather than a mixed methods approach as defined by Alise and Teddlie (2010, p. 112). The development of methods used did not follow a clear linear path, but was a process developing through constant referral between ethnographic experiences and the theoretical framework.
The main method used in my research was the semi-structured group interview. Interviews although an artificial way of obtaining information, are a valid means of data collection for the ethnographer (O' Reilly, 2005, p. 112). They are not merely a conversation initiated by the researcher to obtain information, but should be seen as a source of data generation. They are a social interaction between two humans, where ‘knowledge [is] generated’ (Kvale, 1996, p. 42). They are the means of allowing the interviewees the opportunity to share their ‘stories of the lived world’ and of providing the participants with a voice, and thus recognising their knowledge of their own stories and views (Cisneros- Puebla, Faux, & Mey, 2004). The ethnographic interview ‘facilitates, catalyses and exemplifies…self- narration’ (Nesbitt, 2004b, p. 123). It is no longer a mere means of data collection, but a process more fluid and multi layered, changing both researcher’s and participants’ knowledge and understanding of the topic under discussion. This requires a greater degree of reflexivity; the researcher needs to reflect not only on the variables in the life of the interviewee, but also in his/her own life.
The key characteristics of the qualitative research interview (Kvale, 1996, p. 29) are that it should be situated in the life world of participants, and attempt to obtain descriptions of different aspects of participants’ lives focusing on specific situations, with the interviewer interpreting the meaning of what is said. The knowledge generated, although interpreted by the researcher, is just one description of a particular specific situation in time and place and not a definitive description of the participants’ world reality. Group interviews have value within
their views on an issue, which add to the knowledge of the issue amongst the participants, as well as in the research community. There are many different types of research interview, (Bryman, 2004, p. 319; L. Cohen, et al., 2000, p. 270; Hitchcock & Hughes, 1995, p. 156). The three relevant types here are the informal individual, the semi-structured individual and the semi-structured group interview.
The unscheduled informal interview that takes place within the context of participant observation is useful, when the research is already underway and the researcher is able to ask more specific and relevant questions (Agar, 1980). Burgess (1983) gained many insights through informal conversations with staff in the course of his participant observation. Informal interviews can also reinforce or challenge data generated in more formal interviews.
The semi-structured type of interview is described as a ‘ high preparation, high risk, high gain, and high analysis operation’ (Wengraf cited in Seale, 2004, p. 186). It is most appropriate within my research, as it enables the researcher to focus the interview on a specific topic, and yet allows for flexibility. It gives scope for questions to be asked out of sequence, taking the form of a ‘controlled conversation’ (Burgess, 1984b, p. 165). The relationship between interviewer and interviewee is more equal, neither having complete control over the conversation, which can be a risk for both parties. The interviewer takes a stance of ‘deliberate naiveté’ (Kvale, 1996, p. 30), being open to new and unexpected ideas, while the interviewees’ views are often ambiguous and open to change through the
are that it is time-consuming (although less so than the unstructured interview), not easily codified (as the outcomes are so varied), and yet still, to some extent, an artificial construct.
There is no one right way of interviewing, no single correct format that is appropriate for all situations, and no single way of wording questions that will always work…Therein lie the challenges of depth interviewing: situational responsiveness and sensitivity to get the best data possible. (Genzuk, 2003, p. 6)
One way of overcoming the issue of the artificiality of the interview situation is the semi-structured group interview; it falls between participant observation and the individual interview. Although still an artificial device, it is more natural than the individual interview (Seale, 2004, p. 197), emphasising the ‘interactive and communicative nature of social action and social meaning’ (p.198). Group discussions are more like normal everyday conversations, with known rules and norms (O' Reilly, 2005, p. 129), where the presence of the researcher becomes less obtrusive, and participants may be more open (O' Reilly, 2005, p. 131). Lewis (1992) has undertaken a detailed analysis of the use of group interviews with children and how they differs from individual interview. Group interviews tend to generate information, which is more in line with the group norms (Lewis, 1992, p. 414). The comfort of a group may encourage children to voice views that they would not have done individually, and to stimulate new ideas. It also avoids the problems resulting from a child being reluctant to answer questions, or feeling that he or she has to supply an answer however nonsensical. A good example of
exploring young people’s relationship to local space (Seale, 2004, p. 198). A disadvantage of group interviews is that they may reinforce a group identity, a ‘preferred identity’ (Dunlop, 2008, p. 26). However, within the context of my research, group interviews are of value as what is sought is the view of the pupils as members of a Catholic community rather than as individuals, so as to build a picture of how the school community is perceived.
The size of the group is important, and groups of four to six children are most effective (Lewis, 1992). They often work better where the participants are already at ease with each other (Lewis, 1992, p. 418). The interviewees in my research were chosen in consultation with the gatekeepers in the three main schools. In St Julian’s the school already had an established process for selecting students for focus groups. Indeed such groups were used in interviews of prospective staff and to evaluate aspects of school life. The deputy head, the gatekeeper, made arrangements for the groups to be drawn from across the school, including staff. When I sought to follow up ideas from the initial interviews, the gatekeeper arranged for particular groups to be available, for example to interview students involved with chaplaincy. In St Margaret’s and St Catherine’s the head of RE selected students to participate in interviews. The St Julian’s participants are not a random sample; there is the likelihood that the deputy headteacher who acts as gatekeeper, will select participants who reflect a positive view of the school. This should not affect the validity of my research, as long as any analysis or generalisability of the research findings reflects an awareness that the chosen sample of young people within the school may have been chosen because they are
likely to represent a positive view of their Catholic school. The groups and numbers of interviewees are set out appendix B.
Informed consent is a particular issue when researching in a school and must contain four elements, competence, voluntarism, full information, and comprehension (Diener and Crandall [1978] cited in L. Cohen, et al., 2000, p. 51). Consent was sought from not only the head teacher, but also the parents or guardians in the case of children and the young people themselves (British Sociological Association, March 2002 paragraph 30).Burgess (1984a) explained that in his research the head teacher gave permission for him to interview the staff and pupils, but the participants had the right to refuse to take part. All participants in my study were given the opportunity to withdraw from interviews. Although only two students in the whole study did, I ensured that for all interviews the school had arrangements in place for the students to return to class or be supervised elsewhere.
A key to the success of an interview is the relationship between interviewer and interviewee and it depends on a rapport being established between participant and researcher (Bryman, 2004, p. 119), which is a skill not always easily acquired (Lofland 1971 p.90 cited in Hitchcock & Hughes, 1995). The rapport or lack of rapport needs to be reflected upon within the research as it forms part of the whole picture of the interview. In interviewing young people in a school setting, the gulf between teacher and pupil is always going to be present, but that does not necessarily negate the possibility of establishing a rapport. I always introduced
was already known as a supply teacher, while in St Julian of Norwich, some of the staff interviewed were aware of my role as a governor. Although I was always very aware of the conflict between my role as a teacher and researcher, the students were not, indeed, one year 12 student in St Julian’s remarked at the end of the interview how much they had enjoyed the discussion, but they would not have been able to talk like that with a teacher present.
The power within any research study lies with the researcher, in particular when the majority of participants are children (see Chapter 5). This imbalance can be mitigated to some extent by constant awareness and reflection of this issue throughout the research process (Østberg 2003b, p. 27). It is a privilege for others to allow you insights into their world, and the responsibilities to these participants must be recognised and the imbalance of power cannot be ignored. Research should be open, but there are times when being open will not result in any meaningful data (Burgess, 1984b, p. 199). For example when researching potentially sensitive or controversial issues such as treatment of other faiths in a Roman Catholic school, being open to all participants may result in the interviewer being told what the participants feel to be the ‘correct view’. Another issue is that at times the researcher might be required not to be totally truthful, (L. Cohen, et al., 2000, p. 65). For example when interviewing the students I denied knowledge of various aspects of school organisation, to allow them to tell their perceptions rather than just substantiate mine.
ended and indirect questions, questions that will encourage a detailed response and the sequencing of questions within the interview is important. Patton (1980) argues that less threatening questions should be at the beginning of the interview, to put the interviewee at ease (L. Cohen, et al., 2000, p. 280). Thus it is useful to begin with direct questions requiring factual answers and to move from this to more indirect, non-specific questions. It is however the indirect, non-specific questions, which the interviewer hopes are going to produce the most detailed and relevant responses. As Kvale (1996, p. 145) points out, the aim is for short questions, which will produce long answers (see appendix B for interview question schedule).
The focus of an interview can be maintained not just through questions, but also through other prompts. When interviewing, visual prompts, such as pictures, photos and concept maps may be very useful; Dunlop’s (2008) study in Eastern Europe used photograph elicitation very successfully to generate data. In a study of Muslim schools, Gent (2005, p. 48) used photographs as ‘prompts’ enabling the participants to describe what was happening in ‘hfiz’ classes. On a practical note, when interviewing children visual aids help to minimise boredom and maintain a child’s concentration. In three of my early interviews in St Julian’s, photographs were used as prompts, but this was not wholly successful as students became distracted, as the images chosen were not immediately recognisable to the students, for example the two photos showing a nun in a habit and a Muslim woman in a chador. The interview became side tracked into an explanation as to who these people could be, as neither were recognisable to the students, although
Where the interview takes place has some influence on how it is conducted and this needs to be taken into account when interpreting the results. The rooms allocated within the schools, included a plain anonymous meeting room, RE classrooms, free spaces between classrooms, and the chaplain’s office. Being able to talk freely without interruptions is ideal, but in a busy school this was rarely achieved for the length of the interview. Interruptions from the school bell, people walking through the room and noise from other pupils at break or lesson changeover time were inevitable. As to the length of interview, school-based research has to fit in with school timetabling, but the aim for the length of the interview was 45 minutes; any shorter and there is the risk of not having time to establish the rapport essential for the conversation, any longer and the transcribing becomes a difficulty.
Any data generated by group or individual interview needs to be shown to be credible and to be validated (Seale, 2004, p. 77). The data generated has to have some connection with or relevance to the external world, otherwise the recorded and transcribed interview becomes an account of no more value than a fictional account of the same incident. Within interviews, there is the danger that participants are always going to be tempted to tell the interviewer what they think the interviewer wants to hear. In addition, their perception of reality might not be substantiated by subsequent participant observation, where the researcher observes whether what is said is in fact what is practised. Stringer (1999, p. 52) argues the researcher must always be aware of the different discourses, ‘ what
several ways data can be validated, through internal triangulation (obtaining the same data from the same person using a different method), external triangulation (comparing that data with similar from other participants) or through participant observation (O' Reilly, 2005). The data generated in this research was validated by external triangulation and through observation.