CHAPTER 3: RESEARCH METHODOLOGY 77
3.3 RESEARCH RPOCESS 86
3.3.2 Semi-structured Interview 86
3.3.2.1 Purpose of Semi-structured Interview
Cannel and Kahn (1968) defined an interview as “a two-person conversation initiated by the interviewer for the specific purpose of obtaining research-relevant information, and focused by him on content specified by research objectives of systematic description, prediction, or explanation”. Interviews are particularly useful for pursing in-depth information around the research topic. There are three types of interview categories: structured, semi-structured, and unstructured. The differences among these types lie in the interviewing process and question patterns.
As one of the most commonly used methods of data collection, semi-structured, in- depth interviews are characterised by more or less open questions being brought to the interview situation in the form of an interview guide (Flick, 1998). Semi- structured interviews are more likely to evoke the interviewees’ viewpoints than standardised interviews and questionnaires, which may restrict, rather than illuminate, the interviewee’s standpoint (Kohli, 1978). Only a number of pre- determined questions which are relatively open are designed in advance, while the subsequent interview questions are raised during the interview itself (Wengraf, 2001). Semi-structured interviews are designed to have a number of interviewer questions prepared in advance, but such prepared questions are designed to be sufficiently open so that the subsequent questions of the interviewer cannot be planned in advance but must be improvised in a careful and theorised way (Wengraf, 2001).
The semi-structured interview is a flexible tool, allowing interviewees to express their opinions freely while interviewers can also give appropriate interventions at the necessary moment, and new questions can be posed during the interview as a result of what the interviewee says. These characteristics give rise to the justification of semi-structured interviews in this study. Consequently, in order to deeply and better understand the organisational management and sustainability issue in Australian universities, semi-structured in-depth interviews were constructed as the initial data collection. In this phase, the researcher intended to utilise semi-structured interviews to elaborate organisational elements and identify organisational obstacles which deter the adoption of green technology innovation such as the Green Roof and Living Wall.
The semi-structured interviews aimed:
To obtain the general recognition and understanding about sustainability in higher education
To investigate the common organisational environment and explore organisational issues in Australian universities
To inspect the project management system and benchmark the project delivery process in universities
Overall, the completion of the targets above provided rich and meaningful seed information for the following Delphi study.
3.3.2.2 Selection of Interview Respondents
Choosing appropriate samples is an important step in qualitative research as answers lie in the samples. Morse (1991) suggested that there are four types of sampling used in qualitative research: the purposeful sample, the nominated sample, the volunteer sample, and the sample that consists of the total population. For the interview method in this research, purposeful sampling and the snowball technique were used to contact key informants and participants.
Purposeful sampling is the choice of “informants with a broad general knowledge of the topic or those who have undergone the experience” (Morse, 1991). Thus, in contrast to picking up random samples from the population, the criteria to select participants for this study depended on the key informants being able to provide rich and diverse information about sustainability deliverables in universities. As this research was developed to specifically target the Australian higher education system, sample sites were selected from 39 Australian universities. The choices of universities were mainly based on the degrees of differentiation in the universities’ advance in the sustainability area for the purpose of drawing a full picture of Australian universities’ involvement with sustainability deliverables. According to the signatory list of the Talloires Declaration which is renowned worldwide for leading the trend of “sustainability in higher education”, 21 Australian universities have signed the declaration to make commitments to sustainability. With reference to this indicator of international influence, the universities chosen in this research project included some on the signatory list and others which were not in order to present a full picture.
In the next step, interviewees were selected based on their professional expertise, academic background and working experiences related to sustainability programs in universities. In this research, targeted informants were the group who dealt with sustainability issues in universities on a daily basis, such as sustainability managers, environmental managers, facility project managers and senior administrators. Because all interviewees occupied key positions of dealing with sustainability programs in Australian universities for many years, their responses could then be assumed to be creditable and reliable.
With purposeful sampling, subjects may be able to recommend useful potential candidates for study, which is snowball technique (Martin, 1996). Therefore, at the end of each interview, the researcher asked the interviewee to recommend another person whom he or she thought might be suitable for the research. As Lindlof and Taylor (2002) stated, a snowball sampling technique “is well-suited to studying social networks, subcultures, or dispersed groups who share certain practices or attributes”. The sum of individual participants and key informants could have been as few as 10 people or as many as 40 people (Strauss & Corbin, 1998). Thus, the actual sample size could not be pre-determined in this study, but the collection of samples would cease at the moment the information is saturated when new categories, themes or explanations stop emerging from the data (Martin, 1996). Figure 3.3 illustrates the process of choosing qualified interviewees.
Figure 3.3:Logic of Identifying Interviewees
After the sample size is decided, the next important step is to choose the way the researcher gets access to the targeted respondents. Before the real action of research, targeted samples will be contacted by an email or phone call. In principle, researchers and their study participants must agree on the expectations during the study process, particularly the expectations each party has of the other (Miles & Huberman, 1994). The participants should be informed that the participation is voluntary, and that they can withdraw at any point during the process. In order to develop a trustful relationship between the participants and researcher, the full explanation about the confidentiality and anonymity should be provided. A cover letter which includes all these key points was sent to potential participants to gain their agreement to participate in this study. The consent form for the interviews
39 Australian Universities
21 Australian Universities which are on the signatory list of Talloires Declaration
18 Australian Universities which are not on the signatory list
Sustainability managers Environmental managers Facility project managers Senior administrators (based on knowledge and experience)
stated that every interviewee has read the purpose of the study and accepted the requirements and agreed to participate in the interview process.
3.3.2.3 Interview Process
In order to reduce the researcher’s bias in the questions, to enhance the internal validity, and to guarantee that the initial interview can cover as many as of the topics as possible, the author used a pilot study to refine the interview questions. To start with, open-ended questions for semi-structured interviews were designed and distributed to the panel involving a selected group of experts and professionals. During this testing period, the feedback about whether these questions were appropriate or where other questions should be added helped the researcher to guarantee the validity of the interview questions.
Interviewees were contacted by phone or email and they all signed the consent information sheet. Due to the constraints of time and financial resources, face-to-face interviews were conducted when participants were in the Queensland area; other interviews were conducted by phone or webcam. Each interview lasted for 45 minutes or 1 hour depending on the specific condition, and was tape-recorded with the permission of interviewees. All the interviews were transcribed verbatim for the preparation of coding. Common themes emerged after the process of coding and categorising. The information derived from interviews was used in the next stage, namely, the Delphi study.
3.3.2.4 Interview Data Analysis
Qualitative data analysis “is the process of systematically searching and arranging the interview transcripts, field nots, and other materials that accumulate to increase own understanding of them, and to enable to present what have been discovered to others” (Bogdan & Biklen, 2006). Specific to this research project, qualitative content analysis was chosen to process the interview data, which is described to be analytic approaches ranging from impressionistic, intuitive, interpretive analyses to systematic, strict textual analyses (Rosengren, 1981).
Qualitative content analysis is defined as:
“a research method for the subjective interpretation of text data through the systematic classification process of coding and identifying themes or patterns” (Hsieh & Shannon, 2005)
“an approach of empirical, methodological controlled analysis of texts within their context of communication, following content analytic rules and step by step models, without rash quantification” (Mayring, 2000) “any qualitative data reduction and sense-making effort that takes a
volume of qualitative material and attempts to identify core consistencies and meanings” (Patton, 2002).
Content analysis can provide knowledge and understanding of the phenomenon under study (Downe-Wamboldt, 1992) from the text data obtained from interviews, narrative responses, open-ended survey questions, focus groups, observations or print media such as articles, books, or manuals (Kondracki & Wellman, 2002). Content analysis is regarded as a flexible method for analysing text data (Cavanagh, 1997). There are three main approaches to content analysis, namely, conventional, directed, and summative content analysis. Conventional content analysis directly develops coding categories from the rich data. In a directed content analysis, initial coding scheme is developed through existing theory or prior research before analysing the data (Kyngas & Vanhanen, 1999). The summative approach starts with identifying and quantifying certain words or phrases (frequency count) and later interprets the contents within which explicit or euphemistic words or phrases are used.
Different approaches are applied in response to the theoretical and substantive interests of the research and the problem being studied (Weber, 1990). Conventional content analysis is usually appropriate when existing theory or research literature on a phenomenon is limited. As an inductive category development (Mayring, 2000), new categories and insights emerge from the data. In comparison, directed content analysis is referred to as a deductive application (Mayring, 2000) which begins coding immediately with the predetermined codes. By showing the rank order comparison of frequency of codes (Curtis et al., 2001), the discussion might offer a contradictory view or further refine, extend and enrich the theory. Summative content analysis has a quantitative focus on counting the frequency of specific words or phrases. This counting can help identify patterns in the data and contextualise the codes (Morgan, 1993) which allows for interpretation of the context associated with the use of word or phrase (Hsieh & Shannon, 2005). Although these three content analysis approaches have differences, they share a similar analytical process that can
be summarised into seven steps: formulating the research questions to be answered, selecting the sample to be analysed, defining the categories to be applied, outlining the coding process, implementing the coding process, determining trustworthiness, and analysing the results of the coding process (Kaid, 1989).
In this research, conventional content analysis was chosen due to the need to explore new insights into organisational issues, the project management process and the specific topic of GRLW implementation in universities. A set of systematic procedures for processing data is discussed according to each of the steps.
1) Step 1: Data preparation
To start with, targeted questions with predetermined directions were asked throughout the interviews. For example, the research probed specifically to explore participants’ opinions about or experience of organisational structure, decision- making or organisational culture. The interviews tapes were transcribed into text documents using Microsoft Word software. Significant non-verbal and para- linguistic communications were also noted as much as possible besides the literal statements (Hycner, 1985). Most interviews in this research were conducted over the phone which didn’t allow the researcher to observe the participants’ facial expression or body language. However, the researcher made notes about the participants’ tones, pauses or repetition during the conversation. Once the interpretations were finished, the researcher listened to the entire tape several times and read the transcription repeatedly to achieve immersion and obtain a sense of the whole (Tesch, 1990). This can ensure the accuracy of narratives and help to avoid missing any new emerging codes.
2) Step 2: Developing coding schemes and categories
The initial stage of qualitative analysis starts from “open coding” (Murphy et al., 1998). Within this process, the data is broken down, conceptualised and categorised, and similar incidences, claims and discursive practices are grouped together (Strauss & Corbin, 1998). Labels for codes emerge to reflect key thoughts, which become the initial coding scheme (Hsieh & Shannon, 2005). A coding scheme is a translation device that organises data into categories (Poole & Folger, 1981). The constant comparative method (Glaser & Strauss, 1967) is encouraged to allow the emergence of categories through constantly comparing the current transcript with previous ones.
In this study, codes were grouped into categories based on how different codes were related and linked. As more and more codes appeared, the number of categories increased and the contents under each category became large.
3) Step 3: Integrating mega-themes
The relationship between a category and its sub-categories can be identified based on their concurrence, antecedents or consequences (Morse & Field, 1995). These emergent categories are sorted into meaning clusters (Coffey & Atkinson, 1996; Patton, 2002), and further integrated to determine if there are one or more themes which express the essence of these clusters.
4) Step 4: Data presentation
When presenting qualitative content analysis results, there should be a balance between description and interpretation. Sufficient description provides readers with rich and thick background and context (Denzin, 1989), while sufficient interpretation gives the researcher’s personal and theoretical understanding of the phenomenon to allow readers to understand the description (Patton, 2002). In this thesis, tables of interviewees’ quotations were provided to demonstrate the description before the discussion of organisational environment, project management system and GRLW application in universities. Additionally, key findings for each section of interviews were summarised to make a clear list for readers’ understanding. The result of a conventional content analysis is concept development or model building (Lindkvist, 1981), because the advantage of the conventional approach to content analysis is to gain participants’ unique perspectives from the actual data without imposing preconceived categories or theoretical perspectives (Hsieh& Shannon, 2005).