• No results found

Research design of Phase 2: Interviews

Chapter 3 Methodology

3.5 Research design of Phase 2: Interviews

After gaining some understandings of the disciplinary characteristics of geography through web-based content analysis and bibliometrics, data collection of Phase 2 was carried out between September 2014 and November 2014 using semi-structured interviews. The purpose of phase 2 was to investigate researchers' research practices, understandings and attitudes of data sharing and thus was related to research sub-questions 1, 2, and 3. This phase was a key part of this research.

A research interview has been described as “a conversation with a purpose” (Berg, 2009; Kvale &

Brinkmann, 2009; Marshell & Rossman, 2006; Kahn & Cannel, 1957:149). In a research interview, the interviewer seeks to elicit information from the interviewee (Bryman, 2016). The interview is one of the most popular methods in qualitative research (Bryman, 2016; Packer, 2011). Indeed, the interview has become the method of choice for data collection in a diverse range of qualitative research approaches, including ethnography, phenomenology, case study and grounded theory (Creswell, 2013; Packer, 2011). It can be used as an overall strategy or as one of the methods employed in a research project (Marshall & Rossman, 2006). Many qualitative research projects, in fact, use only interviews to collect empirical data (Packer, 2011).

In this research, interviews were used as a core data collection method. According to May (2011) and Kvale & Brinkmann (2009), qualitative interviews seek to understand the world from

interviewees’ perspective. Using interviews in this phase is in line with the qualitative interpretative philosophical approach of the research. As the purpose of Phase 2 was to understand researchers' research practices and their attitudes towards research data management, using interviews was a appropriate method which can provide rich insights into interviewees’ experience, views and attitudes related to research data.

3.5.1 Types of interviews

Semi- structured interviews are employed to obtain qualitative data in this research. There are three main types of interviews: the structured interview, the semi-structured interview and the

unstructured interview. Typically, the structured interview is associated with quantitative research, while the un-structured interview and semi-structured interview are associated with qualitative research (Bryman, 2016). The differences between these interview types are not only their degree of flexibility in connection with interview structure, i.e. the wording and sequence of the interview questions but also their underlying assumptions (Berg, 2009; Gillham, 2000).

57 Structured interviews

In a structured interview, the researcher uses a formally structured interview schedule. Questions in this type of interview are usually very specific and close-ended (Bryman, 2016). Each interviewee is asked identical interview questions in the same wording and order. Thus, answers can be

aggregated, standardised and compared (Berg, 2009; Bryman,2016; Denscombe, 2010). Research using structured interviews is usually based on a few assumptions. First, researchers using this type of interview have solid ideas of what things they want to reveal and what kinds of answer they need in the interviews (Gillham, 2000). It is assumed questions used in interviews are sufficient to gather all information for their study (Berg, 2009). Second, it is assumed that there are sufficient common vocabularies between the interviewer and interviewees, so interviewees can clearly understand what they are being asked (Berg, 2009). Third, it is assumed that the meaning of each question is the same for every interviewee (Berg, 2009; Frankfort-Nachmias & Nachmias, 2000). A structured interview approach is not suitable for this research, in particular because it cannot be assumed that there are enough common vocabularies to formulate the questions so that they are understood in the same way. It is expected that terms such as research data, research data management need to be clarified during interviews, for instance, research data can mean different things to different researchers. Overall, this type of interview does not allow enough flexibility to gain insights into researchers’ attitudes and practices towards research data management.

Un-structured interviews

In contrast to the structured interview, the un-structured interview does not have an interview schedule. It is like a free-flowing conversation which is started by a theme or a topic (Bryman, 2016).

Unlike the structured interview, the underlying assumption here is the researcher does not know all the necessary questions prior to the interview (Berg, 2009). Thus the researcher does not have a full list of interview questions. Interviewers need to develop and generate appropriate questions and follow-up probes in the interviews (Hakim, 2000). Researchers using structured interviews are interested in the interviewees’ views. Interviewees are encouraged to use their own words to describe their own experiences, beliefs or issues etc. (Frankfort-Nachmias & Nachmias, 2000).

Semi-structured interviews

Similar to un-structured interviews, semi-structured interviews also place emphasis on the

interviewees’ point of view (Bryman, 2016). Semi-structured interviews involve using an interview guide. The interviewer will have prepared a list of questions. However, unlike the structured interview, interviewers do not necessary follow the exact wording or the order of the questions. In addition, follow-up questions and probes are used to get more details. In semi-structured interviews,

58 researchers assume that in order to standardise questions, questions need to be expressed in words that are familiar to the interviewees (Berg, 2009). This assumption shows that researchers are actually aware of different interviewees understand the world differently (Gubrium & Holstein, 2003). Therefore, researchers understand and interpretive the world from interviewees’ viewpoint.

In this research, the semi-structured approach was chosen because this approach provides flexibility to gain insights into researchers’ research practices, data management practices and their

disciplinary cultures. Yet, the semi-structured approach also provides some structure to allow comparability, for instance, to compare differences between different sub-disciplines, which un-structured interviews cannot offer.

3.5.2 Sampling

A total of 23 interviews with academic researchers in geography were conducted during the period 23 June – 7 Oct 2014. The first interview served as pilot. The pilot resulted in the minor modification of one or two questions, but did not necessitate any major revision to the interview protocol, therefore the first interview was included as part of the dataset. In a review paper, Baker and Edwards (2012) investigated the question of ‘how many qualitative interviews is enough’. The answers suggested by experts in the paper range from 1 to 60 interviews, with 30 being the mean.

As Baker and Edwards (2012) and Pratt (2009) point out, there is no right answer and it depends, for example, on the research questions, objectives and resources. More interviews does not necessary mean better. Qualitative research is more focussed on ‘richness, complexity and detail’ than

quantity (Baker and Edwards, 2012:5). Tracey Jensen advised that the quality of analysing interviews is actually more significant than the quantity of the interviews (Baker and Edwards, 2012). In this research, the sample was 23 researchers because the number is not large enough to enable a comparison across human geography and physical geography.

Purposive sampling was used to select participants, which was consistent with the overall qualitative approach of the research. Sampling in qualitative research is quite different from in quantitative research (Kumar, 2011). Quantitative research usually involves probability sampling in which a sample is selected randomly to ensure each unit in the population has an equal chance to be selected (Bryman, 2016). In contrast, most sampling in qualitative research involves purposive sampling. Purposive sampling is a form of non-probability sample. Researchers do not select a sample using a random selection approach but rather select sample participants or cases strategically in order to ensure the sample is relevant to the research goals (Bryman, 2016).

Typically, researchers select a wide range selection of types of individual to represent factors that might affect responses to the research questions. A major problem with purposive sampling is that

59 researchers are unable to generalise findings to a population (Bryman, 2016). Nevertheless, as Polit

& Beck (2010) point out qualitative researchers are more interested in gaining rich, contextualised understandings about issues, individuals or a situation rather than generalising findings, which is the goal of most quantitative research.

As this qualitative research is exploratory by nature, suitable research fields and interviewees were identified after Phase 1. The criteria for selecting suitable participants were the research area of researchers, the volumes and types of research data they produced and the ranking of the university.

3.5.3 Interview guide

An interview guide was developed (Appendix 10). The interview guide not only developed detailed interview questions but also determined the purpose of each question. Interview questions were divided into few topic areas, such as, research areas, data practices, data sharing and data re-use.

All interviews were recorded and notes were taken for important things during interviews. Data were then transcribed and imported to Nvivo for categorising and coding.

3.5.4 Data analysis

Thematic analysis was employed to analyse the qualitative interview data. Braun and Clarke (2006:79) describe thematic analysis as:

“a method for identifying, analysing, and reporting patterns (themes) within data. It minimally organises and describes the data set in (rich) detail. However, frequently it goes further than this, and interprets various aspects of the research topic.”

Although thematic analysis is widely used, there is no clear standardised agreement about the definition of thematic analysis or the way to produce it (Bryman, 2016; Braun & Clarke, 2006;

Boyatzis, 1998). It is sometimes argued that this approach is not an identifiable method of analysis, as searching and categorising themes featured in thematic analysis are essential in a lot of

approaches of qualitative data analysis, such as grounded theory, qualitative content analysis, and narrative analysis (Bryman, 2016). Nevertheless, this approach was chosen for this research because of its flexibility. As Braun & Clarke (2006) point out many other analytic approaches are theoretically bounded. For instance, interpretative phenomenological analysis is bounded to phenomenological epistemology, while grounded theory requires detailed technological knowledge of the approach in order to develop theory which is grounded in data (Bryman, 2016). Thus, thematic analysis provides a theoretically-flexible approach. It is an alternative to researchers who do not want to commit totally to produce a fully grounded theory analysis (Braun & Clarke, 2006).

60 In this research, data was analysed in six phases. The analytical approach was adopted from Braun and Clarke (2006). It is noted that the process of analysis is not a linear set of steps but a recursive process (Braun & Clarke, 2006): It involves moving back and forward between phases.

Familiarisation with data

The analysis started with familiarisation with the content of interview data, including, reading and re-reading whole transcripts and field notes, taking notes for ideas, annotating and highlighting significant quotes. Previous disciplinary knowledge gained from web-based content analysis and bibliometric work during the phase one of the study were also used to be sensitive to the context of the interview data. This stage of the analysis was used to gain a holistic understanding of the whole set of empirical data. It was an initial exploration of the data before coding (Saldana, 2013; Bazeley, 2013). This preliminary stage of data analysis was carried out throughout and after the data

collection process.

Generating initial codes

Second, a first cycle coding was carried out (Salanda, 2013). Codes were applied to a short section of data. According to Salanda (2013:3), a code in qualitative research is

“a word or short phrase that symbolically assigns a summative, salient, essence-capturing, and/or evocative attribute for a portion of [..] data”.

Coding can be seen as a process of indexing. As Bryman (2016) suggested, coding creates an index of terms which help interpret and theorise data. Thus, this stage of analysis of this research was about organising data in a meaningful way. Interesting aspects of the data were coded systematically to generate initial codes from the data. Table 3.4 below shows some initial codes generated from interviews of the research. As this stage of analysis was exploratory in nature, it can be seen that codes were not organised and some codes in some cases overlapped.

61

Culture of data sharing in different communities Data after the completion of research project Data citation

Research data produced by researchers

Table 3.4 Initial codes

Refining codes and theorising concepts

The third stage of the analysis was to review and refine identified codes and develop themes or concepts. Second cycle coding was conducted. As Saldana (2009) stated, the purpose of the second cycle coding is to further develop a conceptual, thematic and theoretical organisation from the first cycle codes. Annotations and analytical memos were to help identify conceptual connections

between codes and reflect how codes can be integrated into themes. Codes were then reviewed and evaluated to ensure the codes work with the coded data extracts as well as the entire data set.

Definitions of codes and a reviewed code structure of this stage are shown in Appendix 3 and 4. At the end of phase five, a thematic map of the analysis and an initial plan of the write up of the interview findings were produced.

Writing up the analysis

The final stage was to conduct the final analysis and write up the report. At this final stage, work was focused on linking the analysis to the literature and research questions in order to produce a

coherent and logical report of the analysis (Braun & Clarke, 2006). The findings of the interview are reported in chapters 7-10.

62