This section describes the data used in the study. It offers an overview of the context of the study as well as the sample that was selected. The size of the sample is explained and justified and the ethical implications of the data are discussed.
4.1. Context of study
The research includes successfully examined doctoral dissertations from (primarily) four research-intensive universities in South Africa5. South Africa presents a particularly interesting context in which to carry out this research, given the unequal educational history of the country. Since democracy in 1994, higher education has undergone significant massification and diversification as access to previously excluded candidates opened up. In a seminal text on the transformation of higher education in South Africa, Wally Morrow (2009) distinguishes between ‘formal access’ (i.e. registering a student at a university) and ‘epistemological access’ (p.77). The latter form of access, which he defines as ‘learning how to become a successful participant in an academic practice’ (p. 78), is of particular significance to this research. In order for students to succeed in higher education, they must achieve ‘epistemological access’
5 Research-intensive universities are the more traditional universities offering Bachelor’s to doctoral degrees
and which place strong emphasis on postgraduate and research activities. Different to comprehensive universities and universities of technology which adopt a stronger focus on undergraduate qualifications and
to the academic practices of their discipline – in effect, to access knowledge. One way of ensuring this access, is to make academic practices explicit and demonstrable to students. In the South African context, the need to ensure epistemological access is heightened given that a significant number of first-generation or previously disadvantaged candidates are now entering the system who require extra support (Council on Higher Education, 2016; National Planning Commission, 2012). There is thus a growing need to make academic practices more explicit so that these learners can thrive in higher education environments. The creation of tools for analysing knowledge in dissertations – for the purpose of developing doctoral writing pedagogy – is therefore a timely and significant step for ensuring ‘epistemological access’ in this context.
In addition to the heightened pressure on academics to accommodate a greater number and more diverse cohort of candidates, South African academia is also facing an aging professorial population (Cloete, Mouton, & Sheppard, 2015; Council on Higher Education, 2016; National Planning Commission, 2012). There is therefore a need to not only make elusive practices explicit to candidates but also to new supervisors entering the field who are expected to teach candidates how to write their dissertations. These historical and situational pressures make the choice of using South African data an interesting and appropriate case study to develop conceptual tools.
The choice of South African universities included in the study was informed by purposive sampling, whereby specific institutions and informants that were judged to possess a particular characteristic of interest to the researcher were targeted (Cohen et al., 2011). Given that research-intensive universities have a strong focus on research, these institutions were deemed the most likely to produce top achieving doctoral candidates and subsequently high quality dissertations. It was also assumed that such institutions would attract top academics who would have been exposed to a wide range of doctoral research, either through their own supervising, thesis examining or involvement in the scholarly community. They were thus considered to be the best judges of what constituted an ‘exemplary’ dissertation, as described in the next section.
4.2. Sample selection
Dissertations collected for the study come from what is known as the ‘social sciences’ and ‘humanities’. This decision was informed by the aim of creating appropriate tools for
discourse-based doctoral dissertations. Different semiotic systems (such as mathematical equations, numbers and images) typical of dissertations in the natural sciences would require a different set of tools to understand the knowledge being enacted in the writing. Such a focus is beyond the aims of the study, as well as being beyond my level of expertise in mathematics and science. Developing tools for discourse-based dissertations thus provides a starting point to see how the knowledge work in doctoral writing can be better understood.
In order to create tools with pedagogic potential, I wanted to work with particularly good dissertations as these would, in theory, enact exemplary knowledge practices. Due to doctoral research not receiving an assessment grade (i.e. there is no easily quantifiable way to distinguish high achieving dissertations to lower achieving ones), I decided to approach select experienced academics for recommendations. As such, reputational case sampling, a sub-type of purposive sampling, best describes the data gathering method as dissertations were ‘selected by key informants, on the recommendation of others or because the researcher [was] aware of their characteristics’ (Cohen et al., 2011). Such an approach to gathering exemplary dissertations has been used in other studies of doctoral theses, such as that conducted by Paltridge, Starfield, Ravelli, and Tuckwell (2012). The ‘key informants’ included Heads of Departments from the selected subject areas, as well as a selection of professors. Professors who held research or funding chairs were specifically targeted, given the assumption that these positions of authority gives credibility to the academics’ disciplinary knowledge and experience in the field.
The key informants were contacted via their publicly available university emails. The initial email explained the aims of the research and asked the individual to recommend any exemplary dissertation(s) which they had either supervised or examined. No date limitation was given. If no response was received, two follow-up email reminders were issued at four-week intervals. In total, 71 academics were contacted. Of this cohort, 24 issued responses, and of those, 17 gave recommendations.
In total, 25 doctoral dissertations were included in the study. This generated a considerable data set of approximately two million words. The breakdown of dissertations according to subject area and reference number used in the thesis is provided in Table 4.1. All but one dissertation come from traditional research-intensive universities in South Africa; however, all
universities. The sample comprises four dissertations from Rhodes University, ten dissertations from the University of Cape Town, three from Stellenbosch University, one from the University of the Western Cape, and seven from the University of the Witwatersrand.
Table 4.1. The sample of dissertations
REF. NO. SUBJECT
1 History 2 History 3 Psychology 4 Psychology 5 Anthropology 6 Political studies 7 Sociology 8 African studies 9 Anthropology 10 Anthropology 11 English literature 12 English literature 13 English literature 14 Political studies 15 Sociology 16 Sociology 17 Anthropology 18 Archaeology 19 English literature 20 History 21 History 22 Political studies 23 Psychology 24 Psychology 25 Sociology
Reference numbers (e.g. ‘Dissertation 1’) were allocated to the dissertations in the sample primarily as an organisation system. The use of a more detached referencing system was also intended to create distance between individual authors, institutions and subject areas in the research, to enable a focus on just the written product (i.e. the dissertation itself).
For triangulation purposes, data also included candidate examiner reports. These were used to confirm the exemplary status of the dissertations by providing a second perspective to that of the academic who made the recommendation. 6
4.3. Sample size
A selection of 25 dissertations – or two million words – is argued to be an appropriate size for meeting the aims of the research. Given the exploratory nature of the research, the sample provides a good breadth of different disciplines as well as including enough dissertations within the same subject areas to reveal intra-disciplinary variation. Furthermore, the sample size provided an appropriate amount of data for conducting detailed analysis at both macro and micro levels while remaining practically manageable. The sample is not intended to be representative of all doctoral dissertations in the ‘humanities’ and ‘social sciences’ in all South African universities. Working with such a sample is not necessary for the goals of the research and it would be beyond the scope of an individual doctoral study.
4.4. Ethical considerations
Ethical considerations for any piece of research considers three main issues relating to research practice: risk (both to the participants and the researcher her/himself), benefit and
consent (The National Health and Medical Research Council, The Australian Research Council, & The Australian Vice-Chancellor's Committee, 2015). Due diligence was taken in this study to ensure that all ethical concerns were addressed. The doctoral dissertations that form the sample are part of the public domain: they are freely accessible online via their institutional repositories. Given that the universities’ repositories are open access, with the explicit intention of the resources being available for use by other researchers (see, for example,
6 Apart from providing triangulation for the quality of the sample, the examiner reports were not used in the
http://open.uct.ac.za/about.html), the University of Sydney Human Ethics Committee (Protocol no: 2017/422) deemed the research to not require ethical approval, as it was not classified as human research.
Further ethical clearance was sought at a later point in the study to use examiner reports. This was granted by the University of Sydney Human Ethics Committee (Protocol no: 2017/897). The ethics application process included the approval of a ‘participant information sheet’ outlining the research aims and an explanation of how the examiner reports would be used in the study, as well as a ‘participant consent form’ which candidates were asked to sign and return to provide permission for the reports to be included in the study.