PhD thesis quality: The views of examiners Keynote address: Sid Bourke, University of Newcastle. Australia 1 Stellenbosch University, South Africa Abstract The PhD is clearly the pinnacle of formal education qualifications internationally, representing excellence and attracting both resources and prestige to universities. The expectation is that each PhD project will provide an original and significant research outcome, presented in the form of a thesis. In Australia, normally the only assessment for the award of a PhD is through the medium of written reports on the thesis by either two or three external examiners. The examiners effectively set PhD standards. The major focus of our work has been on the content of the examiner reports in providing indicators of thesis quality, and the bases for their recommendations on the fate of the thesis. These reports were also linked with candidature and examiner information to provide background information thought to be relevant to the thesis examination process. The study discussed here adds considerable empirical information to the literature in the important and neglected area of doctoral assessment. On the basis of 2121 examiner reports for 804 theses across all discipline areas at eight Australian universities, discipline differences, examiner consistency, and thesis quality are discussed. The incidence and proportion of examiner comments devoted to (1) themselves and the context of their examination of the thesis, (2) to assessable areas covered, (3) dialogic elements of the reports, and (4) to the evaluative elements of their reports are described. Information about what examiners comment on in their reports, and therefore presumably value, should be of considerable use to candidates and their supervisors in assisting them to meet examiner expectations of theses more effectively. In general, the nature of examiner comments do not differ by discipline, although some candidature characteristics do differ in this way, eg, nature of enrolment and time taken to completion. Examiners of the same thesis are usually quite consistent in the nature of the comments they make, although they more often diverge in their final recommendations. It would seem that the same issues in a thesis are identified by examiners, although the extent of change they recommend or demand may vary rather more. Because examiner recommendations are intended for another specific purpose, there are problems in identifying thesis quality from the recommendations, except in distinguishing between the best and the marginal theses. However, more fine-grained distinctions between theses would seem to be possible from a detailed examination of the written comments. Our work is proceeding along these lines. 1 This paper is based on results of an Australian Research Council Discovery Grant awarded to Allyson Holbrook, Sid Bourke and Terence Lovat, and subsequent work undertaken by Sid Bourke. INTRODUCTORY REMARKS Despite differences in the nature of PhD programs and examination internationally, the doctorate is clearly recognised as the pinnacle of our formal educational systems and has been referred to as the ‘gold standard’ (Scott et al, 2004, p.149). And the quality of the PhD remained largely unquestioned until recently when there has been a large increase in the numbers undertaking doctorates in what has been termed the ‘massification’ of doctoral qualifications. As there is no PhD curriculum as such, it is the thesis examiners who set the standard of what is acceptable as a thesis (or dissertation), and consequently set the standard for the award of a PhD. To a large extent the examiners also influence what style of reporting may be used and what should be included. What examiners do in examining is crucial. A few years back it could be said, with authority, that little research had been undertaken into the written assessments of PhD theses (Morley, Leonard & David, 2002; Mullins & Kiley, 2002; Tinkler & Jackson, 2000). In the UK, the major focus in doctoral assessment has been on the oral examination or viva (see, for example, the work by Trafford, 2003 and by Tinkler & Jackson, 2004). With respect to assessment of the written thesis itself, there is a recent and developing literature on examination criteria and process based on interviews with either experienced examiners (Mullins & Kiley, 2002), inexperienced examiners (Kiley & Mullins, 2004) or on questionnaires completed by examiners (Winter et al, 2000). Interest in the USA concerning dissertation quality (Lovitts, 2007) and doctoral assessment (Maki & Borkowski, 2006) is even more recent. But there were a few prior studies of thesis assessment based on examiner reports (Ballard, 1996; Hansford & Maxwell, 1993; Johnston, 1997; Nightingale, 1984; Pitkethly & Prosser, 1995). In the main these were single-institution studies. Only Hansford & Jackson (1993), which examined reports on Masters level theses, identified the literature review and conceptual framework as substantial themes. In this study 34 per cent of examiners made critical comment on the literature review. The two ‘most common criticisms’ were the ‘failure to use recent literature and the inability to critically assess the existing literature’ (p.179). The study drew on 125 students in Education in one institution over a four year period. The authors noted that if their study had ‘endeavoured to identify the nature of examiner comment’ relating to high quality theses, their ‘belief’ was that a number of such comments would refer to aspects of ‘theoretical or conceptual frameworks’ (p.181). Despite these few studies of various aspects of thesis examination, there remains a dearth of direct information about what examiners value as indicators of thesis quality. The most direct source of such information is the content of the written examiner report itself – these reports can be analysed in detail to determine how important each component is in the assessment regime. The study reported below involved direct and detailed examination of examiners’ reports on PhD theses. OUR STUDY A total of 2121 examiner reports on a total of 804 PhD theses across all discipline areas (Broad Fields of Study) at eight Australian universities (with varying research intensiveness) were collected. The associated examiner recommendation on the thesis and information about the candidates, their candidature and their supervision were also collected. The content of the reports were coded across specific categories (described below), and the report length was also recorded. I will return to the examiner reports shortly, but first I will describe the samples of candidates and examiners and relate these characteristics to the examiner recommendations made. The distributions of candidates and examiners across disciplines are shown in Table 1. INSERT TABLE 1 ABOUT HERE There were initially 10 Broad Fields of Study (BFOS) recorded. However, as the numbers of theses in some areas, notably Architecture, Law and Veterinary Science, were small, theses in these areas were included with the fields with which they had the most affinity, Built Environment, Business and Science respectively. It will be noted that the broad field of Science had the largest number of theses included, followed by the Arts, Humanities and Social Sciences grouping. The majority of Australian universities use three external examiners for PhD theses, others use two. In the main, reports are provided to the university independently by each examiner, without prior consultation between examiners. This is a strength of the thesis examination regime – the university has multiple and independent assessments available to it when making a final decision on the fate of the thesis. Each examiner recommendation on the thesis and the University decision were recorded for this study. In most cases the decision is made by a university-wide committee, in other cases where there is little difference between examiner recommendations, the decision may be made by the Dean of Graduate Studies (or equivalent). Other information collected included candidate individual characteristics and some information on the candidate’s supervisors, the pattern of supervision and the examination panel. Although there are some minor variations across Australian universities, essentially the range of recommendations available to examiners and decisions of the university committee can be consistently aggregated to the five categories shown below: • • • • • Accept the thesis as submitted Accept the thesis following minor (or invited) corrections Accept the thesis subject to more major (required) corrections Require the thesis to be revised and resubmitted for further examination Fail the thesis without the opportunity for revision The distribution of examiner recommendations across these five categories is shown in Figure 1 by BFOS. Overall, the proportions of recommendations are: Accept (31%), Invite (minor) correction (40%), Require (more major) correction (22%), Revise and resubmit (6%), and Fail (1%). INSERT FIGURES 1 AND 2 ABOUT HERE The distribution of university decisions by BFOS on the same five categories is shown in Figure 2. The proportions are: Accept (13%), Invite (minor) correction (47%), Require (more major) correction (38%), Revise and resubmit (3%), and Fail (one thesis only). In considering the very high ‘pass’ rate for PhD theses it should be noted that the vast majority of theses has already been scrutinized by at least one supervisor and declared to be ready for examination before being actually submitted for external examination. It will be noted that the distribution for the university decision differs in two ways from that for the individual examiners. First, differences between BFOS are reduced and secondly, there is a shift from the extremes to two middle categories Invite or require (minor) change (from 40% by the examiners to 47% by the committee), and Require (more major) change (from 22 to 38%). THESIS QUALITY It has been argued elsewhere (Bourke, Holbrook & Lovat, 2005) that examiner recommendations on a thesis, although in the main serving a different purpose, provide at least a general guide to thesis quality. The consistency of recommendations is particularly marked for the best and the worst theses. Using examiner recommendation as a measure of thesis quality, thesis quality was not related to any of the following: • • • • • • • • BFOS. Despite the differences observed in Figure 1, the distributions of recommendation by discipline area are not statistically significant. Research intensiveness of the university attended was not related to thesis quality. The universities were grouped into three levels of research intensiveness: high (consisting of the traditional ‘sandstone universities’), medium (consisting of newer universities with a noted research focus), and low (consisting of new generation universities still in the process of developing a comprehensive research record). Gender. There were no differences in examiner recommendation related to any of candidate, supervisor or examiner gender. It should be noted that almost half (47%) of the candidates were female. Whether the candidate was a native English speaker (78% of all candidates were native English speakers). Qualifications on entry to candidature. The most common qualifications for PhD entry were having a bachelor degree with honours (46%), and a master level qualification, either a coursework degree (27%) or a research degree (17%). Nature of enrolment. Whether candidates had been enrolled full-time throughout their candidature (53%), enrolled part-time throughout (15%) or had a mix of full and part-time enrolment (32%) was not related to examiner recommendation on the thesis. Leave taken. Approximately 28% of candidates took leave during candidature. Supervision. None of the following was related to examiner recommendation – number of supervisors (which ranged from one (21%) to four (4%) with 49% having two supervisors), supervisor experience, having had a change in supervision arrangements during candidature (30% did have such a change). By contrast the following were related to examiner recommendation on the thesis: • Candidate age. The relationship here was complex – on average, candidates aged between 25 and 30 years on entry received the best recommendations and candidates aged more than 45 years received the worst. • • • Whether a scholarship was held during candidature. Candidates who had held a scholarship (59%) received more favourable recommendations. A problem reported during candidature. On average, these candidates (34%) received less favourable recommendations. Examiner location. Almost half the examiners of these theses were located outside Australia (47%), being scattered over the rest of the world with the next largest groups being located in the USA (19%) and the UK (9%). The Australian examiners (and the small proportion of New Zealand examiners, ie, 3%) made less favourable recommendations on the theses than examiners in all other countries and regions. EXAMINER REPORT TEXT The complete texts of the examiner reports on the PhD theses submitted by the 804 candidates at the eight universities were standardised to a given line length before being coded, with lines as coding units, in a framework developed across five parent-level categories using QSR N6 software. Multiple coding of a line (or text unit) was possible and, in fact, it was common. For example, a positive judgement might be made (and coded) about the literature coverage (also coded) in the thesis to give two individual codes for each line of the judgement. In all there were 36 sub-categories identified in the coding of report content, 30 of which were suitable for possible inclusion in the text constructs to be developed. Categories labelled ‘Other’ were considered too general and were excluded from any further consideration in the present context. The correlation coefficients were calculated for each sub-category between the proportion of text and the examiner recommendation made. The data for each text code and report length are shown in Table 2. INSERT TABLE 2 ABOUT HERE The first category, Examiner and process, is not directly related to the thesis assessment but records occurrences where the examiners are writing about themselves, their capacity to examine the thesis, and about what they consider relevant to the process of examining. Although each of the sub-categories in this area constitutes a small proportion of the overall report, two of the sub-categories are represented in at least half the reports. All sub-categories are significantly related to the examiner recommendation, mostly positively. However, where examiners write more about their application of the anticipated criteria for a PhD thesis, they tend to have a less favourable view of the thesis, in some cases expressing disappointment that the criteria were not well met. The second category, Assessable areas covered, clearly does relate to the thesis assessments made by focussing on comment on the different areas – the scope, significance of the thesis, the literature review, the approach taken and methods used, and issues of communication. By far the most comment was made on the analyses and reporting of findings which appeared in 95% of the reports and, on average, constituted 37% of reports. This sub-category was significantly negatively related to the examiner recommendation, indicating that such comment tended to be a criticism of what was done. The next most common area commented on was the approach taken in the thesis. However, this sub-category was not significantly related to examiner recommendation, because there was as much positive as negative comment on this aspect. The third most common area was the significance and contribution of the thesis, and this sub-category had the strongest relationship with examiner recommendation. The relationship was positive indicating that there was much more positive than negative comment for this sub-category. Two of the sub-categories concerning the literature review (breadth/depth/recency and use/application) require further explication, and these have been investigated in detail by Holbrook et al (2007). The Dialogic elements category does not directly relate to assessments made of the thesis, but is an indicator of tone in the examination report. It relates to the extent to which the examiner becomes personally involved in the thesis and its examination. The most common sub-category is use of the first person by the examiner, and this is positively related to examiner recommendation. When a thesis is found wanting, examiners are more likely to stand back from it and comment in the third person. Similarly, although this is not as prevalent, examiners tend to engage intellectually with what they see as better theses. The Evaluative elements category identifies all those instances where examiners are making some form of evaluative statement about the thesis, whether positive, negative or indeterminate (called neutral here). One of the most prevalent comments is a summative positive assessment of the thesis as a whole or a significant part of the thesis. As would be expected, this sub-category has a very strong relationship with examiner recommendation. The next most extensive comment in terms of the proportion of text is formative instruction, coded when the examiner was attempting to teach the candidate something in order to improve the thesis or the candidate’s subsequent research or publication. As such, the use of formative instruction indicates that something could or should have been done better and, as a consequence, this subcategory is negatively related to the examiner recommendation. The sub-category prescription was found in most examiner reports and also accounted for more than 10% of report text. Prescription was coded when an examiner did not attempt to teach the candidate, but simply told the candidate what needed ‘fixing’ and how it should be ‘fixed’. In some cases prescriptive comment was related to minor shortcomings, but not always. Overall, prescription was negative correlated with examiner recommendation. The final row of Table 2 indicates that, on average, examiners wrote reports of 132 lines (the range was from one line to 1272 lines) and longer reports were associated with theses that received less favourable examiner recommendations. Clearly it takes more text to point out and explain why something was in error than simply to state something was well done. In any case, examiners would feel a need to justify negative comment more fully than positive comment. PRACTICAL MEASURES OF THESIS QUALITY We now need to make a leap in this presentation from simple percentages of subcategories and correlations with examiner recommendations to further analyses undertaken involving the intersection of assessment areas with evaluative elements, the latter focussing on clearly positive and negative evaluations. As described by Bourke et al (2005), there was a total of ten such cross-classifications of assessment areas and evaluations that could be distinguished as being strongly related to examiner recommendations, particularly for theses at the extremes. These cross-classifications are referred to henceforth as indicators of thesis quality. The ten indicators could be grouped into four main assessment areas – namely contribution, literature, methodology and presentation, as shown in Figure 3. This figure also includes examples of typical text for each indicator found in examiner comments on PhD theses at two disparate levels – those of notably higher and lower quality. INSERT FIGURE 3 ABOUT HERE Thesis contribution was assessed by examiners with reference to originality, making a substantive contribution and impact on either the discipline or practice or both. When broken down into positive and negative comments, all of these indicators distinguished between theses of higher and lower quality. For the literature review, positive examiner comment on literature use and application was the most closely linked with higher quality theses while comment on inaccuracy in the review was most closely linked with lower quality theses. Literature coverage, perhaps because it was simply expected, had weaker links with quality, and requires further explication (see Holbrook et al, 2007). With respect to methodology, positive examiner comments on both appropriateness and application were particularly strongly related to theses being identified as high quality. Finally, examiner comment on the substantive issue of communication and logic in presentation was related to thesis quality, for both higher and lower quality theses. The issue of grammar and expression basically was raised by examiners only when there were problems in this area, and so it was strongly related to theses identified as being of lower quality. Our work to date suggests that the 10 indicators of thesis quality in the four areas mentioned above, and illustrated in Figure 3, would be useful in any endeavour to distinguish reliably between PhD theses of different quality. However, it should be noted that these indicators were based on their effectiveness in making distinctions between theses at the extremes of high and low (although passable) quality. The ability of these indicators to make distinctions between theses across the spectrum of quality is yet to be established. IN CONCLUSION This is where the story ends to date. Next we intend to trial the ten indicators described by asking examiners to rate a thesis they recently examined on each of the ten indicators as well as an overall assessment of thesis quality. We anticipate that the indicator list may require further selection, refinement, addition and subtraction to enable us to more closely differentiate between all theses, not only those at the extremes. Once the final list of indicators was agreed, they could become a standard component of the thesis examination process. The centrality of examiners to PhD thesis assessment is unquestioned. Some of our earlier work investigated the reasons given for the selection of 300 PhD thesis examiners across all discipline areas. Not surprisingly, this study found that examiners were chosen for multiple reasons, the most common criteria being in descending order, expertise (77%), reputation (59%), publications (40%) and experience (36%) (see Bourke, Scevak & Cantwell, 2001). Perusal of other university guidelines for examiner selection suggests these criteria are in general use. Finally, if such an assessment system were adopted for PhD thesis examination, it would be a practicable one. That is, it would be possible to implement as simply a refinement of the current examination system used in Australia, but we would have a justifiable method of distinguishing more reliably between theses of different quality. As the quality indicators that examiners were asked to respond to would have been developed from what examiners wrote in their reports, the assessment would still be based on examiner judgement. The examiners would remain the arbiters of PhD thesis quality and, given the criteria and international basis on which examiners of Australian PhD theses were selected, I maintain that is as it should be. REFERENCES Ballard (1996). Contexts of judgement: an analysis of some assumptions identified in examiners’ reports on 62 successful PhD theses. Paper presented to the Conference on Quality in Postgraduate Research, Adelaide. Bourke, S., Scevak, J. & Cantwell, R. (2001). PhD examination and examiner characteristics. Paper presented at the AARE Annual Conference, Fremantle, December. Available at http://www.aare.edu.au/01pap/bou01588.htm Bourke, S., Holbrook, A., & Lovat, T. (2005). Using examiner reports to identify quality in PhD theses. Paper presented at the AARE Special National Focus Conference: Quality in Educational Research, July 4 -5, 2005, Cairns, Published at: http://www.aare.edu.au/05papc/bo05011y.pdf Hansford, B.C. & Maxwell, T.W. (1993). A masters degree program: Structural components and examiners’ comments. Higher Education Research and Development, 12, 2, 171-187. Holbrook, A., Bourke, S., Fairbairn, H. & Lovat, T. (2007). Examiner comment on the literature review in PhD theses. Studies in Higher Education, 32, 3. 337-356. Holbrook, A., Bourke, S., Lovat, T., & Dally, K. (2004). Investigating PhD thesis examination reports. International Journal of Educational Research, 41, 2, 98-120. Johnston, S. (1997). Examining the examiners: an analysis of examiners' report on doctoral theses. Studies in Higher Education, 22, 3, 333-347. Kiley, M. & Mullins, G. (2004). Examining the examiners: How inexperienced examiners approach the assessment of research theses. International Journal of Educational Research, 41, 2, 121-135. Morley, L., Leonard, D. & David, M. (2002). Variations in Vivas: quality and equality in British PhD assessments. Studies in Higher Education, 27, 3, 263-272. Mullins, G. & Kiley, M. (2002). 'It's a PhD, not a Nobel Prize': how experienced examiners assess research theses. Studies in Higher Education, 27, 4, 369-386. Nightingale, P. (1984). Examination of research theses. Higher Education Research and Development, 3, 2, 137-150. Pitkethly, A. & Prosser, M. (1995). Examiners’ comments on the international context of PhD theses. In C. McNaught & K. Beattie (Eds), Research into Higher Education: Dilemmas, Directions and Diversion. (pp. 129-136). Melbourne: HERDSA. Scott, D., Brown, A., Lunt, I. & Thorne, L. (2004). Professional Doctorates: Integrating Professional and Academic Knowledge, Berkshire England: Society for Research into Higher Education and 0pen University Press. Tinkler, P. & Jackson, C. (2000). Examining the doctorate: institutional policy and the Ph.D. examination process in Britain. Studies in Higher Education, 25, 2, 67-180. Tinkler, P. & Jackson, C. (2004). The Doctoral Examination Process: A Handbook for Students, Examiners and Supervisors. Berkshire: Society for Research into Higher Education and 0pen University Press. Trafford, V. (2003), Questions in doctoral vivas: views from the Inside. Quality Assurance in Education, 11, 2, 113-21. Winter, R., Griffiths, M. & Green, K. (2000). The 'academic' qualities of practice: what are the criteria for a practice-based Ph. D.? Studies in Higher Education, 25, 1, 25-37. TABLE 1: The samples of candidates and examiners By discipline across 8 universities DISCIPLINE AREA Agriculture No. of candidates No. of examiners 41 111 183 487 Business & Law 78 221 Education 84 241 Engineering & Built Environment 71 180 Health 131 320 Science, IT, Veterinary Science 216 561 OVERALL 804 2121 Arts, Humanities, Social Sciences TABLE 2: The distributions of examiner report text codes and correlation with examiner recommendation EXAMINER REPORT TEXT CODE INCLUDED IN TEXT % PROPORTION OF TEXT % CORREL. WITH RECOMMEND. Examiner and process 21: Personal & professional context 50 5 .174** 22: Application of anticipated criteria 61 5 -.053* 23: The model PhD 14 1 .120** 24: Mention of supervisor(s) 22 1 .052* Assessable areas covered 311: Scope of thesis 73 8 .113** 312: Significance and contribution of thesis 82 10 .314** 313: Publications arising 44 3 .243** 314: Existing publications 13 1 .083** 321: Lit. review: Breadth/depth/recency 70 6 -.051* 322: Literature review: Inaccuracies 30 2 -.125** 323: Literature review: Use/application 38 3 NS 33: Approach: method/design/execution 81 13 NS 341: Analysis/reporting of findings 95 37 -.109** 342: Topic related issues 28 5 NS 351: Communication: Substantive issues 79 6 NS 352: Communication: Editorial issues 57 8 -.121** Dialogic elements 41: Intellectual engagement 21 3 .064** 42: Conversation with the reader 30 3 NS 43: Use of first person 89 17 .119** Evaluative elements 511: Summative assessment: Positive 87 19 .463** 512: Summative assessment: Neutral 47 4 -.085** 513: summative assessment: Negative 33 2 -.368** 52: Formative instruction 63 21 -.369** 531: Other instruction: Commentary 64 10 NS 532: Other instruction: Prescription 84 11 -.361** 541: Other judgment: Positive 72 10 .368** 542: Other judgment: Neutral 30 2 .094** 543: Other judgment: Negative 32 2 -.198** . 132 -.382** Total report length (lines) * = correlation significant at the 0.05 level; ** = correlation significant at the 0.01 level; NS = correlation not statistically significant FIGURE 1. Examiner recommendation by Broad Field of Study BFOS Recoded Agriculture Arts, Hum, SS Business & Law Education Engin & Archit Health Science & VetSc 50.0% Percent 40.0% 30.0% 20.0% 10.0% 0.0% Accept Invite (minor) Require (major) Revise & resubmit Examiner Recommendation Fail Figure 2: University decision by Broad Field of Study BFOS Recoded 60.0% Agriculture Arts, Hum, SS Business & Law Education Engineering & Architecture Health Science & Vet. Science 50.0% Percent 40.0% 30.0% 20.0% 10.0% 0.0% Accept Minor change Major change (Invite) (Require) Revise & resubmit University Committee Decision Fail EXAMINER COMMENT HIGHER QUALITY LOWER QUALITY CONTRIBUTION 1. Originality … highly original and thought-provoking. It is hard to see anything new or original about this thesis. The reader is eventually left with the feeling of ‘so what?’ Although sound methodologically, there is little of substance in this thesis. … conclusions, potentially misleading to practitioners. … not likely to have practical outcomes 2. Substance … impressive scholarly contribution, rich in insight … … a major contribution to the discipline of … 3. Impact Provides a helpful heuristic for advancement of knowledge in the field. Practitioners will welcome the guidance LITERATURE 4. Coverage The literature review is most comprehensive and probably definitive. 5. Use & application There is clear evidence that the literature has been fully consulted and understood. … thoughtfully and carefully analysed the literature, and integrated it where appropriate 6. Accuracy … scholarship that is carefully and accurately summarised METHODOLOGY 7. Methodologically the thesis is impressive, Appropriateness evidencing insightfulness, originality and sensitivity. The methodology is well-justified, sound and clearly described. 8. Application … high level of perceptiveness and acumen in using the materials to draw out appropriate and interesting insights and conclusions … always an appropriate method, fully justified … PRESENTATION 9. Cogent and straightforward … easy to read … Communication occasionally eloquent & logic All the necessary steps in the argument supporting this essential point were made crystal clear 10. Grammar & Almost flawless expression and presentation expression Perhaps the most error-free thesis I have ever read. The now quite rich literature on the topic of [X] to enhance biological control is represented by only two articles. … fails to recognise useful and reject non-useful research … It is rare to find a reference in this thesis that has been critically appraised. … has not engaged adequately with a body of relevant knowledge This is not what the author concluded – in fact the opposite. … an extremely misleading interpretation which needs to be corrected The method chosen is quite inappropriate as a means of approaching the research question. … there clearly are better ways of addressing the issues … … offers no critical analysis of the information presented … understanding of the methodology used seems to be rudimentary, at best It was not at all clear what the aim of the study was … There was no logical flow in several places leading to poor communication of ideas. … so careless and error-ridden as to almost defy interpretation I gave up on correcting expression after the first 20 pages. FIGURE 3: Examples of typical examiner text by category for higher and lower quality theses.