Old people often complain that the world is going to hell in a hand-basket, that standards are falling, and it used to be better in our day. Having examined over 40 doctoral students over the last 45 years, often as the external examiner, it would be easy for me to fall into that trap. On the contrary, though, I am impressed with the quality of theses I have been examining recently, partly because of the quality of the students, partly because of the quality of the supervision, and partly because online learning and educational technology in general have matured as a field of study.
However, one advantage of being old is that you begin to see patterns or themes that either come round every 10 years or so or never go away, and that certainly applies to Ph.D. theses in this field. So I thought I might offer some advice to students as to what examiners tend to look for in theses in this field, although technically it should be the supervisors doing this, not me.
Who’s being examined: student or supervisor?
When I have failed a student (which is rare but has happened) it’s ALWAYS been because the standard of supervision was so poor that the student never stood a chance. Somewhat more frequently (although still fairly uncommon), the examiners’ recommendation was pass with substantial revision, or ‘adequate’ in some European countries. Both these classifications carry a significant message to the academic department that the supervisor(s) weren’t doing their job properly. (Although to be fair, in at least one case the thesis was submitted almost in desperation by the department, because the student had exhausted all his many different supervisors, and was running out of the very generous time allowed to submit.)
So the good news, students, is that, despite what might appear to be the opposite, by the time it comes to submitting your thesis for exam, the university is (or should be) 100 per cent behind you in wanting to get you through. (In recent years, this pressure from the university on examiners to pass students sometimes appears to be almost desperate, because a successful Ph.D. may carry a very significant weight towards the performance indicators for the university.)
Criteria for success
So at the risk of over-simplification, here is my advice for students, in particular, on what I, as an examiner, tend to look for in a thesis, starting with the most important. My comments apply mainly, but not exclusively, to traditional, research-based theses.
I have three main criteria which MUST be met for a pass:
- is it original?
- does it demonstrate that the student is capable of conducting independent research?
- does the evidence support the conclusions drawn in the thesis?
The minimum a doctoral thesis must do is tell me something that was not already known in the field. Now this can still be what students often see as a negative outcome: their main hypothesis is found to be false. That’s fine, if it is a commonly held hypothesis in the field. (Example: digital natives are different from digital immigrants: no evidence was found for this in the study.) If it disproves or questions current wisdom, that’s good, even if the result was not what you were expecting. In fact, that’s really good, because the ‘null hypothesis’ – I’m trying to prove my hypothesis is false - is a more rigorous test than trying to find evidence to support something you actually thought to be true before you started the research (see Karl Popper (1934) on this).
Competence in research
For students, there are three good reasons for doing a Ph.D.:
- because you want an academic position in a university or college
- because you want to work as a full-time researcher outside the university
- because you have a burning question to answer (e,.g.: what’s best done face-to-face, and what online, when teaching quantum physics?)
However, the main purpose of a Ph.D. (as distinct from other post-graduate qualifications) from a professional or institutional perspective is to enable students to conduct independent research. Thus the thesis must demonstrate this competency. In a sense, it is a trust issue: if this person does research, we should be able to trust him or her to do it within the norms and values of the subject discipline. (This is why it is stupid to even think of cheating by falsifying data or plagiarism: if found out, you will never get an academic job in a university, never mind the Ph.D.)
My emphasis here is on ensuring that appropriate conclusions are drawn from whatever evidence is used (which should include the literature review as well as the actual data collected). If for instance the results are contrary to what might be expected from the literature review, some explanation or discussion is needed about why there is this difference. It may have to be speculative, but such contradictions need to be addressed and not ignored.
Normally (although there will be exceptions) a good thesis will also meet the following criteria:
- there is a clear narrative and structure to the thesis
- there is a clear data audit trail, and all the raw/original data is accessible to examiners and the general public, subject to normal privacy/ethical requirements
- the results must be meaningfully significant
Narrative and structure
Even in an applied thesis, this is a necessary component of a good thesis. The reader must be able to follow the plot – and the plot must be clear. The usual structure for a thesis in our field is:
- identification of an issue or problem
- review of relevant previous research/studies
- identification of a research question or set of questions
- conclusions and discussion.
However, other structures are possible. In an applied degree, the structure will or should be different, but even so, the reader in the main body of thesis should be able to follow clearly the rationale for the study, how it was conducted, the results, and the conclusions.
Most – but not all – theses in the educational technology field have an empirical component. Data is collected, analysed and interpreted. All these steps have to be competently conducted, whether the data is mainly quantitative, qualitative or both. This usually means ensuring that there is a clear trail linking raw data through analysis into conclusions that can be followed and checked easily by a diligent reader (in this case, the examiners). This is especially important with qualitative data, because it is easy to cherry-pick comments that support your prior prejudices or assumptions while ignoring those that don’t fit. As an examiner, I do want access to raw data, even if it’s in an appendix or an online database.
However, I am also willing to accept a thesis that is pure argument. Nevertheless, this is a very risky option because this means offering something that is quite original and which can be adequately defended against the whole collective wisdom of the field. In the field of educational technology, it is hard to see how this can be done without resorting to some form of empirical evidence – but perhaps not impossible.
Significance of the research question and results
This is often the best test of how much the thesis is mainly the work of the supervisor and how much the student. A good supervisor can more or less frogmarch a student through the various procedural steps in doing a doctoral thesis, but what the supervisor cannot – or should not – provide is the original spark of a good research question, and the ability to see the significance of the study for the field as a whole. This is why orals are so important – this is the place to say why your study matters, but it also helps if you address this at the beginning and end of your written thesis as well.
Too often I have seen students who have asked questions that inevitably produce results that are trivial, already known, or are completely off-base. Even more tragic is when the student has an unexpected but important, well-founded set of data, but is unable to see the significance of the data for the field in general.
The problem is that supervisors quite rightly drill it into students that they must chose a research question that is manageable by an individual working mainly alone, and that their conclusions must be based on the data collected, but this does not mean that the research question needs to be trivial or that once the conclusions have been properly drawn, there should be no further discussion of their significance for the field as a whole. This is the real test of a student’s academic ability.
Tips for success
There are thousands of possible tips one could give to help Ph.D. students, but I will focus on just a few issues that seem to come up a lot in theses in this area:
1. Do a masters degree on online learning first
This will give you a good overview of the issues involved in online learning and should provide some essentially preparatory skills, such as an introduction to research methods and extensive writing.
Do this prior to starting a Ph.D. See: Recommended graduate programs in e-learning for a list of appropriate programs.
Do it online if possible so you know what its’s like to be an online student.
At a minimum, take a course on research methods in the social sciences/online learning.
2. Get a good supervisor
The trick is to find a supervisor willing to accept your proposed area of research. Try to find someone in the local Faculty of Education with an interest in online learning and try to negotiate a research topic of mutual interest. This is really the hardest and most important part. Getting the right supervisor is absolutely essential. However, there are many more potential students than education faculty interested in research in online learning.
If you find a willing and sympathetic local faculty member with an interest in online learning but worried they don’t have the right expertise to supervise your particular interest, suggest a committee with an external supervisor (anywhere in the world) who really has the expertise and who may be willing to share the supervision with your local supervisor. Again, though, your chances of getting either an internal or external supervisor is much higher if that person already knows you or is aware of your work. Doing an online masters might help here, since some of the instructors on the course may be interested in supervising you for a Ph.D., especially if they know your work through the masters. But again, good professors with expertise in online learning are already likely to have a full supervision load, so it is not easy. (And don’t ask me – I’m retired!)
This means that even before applying for a Ph.D., you need to do some homework. Identify a topic with some degree of flexibility, have in mind an internal and an external supervisor, and show that you have done the necessary courses such as research methods, educational theory, etc., that will prepare you for a Ph.D. (or are willing to do them first).
3. Develop a good research question
See above. Ideally, it should meet the following requirements:
a. The research is likely to add something new to our knowledge in the field
b. The results of the research (positive, negative or descriptive) are likely to be significant/important for instructors, students or an institution
c. You can do the research to answer the question on your own, within a year or so of starting to collect data.
d. It can be done within the ethical requirements of research
It is even better if you can collect data as part of your everyday work, for example by researching your own online teaching.
4. Get a good understanding of sampling and the level of statistics that your study requires
Even if you are doing a qualitative study, you really need to understand sampling – choosing subjects to participate in the study. The two issues you need to watch out for are:
1. Bias in the initial choice of subjects, especially choosing subjects that are likely to support any hypotheses or assumptions you may already have. (Hence the danger of researching your own teaching – but you can turn this to advantage by taking care to identify your prior assumptions in advance and being careful not to be unduly influenced by them in the design of the research).
2. Focusing too much on the number of respondents and not on the response rate, especially in quantitative studies. Most studies with response rates of 40 per cent or less are usually worthless, because the responders are unlikely to be representative of the the whole group (which is why student evaluation data is really dangerous, as the response rate is usually biased towards successful students, who are more likely to complete the questionnaires than unsuccessful students.) When choosing a sample, try to find independent data that can help you identify the extent of the likely bias due to non-responders. For instance, if looking at digital natives, check the age distribution of your responders with the age distribution of the total of the group from which you drew the sample, if that is available. If you had a cohort of 100 students, and 20 responded, how does the average age of the responders compare with the average age of the whole 200? If the average age of responders is much lower than non-responders, what significance does this have for your study?
Understanding statistics is a whole other matter. If you intend to do anything more complicated quantitatively than adding up quantitative data, make sure you understand the necessary statistics, especially what statistically different means. For instance, if you have a very large sample, even small differences are likely to be statistically significant, but they may not be meaningfully significant. Small samples increase the difficulty of getting statistically significant results, so drawing conclusions even when differences look large can be very dangerous from small samples.
5. Avoid tautological research design or quantitative designs with no independent variables
Basically, this means asking a question, stating a hypothesis, or designing research in such a way that the question or hypothesis itself provides the answer. To elaborate, research question” “What is quality in online learning?’ ‘Answer: “It is defined by what educators say makes for quality in online courses and my research shows that these are clear learning objectives, accessibility, learner engagement, etc..” There is no independent variable here to validate the statements made by educators. (An independent variable might be exam results, participation rates of disabled people, etc.). Education is full of such self-justifications that have no clear, independent variables against which such statements have been tested. Merely re-iterating what people currently think is not original research.
For this reason, I am very skeptical of Delphi studies, which merely re-iterate already established views and opinions. I always ask: ‘Would a thorough literature review have provided the same results?’ The answer is usually: ‘No, you get a far more comprehensive and reliable overview of the topic from the literature review.’
6. Write well
Easily said, but not easily done. However, writing that is clear, well-structured, evidence-based, grammatically correct and well argued makes a huge difference when it comes to the examination of the thesis. I have seen really weak research studies get through from the sheer quality of the writing. I have seen other really good research studies sent back for major revision because they were so badly written.
Writing is a skill, so it gets better with practice. This usually means writing the same chapter several times until you get it right. Write the first draft, put it away and come back to it several days later. Re-read it and then clarify or improve what you’ve written. Do it again, and again, until you are satisfied that someone who knows nothing about the subject beforehand can understand it. (Don’t assume that all the examiners will be expert in your particular topic.) If you can, get someone such as a spouse who knows nothing about the subject to read through a chapter and ask them just to put question marks alongside sentences or paragraphs they don’t understand. Then re-write them until they do.
The more practice and feedback you can get on your writing, the better, and this is best done long before you get to a final draft.
Is the Ph.D. process broken?
A general comment about the whole Ph.D. process: while not completely broken, it is probably the most costly and inefficient academic process in the whole university, riddled with bureaucracy, lack of clarity for students, and certainly in the non-quantitative areas, open to all kinds of challenges regarding the process and standards.
This is further complicated by a move in recent years to applied rather than research theses. In an applied thesis, the aim is to come up with something useful that can be applied in the field, such as the design of an e-portfolio template that can be used for an end of course assessment, rather than the traditional research thesis. I believe this to be a step in the right direction. Unfortunately though education departments often struggle to provide clear guidance to both students and examiners about the criteria for assessing such new degrees, which makes it even more of a shot in the dark in deciding whether a thesis is ready for submission.
Other suggestions or criticisms
These are (as usual) very personal comments. I’m sure students would like to hear from other examiners in this field, particularly if there is disagreement with my criteria and advice. And I’d like to hear from doctoral students themselves. Suggestions for further readings on the Ph.D. process would also be welcome.
I would also like to hear from those who question the whole Ph.D. process. I must admit to mixed feelings. We do need to develop good quality researchers in the field, and I think a research thesis is one way of doing this. I do feel though that the whole process could be made more efficient than it is at the moment.
In the meantime, good luck to all of you who are struggling with your doctoral studies in this field – we need you to succeed!
Popper, K. (1959) The Logic of Scientific Discovery London: Routlege