Skip to main content Skip to navigation

Preparing your PhD thesis: Good, bad and ugly approaches

Advice from Robert Freedman given at the MOAC Annual Conference


Functions and readership of the thesis

  • It is a definitive record of your work for posterity
  • It is a report on what you have done and your understanding of it for the examiners
  • It is a data source for as-yet-unwritten papers for you and supervisors
  • It is a source of definitive methods for future PhD students
  • It is a source of inspiration, pride, nostalgia for various people
  • …so it is worth taking trouble to make it accurate and readable!

Examiners are the only guaranteed readers – what are they looking for?

  • Understanding and sophistication: The thesis should read as though it is written by a mature, judicious scientist
  • Clarity: Text, figures, sequence of topics, headings, format should make it simple to navigate and a pleasure to read
  • Coherence: The thesis should tell a story or argue a thesis
  • Quantity: Is the volume of new experimental or theoretical results or of method development indicative of sustained effort for 3–4 years? (Don’t pad, or show ‘failures’, but don’t expect examiners to intuit your challenges and difficulties).
  • Quality: Novelty, originality, significance, publishability
  • Context: Does the candidate understand how the thesis work relates to the immediate field and to wider scientific questions?

Is there only one right way to produce a good thesis?

  • No, there are several ways of doing it well..
  • .. and an infinite number of ways of doing it badly!
  • You should anticipate technical thesis production issues e.g….

    • Flexible bibliography for easy revision and updating

    • Be rigorous and consistent with text version control and backing-up!

    • Don’t be enslaved by your programs!!

Plan the structure of the thesis

  • A thesis is a bigger version of a paper with typically: Introduction, Methods, Results, Discussion and Conclusion
  • Decide what results will be shown, in what format (Figs, Tables, etc.) and in what sequence
  • Show some key primary data, not just derived parameters (but have Appendices for massive datasets)
  • Will you present results in one continuous Results section or several distinct Results chapters?
  • A summary of your contribution to co-authored papers could be included as an appendix: 1 page per paper.
  • Compile a ‘Table of Contents’ for Results chapter(s) and then for complete thesis, with titles for each section
  • Consult your supervisors before going further!!!
  • Subsequently, when asking for supervisor comment on drafts, append agreed Table of Contents

Manage your supervisors carefully to extract maximum benefit

  • Consult them on overall shape and structure
  • Submit draft sections in time for turn-around (2 weeks??) and response – discuss to what they can commit
  • Check on periods of absence or unavailability
  • With multiple supervisors, plan and establish roles – is one main consultant, the other receiving material ‘for information’?
  • When submitting drafts, append agreed Table of Contents, so supervisor can identify context

Present your Results and put flesh on the bones: proceed by...

  • Making draft versions of all Figs, Tables etc. and their legends
  • Draft text of Results to describe, analyse and link the data and tell the story
  • Find a tone that is not too chatty, but not so formal it constrains you
  • Do not adopt unfamiliar constructions and terms that you cannot handle confidently
  • Check consistency of voice (active/passive), tense (past/present) and person (first/third)

Give a clear account of Methods

  • Once you have draft Results, compile Materials/Methods
  • Material/methods should enable a successor to reproduce your work. Go back into results and cross-reference to methods.
  • Ensure that Methods text, Results text and Legends provide complete, consistent information with minimal repetition
  • Take care in citing ‘standard’ Methods papers rather than reporting your method in full: is your actual method a remote derivative of the published classic?
  • Any validation of bought-in materials, equipment or software should be reported in ‘Methods’…
  • ..whereas validation of materials, equipment or software developed by you should be reported as Results

Decide how you are going to handle Discussion

  • Discussion can be integrated with Results or in separate Discussion sections – usually chapter by chapter but the discussion could be in one distinct Discussion Chapter?
  • Discussion should address challenges (Are the results valid? What are their limitations?) but should focus on what the results mean and how they relate to current knowledge in the field
  • A sentence with 'could' at the beginning indicates that it is a hypothesis, so does not need further qualifications such as 'may' and 'possibily' etc.
  • Don’t apologise – state clearly what the results mean and why they are significant.
  • Do not over-claim. Demonstrate your judgement.


Write a conclusion section that summarizes what you have done and is pretty much stand-alone.

Tackle the Introduction last and with a strict timetable!

  • Introduction is not a comprehensive review of the field
  • It is a work of rhetoric that introduced YOUR work; anything you write is fair game for your examiners
  • The functions of the introduction are to persuade the reader a) that there is an important research question to be addressed, b) that you understand the question and the background and c) that you have identified a suitable research approach
  • The aim is to make the reader impatient to know what you have done and capable of understanding what it means
  • Writing the Results and Discussion first will clarify for you what it is that you must convey in the Introduction
  • Allow time for this, but do not make it a life’s work!

Advice from Hugo van den Berg on how to write well! This presentation was given by Hugo at the MOAC DTC annual conference in Arnside in 2011. It makes essential reading for all students who want to learn how to write well.