Notes on the Structure of a Dissertation

James Atherton

These notes originate from my efforts to help an old friend undertaking a taught Master's course, at a university which seemed to devise its regulations around the key principle that under no circumstances should, or could, a supervisor be helpful.

So my reaction to his first almost complete draft was to say that it needed to be completely re-organised; it was great content, but did not follow the conventions of a dissertation. Quite reasonably he asked how he was supposed to know that? This is a game he would only play once, whereas supervisors and assessors and external examiners are familiar with it.

He suggested,

"I should have found (useful) a model outline for the MA, like
  • Intro could include..........and should not......
  • Literary review could..... and should not..............
  • Further Chapters could.... and should not...........
  • Conclusion could ...and should not...... "

So this is my attempt to provide just that. I had just assumed that it already existed in practically every text on Research Methods, but on the basis of the sample I have consulted. it doesn't.

I've already modified it in response to some very useful suggestions from people on lists I belong to, and plan to continue to do so—so please write to me (address in the footer) and together we can improve it further. Thanks.

Note that these are general remarks, and that they are trumped by any specific rubrics from the awarding institution. As discussed here, they apply most clearly to dissertations within broadly social studies and to a lesser extent humanities, where, however, there may be more scope for variation.

University and/or faculty (in the sense of organisational unit, rather than academic staff) libraries will contain dissertations from previous years; even if the rules are quite explicit, it makes sense to go and have a look at half-a-dozen or so, to see what has been deemed acceptable in practice. (I was amazed, when doing this prior to submitting my M.Ed dissertation many years ago, to find a completely hand-written piece—beautifully done—but then it was about the teaching of hand-writing.)


This heading is my own portmanteau term—you will not find it anywhere in an actual dissertation.)

Regulations may specify what is to appear in the header and footer of each page, and other details of layout. Pagination, for example, may need to be in a standard form; roman numerals up to (and sometimes including) the Introduction, and standard (arabic) thereafter; or standard and continuous throughout.

This is the part of the dissertation which is probably most precisely prescribed by the university, which may, for example, have a set layout for the title page. There are several short components to these preliminaries, probably including;

Title page, including


The Abstract is a very short summary or digest of an article or dissertation whose basic task is to tell a potential reader, searching for scholarly or research-based material by topic or title, whether or not this is what she is looking for. Writing a good one is quite a craft and there is no substitute for reading lots of abstracts to develop the knack of summarising and selecting the key points.

Indeed, drafting and re-drafting the abstract is a very useful exercise for an author, as one has to be rigorous about priorities when there may be a word-limit as short as 200 words.

Do check the regulations; some universities which specify a word-limit for the abstract may refuse to accept the dissertation if the abstract is one word over. It makes sense to give a word count at the end of the abstract. Check too just where it needs to be located.


Generally speaking, the Introduction will set out;

  1. The aims and objectives of the research; how tightly these can be specified will vary from discipline to discipline but they should have been defined and articulated at the very beginning of work on the dissertation. I won't say more on this because this is not about how to do your dissertation but about how to write it up, but suffice it to say that aims and subsidiary aims or objectives are critical.
  1. The context of the work; the reader needs to get a handle on what this is about as soon as possible and the author’s reasons for engaging with the topic. There is no premium on a marker asking herself, “What is this person on about?” twenty pages in. ...and you might be interested in this link (via my blog) to an interesting reflective piece by Keith Lyons; it's more to do with approaching research rather than writing it up, but it's all part of the process
    • The context may be academic; “Building on Aardvark and Molestrangler’s seminal 2002 work on...” —in which case resist the temptation to go into detail because the place for that is in the Literature Review.
    • It may be historic; “The specification of underwater knitting as a core competence by the Institute of ... has led to...”
    • It may be narrative; “Based on the author’s experience of teaching English as an additional language in Korea, this study uses a grounded-theory approach to generate alternative ethnographic accounts of ...”
    • Regardless, it should be as simple and clear and practical as possible. (The context of the choice of research methods is considered separately under Methodology.) It may well lead into...
  2. A more detailed exegesis (OK! unpacking) of the title; which may include reasons for the choice of certain words, the reason for the part after the colon (which is generally to qualify and restrict the aspirations of the main title);
    • “Generating a theory of everything: necessity and sufficiency in explanatory accounts of the physical world by seven and eight year old children in an inner-city school.” Why that grandiose first part? What it meant by “necessity and sufficiency” in this context? In turn this may lead into...
  3. The more specific research hypotheses to be tested or questions to be answered. Each can be spelt out and then commented on for a paragraph or so. Do this with a view to re-visiting them in the conclusion. (In practice you may well be writing this after writing the conclusion, of course.) Tie these in to the aims and objectives—it may of course make more sense to re-arrange these items in order to make the links clearer. There is nothing sacrosanct about the order in which they are presented here.
  4. Exclusions: you have to get these in somewhere, and up-front is the best place;
    • “The scope of the study does not extend to a consideration of... because of lack of time/resources/space...” “
    • As discussed in the Literature Review below, most previous work in this area has concentrated on... On this occasion, however, attention is directed at...”
  5. The shape of the dissertation; outline, chapter by chapter, how the argument fits together, and mention the material which has been relegated to the appendices.
  6. Conventions adopted;
    • “Because of the nature of the action-research process, the convention of the author referring to herself in the third person makes for convoluted expression and hard reading. After consultation, I have decided to adopt a first-person narrative voice...” (“After consultation” is important—you are less likely to get hammered if your supervisor agreed to it.)
    • “For simplicity, and where it does not affect the sense, reported interviews refer to the interviewer and interviewee as of opposite sexes...”
    • "All transcripts of interviews have been translated into English; original language versions are available if required, but will be destroyed (as per the ethics policy) as soon as assessment formalities have been completed...”

Literature Review

This is where you outline previous work on the topic, and organise it so as to inform the empirical work which will come later, and its discussion. Planning the literature (or research) review calls for careful thought. It needs to be comprehensive, but obviously in areas which have already been well-researched, it is not going to be possible to include everything ever written about the topic. Indeed, as a marker I am inclined to distrust too many brief allusions to (sometimes obscure) studies; they may well suggest an over-reliance on secondary sources, and it is difficult to do justice to research and scholarship at Master’s level and beyond on that basis.

As (of course) Mark Twain is reputed to have said, “The researches of many commentators have already thrown much darkness on this subject, and it is probable that if they continue we shall soon know nothing at all about it.”

The obvious temptation is to set out the review in the order in which you came across the material; you have put a lot of work into reading a lot of stuff and so you are going to show how much work you have done by referring to all that stuff. That is not a good basis, unsurprisingly. It may make sense to you, but not to anyone coming across it afresh. Alternatives are:Up-date Feb 2015: There is an excellent discussion of how to tackle the ordering of material, with examples in Pinker (2014) ch. 5 “Arcs of Coherence”. And his chapter 2 on the “classic style” is essential reading for anyone embarking on a dissertation.

The last of the above reminds us that this chapter is the literature or research review. It is a warm-up for the main empirical show. Generally speaking, apart from the “where your contribution fits” points, and possibly “this particular finding is explicitly tested in the current research reported here...” this chapter is not about your work. But it needs to be structured so that when you do get to your own findings, you can readily refer back to the review;

If the marker is not intimately acquainted with what is by now after all, your specialist area, such signposting is invaluable.

Of course, not all the literature reviewed will be substantive research on the content of the topic. Some of it may involve outlining philosophical or even methodological principles which underpin what you are doing. It is a slightly moot point as to where that discussion should sit; my preference is generally to find it in this chapter (clearly demarcated, of course, and possibly preceding the accounts of substantive research so that you can refer back to the implications of the principles and methods in your evaluation of that research). Sometimes, however, where the points are quite specific and technical and not really up for debate, they can be dealt with comfortably within the Methodology chapter.

If you ever find yourself writing “Another writer says...”, you have lost the plot and descended into an amorphous list.

And you are not expected to provide a précis of every article you have read; as we will discuss under Methodology next, the important issues concern the influence of this previous work on the current study, so confine yourself to that.


It is now standard practice on the Research Methods module of many Master’s programmes to have two assignments, of which the second is the research proposal for the dissertation. This has the advantage of allowing the student to concentrate on the methods she is actually planning to use, and usually to get formative feedback before working on the dissertation itself. It also has an unintended consequence; very often those second assignments re-appear in minimally-edited form as the Methodology chapter of that dissertation. I can usually tell when the chapter starts with, “Research methods fall into two main groups, quantitative and qualitative. Quantitative methods are defined by Cohen, Manion and Morrison (2007) as...” followed by a gratuitous quotation. Strangely enough, I do know that! And I already know a lot of other stuff people trot out which is not worthy of getting within a mile of a dissertation.

It is not the task of the Methodology chapter to revisit social science methodology overall—it is its task to explain and justify the specific choices made for the methods used in this particular study.

One experienced supervisor and assessor commented in response to an earlier draft of this paper ; “A rule of thumb for me here is ‘Could someone else replicate this study based on the information here and in the appendices’?”

The principal—and all too frequent—failing of this chapter is that it is insufficiently specific, and not tied in to the Findings which follow. The chapter has to explain how you got the findings, why they can be trusted, and how they answer the research questions/test the hypotheses. The shape of the chapter is—in almost all cases—a zoom-in;

  1. The chosen research paradigm. I’m not going to go into that—this is a tactical rather than strategic guide—but it may be (there is no consistency in the labels) about:
    • Uncovering the “facts”
    • Making sense of them
    • Understanding what they mean to different actors
    • Locating them in a political or other context... etc...

    Which leads to...

  2. A consideration of the context and indeed constraints of the research. That may mean;
    • Its academic or professional nature (is it about creating knowledge for its own sake, or about suggesting how a problem might be tackled, or evaluating an initiative, for example?)
    • Its organisational context and how that might influence the methods available. Is it undertaken on behalf of management or a sponsor, and what say do they have in it? Is it subject to any additional ethics approval beyond the university’s internal procedures?
    • And the practicalities of access to people and resources.

    In the light of which you can get down to...

  3. The selection of a research strategy, such as (primarily);
  4. Use of published statistics

    Documentary / primary sources

    Commentary (docs + journalism etc.)



    Structured interviews

    Unstructured interviews

    Focus groups

    Records of participant observation

    Video or audio observation/tracking

    ...and any combination of the above.

    It needs to be justified as fit for purpose, and also in ethical terms. And so to;

  5. The research design. This is getting rather technical for a guide like this, so what follows is indicative for a survey-based project, because the protocol for that is the most formal:
    • Sample construction
    • Survey method; on-line, phone, personal contact, mail... etc.
    • Consent and ethical considerations
    • Addressing sampling/ non-response bias
    • Instrument construction:

        i. Addressing independent variables; questions on age and sex (and occupation, ethnicity) and how (and when) they are posed ...

        ii. Addressing dependent variables.

    • Choice of method;

      What questions did you ask and why and how would the answers, particularly cross-tabulated and significance tested, contribute to addressing the research questions/hypotheses? This is the crucial design question: it closes the loop, and you need to lay it out clearly. You do not need to go into detail for every question in the questionnaire or interview schedule, but the questions on the dependent variables need to show a clear line of accountability from the original research questions or hypotheses. How people can leave this out beats me, but many do, and inevitably suffer for it.

    • Arrangements for piloting and any changes made as a result.
  6. Of course in the real world the actual survey or round of interviews or video samples would have taken place weeks or months ago, and so you will probably be evaluating with hindsight rather than designing for the future. But it is best if you can start the writing and re-writing of this chapter while you are actually designing (no, what you did for Research Methods is not a substitute) as well as reviewing after the event; that way the chapter becomes a device for checking progress and structured evaluation.
  7. Finally, outline the methods used for processing the data. The actual results belong in the next chapter, but the choice of processing methods belongs here. Include inter al.;
    • Statistical techniques adopted, including significance testing—of course the instruments will have been designed with these in mind (won’t it?) so this might fit in earlier.
    • Processing methods for qualitative data such as interviews, use of particular packages for tagging and identifying themes. (Even non-standard and suggestive devices like using Wordle.) There are conventions and procedures; show how they have been adopted.
    • Ditto for audio- or video-recordings.
    • Concessions to practicality, such as selective transcription.
    • Use or non-use of independent judges for qualitative data. (People very rarely have the resources to use judges at this level, but better to admit that you didn’t, much as you would have liked to, rather than ignore the possibility. Markers do understand such things.)

    The structure will of course vary from discipline to discipline; one correspondent from medical science, for example, differentiates more than I have done between the methodological considerations—as theoretical isues to be taken into account in the research design—and the actual practical methods and materials used, as she says, "the recipe of the research", so that it can be replicated easily elsewhere. These need to be set out in a clear and distinct section at the end of the chapter.

There are some excellent points about different approaches to writing the Methods Chapter in Pat Thomson's blog.

Findings and Discussion

Whether you deal with these separately or together will depend very much on the kind of research you have undertaken, and so it is harder to be prescriptive about the structure of these chapter(s); what follow are mostly simply suggestions.

What is more than just a suggestion, however, is that you signpost clearly in an introductory paragraph just how you are going to present the material. Raw findings easily become fragmented and hard to follow, which is why you may need to discuss and evaluate them as you are going along rather than delay the findings until you have set out all your wares.

Findings themselves

On the whole, deal with the boring stuff first—but do deal with it. Did you include questions about independent variables such as age and sex etc., so that you could determine the fit between your sample and the overall population? It’s important but not exactly exciting; get it out of the way. We shouldn’t have to say this, but even in otherwise good dissertations the use and presentation of statistics is often abysmal. It is way beyond the brief of this paper to go into reasons and solutions, but frankly many non-scientists appear too frightened of maths to engage properly with statistics and failure to make appropriate use of quantitative techniques accounts for the poor quality even of some published research in the social “sciences”.

Whether or not charts and tables (or interview transcripts) are presented “in-line” as part of the narrative of the chapter, or separated into appendices is an important choice, but it does not have to be all-or-nothing; just make it clear at the start of the chapter what rule you are going to apply. The referencing of material in appendices is critical, and it is worth getting to grips with automatic cross-referencing in your word-processor, so that when you edit you do not lose track and point the reader to the wrong table or chart. And for the marker’s sake remember that continually flipping from main text to appendix and back can detract from concentration on the argument.

There are many ways of picking a route through the data;

And of course by this stage in the game you may have realised that the original research questions were misconceived (see below under Discussion).

Alternatively, you may simply work on a set of emergent issues which have become apparent as you scrutinised the data. That is fine, as long as you signpost the strategy.

As a rule of thumb, confine the information to what you yourself have found out, and leave the link with the Literature Review for the Discussion section.

Discussion itself

This is IT! This is the heart of the dissertation. This is where you tie together the research questions or hypotheses, the data you have unearthed, and the previous research and models and arguments. In a sense, anything goes in this chapter, except that if it is separate from the Findings, there should be no new information or data. It is all about the potential meaning(s) of data you have already reported, whether yours or that of previous researchers.

What must run through the chapter, though, is this continual knitting of the present material and the previous research. Moving away from the mundane aspects, this is a conversation between the present and the past, to adopt Oakeshott’s metaphor (1989) and both sides need to be heard.

You can now speak in your own voice, as it were (although don’t switch from third-person to first-person if you are using that convention). You have led the reader systematically through the research process, and this is where you can point out what it all adds up to: if you have done it properly, you are now expert and you have earned the right to be heard.

This is also where you should in some measure evaluate your efforts and their limitations. If the research design did not prove up to the task, say why and how that qualifies the results; if the survey suffered from a poor response rate, don’t try to cover it up, but discuss how this might have upset the sampling. The marker will already have noticed these limitations anyway, so there is no point in trying to conceal them, but you show your professionalism by the way you address them.

The most critical evaluation issue is the “If I were trying to get to there, I shouldn’t start from here” problem; you find you have been asking the wrong questions (or you have proved the null hypothesis). That is not necessarily a Bad Thing; you will not be randomly wrong, and the questions will have been formulated in relation to the previous research and scholarship, so the results can still be interesting and useful—just not quite in the way you first thought. But once again, do not try to fudge the issue: tackle it head on.

And of course this may be where the unexpected result may show up. That’s great! That is the climax!


Unless the requirements call for Recommendations as they may in the case of professionally-based programmes, the conclusion should be relatively brief and to the point, usually based on a re-visitation of the research questions (not in enormous detail) and a summary statement of what we can now say about the title. The loop is closed.

But the story does not end there. Where to from here? What questions are posed which are worthy of further study? For most people, questions need answers; for academics, questions pose more questions. The traditional conclusion for a dissertation is as ever, “More research is needed.”

"The outcome of any serious research can only be to make two questions grow where only one grew before." Veblen (1908)


...or separate References and Bibliography, if called for by the regulations.

I’m not going to go into formatting! It should be second nature by the time you get this far.


There are policy matters to resolve about how much evidence to include in appendices, and guidance from your supervisor is important here (as of course it is throughout). In general, however, be guided by the common-sense questions;

In all probability this will mean that you do not have to include original completed questionnaires (although of course you will include a blank version and its covering letter) or complete transcripts of interviews or focus groups, but should include all original statistical calculations in tabular format.

Video or audio material may be included; make sure that it is presented appropriately so that specific sections can easily be accessed.

You may need to include details of particularly arcane technical procedures with which a generic reader might not be familiar. You may also need to include some primary source material not readily available to a marker; if so, it will need to be redacted and anonymised. (If you need to anonymise, the most reliable procedure is to write the whole thing using real names and then Find and Replace throughout.)

And a quick housekeeping job before you put it away; read the appendices in conjunction with the main text and make sure that they really do what you claim for them, and their rationale is apparent.


No authors were harmed in the preparation of this paper; apart from those below all references are fictional;

And no, I can’t source the claimed Mark Twain quotation—and boy, have I tried!

Simple, really!

A .pdf of a slightly earlier version of this paper is available to download here and see here for a view from the other side of the fence.

Most of the other stuff on the web which supposedly covers this stuff, incidentally, is a come-on for cheat and plagiarism sites. This site does not accept advertising (at the time of writing [08.10]—I may have to go there some time and forget to take down this self-righteous posturing :-) ) or any external funding.

But do check out Brian Whalley's take on how not to do it! as well as his more general helpful advice and resources here and Pat Thomson (again) on the final check-list before handing in—tuned to the requirements for the PhD.

up-dated 16 February 2015