A few notes on dissertation structure. I suppose this is primarily aimed at people writing up a basic five-chapter dissertation: introduction, lit. review, methods, results, discussion.
And the abstract should be included, too.
The structure of these sections is determined by a logic in which each piece is set to play a role in telling the reader of your work.
The pieces work together and in a way they all mirror each the main project and each other--but they are distorting mirrors, if you will: they exaggerate different things as they reveal the project.
The abstract is a mirror that makes the image small: the abstract tells the reader about the entire project--its motivation, context, basic premises, method, results and conclusions.
The Introduction does the same job as the abstract, but at the next order of magnitude, and with exaggerated concern with the issues the motivate and contextualize the project. Methods are briefly discussed; lengthy discussion of the methods is saved for the methods chapter. The results and discussion/conclusion are not mentioned, but obviously the issues discussed in the introduction are the ones that set up the later discussion: the research questions set up the character of the research results.
The literature review, I think, is poorly understood. It's not just a compendium of material written about your subject; it's a description of the sources that explain your position and your interests. The literature review discusses the material that helps explain why you believe or expect what you believe and expect. If you are studying some phenomenon, what material in the literature defines the phenomenon? what literature explains it? What literature sets up the expectations that drive your study? The literature review has a responsibility to attempt to show material beyond that which you are using most immediately, especially to show positions/arguments that contradict or refute the premises that you are using. But primarily its purpose is to show the reader what you are talking about. In this way it is another reflection of the same large project: it is the part where you share with the reader your background.
The methods, results and discussion chapters all also share this sense of mirroring a larger project: each has a position in which it exaggerates one part of the whole, but still keeps the whole in view so that the reader always understands how the piece under current view is still kept in relation to the larger whole.
Structure isn't as easy if you can't follow an easy formula, like that for an empirical study. However structure, whatever structure you use, is built out of a desire to share one large image with the reader. There ought to be a coherent conception to the dissertation.
I know, from my own experiences, and from speaking with people working on dissertations in the humanities that often one is encouraged to write one chapter at a time and then try to bring the work together as a whole when the chapters are much more complete. There is some justification for this, given the difficult nature of writing up research in the humanities, where the ideas cannot be so easily reduced quantities or codes, but rather interpretive and analytic subtleties are being explored.
But I believe that understanding a larger structure and having a larger vision in sight, it is easier to write the pieces of the work. Why is it easy to write a one-page essay? Partly it is because you can see, rather easily, just how the whole piece will be structured: you can see that you have to briefly introduce the subject, present your evidence and discuss the conclusion that you want the reader to take away--and that, in itself, is most of the work: writing a coherent one-page piece can be the work of fifteen minutes (granting, of course, that a brilliant one-page piece might require many revisions).
When the pieces get larger, it gets harder to see what to put where. Yes, we need to start by introducing the idea, and we need to finish by concluding something, but there's a lot more space in between the opening and closing. What to put where? Having a sense of purpose related to large structure can help with this: if you can see the overall project, and can see how different presentations of important ideas can reach the reader, then you can make choices about where to put material, and what to goal of the material is.
It may well seem that it is harder to try to impose a larger view of the project and a structure to which each little piece must respond. It may well be different for different people, but I know that for myself, if I know where the pieces are going and what I will try to accomplish, it makes it easier. For the written portion of my qualifying examinations, I was required to answer five questions from the members of my examination committee. When I was given the five questions I asked if they needed to work together; My chair said they did not and even suggested that it would make it more difficult, but I was already seeing how each question could be useful in contextualizing what I wrote for the other questions. By making one question into an introduction, another into a conclusion, two more into discussions of different aspects of theory, and a final question into a discussion of research methods, the pieces supported each other, but better, I knew what I had to put where: some material had to set context, some material had to aim the discussion into the future and so forth. By giving roles of this sort to the different questions, I had reduced the possibilities for what I could write on each question.
Structure actually helps. If you can see structure for the way it helps organize the material, for the different roles that you can give to each part of structure, and so forth, the work as a whole makes more sense, and thereby becomes easier to write. Without a vision of overall structure, which reflects your purpose, it becomes more difficult to see what possible choices are most useful.