How to write a good PhD literature review?

Question

I am in the early stages of my PhD and I am trying to put together a literature review of my topic. My intention is that this review will eventually be included in my thesis, something like "the chapter that describes the state of the art in the domain."

My subject is soft matter physics and a large number of theses have already been written this topic. When looking at the literature reviews included in these other works I've noticed that:

• The sequence of ideas is often the same (this makes sense, so why not?)
• Cited papers are always the same, even though 20 years elapsed between the first and last thesis. Moreover, papers from the 30's are cited but no one in my lab seems to have a copy available?

My hypothesis is that these theses are derived from some sort of a Mother of All Thesis, and that paraphrasing the work of the previous student is OK.

So, what makes a good literature review?

How do you notice that a literature review is sloppy?

PS: I posted this question because I assume many of you have experience with this process either in your own PhD work, or as a supervisor where you have had to deal with paraphrasing of your students.

PS2: I am not asking about making proper citations, LaTeX+BibTeX handles that like a charm.

Answer 1

The purpose and expectations of a PhD literature review is likely to vary from field to field. My PhD was in Physics, but my views might be taken to apply generally.

There is likely to be some repetition or paraphrasing between students in the same research group, when it comes to the literature review. However, perhaps the comments below might help.

A literature review should be an enjoyable to read (!) introduction, survey and guide to the state of the art. You want to introduce your reader to the field (assuming a clever, but non-expert reader), setting out what has gone before, and perhaps to some extent showing where the gaps are in the research - raising the opportunity for you to present your research as that which fills a gap (Next chapter: "Aims and Objectives", or similar).

My primary hallmarks of a poor, or sloppy literature review is that papers are listed without any helpful context. A dreary literature review, to me, is a listing of papers that we all already know about, without any guide to the reader why the trail leads me to hold the present thesis. No-one much likes reading a dry chronology of papers. I personally, want to be told the story of the research and the literature review plays an important part of that story.

The opportunity exists here to review the field - what are the general trends in the literature? For example, Paper A was the first to introduce the theory that drove the authors of Paper B to perform experiment Z which is now the standard technique. However, Paper C suggests that an alternative method may be more effective, etc.

Note that your review shouldn't attempt to be a complete review of the field - whole standalone papers are written on this, usually by invitation. Your PhD literature review should be more focused, but still a recounting of the Story So Far.

Since your literature review is to be a nice, focused review of the path to your contribution, it is likely that you will read far more papers than you will need to cite in your literature review. Those papers that do not contribute to the Story So Far can be excluded from your literature review. Going off on a tangent, like in any story, can lose and confuse your readers. If you feel a need to refer to these papers, perhaps you can refer to a decent review article which discusses them in detail, for the interested reader.

Answer 2

I think Nikolas' answer is already pretty great. I'm not doing physics, but I'll try and stay as general as I can. Here's some specific advice I got from my supervisor and things I realized while doing my own lit review:

• It is normal for a big section of referenced papers to be the same across a lot of survey / literature review papers. Those would be the papers that first introduced a problem, a concept, an approach.

• In addition to the seminal papers from your (sub)field, you usually want to describe the current state-of-the art. This would be based on current papers based on the original problem, concept or approach that adopt the problem for a different environment, apply a concept for a different purpose or represent an improvement to the approach.

  For example, in Computer Science, it would be okay to talk about a structure or a problem (seminal paper) and then talk about the current best algorithm(s) to solving the problem (state-of-the-art) without mentioning every single "evolution step" of the algorithm.

  _{^{Basically, to sum up and dump up these two points: you cite the "first" and the "last" paper dealing with the same thing.
  Of course, there's exceptions to this: if there's any groundbreaking papers between the "first" and the "last" paper, sometimes intermediate papers can also be viewed as "seminal papers" for the subject/field.}}

• This might depend slightly on the type of document you want to produce, but usually it is okay for you to explain the technique / method in detail, while for practical uses of the technique, you just mention (and cite) several successful applications of the technique without going in to detail about how exactly the method was adopted.

• Finally, if your goal is to publish your literature review as a survey paper (which is usually worth a shot), you should think about how to "get a new spin on things".

  Every paper, including survey papers, is supposed to be a scientific contribution. That means that you have to find something that makes your survey useful, or in some context better, then all the existing surveys. This might be a change of context in which the methods are examined, it might cover more material, offer new classifications of the methods or new links between them.

  I would say you have to think of at least one type of reader (a reader with one type of goal) who will take your review and say: "That's it!", while he can not say that any of the current surveys out there are exactly "it" for him.

Answer 3

References and bibliography are to be read and digested in a progressive manner. References that might have been not so intuitive become useful over time as we gain more experience. One needs to document them in any suitable way and:

1. Add new references and connections with current work.
2. Track these references and revisit them when and where you touch base again with them.
3. Revise the entry with new information or clearer understanding of the subject.
4. Remove any parasite or related paper that you think is no more directly related to your work - clear clutter up - this is important to stop accumulating lots of bibliography which can become non specific!
5. All of the references you might accumulate may not be useful for the final bibiliography. There is need to sort or classify these references as biblio, self learning references, state of art, related (first order, second order) and so on and so forth.

Doing this using a wiki would be advisable - and if there is a team involved group updates would be preferable! Basically one needs a good sense of organization while writing the thesis.

Answer 4

There are couple of points i like to make from my perspective.

1. Bibliography is something which evolves during your PhD. I would recommend not to write it at the first place. As you read more and more, relevant to your area in Soft matter, you can keep adding it to your bibliography

2. Soft matter is a really huge area. One who works on Molecular dynamics may not even touch crystal defects while writing his bibliography. In that way it is really topic specific and not the entire area.

3. You said, you saw some say 30 papers in every thesis. This is not because of magic, this is only because they are path breaking. If you are in MD area (which is mine), and you are using a thermostat, it is 99.9999 % sure that you cite Nose-Hoover paper. It is no magic.

4. There is no point in saying none from my lab is cited. It depends on how many groups are working in that "specific" area and what impact had the papers published from your lab made in their research.

This all points out to the fact that one does not simply write a bibliography of an entire area :)