401. Doing a literature review
Here is some advice for PhD students on doing a literature review as part of their research for inclusion in their thesis.
In a traditional thesis in the Australian PhD system, you needed to include one or two large literature-review chapters.
These days, most PhD students in Australia do a thesis by publications, and in those, most of the literature review occurs within the individual papers. At UWA, the system is quite flexible and you can choose to include an additional separate chapter of literature review if you wish. I’m not sure about other universities.
Your literature review should not be anything like one particular PhD thesis that I examined, where the student had summarised many articles and stuck all the summaries together in sequence, one article per paragraph.
The ideal literature review builds up to something. It helps the reader (and in fact the researcher) reach some conclusions about:
- what are the un-researched or under-researched issues in the area?
- which of them are the key ones for further research and why?
The idea is that, having answered those questions, the researcher heads off and researches those key under-researched issues. It may not really happen like that in reality, but in writing up a literature review, you should aim to portray it as if it did. It is best if it DOES happen somewhat like this caricature, but it is OK if it doesn’t.
How many papers do you need to review? Generally, you do not have to review every slightly relevant paper ever written. You do of course need to cover all the famous and important papers in the area. Beyond that, it depends on how big you want or need the review to be. If it’s a stand-alone dedicated lit review, it might be quite big and so include a lot of papers. There is judgment required when you decide where to set the boundary when thinking about which papers to include. In many cases, the answer may not be obvious.
Make sure your references are not all old ones and not all new ones (I once examined a PhD with almost no references less than 10 years old!)
When you write the review, you should discuss the key papers in some depth, but most of the papers you cite will be covered at a relatively shallow level. You might cite a paper just to support one narrow point.
To avoid the temptation to do the wrong thing, don’t write up any text summaries of the papers you read. Just keep dot point notes about them, and no more than about half a page of those in most cases, sometimes much less.
Your review should be sharp, well structured, to the point, not waffly, not excessively long, and most importantly highly relevant to the thesis topic. Just because you read a bunch of literature, don’t feel that it has to go into the thesis unless it is sufficiently relevant. Quite a few of the papers you look at might not make it into the thesis.
A key question is how to organise the references into a good structure for your lit review. Ask yourself, how can you bundle the papers together in groups in interesting and helpful ways? Is it useful to consider differences in research done at different scales, or different locations, or using different methods, or whatever? If you choose a particular criterion to use to bundle up papers, which aspects have been well researched, and which poorly. Have the findings been consistent or inconsistent? If inconsistent, why? Bad research methods? Diversity in the real world? What are the main results in each bundle of papers? Are there things that haven’t been done, or have been done poorly? If you can see faults, by all means be critical but politely.
What you are really looking for in the papers are trends and gaps, areas of consensus and differences.
A really good literature review might also include a conceptual framework to help your organise the discussion about the different studies and their findings. The different studies you review might relate to different parts of the conceptual framework. That might be one of the criteria you use to chunk the literature.
A really good literature review is not just a summary of the literature. It includes quite a lot of judgment and insight from the author(s) that adds value to what is in the existing studies. It identifies problems with the existing literature and opportunities to create new knowledge.
Each set of papers that you’ve decided to handle as a bundle becomes a section (or maybe a paragraph in a short lit review), and you look for trends-gaps-consensus-differences within each section. Make some brief notes about what the trends-gaps-consensus-differences are. They become your key messages. If you’ve read the literature, you should be able to do this without having to go back and look at a single paper while you are doing it.
Think about the order of your sections, and then you’re ready to write it.
When thinking about what you write, concentrate on the story you want to tell, not the results of individual studies. It may be that, at the start of a section, the text is mainly the author (i.e., you) talking, explaining what’s in this section, and relating it to the conceptual framework. It might not include a high frequency of references at the start. These increase as you get into the details.
Once I get into the details, I still tend not to look at individual papers, I’ll just write the story and leave gaps for references. Sometimes you may have to check what’s in a paper or two as it affects the storyline. In telling the story, you cite examples of papers that illustrate the point you are making. Many would be cited only in passing, to support a point you are making. Only a minority would have any sort of summary of their findings. That might be because they are the only paper in a particular category or because they are particularly important.
It can be useful to include some quotes from papers, but do it so as to best support the general point you are making.
Literature reviews vary enormously in style and structure. That’s fine, as long as you meet the basic needs.
University of Queensland. How to write a literature review.