Literature Reviews for Busy People

This is a guide to literature reviews. I cannot claim impressive credentials, but I can claim annoyance. Professors and researchers quip, "Nobody ever reads papers from start to end. They skim them. I wish I figured this out sooner.", but what does this mean? You know you've read a paper when you hit the last word and you know you understand a paper when you understand it, but when do you know you're done skimming?

After years of annoyance, I found a solution. Thomas Deetjen, an energy systems researcher at the University of Texas, compiled his research experience into a small paperback: a cross between a self improvement book, an experiment planning guide, and a writing manual. The book is instructive; it tells you exactly what to do. It goes as far as to include flow charts. Science can be tedious. When it is, being able to turn off higher decision making and still progress is a godsend. Deetjen provides that. If you're in academia or industry and struggling to publish, this book is for you.

This post is not meant to be a book review. Instead, I'm picking out the literature review part of Deetjen's book and rewriting what I found important in my own words. This will make it easier to explain to others how I'm doing a literature review and why literature reviews are relevant.

In the remainder of this document I give the benefits of literature review, the difficulties, and ways to mitigate the difficulties. The reader should end this document motivated and capable.

Motivation

First, why do we review literature?

There are three main reason:

  • We review to find motives: why the research is useful to society.
    • If we reuse existing motives, we don't have to justify our own motives from scratch.
    • If our research lacks motives, then the research isn't worth funding.
  • We review to find results: what has already been discovered.
    • If we look at existing results, we can find gaps for future work.
    • If our research has already been done, then the research isn't worth funding.
  • We review to learn methods: how experiments are done.
    • If we reuse existing methods, we don't have to develop and justify our own methods from scratch.
    • If our research methods cannot be justified, then the research isn't worth funding.

Each of the three reasons corresponds to funding and manuscript acceptance.

Inadequate literature review leads to disasters:

  • If your research lacks motives, then you will struggle to publish. (How did you even get funding?) If you do publish, nobody will read your publication. It's possible to fix this without doing a new experiment.
  • If your research reproduces existing results, you will not be able to publish. Reviewers will jeer at you. You will have to do a new experiment.
  • If your methods are not acceptable, you will not be able to publish. Reviewers will jeer at you. You will have to do a new experiment.

Difficulties

Literature review is hard for several reasons:

  • Many papers must be read. A rule of thumb is 50 papers. If your papers average 10 pages, that's 500 words, the size of a meaty textbook.
  • Fully understanding a technical paper requires multiple focused readings and background research. To understand a single paper, you may need to consult several textbooks.
  • When to put a paper aside is unclear.
  • Diligent notes are required.
  • Finding papers is a context switch from reading them.

Solution

The solution is a methodical process that reduces all ambiguity to decision questions and templates.

  • Instead of skimming a paper at random, skim to find motives, results, and methods. Write them down. Once you have key information recorded, you're done skimming the paper. With a clearly defined end, you will complete more papers.
  • Instead of noting anything that looks interesting, use a template with sections for motives, results, methods, and identifying information. Deetjen provides a literature review template. With a template to record information in, you will skim faster.
  • Instead of picking papers at random, keep them all in a read queue folder. When you finish a paper, move it to a done folder. With a constant supply of material, you will dawdle less.
  • Instead of switching between reading and finding papers, set aside time to fill your read queue with papers to get you through the week. With a constant supply of material, you will dawdle less.

None of these are ironclad rules, they are defaults. If you find an interesting paper while reading, throw it in your queue. If it's extremely interesting, start reading it. This process is an efficient way to deal with mundane literature review. If your experiment is based very closely on the methods in some paper, you will need to return to that paper, read it, and understand it; skimming is not enough.

Conclusion

I have inflected these techniques on myself and others, and it's made mounds of papers dissapear; the techniques work. I hope you find them useful.

Personal Setup

My typical setup is a single document for all notes per project. I use a word processor so I can paste figures easily. When I'm performing a literature review, I aim for 2 papers a day with 45 minutes or less dedicated to a paper. Each Friday, I ensure I have 10 papers to read for the next week. Having these notes has been invaluable when writing white paper. A paper alluding to concrete research comes out very sincere.