Last updated: 2026-10-06

U
Undergraduate level

Literature Reviews That Do Analytical Work

A literature review is often treated as a summary of papers, written in order, with one paragraph per source. That structure produces a document that describes the field without saying anything about it. A review that does analytical work groups the sources by the ideas they share and dispute, and uses that grouping to justify the project. It shows where the field stands, what it has not yet settled, and where your project fits.

This page covers how to plan a review around concepts rather than papers, how to read sources for what they claim and what they assume, and how to use the review to check the question. The review's place in the wider project is described in Project Navigation and the Art of Triangulation. The question it tests is set out in Finding a Project Idea and Refining the Question.triangulation: cross-check your claim with other sources

Why the Review Matters

Webster and Watson argue that an effective review gives a firm foundation for advancing knowledge and supports the development of theory[1]. For a student project the practical version of that claim is simpler. The review tells you what is already known, so you do not spend the project rediscovering it, and it tells you where the open questions are, so you can choose one of them.maps the known to the unknown to find the gap

A review that does neither is a chore. A review that does both becomes the project's reference point, and it is the part you will return to most often.

Organise by Concept, Not by Paper

Start by listing the concepts the question depends on. For a question about recommender systems, these might be "cold-start", "evaluation metric", and "user trust". Each concept becomes a section, and each section draws on whichever papers address it. A paper may appear in several sections, which is expected. The structure makes the relationships visible.

Webster and Watson's account of writing reviews in information systems is directed at the same goal from the other direction: a review should build on prior work and make the state of the field usable[1]. A concept-based outline does this in practice. It also tells you early where the literature is thin, which is often where a project can contribute.thin spots are where you can contribute

Read for Claims and Assumptions

When you read a source, record three things: what it claims, the evidence for the claim, and the assumptions the claim depends on. Assumptions are the part students skip, and they are often the most useful. A result that holds for one dataset size, one class of users, or one hardware setting may not hold for yours. Knowing the assumption tells you whether the source supports your question or only resembles it.the hidden scaffolding of any argument

Keep a short note on each source in the same form, so they can be compared later. A template is enough: the citation, the question the paper answers, the claim, the evidence, the assumptions, and one sentence on how it bears on your project.

Analysis Rather Than Description

The University of Reading's guidance on writing styles distinguishes descriptive, analytical, and reflective writing. Descriptive writing answers "what?", "when?", and "who?". Analytical writing answers "why?", "how?", and "so what?", by reviewing what you have read in the light of other evidence and discussing its implications[2]. A literature review is mostly analytical. Description has its place, but it should be brief, and each paragraph should move from what a source says to what that means for the question.

Don't do this. Write a paragraph that begins "Smith (2019) proposed X" and ends with the next paper. Each paragraph should answer a question the reader has, such as "is this approach established?" or "what does the evidence say about this assumption?"

Using the Review to Check the Question

The review has a second job, which is to test the question. Once the main sources are read, ask three things:

  1. Has this been answered? If so, what is still open, and does your project address it?
  2. Has this been asked in a different way? A parallel literature may use other vocabulary for the same problem. Search for synonyms and for the community that uses them.
  3. What do the best sources assume that you cannot assume? If a method requires a condition your project will not meet, the method may not carry over.

If the answer to the first question is "yes, and it has been answered well", the question needs to change, not the review. Record the change and its reason, as described in Scope, Feasibility, and Defining Success.

Keeping the Review Alive

A review written once in week three and never revisited will be out of date by the time the project is assessed. Schedule a short check at each milestone to search for new work, and record each new source in the same note format. Papers that change the direction of the project belong in the discovery log, not only in the bibliography.

References

  1. Webster, J., & Watson, R. T. (2002). Analyzing the past to prepare for the future: Writing a literature review. MIS Quarterly, 26(2), xiii–xxiii.
  2. University of Reading. (n.d.). Descriptive, analytical and reflective writing. Academic writing guides, LibGuides. https://libguides.reading.ac.uk/writing/stylesofwriting