To write a literature review, define a focused question, search several databases with recorded search terms, read and log each source in a matrix, group the sources by theme rather than by author, and write a synthesis that compares what they found, where they disagree and what is missing. A literature review is an argument about the state of knowledge, not a list of summaries.
What a literature review is for
It shows what is already known about a topic, how researchers have studied it, where they disagree and where the gaps are. In a dissertation, it justifies your research question. As a standalone piece, it maps a field for the reader. Check your department's guidance on length and format, which varies a lot.
Step 1: define a narrow question
"Social media and mental health" is too wide to review. "Associations between daily social media use and depressive symptoms in adolescents" can be finished. Write the question, the population, the period you will cover and the types of source you will include. These are your inclusion and exclusion criteria, and they stop the search from growing forever.
Step 2: search systematically
- Pick databases for your field: for example PubMed for health, ERIC for education, JSTOR for humanities, Scopus or Web of Science for broad coverage, and Google Scholar or OpenAlex for open searching.
- Write a list of search terms, including synonyms and spelling variants, and combine them with AND and OR.
- Record what you searched, when, and how many results you got. You may need to report it.
- Use citation chasing: look at the references of good papers (backward) and at who has cited them since (forward).
- Start with recent review articles; they save weeks.
Step 3: screen and read
Screen titles and abstracts against your criteria first. Then read the full text of what is left, using the first two passes of the approach in our guide to reading a research paper. You will not read every source deeply. Be honest about which ones are central and which are context.
Step 4: build a literature matrix
A matrix is a table with one row per source and columns you choose. It is the tool that turns a pile of PDFs into a review.
| Source | Question | Method and sample | Key findings | Limits | Theme |
|---|---|---|---|---|---|
| Author, year | What it asked | Design, N, setting | Main results | Weaknesses | Which theme it fits |
When you fill the Theme column, patterns appear. Sort by it and you have the skeleton of your review.
Step 5: choose a structure
- Thematic: sections by concept or issue. Best for most reviews.
- Chronological: by how the field developed. Useful where history matters.
- Methodological: by approach, such as qualitative against quantitative.
- Debate-based: by the competing positions on a question.
Thematic is the safest. Whichever you choose, tell the reader the plan in the introduction.
Step 6: synthesise, do not summarise
A weak review reads: "Smith (2019) found X. Jones (2020) found Y. Lee (2021) found Z." A strong one reads: "Three studies report X, though the effect is larger in samples recruited online (Smith, 2019; Lee, 2021), possibly because of self-selection. Jones (2020) found no effect using a longitudinal design, which suggests the earlier results may reflect..." Compare, contrast, weigh the quality of evidence, and say what you conclude.
Step 7: write, then revise
- Introduction: the topic, why it matters, scope, and the plan.
- Body: one theme per section, with a topic sentence stating the claim of the section.
- Conclusion: what is known, what is disputed, what is missing, and how your project responds.
- Revise for argument first, then citations, then language.
Common mistakes
- Describing studies one at a time instead of grouping them.
- Leaving out opposing evidence.
- Quoting too much. Paraphrase and cite.
- Relying on abstracts you did not read in full.
- Poor reference management. Use Zotero, Mendeley or similar from day one.
Keeping track of what you have learnt
A review depends on you remembering how studies relate. Short retrieval sessions help: after reading each source, write its question, method and finding from memory. If you are new to the field, you can use a tool such as Type to Study to build a course from a set of PDFs or from a topic researched from OpenAlex papers, so you meet the key terms and ideas on a schedule. It cites the pages and sources it used, but you still need to read the primary literature yourself and not cite the course.