A literature review does more than list sources — it synthesises, evaluates, and maps the existing knowledge on a topic. This guide walks through database searching, organisational structures, synthesis techniques, and how to identify gaps your research will fill.
A literature review is a critical, structured overview of published research on a specific topic. Its purpose is not to describe every paper you read — it is to build a coherent argument about what is known, how it is known, where the evidence conflicts, and what remains unresolved.
Literature reviews appear in two main forms:
In STEM, standalone literature reviews are highly valued because they synthesise evidence for practitioners who cannot read every primary paper. Systematic reviews and meta-analyses are the gold standard in medicine, environmental science, and psychology.
A scoped literature review covers a well-defined question. Before opening a database, answer these questions:
Use the PICO or PECO framework (Population, Intervention/Exposure, Comparison, Outcome) to define your scope precisely. This is the standard approach in systematic reviews and helps you construct database search strings efficiently.
For an academic literature review, use peer-reviewed databases — not Google. Key databases for STEM:
| Database | Best For |
|---|---|
| PubMed / MEDLINE | Medicine, clinical science, pharmacology, nursing |
| Web of Science | All STEM disciplines — high-impact journals |
| Scopus | All STEM + social sciences, citation tracking |
| IEEE Xplore | Engineering, electronics, computing |
| Google Scholar | Broad coverage — use for citation chaining, not primary searching |
| ACM Digital Library | Computer science, software engineering |
| ScienceDirect (Elsevier) | Physical sciences, chemistry, materials |
Build a search string using Boolean operators:
A broad search may return thousands of papers. Screen in stages:
Use a reference manager. Zotero, Mendeley, or EndNote allow you to import search results directly, remove duplicates, store PDFs, and generate citations automatically. Start using one before your first search — not after you've accumulated 200 unsorted PDFs.
Our STEM specialists research, synthesise, and write full literature reviews with systematic search documentation and APA/IEEE/Harvard referencing.
How you organise the literature depends on what makes the argument clearest. Three common structures:
Group sources by the themes, concepts, or variables they address — not by who wrote them or when. Best for most STEM literature reviews because it synthesises rather than lists.
Organise by how the field has evolved over time — useful when the development of a technology or theory is itself the story.
Group studies by the research design or method used — quantitative vs qualitative, experimental vs computational, in vitro vs in vivo. Useful when you want to compare what different methods reveal about the same question.
The most common failure in literature reviews is writing summaries (one source per paragraph) rather than synthesising (multiple sources combined to make a point). Examiners reward synthesis.
Chen et al. (2022) found that deep learning outperformed traditional methods. Rodriguez et al. (2023) also found that deep learning performed better. Kim and Park (2023) conducted a study on deep learning.
Multiple recent studies confirm the superiority of deep learning over traditional image classification methods in clinical diagnostics (Chen et al., 2022; Kim & Park, 2023; Rodriguez et al., 2023), though performance gains vary substantially with dataset size and imaging modality.
A good literature review does not treat all sources as equally valid. You should evaluate:
After synthesising what is known, you must identify what is not known or not yet resolved. This gap justifies your study. Common types of gaps:
State the gap explicitly: "Despite the volume of research on deep learning in radiology, no study has examined its performance specifically on low-resolution X-ray images from resource-limited settings. This study addresses that gap."
There is no fixed number — it depends on the scope and the depth of your topic. Undergraduate literature reviews: 20–40 sources is common. Master's level: 60–100. Systematic reviews: whatever the search protocol yields — this could be 500+ initial papers screened down to 20–50 included studies. Quality and relevance matter more than quantity.
It depends on the purpose. For systematic reviews in medicine, grey literature is important because it helps prevent publication bias (negative findings are often not published in journals). For most STEM assignments, peer-reviewed journals and books are the primary sources. Always check your marking criteria.
A narrative review is structured by the author's expertise and judgement — sources are selected to build a coherent argument. A systematic review follows a pre-registered, reproducible protocol: explicit inclusion/exclusion criteria, multiple databases searched, duplicate reviewers, risk-of-bias assessment. Systematic reviews are higher in the evidence hierarchy but require significantly more time and rigour.