How to Write a Literature Review

How to Write a Literature Review: Quick Answer

Name the assignment first. A course paper, a thesis chapter, a standalone narrative review, a PRISMA 2020 systematic review, and a PRISMA-ScR scoping review are different products. The official PRISMA site still names PRISMA 2020 as the main reporting guideline; PRISMA-C 2026 is a paediatric extension, not a replacement of that checklist. Then write a question, search more than one database, screen with stated criteria, read for themes rather than for a pile of summaries, and organize the body around ideas. University of Toronto Writing Advice calls a paragraph that starts with a new researcher name a warning sign. Ask your instructor how many sources and how long the paper should be before you invent a quota.

Capture the bibliographic header while the PDF is still open. The CitationEasy generator can style the string later; it cannot recover a source you never identified. Use how to find scholarly sources for the search and how to evaluate sources for the screen.

What a Literature Review Is

The UNC Writing Center opens by killing a common mix-up: "literature" here is the published research on a topic, not novels you rate. Their definition is the one this page uses.

University of Toronto Writing Advice, in The Literature Review: A Few Tips On Conducting It, restates the same job as an account of published work by accredited scholars, defined by a guiding concept. Their next sentences are the failure mode this page is written against, and the four jobs the finished prose has to do.

UNC adds that a review may evaluate sources and advise the reader on what is most pertinent. That evaluation is critical analysis, not a personal like-or-dislike. If you only needed a list of sources with a note under each one, you were assigned an annotated bibliography.

Literature Review vs Research Paper vs Annotated Bibliography

UNC draws a hard line between a standalone review and a research paper that contains a review. Mixing the two is how a methods chapter gets written as if it were the whole project.

The same handout is equally blunt about the annotated-bibliography trap.

If the prompt says "review the literature on X," you are writing this page. If it says "write a research paper on X," use how to write a research paper and keep the review as one section. If it says "annotate these sources," stop here and open the bibliography guide.

Which Review Type You Were Actually Assigned

Maria J. Grant and Andrew Booth's 2009 paper A typology of reviews (Health Information and Libraries Journal, 26(2), 91–108) mapped 14 review types against a Search, Appraisal, Synthesis, and Analysis (SALSA) framework because the words review, systematic, and overview were already being used interchangeably. Course assignments still do that. Name the type before you search.

A course "lit review" and a Cochrane review cannot share one method section. Grant and Booth quote the MeSH scope note for the generic literature review; that breadth is why the label alone does not tell you the method.

If the assignment borrows systematic steps without the full protocol, Grant and Booth's honest label is systematized review, not systematic review.

The Cochrane Handbook for Systematic Reviews of Interventions version 6.5 (2024), Chapter 1 by Lasserson, Thomas, and Higgins (last updated August 2021), is the methods book behind a true systematic review. PRISMA 2020 is the reporting checklist for that product, not a substitute for the handbook.

Page et al. published PRISMA 2020 in several journals in 2021, including PLOS Medicine. The official site still treats that 27-item statement as the main guideline. Box 1 is the definition to keep.

A 2026 date on a PRISMA page does not mean the core checklist was replaced. PRISMA-C 2026 (Baba, Farid-Kapadia, Smith, Hartling, Moher, Hooft, and Offringa, BMJ 2026;393:e088561) extends PRISMA 2020 for reviews of interventions in people aged 0–19 years, mixed child–adult populations, and parent/child dyads. Use it only when the population matches. Do not paste it onto a course narrative review.

A scoping review is a different product. The PRISMA-ScR page and Tricco, Lillie, Zarin, et al. (Annals of Internal Medicine, 2018, 169(7), 467–473; DOI 10.7326/M18-0850) treat it as a map of evidence, not a verdict on effectiveness. Grant and Booth's scoping sentence is the size-and-nature test.

Grant and Booth point to a real published scoping review as their specimen: Weeks and Strudsholm (2008), "A scoping review of research on complementary and alternative medicine (CAM) and the mass media: Looking back, moving forward," BMC Complementary and Alternative Medicine, 8, 43, DOI 10.1186/1472-6882-8-43. Open that paper to see what a map looks like. Do not cite it as evidence on your topic unless your question is CAM and the mass media.

If your instructor asked for a narrative review, do not paste a PRISMA flow diagram to look rigorous and then hide an unsystematic search. If they asked for a systematic review, do not write UNC's course-paper structure and call it PRISMA. Cochrane will not treat a one-person student paper as a Cochrane review. Grant and Booth's "systematized review" is the honest label for a graduate assignment that borrows systematic steps without the full protocol.

Step 1: Name the Question and the Scope

UNC's first writing move is to clarify the assignment, then narrow the topic. Their handout says instructors will not expect you to read everything, and that a narrower topic is what makes a survey possible. Toronto's first self-check is the same: what thesis, problem, or research question does this review help define? For a systematic review of an intervention, Cochrane Chapter 1 points to PICO — Population, Intervention, Comparison(s), and Outcome — as the usual question frame.

Write those answers down before you search. A review of "social media" is not a scope. A review of experimental studies on evening social-media use and sleep onset in adolescents, 2015 to the present, is a scope. UNC also warns that currency is field-specific: in the sciences, two-year-old treatment evidence can be obsolete; in the humanities, a history of the debate may be the point.

• Question: what conversation are you entering, in one sentence?
• Boundaries: which populations, methods, languages, and dates are in or out?
• Source types: peer-reviewed articles only, or also books, dissertations, and grey literature?
• Function: standalone paper, thesis chapter, or the front of an empirical article?
• Review type: narrative, systematized, systematic, or scoping — named before the first database run

Step 2: Search the Literature Before You Draft

Toronto lists information seeking as the first skill a review demonstrates: scan efficiently, with manual or computerized methods, and identify a useful set. That is a documented search, not the first ten Google Scholar hits. For database choice and Boolean syntax, use how to find scholarly sources and library databases. This page only covers what a review writer has to record.

1. List concepts and synonyms for each part of the question, then combine them with AND / OR / NOT.
2. Search more than one database. Multidisciplinary indexes (Web of Science, Scopus, Google Scholar) miss field collections (PubMed, PsycINFO, ERIC, MLA International Bibliography, IEEE Xplore).
3. Use citation chaining. UNC recommends the reference lists of sources you already trust, plus "who cited this?" in the database or Scholar.
4. Decide what counts as current in your field, then keep seminal older works when the debate requires them.
5. Write the search down the day you run it: database, date, string, filters, and hit count.

PRISMA 2020 expects systematic reviewers to report the full search. A narrative course review is not PRISMA, but the same log stops you from "remembering" a search you cannot defend. If the assignment is a systematic or scoping review, write or post a protocol before the screen — Cochrane Chapter 1 treats the protocol as the bias-control step — and use the PRISMA 2020 flow diagram rather than a homemade funnel.

Step 3: Screen, Select, and Evaluate

Toronto's second skill is critical appraisal: identify unbiased and valid studies. UNC's "be selective" rule is the writing half of the same job — mention only the points that serve the review's focus. How to evaluate sources is the CRAAP walkthrough; the peer-review process explains why a published article can still be weak. Do not treat peer review as a quality stamp and then skip methods.

1. Title screen against the written question, not against how interesting the PDF looks.
2. Abstract screen for population, design, and whether the paper actually studies your concept.
3. Full-text screen for methods you can describe, not just a useful sentence in the discussion.
4. Appraisal: who funded it, what was measured, who was left out, and whether the conclusion outruns the design.
5. Keep contrary studies. Toronto asks, explicitly, the question below.

Grey literature — reports, dissertations, preprints, conference papers — can fill a gap the journals have not published yet. Cite it as grey literature, and do not let a blog or a Wikipedia article do the work of a primary study. Wikipedia is a map to sources, not a source; see Wikipedia for research.

Step 4: Read for Themes, Not for a Stack of Summaries

A review written from highlighted PDFs becomes a list. Read with Keshav's three-pass method so most papers stop after a first or second pass, then take notes you can sort. The note fields below are the ones a synthesis paragraph actually uses. Capture quotations with the labels in how to take research notes.

• Full citation, DOI, and the database you found it in
• Question or purpose, in the authors' terms
• Population, setting, and design
• Key result, with the actual statistic or claim, not your memory of it
• Limitation the authors admit, plus any you see
• Theme tags you can sort on later (method, theory, population, finding)
• One sentence on how it talks to two other sources you already have

A literature matrix — one row per study, columns for question, sample, method, finding, and limitation — is still the most reliable way to see a theme before you outline. Spreadsheet, Zotero note, or paper table does not matter. How to organize research covers the filing system; this page only needs the matrix to exist.

Step 5: Choose an Organization

UNC says a review may not have a traditional argumentative thesis, but you still have to tell the reader the organizing principle. They give two model sentences, both real patterns rather than fake citations.

Then pick one body pattern. UNC's three named methods are chronological, thematic, and methodological. Grant and Booth add conceptual or theoretical analysis as another common cut. Theoretical organization — competing frameworks rather than competing findings — is the right cut when the assignment is a theory paper.

Chronological

Trace how the conversation changed. UNC's sperm-whale walkthrough is the caution: ordering by publication year alone can scramble subjects that happen to share a date. Prefer eras inside a trend (for example, pre-regulation / post-regulation) over a year-by-year dump.

Thematic

Group by issue. UNC calls this the usual structure for a complex topic. Time can still appear inside a theme. A thematic review that never compares two sources in one paragraph is an annotated bibliography with headings.

Methodological

Group by how the studies were done — qualitative versus quantitative, lab versus field, survey versus experiment — when the debate is about evidence quality rather than about a concept. UNC's point is that method, not content, is the focusing factor.

Theoretical

Group by framework when the literature is a fight among theories. Toronto asks whether you are looking at issues of theory, methodology, or policy; if the answer is theory, do not bury the frameworks inside a chronology of publication dates.

Step 6: Write Introduction, Body, and Conclusion

UNC requires the same three parts as most academic papers. Their labels are short on purpose.

Introduction

State the question, the scope, the review type, and the organizing principle. If the review sits inside a larger paper, say what the synthesis is for. A student paper can put a take-home sentence near the end of the introduction and check it after the draft so it still matches the body. That is a course-paper convention, not a PRISMA heading.

Body: synthesize in the paragraph, not after it

UNC reprints a real paragraph from Erika Falk and Jordan Mills, "Why Sexist Language Affects Persuasion: The Role of Homophily, Intended Audience, and Offense," Women and Language 19(2). Gastil (1990) and Hamilton (1988) are the studies those authors discuss. Use this paragraph as the model, not as a source on your topic.

UNC's commentary on that paragraph is the writing rule. They note that Falk and Mills use no long quotations, recap Hamilton, then say what the finding does in their argument, and start and end in their own voice.

Paraphrase unless the exact wording is the evidence. Cite at the first sentence a source governs, and do not recite the same paper in every clause. For the difference between paraphrase and quotation, see paraphrasing vs quoting.

Conclusion

Toronto wants the review to say what is known, what is contested, and what still needs research. If the review fronts your own study, the last paragraph is the handoff: the gap is specific enough that a methodology section can start. If the review stands alone, stop at the gap. Do not announce a finding you did not study.

Synthesis Sentences You Can Reuse

An earlier version of this page invented a remote-work paragraph (Smith, Jones, Brown, Zhang). Those citations were not real. Use the UNC Falk and Mills paragraph as the worked synthesis example, Weeks and Strudsholm as the worked scoping-review specimen, and the templates below only as syntax.

Failure Modes Students Actually Hit

• Source-by-source order. Toronto's "every paragraph begins with a name" test. Re-outline by theme before you revise sentences.
• Summary with no verdict. UNC wants evaluation of pertinence. A paragraph that only restates abstracts has no writer.
• Cherry-picking. Toronto asks whether contrary studies appear. A review that only supports your hoped-for finding is advocacy.
• Scope creep. Including a tangentially related paper because you already annotated it dilutes the question.
• Fake currency. Dropping every source older than five years erases foundational work. UNC says to check what your field treats as current.
• Quote collage. UNC tells you to paraphrase. A page of block quotes is not a review.
• Undocumented search. If you cannot say which database and which string produced a key paper, you cannot defend coverage.
• Wrong review type. A narrative course paper is not a systematic review. Do not add a PRISMA diagram to a search you could not replicate, and do not call PRISMA-C 2026 a replacement for PRISMA 2020.
• One-person "Cochrane" claim. Cochrane Chapter 1 requires a team and will not publish a single-author review. A graduate assignment that borrows two or three systematic steps is a systematized review.
• Missing "so what." The conclusion has to name a pattern or a gap. "More research is needed" is not a finding unless you say which research.
• Unformatted citations. Inconsistent style is a separate failure from weak synthesis. Pick the style the course named and keep a complete reference list.

Format the Citations While the Files Are Open

APA 7, MLA 9, and CMOS 18 disagree about author lists, titles, and DOIs. They agree that a review with broken references is unfinished. Use which citation style if the course did not name one, and how to write a bibliography for the end list. Format each record as you decide to keep the paper, not the night before submission.

Related Guides

• How to Write a Research Paper
• How to Write a Methodology Section
• Annotated Bibliography
• How to Find Scholarly Sources
• How to Evaluate Sources
• How to Read Academic Papers
• How to Take Research Notes
• How to Organize Research
• Grey Literature