How to Write a Literature Review: Step-by-Step Guide with Examples
Ryan McCarroll
2 min read
Last updated: September 22, 2026
A literature review explains what reliable research says about a focused question, where studies agree or conflict, and what remains unknown. To write one well, define the question, search systematically, screen sources, extract comparable evidence, organise it around an argument, and revise until each paragraph synthesises several studies rather than listing them.
TL;DR
A literature review analyses and connects existing studies. It is not an annotated list of summaries.
Choose a narrative, scoping or systematic approach before you search.
Organise the body thematically, chronologically or methodologically, according to the question.
Record methods, samples, findings and limitations in a synthesis matrix before drafting.
AI can accelerate searching and extraction, but every citation and claim still needs human checking.
What is a literature review, and what is it not?
A literature review is a critical synthesis of published research on a defined topic or question. It establishes what is known, compares theories and methods, explains disagreements, evaluates the strength of the evidence, and identifies gaps that justify further work. It can stand alone as a paper or form part of a thesis, dissertation, research proposal or journal article.
A literature review is a reasoned account of a field, not a stack of article summaries. It combines evidence from several sources to explain patterns, disagreements, limitations and unanswered questions. The writer’s contribution is the structure and interpretation: showing how individual studies connect, how much confidence their methods deserve, and what the combined evidence means for the research question.
An annotated bibliography treats sources one at a time. A literature review puts them in conversation. Instead of writing “Smith found X, Jones found Y, and Patel found Z,” ask why the findings differ. The explanation may lie in the sample, setting, measurement, theoretical framework or period studied.
A strong review does five jobs:
Places a research question within the current scholarly conversation.
Shows areas of agreement, debate and change.
Evaluates the credibility and relevance of the evidence.
Prevents unnecessary duplication of earlier work.
Identifies a defensible gap or direction for further research.
Which type should you use: narrative, scoping or systematic?
The right type depends on the question and the standard of reproducibility your project requires. A broad interpretive question may suit a narrative review. A field that has not yet been clearly mapped may need a scoping review. A narrow question about effects, associations or interventions may require a systematic review.
Review type | Best for | Search and selection | Typical output | Main limitation |
|---|---|---|---|---|
Narrative | Explaining concepts, theories and debates | Purposeful and transparent, but not always protocol-led | Interpretive synthesis organised around an argument | Selection can be harder to reproduce |
Scoping | Mapping a broad or emerging field | Broad search with explicit inclusion criteria | Map of themes, evidence types and gaps | Usually does not estimate one combined effect |
Systematic | Answering a tightly defined question | Predefined protocol, documented search and screening | Reproducible synthesis, sometimes with meta-analysis | Time intensive and narrow by design |
A thematic, chronological or methodological review is an organisational choice, not necessarily a separate review type. For a formal protocol, screening flow and reporting standards, use the guide to conducting a systematic literature review.
How should you structure a literature review?
Most literature reviews contain an introduction, a synthesised body and a conclusion. The introduction defines the question, scope, inclusion boundaries and organising logic. The body compares evidence. The conclusion states what the literature supports, where uncertainty remains, and how the gap relates to the next study.
Thematic structure
Group studies around recurring ideas, outcomes or debates. This is usually the strongest choice when the field contains several connected explanations. A review of social media and adolescent mental health, for example, might use themes such as sleep disruption, social comparison, cyberbullying and supportive communities. Within each theme, compare methods and findings rather than describing papers one by one.
Chronological structure
Trace how knowledge changed over time. Use it when turning points matter, such as the arrival of a new theory, technology, dataset or policy. Do not merely sort citations by year. Explain what changed, why later work challenged earlier assumptions, and which questions persisted.
Methodological structure
Group research by design, such as experiments, surveys, interviews, cohort studies or computational analyses. This works when different methods produce different answers. It lets you show that an apparent disagreement may reflect how a concept was measured rather than a genuine conflict.
You can combine approaches. A review may use themes as the main H2 sections, then trace developments chronologically inside each theme. The key test is simple: the structure should help answer the research question, not merely store information.
How do you write a literature review step by step?
1. Define the question and scope
Write one answerable question. Set boundaries for population, concept, context, dates, languages and source types. A topic such as “social media and mental health” is too broad. “How does passive social media use relate to depressive symptoms among adolescents?” gives the search and synthesis a clear centre.
Decide what the review must accomplish. Are you mapping a field, testing whether findings are consistent, comparing methods, building a theoretical model or identifying a gap? This decision controls what evidence belongs.
2. Build and document the search
Break the question into concepts and list synonyms for each. Combine synonyms with OR, then connect concepts with AND. Search databases that fit the discipline and record the database, date, exact query and filters. Follow reference lists and papers that cite key studies, but apply the same eligibility rules to every candidate.
A useful search is iterative. A first pass reveals recurring terminology, authors and subject headings. Add those terms, rerun the search, and stop expanding only when new searches mostly repeat material already found.
3. Screen and evaluate sources
Remove duplicates, screen titles and abstracts, then assess full texts against the inclusion criteria. Keep a reason for each exclusion if the review needs a transparent record. Evaluate relevance and quality separately: a paper can be closely related to the topic yet provide weak evidence.
Check the research design, sample, measures, analysis, conflicts of interest and whether the conclusion matches the results. Give greater weight to evidence with methods suited to the question. Do not treat publication in a journal as proof that every claim is sound.
4. Extract comparable evidence
Use a consistent matrix with columns for the research question, method, sample, context, main finding, limitation and your analytical note. This reduces repeated reading and makes patterns visible. Record page numbers for quotations and verify bibliographic details while the source is open.
Look across rows for agreements, contradictions and blind spots. Ask whether findings change with age, location, measurement, study design or follow-up period. A table of notes becomes synthesis only when you explain these relationships.
5. Synthesise, draft and revise
Create a claim for each section before drafting it. Then choose several sources that support, qualify or challenge that claim. A strong paragraph usually opens with the point, compares evidence, evaluates the comparison, and closes by explaining its significance.
Revise at three levels. First, check the argument: does every section answer the question? Second, check synthesis: are studies connected rather than listed? Third, check accuracy: does every citation support the exact sentence attached to it? Remove repetition, unsupported generalisations and detail that does not change the conclusion.
What does synthesis look like in a worked example?
Before: source-by-source summary
Smith studied university students and found that retrieval practice improved test scores. Lee also studied retrieval practice and found a benefit. Ahmed examined online learners and reported mixed results. These studies show that retrieval practice can help learning.
After: analytical synthesis
Across classroom and laboratory settings, retrieval practice generally improves later recall, but the size and consistency of the benefit depend on how practice is spaced and how closely the test matches the practised material. Smith and Lee report positive effects in university samples, while Ahmed’s mixed online results suggest that unsupervised completion and uneven spacing may weaken the intervention. The evidence therefore supports retrieval practice as a learning strategy, but not as a context-free solution.
The revised paragraph does four things the first version does not. It leads with a cross-study claim, names the pattern, explains a disagreement through method or context, and ends with a bounded conclusion. The sources become evidence for an argument rather than the paragraph’s organising units.
Use the same method when findings conflict. Put the studies side by side, compare samples and measures, and state which explanation is best supported. If the evidence cannot resolve the conflict, say so directly and identify what a future study would need to test.
Where can you find a template and annotated examples?
Start with the free literature review template and outline when you need headings, prompts and a synthesis matrix you can copy. Then examine the annotated literature review examples by discipline to see how published writers frame scope, compare evidence and move from findings to gaps.
A template is a scaffold, not a fixed formula. A short course assignment may need only an introduction, two or three thematic sections, and a conclusion. A dissertation chapter may need a longer methods explanation, theoretical framework and explicit account of the gap. Keep the logic, then adapt the section lengths to your question and discipline.
How can you use AI without creating fake citations?
Use AI to accelerate bounded tasks, not to replace source checking or scholarly judgement. It can help refine a question, propose search terms, organise notes, compare extracted fields and improve a draft’s clarity. It should not be trusted to invent a bibliography, quote a paper it cannot access, or decide that a gap exists without checking recent research.
Follow four safeguards:
Work from identifiable papers and open the original source behind every important claim.
Confirm authors, title, year, journal, DOI and quotation wording before citing.
Separate extracted evidence from your interpretation in your notes.
Keep your search record and inclusion decisions so another reader can understand how the evidence was chosen.
Doing this with AnswerThis
Open the AnswerThis literature review tool and enter a focused research question. Select the databases and date range that fit the project, then review the generated answer beside its cited sources. Open the relevant papers, compare methods and findings in the source table, save useful evidence to your working notes, and verify each claim before moving it into your draft.
After mapping the recent literature, use the research gap finder to generate candidate gaps. Test each candidate against the newest relevant papers before treating it as a defensible research gap.
AI output is a starting point for analysis. Your review still needs your inclusion logic, evaluation of methods, interpretation of disagreements and academic voice.
FAQ
How long should a literature review be?
The required length depends on its purpose, discipline and assessment rules. A section in a short paper may be compact, while a dissertation chapter or standalone review needs enough space to explain the search, evaluate evidence and develop a sustained synthesis.
How many sources should a literature review include?
There is no universal number. Include enough relevant, credible research to represent the major evidence and debates within your defined scope, then follow any requirement set by your journal, department or supervisor.
Can a literature review include older sources?
Yes. Include foundational work when it defines a theory, method or turning point, then connect it to current evidence. For fast-moving topics, prioritise recent research without erasing the studies that shaped the field.
What is the biggest mistake in a literature review?
The most common mistake is summarising one paper after another without an argument. Organise around themes, methods or developments, and make each paragraph compare several sources in relation to the research question.
Can AI write a literature review for you?
AI can produce a draft and help organise evidence, but it cannot take responsibility for source selection, citation accuracy or interpretation. Treat generated text as material to verify and revise, not as a finished scholarly review.
When should you choose a systematic review?
Choose a systematic review when the question is narrow enough for predefined eligibility criteria and a reproducible search and screening process. If the aim is to map a broad field or interpret a debate, a scoping or narrative review may fit better.
