Guide · Academic writing

How to Write a Literature Review: A Step-by-Step Guide for Master's and PhD Students

Published 22 October 2026 · 7-minute read · By the CiteLoom team

The literature review is often the hardest chapter of a thesis to start and the easiest to get wrong. Many first drafts read like an annotated bibliography: one paragraph per paper, each beginning "Smith (2019) found…". Examiners want something different: an argument about what the research shows, where it agrees and conflicts, and what is still unknown.

This guide explains how to write a literature review step by step, with practical tips, a structure you can adapt and a final checklist.

What a literature review is (and isn't)

A literature review surveys, evaluates and synthesises existing research on a question. It shows you know the field, justifies your own research by identifying a gap, and positions your work within existing debates.

It is not:

• a list of every paper you read;
• a series of summaries in chronological order;
• a place for your own new data;
• neutral description without evaluation.

There are several types. A narrative review surveys and interprets the literature on a topic. A critical review emphasises evaluation. A scoping review maps a broad field. A systematic review follows a strict, reproducible protocol (such as PRISMA) to answer a narrow question. Most thesis chapters are narrative or critical reviews; check what your programme expects.

Step 1: Define your question and scope

Write one sentence that states what your review covers. Specify:

• Who: the population (for example, rural women, primary-school teachers, SMEs).
• What: the phenomenon or intervention.
• Where: countries or settings.
• When: the date range.
• Type: narrative, critical, scoping or systematic.

Weak: "Mobile money." Strong: "How mobile money use relates to household savings among rural women in sub-Saharan Africa, 2007–2025: a critical narrative review."

A focused scope makes searching manageable and your argument sharper.

Step 2: Search systematically

1. List keywords and synonyms for each part of your question ("mobile money", "M-Pesa", "digital financial services"; "savings", "financial resilience").
2. Search several databases: Google Scholar plus discipline databases (for example Scopus, Web of Science, PubMed, ERIC, JSTOR) and open indexes such as OpenAlex and Semantic Scholar.
3. Use Boolean operators: combine synonyms with OR and concepts with AND.
4. Follow citation trails: check the references of key papers (backward) and the papers that cite them (forward).
5. Record your search: databases, search strings, dates and number of results. Examiners may ask, and it makes your review reproducible.

Step 3: Screen and select

Read titles and abstracts first, then full texts for anything promising. Decide explicit inclusion and exclusion criteria: date range, study types, languages, populations. Keep studies that disagree with your expectations. A review that ignores contradicting evidence is easy to criticise.

Prioritise peer-reviewed research, recent work and seminal studies everyone in your field cites. Note anything retracted and leave it out.

Step 4: Read critically and extract

For each included study, record the same details in a table. This evidence table becomes the backbone of your review:

While reading, ask: How strong is this evidence? Does the design support the conclusion? Would it apply in other settings?

Step 5: Synthesise by theme

Synthesis is the heart of a literature review. Instead of summarising papers one by one, group them:

• By theme: effects on savings, effects on borrowing, effects on women's decision-making.
• By debate: studies that find a strong effect versus those that find little.
• By method: what surveys show versus what experiments show.
• Chronologically, only if how thinking developed over time is itself your point.

For each group, ask: Where do studies agree? Where do they conflict, and why (different samples, methods, settings)? What is missing?

Step 6: Outline an argument

Turn your themes into an outline where each section makes a point. A common structure:

1. Introduction: the topic, why it matters, your scope and how the review is organised.
2. Body sections: one per theme or debate, each comparing and evaluating studies.
3. Discussion or synthesis: what the evidence shows overall and how strong it is.
4. Conclusion: the key gaps and how your research addresses them.

Step 7: Write

Use topic sentences that make a claim, then support it with several sources:

Weak: "Smith (2019) studied mobile money. Jones (2021) also studied mobile money." Stronger: "Survey evidence consistently links mobile money with higher saving (Smith, 2019; Jones, 2021), but because these studies are cross-sectional, they cannot show that mobile money causes the increase."

Good literature-review prose:

• compares ("in contrast", "consistent with", "however");
• evaluates ("a small, self-selected sample limits…");
• signals your voice ("this suggests", "taken together");
• cites accurately, with the claim matching what the source found.

Step 8: Check and revise

1. Check every citation: each reference should exist and support the sentence it's attached to. Narrow any claim that overstates a study.
2. Check balance: have you represented contradicting evidence fairly?
3. Check the gap: does your conclusion clearly lead to your research question?
4. Check flow: clear transitions between sections and paragraphs.
5. Check formatting in your required style (APA, Harvard, Chicago and so on).

Literature review or systematic review?

Students sometimes describe a narrative review as "systematic" because they searched carefully. Examiners treat the term precisely. A systematic review follows a pre-registered or pre-specified protocol, searches defined databases with recorded strings, screens studies against explicit criteria (often with two reviewers), assesses risk of bias, and reports the process with a flow diagram such as PRISMA. A narrative or critical review is more flexible and interpretive. Both are valuable; just label yours accurately and describe your search honestly.

How long does a literature review take?

For a master's thesis chapter, expect several weeks of searching and reading and two to four weeks of writing and revising. For a PhD, the review often evolves throughout the degree. AI tools can shorten searching, extraction and first drafting, but reading key studies and checking claims still takes time.

Using AI to help write a literature review

AI tools can help at several stages: suggesting search terms, finding papers, extracting details into an evidence table, drafting a first version from your chosen sources and checking citations. Use them carefully:

• prefer tools that cite only real papers from scholarly databases or your own library;
• approve the sources and outline yourself;
• check every cited sentence against its source;
• rewrite in your own words and follow your university's rules on AI use and disclosure.

Common mistakes to avoid

• Summarising instead of synthesising. If most paragraphs start with an author's name, restructure by theme.
• Ignoring contradicting evidence. Examiners notice when obvious counter-studies are missing.
• Over-relying on one or two sources. A few authors dominating your review suggests narrow searching.
• Outdated literature. Include recent work unless your review is deliberately historical.
• Overstating findings. Associations reported as causes, or one country's findings generalised to all.
• No clear gap. The review should end by showing exactly what is unknown and why your study matters.
• Citation errors. Real papers cited for claims they don't make, or references that don't exist.

Final checklist

• Clear question and scope stated in the introduction
• Search strategy recorded
• Studies grouped by theme, debate or method, not one per paragraph
• Studies compared and evaluated, not just described
• Contradicting evidence included
• Gaps identified and linked to your research
• Every citation verified against its source
• References formatted consistently

Frequently asked questions

Sources

1. Education News Kenya, Postgraduate enrolment trends (Commission for University Education data).
2. Evaluation of AI tools versus the PRISMA method for literature search and data extraction, PMC.