How to Write a Literature Review: Structure and Synthesis

Move from summarising sources to synthesising a field, search strategy, thematic organisation, critical evaluation, and the gap statement.

The single most common literature review failure is the "annotated list": paragraph one summarises Smith, paragraph two summarises Jones, and so on. A review that earns marks does something harder, it makes an argument about what a whole body of research collectively shows, disputes, and misses.

Step 1: Systematic Searching

Define search strings from your research question's key concepts plus synonyms, joined with Boolean operators: ("employee retention" OR "staff turnover") AND ("remote work" OR "telework"). Run them across Google Scholar plus at least one discipline database (Scopus, Web of Science, PsycINFO, HeinOnline). Set inclusion criteria, date range, peer-reviewed only, language, context, and record them. Methodology chapters increasingly expect a search audit trail. Then snowball: mine the reference lists of the best papers backwards, and use "cited by" to search forwards.

Step 2: Choose an Organising Logic

Thematic (default): sections follow concepts, e.g. "Drivers of turnover", "Remote work and isolation", "Moderating role of management style". Best for most dissertations.
Chronological: only when the evolution of thought is itself the point (e.g. paradigm shifts in a theory's development).
Methodological: grouping by how studies were done, useful when qualitative and quantitative literatures disagree.

Step 3: Synthesise, Don't Summarise

The grammatical tell of synthesis is multiple citations per sentence, organised around an idea:

Summary (weak): Smith (2020) found remote workers felt isolated. Jones (2021) surveyed 300 employees and found similar results.

Synthesis (strong): Isolation emerges consistently as the primary psychological cost of remote work (Smith, 2020. Jones, 2021. Okafor, 2023), although the effect weakens sharply where managers hold structured one-to-ones (Jones, 2021), suggesting the mechanism is managerial rather than spatial.

Useful synthesis moves: agreement ("consistent with"), conflict ("in contrast"), progression ("building on"), and evaluation ("however, these studies rely on self-report measures").

Step 4: Evaluate Critically

For key studies, interrogate: sample size and representativeness, measurement validity, context transferability, and funding/bias. You are not attacking authors, you are calibrating how much weight each finding can bear. Phrases that signal criticality: "the small sample limits generalisability", "causality cannot be inferred from this cross-sectional design", "findings from US retail may not transfer to UK healthcare".

Step 5: Land on the Gap

The review's final section funnels to an explicit gap statement that your study will fill: a population no one studied, a context no one tested, contradictory findings no one reconciled, or a method no one applied. Then state how your research questions address it. A literature review that doesn't end at your study's doorstep has not finished its job.

Quality Checklist

No paragraph is about a single source.
Every section heading is a concept, not an author.
At least a third of sentences contain evaluation, not description.
The gap statement names precisely what is missing and why it matters.
Every source cited has actually been read, not harvested from another author's footnote.

Four Ways to Organise, and When Each Works

Organisation is the decision that most affects whether a review reads as synthesis or as a list. Choose deliberately rather than defaulting to whatever order you read things in.

Structure	Organised by	Best when
Thematic	Concepts and issues	The default for most dissertations. Forces synthesis
Chronological	Time	The development of thought is itself the story
Methodological	How studies were done	Method differences explain conflicting findings
Theoretical	Competing frameworks	Rival schools structure the field

Thematic is right far more often than not, because it makes the review about ideas rather than about publications. Chronological is the most commonly misused, since it slides easily into "and then someone else wrote".

The Synthesis Matrix

The single most effective tool for turning notes into synthesis is a grid with sources down the side and themes across the top. Fill each cell with what that source says about that theme.

The value appears immediately in the columns. A column with many entries is a well-established point. A column where entries disagree is a debate, which is where your best paragraphs will come from. A column that is nearly empty is a gap, and it may be the gap your study fills. Reading down a column gives you a paragraph, whereas reading across a row gives you a summary of one source, which is what you are trying to avoid writing.

Sentence Patterns That Synthesise

Synthesis is visible at sentence level. These patterns do the work.

Convergence: "Three studies converge on X (A, B, C), though each measured it differently."
Divergence: "Where A found X, B reports the opposite, a difference attributable to sample age rather than method."
Extension: "B extends A's model to a context A explicitly excluded."
Limitation: "Both rely on cross-sectional data, so neither can establish direction."
Consensus with a caveat: "The association is well established, though almost entirely in high-income settings."

Each begins with a claim about the literature and cites sources as evidence for it. Contrast that with "Smith (2019) found X. Jones (2020) found Y." Two sentences, two sources, no synthesis.

Being Critical Without Being Dismissive

Critical engagement does not mean finding fault. It means assessing what a study can and cannot support. A landmark paper with a small sample is still landmark, and saying so while noting the limitation is more useful than either uncritical praise or reflexive dismissal.

Useful angles include whether the sample supports the claim made, whether the measure captures the construct, whether the design supports causal language the authors use, whether the conclusion outruns the data, and whether findings have replicated. Applying two of these consistently produces a far more critical review than mentioning all of them once.

How Much to Include

Level	Typical share of the word count	Rough source count
Undergraduate dissertation	20 to 25 per cent	30 to 50
Masters dissertation	25 to 30 per cent	50 to 80
PhD thesis	Often a chapter or more	150 plus

Treat these as orientation, not targets. A review citing forty sources with genuine synthesis beats one citing ninety in a list, and examiners can tell the difference within a page.

Common Mistakes and Fixes

Mistake	Fix
One paragraph per source	Reorganise by theme. Each paragraph should use several sources
Headings named after authors	Name headings after concepts
Description with no evaluation	Assess what each study can support, not only what it claims
Everything agrees	Look harder. Real literatures contain disagreement, and it is your best material
The gap appears only in the final line	Build toward it throughout, so it reads as a conclusion
Citing sources found in another author's footnote	Read the original. Secondary citation propagates errors
No date boundary or scope statement	Say what you covered and why, and what you excluded

Where Ethical Support Fits

Subject librarians will help you build a search strategy and find the literature, which is exactly the service they exist to provide. Discussing how to organise your themes with a supervisor is normal, and proofreading is fine.

What stays yours is the reading and the synthesis. A review is an argument about a field, and that argument can only be made by someone who has read it. Citing a source you have not read is also the mechanism by which errors and, occasionally, entirely fabricated references enter a bibliography.

Frequently Asked Questions

How recent do sources need to be?

Recent enough to show the field is current, which for most areas means substantial coverage of the last five to ten years. Foundational older work should still be cited where it defined the concepts.

Can I cite a source I found in another paper's reference list?

You can find it that way, but read it before citing it. Secondary citation propagates misreadings, and examiners occasionally check.

Should the literature review come before or after data collection?

Usually before, since it justifies the study. Expect to revise it afterwards, because your findings will change which literature matters.

How do I know when I have read enough?

When new searches keep returning sources you already know, and when you can state the main positions in the field and who holds them.

Can I include grey literature?

Yes, where relevant, including government reports, theses and professional guidance. Note its status, since it has not been through peer review.

What do I do with a source that contradicts my whole argument?

Engage with it directly and early. A review that omits the strongest counter-evidence is weaker than one that addresses it, and an examiner who knows the field will notice the absence immediately.

Does the review go in the introduction or its own chapter?

In a dissertation, usually its own chapter. In a journal article, usually folded into the introduction. Follow the format you are writing in.

Your Next Step Today

Build the synthesis matrix. Sources down the side, themes across the top, one afternoon. The columns will show you your paragraphs, your debates and, in the sparsest column, quite possibly your gap.

Trusted Sources

University of Southern California, The Literature Review. Accessed 11 August 2026.
University of North Carolina Writing Center, Literature Reviews. Accessed 11 August 2026.
Paré G and Kitsiou S, Methods for Literature Reviews, NCBI Bookshelf, on review typologies. Accessed 11 August 2026.
University of Southern California, Theoretical Framework. Accessed 11 August 2026.

Requirements differ by discipline and department. Where this guide and your handbook differ, follow your handbook.