How to Write a Literature Review (With a Worked Example)

The review that reads like a receipt#

You have read forty papers. You have a document with forty paragraphs, one per paper, each summarising what the author did and found. It is organised, it is accurate, and a marker will still hand it back with "this is a summary, not a review" scrawled in the margin.

This happens to good students constantly, not because they have not read enough but because nobody explained what a literature review is for. It is not a book report on your reading list. It is an argument about the state of a field, built so your own research question falls out of it as the obvious next step. Get that shift right and everything else gets easier to judge.

Summary, synthesis, and the gap: what these words actually mean#

Two terms get thrown around in every dissertation handbook without being explained properly.

Summarising a source means reporting what it says: this study surveyed 200 nurses and found X. It is source-by-source, one paper at a time.

Synthesising means writing about the topic, using several sources together to support a point you are making. Instead of "Smith (2019) found X. Jones (2021) found Y," synthesis says "there is reasonably strong agreement that X occurs in clinical settings (Smith, 2019; Jones, 2021), though the effect shrinks in community samples (Patel, 2020)." The sentence is now about the phenomenon, and the citations back it up rather than drive it.

A gap in the literature does not mean nobody has mentioned your topic before. It usually means a population, context or method that has not been studied; findings that conflict without being reconciled; or a well-studied area not yet connected to a related one. Cronin, Ryan and Coughlan, whose guidance for nursing and health students gets assigned reading across UK universities, frame the review's job as showing not just what is known but where the uncertainty sits.

Searching and screening sources systematically#

A haphazard search produces a haphazard review, so treat this stage as its own small project.

Turn your question into search terms and their synonyms, then run them through your university's discovery service or a subject database (Scopus, Web of Science, or a discipline-specific one such as PsycINFO). Keep a simple log: search string, database, date, hits, number kept. A supervisor will ask why certain studies are missing, and "I don't remember" is a weak answer.

Worked example. A student researching mobile phone use and sleep in teenagers ran "adolescent* AND smartphone* AND sleep" across two databases and got 340 hits. Screening titles and abstracts against inclusion criteria, published after 2015, empirical study, participants aged 11 to 18, cut that to 46. Reading full text against tighter criteria, excluding studies relying on self-reported screen time only, left 19 sources. That funnel, and the reasons for each cut, went into the methodology section as evidence of a systematic approach, which is what Aveyard and colleagues recommend in their guidance for health and social care students doing literature-based dissertations.

Two habits catch what a keyword search misses on its own. Citation chaining, sometimes called snowballing, means checking a strong paper's reference list for the earlier studies it builds on, then checking its "cited by" count on Google Scholar for later work that responds to it. Reference management software such as Zotero, Mendeley or EndNote earns its keep here too: it stores the PDFs, formats citation lists in whatever style your department requires, and stops you losing track of which version of a paper you actually read once your source list passes a dozen or so entries. Set it up early rather than retrofitting it once the review is half written.

Choosing an organising structure before you start writing#

The single most common structural mistake is starting to write before deciding how the review is organised. Paragraphs then get arranged by whichever order you read the papers in, and readers can always tell.

Most university writing centre guidance, echoing frameworks like Arksey and O'Malley's, points to three common approaches, and the right one depends on your question and discipline.

Most undergraduate and taught postgraduate dissertations work best thematically, because it forces synthesis: you cannot write a themed paragraph without pulling several sources into the same argument. PhD reviews sometimes combine approaches, but that is a deliberate choice, not a default.

Turning summary into synthesis: a worked before and after#

Here is a paragraph in its summary form, the kind that fills a first draft.

Before: "Brown (2018) studied remote working and found it increased reported job satisfaction. Lee (2020) also looked at remote working and found similar results in a UK sample. However, Osei (2021) found no significant effect on satisfaction among younger employees."

Now the synthesised version, built from the same three sources.

After: "Remote working appears to raise job satisfaction in some populations but not others. Studies of established employees report consistent gains (Brown, 2018; Lee, 2020), yet this effect largely disappears among younger workers (Osei, 2021), a discrepancy that has not been explained by tenure, seniority or home working conditions in existing research."

Notice what changed. The paragraph now makes a claim about the topic, and the three citations support one idea instead of three separate ones. The final clause does something the summary version could not: it points at an unresolved question, the seed of your eventual gap.

Naming the gap without overclaiming it#

Students often do one of two things badly here: claim a gap so large it seems unlikely ("no research exists on this"), or hedge so much the gap disappears into vague language ("more research is needed," a phrase examiners are tired of reading).

A workable gap statement names the specific limitation and why it matters: "Existing studies of remote working and satisfaction rely almost entirely on samples over 30, so it remains unclear whether the pattern holds for early-career employees navigating different expectations around office presence." That sentence says what exists, what is missing, and why the gap is worth filling, which is exactly the transition your introduction needs into your own research question.

Common structural mistakes that cost marks#

A few patterns show up repeatedly in markers' feedback at UK universities:

• One paragraph per source, fixable by asking of every paragraph, "what claim about the topic is this making?"
• No signposting between sections, so a thematic review reads as a list of headings rather than an argument that builds.
• Citing sources you have not actually read, often lifted from another paper's reference list. If you cannot describe a study's method yourself, do not cite it as evidence.
• A conclusion that just repeats the introduction, instead of sharpening the gap into your project's specific question.
• Citing only the studies that agree with each other. Reviews built that way look curated rather than critical, and markers notice.

Questions students ask about writing a literature review#

How long should a literature review be?#

It depends on level and purpose. An undergraduate chapter might run 2,000 to 4,000 words, a master's dissertation somewhat more, and a PhD review can fill a full chapter of 8,000 words or beyond. Check your department's handbook rather than a generic word count.

Is a literature review the same as a systematic review?#

No. A traditional or narrative review, the kind most taught courses ask for, uses a less formal search process shaped by the author's judgement. A systematic review follows a pre-registered protocol with explicit inclusion and exclusion criteria, often reported using the PRISMA framework, and aims to be reproducible. Postgraduate students should check which their supervisor expects.

Can I use secondary sources that summarise other studies?#

Sparingly, and not as a substitute for primary sources. Textbooks and review articles help with orientation early in your search, but the analysis itself should draw on the original studies wherever possible.

Do I need a theoretical framework before I start?#

Not always, but it helps. Even a loose framework gives you criteria for what counts as relevant, and stops the "download everything" approach.

How many sources count as enough?#

There is no fixed number. What matters is coverage of the key debates and enough depth to synthesise properly. A shorter review built from twenty sources you understand thoroughly usually scores better than a longer one skimming sixty.

Is it acceptable to use AI tools to help with a literature review?#

Many UK universities now permit AI tools for organising notes or checking clarity, but check your institution's specific policy, since rules vary by department. Using AI to generate citations or summarise papers you have not read is an academic integrity risk, because these tools can fabricate references that look plausible but do not exist.

Should grey literature count towards my review?#

Sometimes. Grey literature, government reports, theses, conference proceedings and policy documents that were not published in peer-reviewed journals, can fill gaps that academic publishing is too slow to cover, particularly on emerging or applied topics. Include it where it adds genuine evidence, but label it clearly as grey literature and hold it to the same questions you would ask of a journal article: who produced it, what method sits behind the claims, and whether it was reviewed by anyone before release.

What is the difference between a literature review and a bibliography?#

A bibliography or annotated bibliography lists and briefly describes sources individually. A literature review builds an argument across sources, which is exactly the distinction behind most "this is a summary" comments in the margin.

Before you write your next draft#

Reread your current draft and ask, paragraph by paragraph, whether it describes a source or makes a claim about the field. Where it is the former, that is where the rewriting starts. A literature review is not proof you read widely. It is proof you can think about what you read.