How to write a literature review: from summary list to argument

The question a committee asks while reading a literature review is always the same: "So what follows from all this?" A chapter in which every paragraph starts with an author and ends with that author’s finding does not answer it. The reader collects facts for pages but has to work out alone why your thesis is needed. The steps below show how to build a chapter that answers the question with the same sources.

Summary list versus argument

A summary list looks like this: "A (2018) found this. B (2020) examined that. C (2021) reached a similar conclusion." Each sentence is true, but no relationship is drawn between them. An argument looks like this: "Studies of this relationship reach different conclusions depending on the age of the sample: those with adults find a strong association (A; B), while the only study with adolescents does not (C)."

In the second version the subject is the idea, not the authors. The sources sit in brackets as support for the claim. That small change turns the whole chapter from a list into a line of reasoning.

Thematic or chronological?

In most theses a thematic structure works best: sections are built around the concepts that make up your research problem, or around debates in the field. In each section the reader sees one question and the different answers given to it.

• Chronological order works when the development of the concept or approach over time is itself the subject of your thesis. Otherwise it turns into "first this was done, then that" and the argument disappears.
• Methodological order suits fields where the debate is mainly about how something is measured or which design is used.
• A mixed structure is also possible: thematic overall, with chronology inside a theme when you need to show how it developed.

Whichever you choose, the order of the sections should say something too. Moving from general to specific, from concept to gap, prepares the reader for your research question.

How to write synthesis sentences

A synthesis sentence places at least two studies in a relationship. If the first sentence of a paragraph is a synthesis sentence, the rest becomes supporting detail. A few patterns that work:

• Agreement: "Studies in different settings largely converge on ..."
• Disagreement: "Two positions stand out on this question: ..."
• Explaining the disagreement: "The difference in results appears to stem largely from the instruments used."
• Shared limitation: "All of these studies rely on cross-sectional data, so they cannot speak to change over time."
• Moving to the gap: "This leaves the question of ... unanswered."

Identifying the gap

The destination of a literature review is the reason your research question needs to be asked. The gap usually takes one of a few forms: a group or context has not been studied, findings contradict each other, existing studies share a methodological limitation, or a theory has not been tested in a particular setting.

The gap should emerge from the studies you have reviewed, not appear suddenly in the last paragraph. Absolute statements such as "this has never been studied" are the first thing a committee challenges. Saying which databases and search terms you used and writing "as far as I could find" is both more honest and easier to defend. The final paragraph should lead straight into your research question, and the same gap returns in the discussion chapter alongside your findings.

How much literature is enough?

There is no number that applies to everyone; the field, the degree level and what your advisor or graduate school expect all matter. The signs that you have reached enough are clearer:

• New studies you find cite sources you already know and do not add a new position.
• You can explain the main positions in the field, the key works behind them and the debate between them in a few sentences.
• Every source in the chapter has a job in the argument; remove it and a claim loses its support.

The last test is the most useful. A source added only because you read it makes the chapter longer, not stronger.

Common mistakes

Outdated sources

Citing the classic works of your field is natural, but if the chapter rests only on older sources the committee will ask what has changed in recent years. Present classic studies together with newer work that tests or extends them.

Only supporting studies

Selecting studies that support your hypothesis and leaving out contrary findings is a gap any committee member who knows the field will notice. State the contrary findings openly and discuss why they may have reached different results; this does not weaken your argument, it makes it credible.

Citation without engagement

Stacking five sources at the end of a sentence does not tell the reader what they show. A more serious problem is citing sources you have not read, copied from another paper’s citations: a review of medical journals found that 16.9% of citations were not supported by the source they cited (Baethge & Jergas, 2025). Reference lists generated with AI carry their own risk, covered in the guide to AI for thesis writing. How to verify your references is in the reference check guide.

A quick check before submission

1. Does the first sentence of each section synthesise, or does it start with an author’s name?
2. Does the order of the sections lead the reader towards your research question?
3. Are contrary findings included?
4. Does the gap follow from the reviewed studies and connect to your research question?
5. Does every citation in the text have an entry in the reference list, and every entry a citation?

Getting an outside reader for this check is not always easy. tezkeys reads your thesis chapter by chapter, then evaluates it as a whole, and marks literature, argument, method and interpretation problems on your own text, with their location and a severity level. It also compares your references with Crossref records, showing citations missing from the reference list, retracted publications and DOIs that point to a different publication.

It does not write or rewrite your text; the report is AI-generated, can be wrong and does not replace your advisor. The reference pre-check is free and needs no sign-up; see what a full report looks like in the sample report.