How to structure the review
The mechanics below apply whatever you write it in.
Pick an organizing principle and say what it is
Thematic organization is the default in most fields: group by idea, not by author. Chronological works when the story is how thinking changed. Methodological works when the disagreement in the field is really about how things were measured.
Tell the reader which one you've chosen in the opening paragraph. It costs one sentence and makes the rest of the chapter easier to follow.
| Structure | Use it when | Watch out for |
|---|---|---|
| Thematic | Several strands of research bear on your question | Themes that are only topics, with no argument |
| Chronological | The field changed direction and the history matters | Sliding into a timeline of who published when |
| Methodological | Findings differ because methods differ | Losing sight of what was actually found |
| Theoretical | Competing frameworks explain the same evidence | Describing theories without testing them against studies |
Write paragraphs that synthesize
A synthesis paragraph opens with a claim about the literature, not about a paper. It then brings in two or more sources that support, complicate or contradict the claim, and closes by saying what follows.
Compare a summary sentence, "Study A surveyed 300 nurses and found burnout was linked to shift length", with a synthesis sentence, "Shift length is the most consistent predictor of burnout across settings, although the two studies that controlled for staffing ratios found a weaker effect." The second one does the work a review is for.
A quick test: if a paragraph starts with an author's name, it is probably a summary.
How many sources is enough
There's no number. A workable stopping rule is saturation: when new searches keep returning papers you've already seen, and new papers stop changing your matrix, you've covered the ground.
Your supervisor or assignment brief may give a range. Treat it as a floor for coverage, not a target to hit.
Earn the gap
The gap you identify at the end should be visible in the matrix: an empty column, a population nobody sampled, a method nobody tried. If the reader is surprised by your gap, the review hasn't set it up.