Two documents, one name.
An annotated bibliography reports what each source says. A literature review reports what the sources say together, including where they disagree and why. Both are legitimate documents and only one of them is being asked for, yet the first is what arrives, over and over, because it is what note-taking naturally produces if nothing intervenes.
The test is mechanical and takes a minute. Read the first sentence of each paragraph in turn. If those sentences name authors, the document is organized by source. If they name findings, tensions or conditions, the document is organized by argument. Only the second kind can support a study, because only the second kind has a shape into which a gap can be cut.
Build the matrix before writing anything.
The reorganization is easier in a table than in prose, and the table is worth building even for a modest review. One row per study, and columns you fill as you read rather than afterward. Themes then appear as patterns down the columns, which is considerably more reliable than trying to remember forty papers at once.
- The citation, entered in your required style at the moment you read the paper rather than at the end
- The question the study asked, in your own words and in one line
- The design: experimental, quasi-experimental, correlational, qualitative, or a review of others
- Sample and setting, since a finding from one professional group in one country is a finding about that group in that country
- How the key variable was measured, because studies reaching different conclusions have often measured different things
- The finding, with an effect size or magnitude where the paper reports one
- The limitation the authors themselves name, which is frequently the most useful cell in the row
Read for design, not only for conclusion.
Two papers can support the same sentence and constitute very different evidence for it. A controlled trial and a cross-sectional survey may agree, but they license different claims, and a review that stacks them as equivalent has quietly overstated its case. Weight what you cite by what the design permits: whether comparison groups existed, how participants were selected, whether the measure has been validated, and whether the analysis reported magnitude alongside significance. That last habit is set out in the note on reading output.
Disagreement between studies is an opportunity rather than an inconvenience. When findings conflict, the explanation almost always lies in population, setting, measurement or timing, and a paragraph that identifies which of the four is responsible is the most valuable paragraph in a review. Students tend to hide such conflicts, or to cite only the side that suits the argument. Committees notice both, and the second is worse.
The shape of a paragraph that carries a claim.
Once themes exist, each paragraph does one job and announces it. Open with the claim, in your own voice, as a sentence that could be true or false. Then bring two or three studies as evidence, saying what each contributes rather than what each is about. Then a sentence naming the limitation of that evidence, the disagreement inside it, or the condition under which it holds. Then a sentence connecting the claim to the study you intend to conduct.
Four sentences of that structure carry more weight than a page of chained findings, and the difference is visible to a reader in seconds. It also solves the citation problem that returns so many drafts: sources appear where they are needed for an argument rather than in the order they were collected, and any source that cannot be attached to a claim can be removed without loss.
A gap is argued, never announced.
The sentence beginning little research has been conducted is doing no work at all unless the preceding pages have made it inevitable. A defensible gap follows from the review the way a conclusion follows from an argument, and it takes one of a small number of forms: a population nobody has studied, a method nobody has applied to this question, a contradiction nobody has resolved, or a context in which a well-established finding has not been tested.
Name which of those you are claiming, then point back to the paragraphs that establish it. A reader should be able to trace the gap to the evidence without your help. Where the review is sound but the chapters around it are stalling, or where the matrix has grown faster than the argument, that is the ordinary work of the capstone and dissertation section, taken theme by theme.
Questions to the desk.
How many sources does a literature review need?
Enough to establish each claim you make and no more, which means the count follows from the argument rather than setting it. Where a program specifies a minimum, treat that as a floor to satisfy, not a target to hit. A review of thirty well-chosen studies organized by theme reads as scholarship; a review of sixty summarized one by one reads as a reading log.
How current do the sources have to be?
Recency rules exist because fields move, and most programs ask for the bulk of citations to be recent. The exception is deliberate: seminal work establishing a theory, an instrument or a definition is cited at its origin regardless of age. Cite the old paper for what it founded and the new papers for what is currently known, and make the distinction visible to the reader.
Can I cite a study I have only read about in another paper?
Only as a secondary citation, clearly marked as such, and only when the original is genuinely unobtainable. The risk is that you inherit the second author's reading, including any error in it. Since the original is usually available through a library, the honest and safer course is to obtain it, read it, and cite it directly.
What does it mean when a review is called descriptive rather than critical?
It means the draft reports what studies found without weighing how well they found it. The repair is to add judgment: comment on design quality, note where samples limit generalization, and say plainly where findings conflict. A critical review is not one that criticizes everything, but one in which a reader can see you evaluating the evidence rather than only collecting it.
