001Catalog versus argument.
The tell is visible from the shape of the paragraphs. A catalog has one paragraph per source: this author found this, that author found that, a third author found something else. Each paragraph is accurate, the citations are correct, and the reader finishes knowing what several studies said and nothing about what the field thinks.
An argument organizes by idea. A paragraph takes a claim or a tension and brings several sources to bear on it, showing where they agree, where they diverge, and what the disagreement implies. The same sources appear, sometimes in several places, and the reader finishes understanding the state of the question.
Rubrics call the difference synthesis, which is an unhelpfully abstract word for a concrete structural choice. The mechanical version of the instruction is this: if you can predict from the first sentence of a paragraph which single study it is about, you are writing a catalog.
002The table that produces the structure.
Before writing anything, build a matrix. Sources down the side, and across the top the dimensions on which they can be compared. This takes an afternoon and it saves the several days usually spent rearranging paragraphs that were never going to cohere.
- Population and setting, because studies that appear to conflict frequently studied different people, which is itself a finding worth reporting.
- Method and design, since a disagreement between a controlled trial and a cross-sectional survey means something different from a disagreement between two of the same kind.
- The key variables as each study defined them, which is where much apparent contradiction dissolves; two studies measuring the same word differently are not really disagreeing.
- The finding, stated in one line without hedging, so that patterns become visible across the column.
- The limitation that matters most, which is what you will draw on later when arguing that a gap exists.
- A theme tag or two of your own, which is the column that eventually becomes your section headings.
003Read down the columns.
Once the matrix is populated, stop reading across rows and start reading down columns. The themes appear on their own. You will see that four studies cluster around one approach while three take another, that the disagreement in the literature tracks a difference in population rather than in method, or that everyone measuring a construct one way found an effect and everyone measuring it another way did not.
Those observations are your paragraphs, and each one is a claim supported by multiple sources rather than a summary of one. The section headings write themselves from the theme column, which is why that column is worth being deliberate about.
This is also where the gap for your own study comes from, and it comes honestly rather than by assertion. A gap identified from a matrix is defensible because you can point at the cells that are empty. A gap asserted because nobody has studied your exact population in your exact setting is the weakest form of the move and committees push back on it, correctly.
004Appraise rather than report.
Graduate rubrics, and every evidence-based practice course, ask you to judge the literature rather than relay it. That means saying whether a study is any good and whether it applies to your question, which are two separate judgments and both are usually required.
Quality first: sample size and how the sample was obtained, whether the design supports the causal language the authors used, what was measured and how well, and what the authors themselves flagged. Then applicability: whether the population, the setting and the intervention resemble yours closely enough for the finding to transfer.
The sentence that demonstrates appraisal usually contains a qualification. This study found an effect, but in a population unlike the one in question, so the finding is suggestive rather than directive. That single construction, repeated where warranted, is most of what separates a review that scores in the top band from one that does not. Where the review belongs to a thesis, it runs through the long-form desk as a stage rather than as a paper.
005Two practical failures worth naming.
Reviewing everything you found. A literature review is not an inventory of your search results, and including a source because you read it rather than because it bears on a theme dilutes the argument. Some of the reading will not appear in the review, and that is correct rather than wasteful.
Letting the review age. In a project running across terms, the review written two years ago is now missing whatever appeared since, and a committee will notice. Refreshing a search is a smaller job than it feels, particularly if the matrix still exists, which is one more reason to build it as a document rather than as a mental exercise. Where a project has stalled long enough for this to be a concern, it is usually one of several symptoms rather than the cause, which a stall diagnosis can separate out.
Questions.
What is the right number of sources for a review?
Follow your assignment where it states a number, and where it does not, the answer is however many the themes require. A review covering four themes with three or four well-used sources each is stronger than one listing thirty sources in sequence. Graders notice depth of use far more readily than count.
Can I organize chronologically instead?
Only when the development over time is itself the argument, which is occasionally true for a field that genuinely shifted. Chronological order chosen by default is just a catalog with dates attached, and it produces the same problem: one study per paragraph and no synthesis anywhere in the section.
What if the studies genuinely contradict each other?
That is excellent material rather than a problem, and handling it well is what top-band reviews do. Say plainly that the literature disagrees, then use your matrix to work out why: different populations, different measurement, different designs. An explained contradiction is a stronger contribution than a tidy consensus you had to force.
Does the matrix go into the paper?
Sometimes, as an appendix or a summary table, and some programs require one. Even where it does not appear, build it, because its value is in producing the structure rather than in being submitted. Keep it as a working document, since it is also what makes refreshing the review later a manageable task.
