Write the search down while you run it
A review is a method, and methods are reproducible. Keep a log from the first search onward, because reconstructing one later is far harder than recording it as you go.
In a proposal this log becomes the search strategy section. At a defense it is the answer to the question about how you know the review is complete.
- Every database searched, with the date you searched it.
- The exact terms and combinations, including the ones that returned nothing.
- The filters applied: years, language, peer review, population.
- How many results each search returned and how many you kept.
- Why anything was excluded, written as one rule rather than as a note per study.
Group by what the studies disagree about
Most weak reviews are organized by study. Each paragraph summarizes one paper and the reader assembles the meaning, which is the work you were supposed to do. Strong reviews are organized by finding, and the most productive organizing principle available is disagreement.
So read for the conflicts. Two studies asking the same question and reaching opposite results are the most valuable pages in your review, because explaining that conflict is synthesis in its purest form. Look at the samples, the measures, the settings and the designs. The explanation is nearly always in one of those four.
Read for design, then for size
A finding is worth what its design can support. For every study, note what kind of evidence it is, how many participants it had, what it measured and how, and what it cannot claim. Two lines per study in a table is enough, and it makes the writing almost mechanical later.
Sample size deserves its own column. A small study agreeing with a large one adds little. A small study disagreeing with a large one is either an important boundary condition or an artifact, and deciding which is exactly the judgment a review exists to make.
Keep that table in the same file as the log. A reviewer asking why one study carries more weight than another is asking a question you already answered while reading, and the table is where the answer is sitting.
The paragraph that carries a claim
One shape, repeated. It reads as authority and it is mostly mechanical.
A paragraph built this way never opens with an author's name, and that single habit repairs more reviews than any other edit.
- Open with the claim the paragraph is making, in your own words.
- Support it with the studies that agree, cited together rather than one per sentence.
- Name the exception, and what distinguishes it from the rest.
- Close with what this means for the question you are asking.
The gap has to be argued
The end of a review earns the study that follows it. A gap is not announced by saying that little research exists, which is a claim a reviewer will test in one search. It is argued: this is established, this is where the evidence conflicts, this population or setting has not been examined, and therefore this question is worth asking.
Reviews come back for two reasons above all others: sources that are current but thin, and a gap the review itself does not support. Both are visible before submission to somebody who has read a few hundred of these, which is the kind of reading that goes into chapter work.
Questions
Where does the review stop and the methods chapter begin?
The review establishes what is known and what is not. The methods chapter says what you intend to do about it. If a paragraph has started describing your own procedures, it has crossed the line. The bridge between them is the gap paragraph, which should read as though the design that follows were the only sensible response to it.
How do I handle a study that contradicts my premise?
Include it and explain it. A review omitting an inconvenient finding is the easiest thing for a committee member to catch, and it costs credibility the rest of the chapter cannot recover. Compare its sample, measure and setting against the studies it contradicts. Most contradictions resolve into a boundary condition, which is often the most interesting sentence in the chapter.
Can I include a study whose full text I cannot get?
Not on the strength of an abstract. Abstracts omit the sample details, measures and limitations that a review is judged on. Request it through your library rather than settling for the summary. Where it genuinely cannot be obtained, leave it out and note the exclusion in your log rather than citing something you have not read.
How do I keep a review from becoming a list?
Check the first four words of every paragraph. If most of them begin with an author's name or a year, the chapter is organized by study rather than by finding. Rewrite each opening as a claim and move the citations behind it. That single pass converts most of a summary chapter into a synthesis without changing the underlying research at all.
How do I know when the search is finished?
When new searches keep returning studies you already hold. That point has a name, saturation, and it is a defensible answer at a defense provided the log shows the searches that produced it. Chasing completeness without recording anything produces the same feeling with no evidence behind it, which is why the log begins with the first search.
