Questions to Ask When Writing a Literature Review
For undergraduate and graduate students with a literature review to write, as a course paper, a thesis chapter or a piece that stands on its own. Some of the questions are put to one study at a time and others to the whole pile, in six groups that run from setting the scope and searching, through reading each study and comparing across studies, to naming the gap and writing it up. The two middle groups are answered in a spreadsheet with a row for every study, which the guide on this page shows how to set up.
Want questions from the whole vault instead? Try the random question generator.
The questions
Each question, and why to ask it
Scope
What question is this literature review supposed to answer?
Why ask it
A review built around a question has a test for what goes in, and a review built around a topic has none. Write the question as one sentence and keep it where you can see it, because every later decision to keep or drop a source gets checked against it.
Is this a stand-alone review, or the opening chapter of a thesis or a study of my own?
Why ask it
A stand-alone review ends with a verdict on where the field stands. One that sits in front of your own study has a second job, which is to make that study look necessary, so its final paragraph has to arrive at your research question.
What kind of review is expected: narrative, systematic, scoping or something else?
Why ask it
The labels carry different obligations. A systematic review needs a documented search and inclusion rules stated in advance, where a narrative review leaves the selection to your judgment. Ask your supervisor or instructor which they mean, since the names are used loosely and departments differ.
Where do I draw the borders: which years, populations, settings and types of study count?
Why ask it
Write each limit down with its reason before the reading piles up. "Studies from 2013 on, because the diagnostic manual was revised that year" is a border you can defend in a sentence of the review. A bare "since 2013" is the first thing a reader will query.
Which terms name my central ideas, and do the authors I am reading mean the same thing by them?
Why ask it
Fields rename things, and one word can cover several. "Burnout" may be scored as exhaustion in one paper and as intention to quit in another. Fix your own definition early and flag each source that departs from it, because a clash in findings later may trace back to this.
How long should the review be, and roughly how many sources do reviews like it cite in my program?
Why ask it
No number holds everywhere, so ask, and then look at two or three theses or course papers from your own program that passed. Their reference lists and section lengths are a better guide than any rule of thumb from outside it.
Who will read the review, and how much of the field can I take for granted with them?
Why ask it
A committee from your own subfield needs no tour of the basics and will notice a missing name at once. A reader from the department next door needs the central terms defined on the first page. Your answer decides how much background the review carries.
Searching
Which databases and search strings have I tried, and did I write them down with the date?
Why ask it
Keep a log: database, exact string, date, number of results. A systematic review has to report this, and in any other kind it rescues you when a supervisor asks how you found your sources or when the search needs rerunning six months on.
Which two or three studies does everyone in this area treat as the starting point?
Why ask it
You will meet them in the opening paragraph of other people's introductions. Read each one in full, because the one-line summaries passed from paper to paper drift away from what the original reported.
Who has cited this key study since it came out, and what did they do with it?
Why ask it
The "cited by" link in a database or in Google Scholar runs the search forward in time. It tells you whether later work confirmed the finding, stretched it to new groups or took it apart, and it is the quickest route to the most recent papers.
Have I read the whole paper, or am I about to cite it from the abstract?
Why ask it
An abstract leaves out the sample, the measures and the limits, which are the things a review compares. If the full text sits behind a paywall, ask your library what it can get for you, since loan services and subscriptions differ from one institution to the next, and check whether the authors have posted a version of their own.
Is there a recent review article on this subject, and what will mine do that it does not?
Why ask it
A published review is both a gift and a problem. Mine its reference list and study how it is organized, then name your difference: newer studies, another population, a narrower question. If you cannot name one, shift the question until you can.
Which kinds of source am I leaving out: dissertations, reports, preprints, work in other languages, neighboring disciplines?
Why ask it
Any of these can reasonably stay out, but each should be a decision you could state in the review, and whether preprints or other unreviewed work may be cited at all is for your instructor to say. The neighboring discipline is the one to test, since the same problem often goes by another name there. Twenty minutes of searching with their vocabulary will tell you.
Have I looked on purpose for studies that found no effect, or the opposite one?
Why ask it
They can be harder to turn up, because a striking result travels further than a flat one. Add words like "no effect", "null" or "failed to replicate" to a search and see what comes back. A review holding only confirming studies reads as a case for one side.
How will I know when I have read enough?
Why ask it
The signal is repetition: the reference list of each new paper is mostly things you already hold. Set a date on which searching stops and drafting starts, and leave an alert on your saved searches to catch anything published afterward.
Each study
What will I record about every study, and where will I keep it?
Why ask it
The usual answer is a synthesis matrix: a spreadsheet with a row for each study and the same columns for all of them, such as question, design, sample, measures, findings, limits and relevance. Loose notes hold the same facts but cannot be read down a column, which is how patterns show.
Is this a report of original research, a theory paper, a review or a commentary?
Why ask it
Only the first kind adds a finding to your matrix. A review or a commentary can frame a section and lead you to studies, but citing one as though it had tested something inflates the evidence, so give the kind of source a column of its own.
What was this study trying to find out?
Why ask it
It tends to sit in the last lines of the introduction, sometimes as numbered hypotheses. Put it in your matrix in your own words. Two studies that look as if they contradict each other often turn out to have asked different things.
What design did the authors choose, and what kinds of claim can that design support?
Why ask it
An experiment with random assignment can speak about cause, a one-time survey about association, an interview study about how people understand their situation. Give design its own column, since it sets how much a finding can carry when you weigh studies against each other.
Who or what was studied, how were they chosen, and how many were there?
Why ask it
Note the number, the setting and the route by which participants or cases got in. When findings part ways later, samples are the first place to look: first-year students at one campus and employees across several firms are not the same evidence.
How did the authors measure the main thing they were studying?
Why ask it
A questionnaire people fill in about themselves, an observer's rating, a test score and an administrative record all capture something different. Write down the name of the instrument. Two studies of "engagement" built on different instruments may not be about the same thing, and saying so is a sentence for your review.
What did the study find, in its own numbers or its main themes?
Why ask it
Take this from the results section and the tables, not from the abstract. For quantitative work record direction and size, and for qualitative work the named themes, so that later you are comparing findings and not the adjectives authors chose for them.
Do the conclusions claim more than the results can carry?
Why ask it
Set the final paragraph of the discussion against the results tables. A correlation written up as an effect, or forty people in one town stretched to cover everyone, earns a qualifying clause when you cite it: "although the design cannot separate cause from selection".
Which limitations do the authors admit, and which have they left unsaid?
Why ask it
The admitted ones are near the end of the discussion. The others are yours to spot: heavy dropout, no comparison group, a single person doing all the coding. When one weakness recurs across many rows of your matrix, it stops being a note on a paper and becomes a paragraph about the field.
Who paid for the study, and do the authors declare a stake in the result?
Why ask it
Look for the funding and competing-interests statements, often printed near the end of the paper or on its first page. A sponsor does not make a finding wrong. It earns a sentence in your review when the favorable studies of a program or a product turn out to share one.
What theory or framework is the study working from?
Why ask it
The introduction names it, or the first citations give it away. Authors working from different frameworks pick different variables and can pass each other without contact. Record it now, since sorting sources by framework is one of the ways the finished review can be organized.
When and where was the study carried out, and does that setting still hold?
Why ask it
Data gathered under an older policy, an earlier technology or another country's school or health system can still belong in the review, with a phrase saying so. Check the year the data were collected, which can be well before the year of publication.
What does this study do for my question: support, complicate, give background or lend a method?
Why ask it
Put one word in the relevance column for every source. A study you cannot label probably does not belong, however much you enjoyed it. Those labels are also your first rough sort into sections.
Could I state what this study contributed in two sentences without opening it?
Why ask it
Try it from memory once your notes on the paper are finished. If you cannot, you do not yet know it well enough to cite it. If you can, those two sentences are close to what will appear in the review.
Across studies
Where do the studies agree, and how many independent teams stand behind that agreement?
Why ask it
Count research groups and datasets, not papers. Five articles drawn from one lab's single dataset are nearer to one piece of evidence than to five, and a phrase like "consistently found" should rest on the smaller number.
Where do the findings conflict, and what could account for the difference?
Why ask it
Work down the matrix columns in order: definitions, measures, samples, designs, dates. One of them often explains the split, and saying which is the heart of synthesis. If none does, report the disagreement as unresolved, which is a finding in its own right.
Which studies are the strongest, and does my review give them more weight than the weak ones?
Why ask it
One sentence each for a long-running cohort study and a convenience survey of a single class tells the reader they count the same. Weight shows in the space you give a study and in your verbs: "demonstrated" for one, "suggested" for the other.
How has thinking on this question changed over time?
Why ask it
Sort the matrix by year and read down the findings. You are looking for a turn: a new method, a critique, a paper after which people framed the problem differently. A real turn can be the spine of a chronological section. Without one, dates are a poor way to organize.
Which methods dominate this literature, and what does that leave unexamined?
Why ask it
Tally the design column. If nearly every row is a one-time survey, the field knows a good deal about what goes with what and little about what changes first. That observation is fair criticism, and it often points straight at a gap.
Who has been studied again and again, and who hardly at all?
Why ask it
Age, country, setting and occupation are the things to read for in the sample column. When one kind of participant fills it, say students at a handful of large universities, nobody can tell how far the findings travel, and the group that is absent may be the reason your own study is needed.
Which camps or schools of thought are there, and who is answering whom?
Why ask it
Watch who cites whom and, just as telling, who never does. Once you can name the camps you can organize a section around their dispute, and the reader sees an exchange between positions instead of a stack of separate results.
What themes appear when I sort the findings by what they say instead of by who said it?
Why ask it
Copy the findings column onto cards or into a fresh sheet with the author names hidden, then group them. Ending up with a dozen piles means some are sub-points of others. Ending up with one means the sorting was too coarse.
What is broadly accepted here, what is contested, and what is simply unknown?
Why ask it
Make three lists. They often give you the arc of the whole review: settled ground covered briefly, contested ground at length, and the unknown held for the end, where the gap goes.
What do most of these authors assume without testing it?
Why ask it
It might be that influence runs in one direction only, or that the standard measure captures the concept. This one comes late, after a lot of reading, and a single careful sentence about it does more to separate a review from a summary than another ten citations.
The gap
What has nobody studied yet, and is that because it is hard, unimportant or answered somewhere else?
Why ask it
An empty space is only a gap if filling it would matter. Search the neighboring fields before you claim it, then write the sentence that says what the field cannot explain or decide while the space stays empty.
Is the gap a missing population, a missing method, a missing variable or a conflict nobody has settled?
Why ask it
Naming the type makes the claim exact. "No study has followed these students past their first year" can be checked by a reader. "More research is needed" cannot, and it appears at the end of so many papers that it carries no information.
What do the authors themselves call for in their closing paragraphs on future research?
Why ask it
Collect those lines in a column of their own. When several unconnected papers ask for the same thing, you can cite them as evidence that the field sees the gap too. Then run a forward search to check that nobody has answered the call since.
Can I back up a claim that something has not been done?
Why ask it
Tie it to your search log and word it with care: "in the studies reviewed here" or "to my knowledge". An examiner who knows of a paper you missed will mention it, and a bounded claim survives that where "no research exists" does not.
What would a skeptical reader say is already known, making my gap smaller than I claim?
Why ask it
Ask your supervisor to play that reader for ten minutes. If the objection lands, restate the gap more narrowly. A small gap that holds is a firmer footing for a study than a wide one that gives way under the first hard question from an examiner.
How does my own study or argument fit the space the literature leaves?
Why ask it
In a thesis or proposal the final section should narrow toward this, so that a reader could guess your research question before you state it. In a stand-alone review, replace it with what the next studies in the field ought to do.
Writing it
Should I organize the review by theme, by chronology, by method or by theory?
Why ask it
Theme is a common default. Chronology suits a field with a clear turning point, method suits one where the design drives the result, and theory suits rival explanations. Sketch headings for two of the options and keep the one in which fewer sources have to appear in three places.
What does each section claim about the literature, in one sentence?
Why ask it
"Parental involvement" is a topic. "Involvement at home predicts more than involvement at school" is a claim a section can be built to support. Write one such sentence under every heading before drafting, since a section without one turns into a list.
Does each paragraph open with a point about the literature, or with an author's name?
Why ask it
Scan the first three words of every paragraph. A run of "Smith (2019) found", "Jones (2021) found" is an annotated bibliography with the line breaks removed. Rewrite so the idea is the subject of the sentence and the authors move into the parentheses.
Where in each paragraph have I brought two or more sources together?
Why ask it
Synthesis is visible at the level of the sentence, in words such as "whereas", "by contrast" and "consistent with". A paragraph given over to one source should be rare and kept for the landmark studies, so mark every paragraph that cites a single paper and ask what it is being compared with.
How much space has each source earned?
Why ask it
A landmark study, or one very close to your question, gets a paragraph with its sample and method. Supporting studies can share a sentence and a grouped citation. Equal space for everything hides from the reader which sources matter.
After each cluster of citations, have I said in my own voice what it adds up to?
Why ask it
Add a sentence of yours at the end of the cluster: what those sources establish together and how firmly. Leave it out and the reader has to do the synthesis, which is the part of the work a grader is looking for.
Have I gone back to the originals to check that each one says what my sentence says it does?
Why ask it
Notes drift, and a claim picked up from another author's summary drifts further. Check the ten citations the review leans on hardest. For a source you could only reach through someone else's account, your style guide has a form for secondary citation, so look it up and use it.
Does the review end by saying where the field stands and what comes next?
Why ask it
The closing paragraph gathers what is established, what is disputed and what is missing, and in a thesis it hands over to your question. Test it by asking whether it could be pasted onto the end of a different review. If it could, it is too general.
What will I do about relevant studies that appear after the review is drafted?
Why ask it
A thesis review written in the first year can be well behind by submission. Put a date in the calendar a few weeks before the deadline to rerun the logged searches, and work what turns up into the sections where it belongs, not into a paragraph bolted on the end.
From a folder of PDFs to a synthesis
Practical guidance for the conversation itself
Set up the matrix first
One row for each study
Open a spreadsheet before you open the second paper. Each study gets a row, and the columns are the questions in the Each study group: kind of source, research question, design, sample, measures, findings, stated limits, funding, framework, and what the study does for your own question. Add a column for the page numbers of anything you might quote.
Fill it in while the paper is open
Complete the row as soon as you finish reading, in your own words, with the paper still on screen. Rows filled in from memory a week later come out thinner, and the findings cell tends to turn into the wording of the abstract.
Keep the cells short
A cell is a phrase, not a paragraph: "survey, one time point", "312 nurses, two hospitals". Short cells are what make a column readable from top to bottom, and reading down the columns is where the Across studies questions get answered.
Add columns when a pattern appears
Partway through you will notice something the starting columns do not hold, such as which country the data came from or how long the participants were followed. Add the column and go back through the earlier rows. That second pass is tedious, and it is often where a comparison nobody else has drawn turns up.
Matching the groups to the stage you are at
Before the first search
Answer the Scope group on one page and show that page to whoever will grade or supervise the review. A wrong assumption about the type of review or its borders is cheap to correct at this point and expensive after forty papers.
While the pile is growing
Alternate between Searching and Each study. Read a batch, fill in the rows, then let what you learned change the next search: a new term, a landmark paper to chase forward, a discipline you had not thought to check.
When new papers stop surprising you
Move to Across studies and The gap, and close the PDFs. These are answered from the matrix. Give them an unbroken afternoon, because seeing a pattern across thirty rows needs all thirty in your head at once.
With an outline, then with a draft
Use the first two questions in Writing it on the outline, before any paragraphs exist. Save the others for a complete draft, and go through them with a pen on a printed copy, where a run of paragraphs starting with surnames is easy to see.
If the deadline is close
Do not skip the matrix, but shrink it to four columns: question, design and sample, findings, relevance. Then answer only the agreement, conflict and gap questions. That is the minimum that turns a set of summaries into a review.
Where a review slides back into summary
A paragraph per paper
This is what comes out when you draft from your notes on each source in turn. Draft from the themes instead: write the claim for the paragraph, then pull in every row of the matrix that bears on it.
Quotation doing the work
Long quotations from sources are a sign that the comparing has not happened yet. Many fields expect a literature review to paraphrase nearly throughout and to quote only a definition or a phrase under dispute. Look at published reviews in your own field to see what is normal there.
Reporting without judging
Listing what each study found, with no word on how well it was found, leaves out the review part of the review. One clause on design or sample beside the finding is usually enough: "in a small sample of volunteers", "across three national cohorts".
Criticizing everything
The opposite failure is a review that finds fault with every source and credits none. Say what the body of work has established before you say what it has missed. The gap you are claiming only makes sense against something solid.
A gap announced in the last line
If the missing piece first appears in the closing sentence, the reader has no reason to believe in it. Let each section end by noting what its part of the literature leaves open, so that the final statement of the gap collects points already made.
Who to ask, and what to bring
Your supervisor or instructor
Take the type of review, the length, the borders and the gap. Bring the one-page scope and later the list of section claims, since both can be read in five minutes. Ask too which authors would be conspicuous by their absence.
A subject librarian
Take the database and search string questions. Which databases you can reach depends on what your library subscribes to, and a librarian for your subject knows the controlled vocabulary each one uses. Bring the search log so they can see what you have already tried.
A classmate from another topic
Hand over the draft and ask only two things: what do I think the literature says, and what is missing from it? If they cannot answer both after one reading, your own voice is not yet on the page, whatever the citations are doing.