What is the difference between a literature matrix and a synthesis matrix?

The terms overlap. A literature matrix, review matrix or summary table usually records descriptive facts about each study: purpose, design, sample, findings. A synthesis matrix is organized by your themes, so reading down a column shows what all sources say about one idea. Many people keep both in one spreadsheet: descriptive columns on the left, theme columns on the right.

Should sources go in rows or columns?

Either works. The original NC State handout puts sources across the top and themes down the side. With more than about eight sources, rows are more practical because spreadsheets scroll and sort better vertically. This guide uses one row per source.

How many themes should a synthesis matrix have?

Usually four to seven. Fewer than three and you are still summarizing. More than eight and most cells will be empty, which suggests some themes are really sub-points. Themes often end up as the subheadings of your review.

When should I choose the themes?

After reading, not before. Skim or read five to ten core sources, note recurring ideas, then set up columns. Expect to rename, merge or split columns as you read more. If you are testing a specific framework, its components can be your starting themes.

Can AI fill in a synthesis matrix for me?

Extraction tools can pre-fill descriptive columns such as sample size or design, and those entries are quick to check against the paper. The theme columns are different: deciding what a source contributes to your argument is the analytical work your review is graded on. If you use a tool for a first pass, verify every cell against the full text and follow your institution's rules on disclosing AI use.

How to Build a Literature Review Synthesis Matrix (With Template)

The most common criticism of student literature reviews is that they summarize one source after another: "Author A found... Author B found... Author C found..." Each paragraph is accurate and the review still fails, because nothing connects the sources. A synthesis matrix fixes the problem at the note-taking stage. It forces you to file what you read under ideas instead of under authors.

What a synthesis matrix is

A synthesis matrix is a table. In the layout used here:

each row is a source,
each column is a theme, question or variable that matters to your review,
each cell holds what that source says about that theme, in a sentence or two, with a page number.

Read across a row and you get one paper's contribution. Read down a column and you get the state of the literature on one idea: who agrees, who disagrees, who never addressed it. Paragraphs come from columns.

The technique has been taught in university writing centres for years. A widely copied handout was written by tutors at the NC State University Writing and Speaking Tutorial Service in fall 2006, and the Florida International University writing centre hosts a copy. Stanford's Program in Writing and Rhetoric and the Johns Hopkins libraries also teach a synthesis matrix. In health sciences a similar idea appears in Judith Garrard's Health Sciences Literature Review Made Easy: The Matrix Method (6th ed., 2022), where the "review matrix" is one part of a system whose other chapters cover planning and managing the search and selecting and organizing the documents.

Set it up in five steps

1. Read a first batch before you build anything

Pick five to ten central sources and read them properly. A quick way to triage which ones deserve that is the three-pass reading method. As you read, keep a running list of ideas that come up more than once: a concept everybody defines differently, a finding that keeps being replicated, a population that is always excluded.

2. Turn recurring ideas into columns

Group the list into four to seven themes. Word them as questions where you can, because a question tells you what to write in the cell. "Effect on retention" is vague. "Does the intervention improve retention after four weeks or more?" tells you exactly what to look for.

Add fixed descriptive columns at the left: full citation, design and sample, and context (country, setting, year of data collection). Add a final column for your own evaluation: limitations, how much weight you give the paper, and how it relates to your study.

3. Fill the cells, briefly and specifically

Write in your own words. Put anything copied in quotation marks immediately, with the page number, so that you cannot plagiarize by accident months later.
Be concrete. "Positive effect" is of little use. A cell that gives the outcome, the size of the difference, the time point and the page can be compared with another cell.
If a source says nothing on a theme, write "not addressed". An empty cell could mean you have not checked. "Not addressed" is a finding.
Keep cells to one to three sentences. Longer notes belong in your reference manager or annotated PDF.

4. Revise the columns as you go

After fifteen or twenty sources, look at the matrix as a whole. A column that is nearly empty is either a gap in the literature or a theme that does not matter. A column where every cell says the same thing may not need its own section. A column with long, mixed entries probably contains two themes. Split, merge and rename. This is analysis, and it is supposed to happen.

5. Sort and filter

In a spreadsheet, sort by year to see how a finding changed over time, by design to see whether the experiments and the surveys disagree, or by context to see whether results from one country dominate. Patterns that are invisible in a folder of PDFs are often obvious in a sorted table.

Template

Copy the block below into a plain text file, save it with a .csv extension and open it in Excel, Google Sheets or Numbers. Replace the theme headings with your own.

Citation (author, year),Design and sample,Context (country / setting / data year),Theme 1: [question],Theme 2: [question],Theme 3: [question],Theme 4: [question],Theme 5: [question],Limitations and quality notes,Relevance to my study,Pages / quotes to reuse
"","","","","","","","","","",""

The same structure as a table:

Citation	Design and sample	Context	Theme 1	Theme 2	Theme 3	Theme 4	Limitations	Relevance to my study
Author (year)	e.g. RCT, n = 240	Country, setting, year	What does this source say? (p.)	...	...	...	Design weaknesses, bias, funding	Supports, contradicts or motivates which part of my project?

If you prefer sources across the top and themes down the side, as in the NC State handout, transpose the table. The logic is identical.

A filled-in example

The four sources below are real, well-known papers on retrieval practice (learning by testing yourself), and each cell is based on the paper's published abstract. A real matrix would be filled from the full text and would carry page numbers. The review question is whether retrieval practice improves retention compared with restudying, and whether the effect holds outside the laboratory. Full references are at the end of this guide.

Source	Design and sample	Effect on delayed retention	Does it hold outside the laboratory?	Does it transfer to new questions?	What do students believe or do?
Roediger and Karpicke (2006)	Two laboratory experiments. Students studied prose passages, then took one or three immediate free-recall tests without feedback, or restudied the material the same number of times. Final test after 5 minutes, 2 days or 1 week	Restudying improved recall at 5 minutes. On the delayed tests, prior testing produced substantially greater retention than studying	Not addressed (laboratory)	Not addressed	Repeated studying increased students' confidence in their ability to remember the material, even though the delayed advantage lay with testing
Butler (2010)	Four laboratory experiments. Subjects studied prose passages, then repeatedly restudied or repeatedly took tests. Final test one week later	Repeated testing produced superior retention relative to repeated studying at one week	Not addressed (laboratory)	Yes. The benefit held for new inferential questions within the same knowledge domain and from different knowledge domains	Not addressed
Roediger, Agarwal, McDaniel and McDermott (2011)	Three experiments with sixth-grade students in a suburban middle school social studies course, using the course's own material and, in most cases, the exams students were graded on	Performance on chapter exams and semester exams improved following quizzing, relative to not being quizzed and to items presented twice	Yes. This is the classroom case: low-stakes multiple-choice quizzes in class, plus a web-based self-quizzing condition in Experiment 3	Not addressed	Not addressed
Karpicke, Butler and Roediger (2009)	Survey of 177 college students about how they study	Not measured	Not measured (survey of real study behaviour)	Not measured	A majority repeatedly read their notes or textbook. Relatively few engage in self-testing. The authors propose that many students experience "illusions of competence" while studying

Reading down the columns

Delayed retention: all three experimental papers agree that testing beats restudying on a delayed test. The first row adds a condition: on an immediate test, restudying wins. The design column also shows a limit. The longest laboratory delay in these abstracts is one week, and only the classroom study measures anything like a semester.
Outside the laboratory: one of the four. It is a genuine classroom study, but with sixth-graders rather than the undergraduates in the other rows, so it supports the effect without settling it for your population. That mismatch is exactly what the design column is for.
Transfer: one paper tests it and finds transfer. Nothing else addresses it, so this is a thin column, not yet a gap in the field.
Beliefs and behaviour: two papers line up. Restudying makes students feel more confident (Roediger and Karpicke), and when surveyed most students choose rereading over self-testing (Karpicke et al.).

Turning a column into a paragraph

A source-by-source version would read like this:

Roediger and Karpicke (2006) found that tested students retained more after one week. Butler (2010) found that repeated testing improved retention and transfer. Roediger et al. (2011) found that quizzing improved exam performance in a middle school classroom.

A synthesized version makes a claim about the literature and uses the sources as evidence:

Laboratory and classroom evidence agree that retrieval practice improves delayed retention more than restudying does. With prose passages, tested students outperformed restudying students on delayed tests, although restudying was better on an immediate test (Roediger & Karpicke, 2006), and the advantage extended to new inferential questions in the same and in different knowledge domains (Butler, 2010). The effect also survives the move out of the laboratory: quizzing improved sixth-graders' chapter and semester exam results in a social studies course, measured on the exams they were actually graded on (Roediger et al., 2011). Students do not seem to expect any of this. Repeated studying raised their confidence in their ability to remember (Roediger & Karpicke, 2006), and in a survey of 177 college students most reported rereading while few reported testing themselves (Karpicke et al., 2009). The classroom evidence here is from middle school, so whether it holds for undergraduates is not settled by these four papers.

The second version has a topic sentence about the field, groups sources by what they show, uses the design column to qualify a finding, and ends on a limit that tells you where to search next. Each of those moves came from the matrix. Note what the last sentence does not say. Four papers cannot establish a gap in the literature, only a gap in your matrix so far. For more on that step, see how to find research gaps.

Matrix or annotated bibliography?

An annotated bibliography stores notes by source. A matrix stores them by idea. They are compatible. Many students write annotations first and then distribute their content across matrix columns. If your assignment is an annotated bibliography that leads to a review, the annotated bibliography vs. literature review guide explains how the two connect.

Common mistakes

Choosing themes from the assignment brief before reading. You end up forcing sources into columns that do not fit them.
Pasting abstracts into cells. An abstract reports what the authors think is important. The cell should contain what matters for your theme.
Leaving out methods. Without design, sample and context you cannot explain why studies disagree.
Never revising columns. A matrix that looks the same at fifty sources as it did at ten has not been used for thinking.
One giant matrix for a whole dissertation. Build one per chapter or per research question.
Dropping page numbers. You will need them for quotations and for checking your own claims.

Tools

A spreadsheet is enough. Google Sheets or Excel handle sorting, filtering and freezing the header row. Reference managers such as Zotero can hold the detailed notes and export citation data to start the first column. Some researchers prefer a visual layout in which papers and highlights are dragged into theme groups. That is the same logic presented as a board instead of a grid.

In Wonders, the organize board works like a visual synthesis matrix. You group saved papers, highlights and notes under themes, and each item stays linked to its source, so the paragraph you write from a theme can be traced back to the pages it came from.

References

Butler, A. C. (2010). Repeated testing produces superior transfer of learning relative to repeated studying. Journal of Experimental Psychology: Learning, Memory, and Cognition, 36(5), 1118–1133. https://doi.org/10.1037/a0019902
Garrard, J. (2022). Health sciences literature review made easy: The matrix method (6th ed.). Burlington, MA: Jones & Bartlett Learning. ISBN 9781284211177.
Karpicke, J. D., Butler, A. C., & Roediger, H. L., III. (2009). Metacognitive strategies in student learning: Do students practise retrieval when they study on their own? Memory, 17(4), 471–479. https://doi.org/10.1080/09658210802647009
NC State University Writing and Speaking Tutorial Service. (2006). Writing a literature review and using a synthesis matrix. Reproduced by the Florida International University Writing Center: https://case.fiu.edu/writingcenter/online-resources/_assets/synthesis-matrix-2.pdf
Roediger, H. L., III, Agarwal, P. K., McDaniel, M. A., & McDermott, K. B. (2011). Test-enhanced learning in the classroom: Long-term improvements from quizzing. Journal of Experimental Psychology: Applied, 17(4), 382–395. https://doi.org/10.1037/a0026252
Roediger, H. L., III, & Karpicke, J. D. (2006). Test-enhanced learning: Taking memory tests improves long-term retention. Psychological Science, 17(3), 249–255. https://doi.org/10.1111/j.1467-9280.2006.01693.x
Stanford University, Program in Writing and Rhetoric. Synthesis matrix (teaching activity). https://teachingwriting.stanford.edu/synthesis-matrix
Johns Hopkins University Sheridan Libraries. Write a literature review: synthesize. https://guides.library.jhu.edu/lit-review/synthesize
University of North Carolina Writing Center. Literature reviews. https://writingcenter.unc.edu/tips-and-tools/literature-reviews/