How to build a literature review comparison table

A literature review is not a list of summaries. It is an argument about what a body of research shows, where it agrees, and where the gaps are. That is hard to see when each paper lives in its own PDF. A table makes it visible.

This guide shows how to set up the table, how to read a paper to fill it, a worked example with two real papers, and how to go from the table to writing.

What a comparison table is

A comparison table is a grid. In the most common layout, each row is one paper and each column is one thing you want to know about every paper: how the study was done, who took part, what it found, what its weaknesses are.

One well-known version comes from the health sciences: Judith Garrard's book Health Sciences Literature Review Made Easy calls it the Matrix Method. Many university libraries teach a version of it. Duquesne University's library guide, for example, suggests every review matrix start with the same three columns: the authors, title and journal; the publication year; and the purpose of the paper (Duquesne University Library).

Two ways to lay it out

Scroll sideways for more columns →

Central Michigan University's library guide describes the theme layout as a synthesis matrix: main ideas down the side, sources across the top, and notes in each cell on how a source relates to that idea (Central Michigan University Libraries).

A practical approach: build a paper matrix while you read, then turn it into a theme matrix when you start to write.

Choose your columns

Start small. Five columns cover most reviews. Add more only if every paper can fill them and your review needs them.

Useful extra columns, depending on your topic:

• Research question or purpose. What the authors set out to find.
• Measures. How the outcome was measured: a test, a scale, a survey question.
• Setting. Country, type of school or hospital, lab or real world.
• Theory or framework. If your field cares about it.
• Your notes. How the paper relates to your own question. Keep this column separate, so your opinion never mixes with what the paper says.

How to read a paper for the table

You don't need to read every paper from start to finish before it goes in the table. Read in an order that fills the columns fast, then go deeper on the papers that matter most.

1. Abstract. Fill in a rough version of every column. You will correct it.
2. Methods. Fix the methods and sample cells. Look for the design, the number of participants, how they were chosen, and what was measured.
3. Results. Fix the findings cell. Go to the tables and figures, not just the text. Note the key numbers and whether a difference was found or not.
4. Discussion. Look for the limitations paragraph. Note what the authors say should be studied next. That is often where the gaps in the field are.
5. Conclusion. Check that your findings cell matches what the authors claim, and note if they claim more than their results show.

Three habits that save a lot of pain later:

• Write a page number in every cell. When you write the review, and when your supervisor asks "where does it say that?", you can find it in seconds.
• Write "Not stated" when a paper doesn't say. An empty cell could mean you missed it. "Not stated" means you looked. A paper that doesn't report its sample size is a finding in itself.
• Separate the authors' limitations from yours. The limitations column is what the paper admits. Your own criticism goes in your notes column, marked as yours.

A worked example

Here is a small table for two open-access papers about retrieval practice: the study method of testing yourself instead of rereading. (Our guide on practice questions and active recall covers the method itself.)

Scroll sideways for more columns →

Now read down the columns, not across the rows:

• Where they agree. Both use experiments with random or within-person comparisons of recall practice against restudying.
• Where they differ. The findings point in different directions. But so do almost all the other columns: children versus university students, word lists versus maths procedures, memory tests versus problem-solving tests.
• The gap. Two papers can't tell you whether the difference comes from age, from the type of material, or from the test. A fair sentence for a review would be: "Retrieval practice improved children's memory for words, but did not improve university students' solving of maths word problems; the studies differ in age, material and outcome, so the reason is unclear."

That last sentence is synthesis. It says what two papers show together, which neither says alone. This is what the table is for.

From table to writing

When you start writing, organise by theme, not by paper. A review that goes "Smith found... Jones found... Lee found..." is a list of summaries. Instead:

1. Read down each column and note patterns. Which findings agree? Which methods are common? Which samples are missing (an age group, a country, a type of school)?
2. Group papers into themes. "Studies with children", "studies of procedural skills", "classroom versus lab studies". These become your sections.
3. Write each paragraph around a claim, then cite every paper that supports or challenges it: "Several experiments with children found..., although..."
4. End each section with the gap. What do the studies leave unanswered? Your own project often lives there.

Some sentence starters that help:

• "Both X and Y found..., but they used different..."
• "Most studies in this area used university students; few..."
• "These results may not apply to..., because..."
• "No study in this review measured..."

Tools

A spreadsheet works well: one row per paper, frozen header row, a column for the PDF's file name. Reference managers such as Zotero let you attach notes to each paper, but a table is still the easiest way to compare across papers. Whatever you use, keep the page numbers.

Doing it with FileAI

FileAI is a tool that reads your files and writes from them. Upload two or more papers as PDFs with selectable text, select them together, and either pick the suggestion "Compare these papers' methods, data and findings" or type your own request, such as "Compare the methods, data, findings and limitations of these papers in a table." You get:

• At a glance: two or three sentences on how the papers relate and the main difference.
• A comparison table with one row per paper and five columns: Document, Methods, Data or sample, Findings, Limitations. Where a paper doesn't say, the cell reads "Not stated".
• A citation in every cell. Each one names the paper and the page. Click it and the page opens with the passage highlighted.
• Where they agree, where they differ, and gaps: short lists, each point citing the papers it mentions.

FileAI compares only what the papers say. It doesn't judge which paper is better unless the papers themselves do.

What it can't do. It can be wrong, so open the cited page for every cell you use and fix the cell if the paper says something different. It doesn't read scanned PDFs yet. It doesn't write your references: you cite the papers yourself, in the style your course asks for. And it can't do the synthesis for you. The table is the first draft; the argument is yours.

Your language. Type your request in the language you know best and the table comes back in it, even if the papers are in English. Quoted text stays as the paper wrote it.

The free plan compares up to two files at a time, with no card needed. Pro compares more papers at once and adds more files and questions each month. You can see a real comparison of the two papers above, with a page link behind every cell, on the compare research papers page, and try the numbered sources on a sample document in the demo.

Sources

• Duquesne University Library, Matrix method for literature review: organize your sources
• Central Michigan University Libraries, How to conduct a literature review: synthesize
• Garrard, J. Health Sciences Literature Review Made Easy: The Matrix Method. Jones & Bartlett Learning (several editions).
• Karpicke, J. D., Blunt, J. R., & Smith, M. A. (2016). Retrieval-based learning: Positive effects of retrieval practice in elementary school children. Frontiers in Psychology, 7, 350.
• Huang, X., Zheng, S., Yu, Z., & Chen, S. (2023). Retrieval practice may not benefit mathematical word-problem solving. Frontiers in Psychology, 14, 1093653.