How to build a literature matrix: from notes to synthesis

When can you use it?

This guide is for authors drafting a thesis literature chapter, research proposal or an article’s theoretical background. You can begin after reading and annotating a few relevant sources, before the first draft. Stanford uses the matrix in a classroom activity built around sources students have already read. This is a teaching sequence, not a universal minimum source count. [Stanford] [GMU: writing]

Treat the matrix as a working document. A newly encountered concept may send you back to earlier rows; a methodological difference may become relevant later. If your question changes, review which notes still help answer it. This flexible workflow is our editorial recommendation.

Write the question above the table

“University students and technology” names a topic but may be too broad for a useful comparison. “What conditions influence university students’ participation in online writing support?” gives the table a more focused starting question. Write the question, context and any date limits above the table.

Keep a separate search log: databases, terms, limits and reasons for selecting sources. It documents how you searched; the matrix compares what you found. VCU distinguishes search documentation from recording relationships between sources. Adapt the record to the kind of review you are writing. [VCU Libraries]

An adaptable five-column starting template

Copy the table into a word processor or spreadsheet. These five columns are an AcademicFix starting suggestion, not a compulsory format or a validated measurement instrument. VCU explicitly permits adapting columns; the NC State handout demonstrates a different, source-by-theme arrangement. [VCU Libraries] [NC State]

Download the blank five-column CSV template. The file contains only column headings and blank rows. It excludes the fictional example.

Source: Use a short identifier that leads back to the complete bibliographic record. A file link or reference-manager key can help distinguish similar author names and publication years.

Question / context / method: Go beyond “quantitative” or “qualitative.” Where relevant to your comparison, note the participants, data collection and period studied. Stanford’s activity draws attention to methods as well as geographic and temporal differences. [Stanford]

Relevant finding and location: Select information that addresses your question rather than copying the whole abstract. Record the page; for an unpaginated text, use a section heading, table number or another retrievable location. This is a practical note-taking recommendation, not a claim that every source has page numbers.

Relationship to other sources: Replace “similar” with the point of similarity: the same barrier, explanation or result under different conditions? If the connection remains uncertain, write “comparison pending.” Do not manufacture consensus or disagreement merely to fill a cell.

Limits / role in the argument: Separate an author’s stated limitation from your interpretation. “Author’s limitation: one institution” and “my use: I will not generalise to other institutions” are different notes. Consider how much support the source can provide for your proposed sentence.

Fictional example: read three sources together

A, B and C below are entirely fictional teaching examples. They are not real publications, data or findings; their page numbers are also invented. Do not put them in a reference list. Suppose the question concerns how written feedback for university students is evaluated.

A source-by-source summary might say: “A interviewed students. B compared drafts. C surveyed preferences.” Readers learn about the sources but must work out their relationship themselves.

A thematic synthesis could instead say: “Evaluating written feedback requires distinguishing student experience, changes in the text and service expectations. A’s interviews concern perceived usefulness; B’s comparison concerns short-term changes in drafts; C’s survey concerns a preference for faster responses. A and C do not establish improvement in the text. B’s observed change does not by itself establish a lasting effect or a causal relationship. The sources contribute different evaluation dimensions.” This paragraph refers only to the invented scenario.

Synthesis here preserves distinctions. It identifies a relationship and states which conclusions the fictional material does not establish. It does not force different kinds of evidence into one verdict.

Turn the matrix into a paragraph plan

Read across the relationships and assign provisional themes: in this example, “perceived usefulness,” “changes in the text” and “response-time expectations.” Write one defensible point under each heading. GMU describes selecting, arranging, grouping and ordering material, then turning theme labels into assertions; UNC also recommends organising around connecting ideas. [GMU: organising] [UNC Writing Center]

Open the paragraph with the point you want to make, then compare the relevant sources. The same source may appear under more than one theme when it supports different points. GMU discusses both practices; UNC explains synthesis within paragraphs. [GMU: writing] [UNC Writing Center]

Inspect empty cells separately. “This source does not address the issue,” “I have not checked yet” and “the information is unclear” are different states. A blank in your own table does not establish a research gap across an entire field. This is an editorial caution about the limits of your notes.

Separate quotation, paraphrase and interpretation

Mark direct quotations with quotation marks while taking notes. Keep the citation when you express a borrowed idea in your own words. Check that paraphrasing preserves qualifications and the group to which the claim applies. Both GMU and UNC discuss distinguishing quoted or paraphrased material and representing sources accurately. [GMU: writing] [UNC Writing Center]

Label a note based only on an abstract “abstract only.” Do not add detailed methods or limitations without checking the full text. Recheck any sentence you move into the draft against its source, especially numbers, direction, measurement and location. Label your own evaluations “my interpretation.” These are our editorial checks.

This matrix supports reading and drafting. It does not, on its own, document a systematic review’s search coverage, determine eligibility or assess risk of bias. Filling the table cannot establish that all relevant research was found or that its reliability was evaluated. Use the separate methods and appraisal procedures appropriate to your review design.

Three practical questions

Should each article have one row?

One row per article is a useful starting arrangement. If a source contributes to several questions, you can add rows while retaining its source identifier. Keep the information you need to compare.

How many sources do I need before starting?

There is no fixed source-count threshold in this workflow. Begin with a few relevant sources and develop the table as you read. The assignment, thesis or research purpose determines the required coverage; a classroom activity’s source count is not a universal rule.

Must the matrix appear as a table in my thesis?

Not necessarily. The matrix is primarily a thinking and drafting tool. If you include it in your thesis or article, consider what it adds for readers and check institutional or journal requirements.