A thesis literature review is read over a year or more and rests, by the end, on several hundred sources. Nobody holds that many in their head, and the ones that go wrong were read in March and cited in November.
This guide, published by EdCitation, is about managing those sources and nothing else. Finding them has its own guide; the rest of the dissertation, from proposal to examiners, is in managing dissertation references from proposal to submission. The method borrows from systematic reviewers, whose handbooks say what to record, how to count and what a duplicate is.
What goes wrong at several hundred sources?
Management errors, not reading errors. The table gives the failures that appear once a list passes a few hundred entries and the habit that stops each.
| Failure | How it shows up | The habit that prevents it |
|---|---|---|
| Read but not recorded | A claim with no page number and no memory of its paper | A matrix row on the day of reading |
| Filed in one place | A paper on two themes found in neither | A tag per theme and per chapter |
| Entered twice | Two spellings of one title, or a preprint and its published version both cited | A duplicate check by identifier at each chapter's end |
| Retraction unnoticed | A withdrawn paper cited as evidence | A retraction screen when cited and again on the whole list |
How do I keep a reading log or synthesis matrix?
One row per source, written on the day it is read, with the same columns in every row. Library guides call it a synthesis matrix; a reading log is the same table in a spreadsheet.
What the library guides say, and where they differ
Utah State University Libraries (2026) puts the paper's main ideas down the side and the sources across the top, each box holding "a paraphrase or direct quote" of what that source says on that idea, with a last column for your own analysis; a main idea with fewer than two sources under it is a gap to fill. Williams College Libraries (2026) draws it the other way round, sources in rows and themes in columns.
The row a thesis needs
At several hundred sources, a source per row is the only layout that fits, and five columns cover it: the identifier (a DOI, PubMed ID or ISBN, never a typed title); the claim you would cite it for; the method and sample, with its size; the finding, with its page; and its place in the argument, the chapter and theme it supports and whether it agrees with the rows beside it. The identifier matters most: a typed title is where duplicates begin.
How should I tag sources by theme and by chapter?
On the day the source is added, with at least two tags: its theme and the chapter it will appear in. The Cochrane Handbook tells reviewers the same, that studies met along the way "should be recorded or tagged as they are encountered" (Lefebvre et al., 2025).
Why a tag beats a folder
A folder holds a source in one place, so a paper on two themes and one chapter is filed under one and lost to the others. A tag is not a place: a source can carry five and be found by any, and a list filtered to one tag is a chapter's reference list. Keep the tags few and fixed: one per theme, one per chapter, one for "read, not yet placed".
Which citation style should the list use?
One, chosen at the start and applied to every entry, with exceptions set on the single entries that need them. The cluster's leader says who chooses.
The entries that need their own style
- A legal case or a statute. The American Psychological Association's (n.d.) guide to the seventh edition says Chapter 11 of the manual carries the legal references and that The Bluebook's guidelines "continue to be the foundation for APA Style legal references, with some modifications". The APA Style site's public examples list no legal case and the chapter was not read for this guide; take the form from the manual and set it on that entry only.
- A standard. The American Psychological Association (2020) gives ISO standards their own page: the organisation that sets the standard is the author, the year is the year it took effect, and the standard number sits in parentheses after the title, not in italics; its example is ISO 45001:2018.
- A source your school's handbook treats differently. Theses, archival material or a department's working papers may have a required form the style lacks. The handbook wins, on that entry.
How do I find and remove duplicate references?
By identifier, at the end of every chapter, before the counts are taken. Three kinds of duplicate appear at this scale.
The three duplicate cases
- The same paper under two spellings. One entry typed from a PDF, one built from the DOI. The Cochrane Handbook's typical selection process begins by merging results in reference management software to "remove duplicate records of the same report", records with the same journal title, volume and pages (Lefebvre et al., 2025). Keep the entry built from the identifier.
- A preprint and its published version. Two reports of one work. Cite the version you read; if you switch, re-read the passage you relied on. The cluster's leader gives the APA rule.
- A conference paper and its journal version. The PRISMA 2020 explanation paper draws the line: records of the same report are duplicates, but records of reports that are "merely similar", its example an abstract sent to two conferences, are unique (Page et al., 2021b). Cochrane asks reviewers to collate such reports so that "each study, rather than each report" is the unit, choosing a primary report and keeping the others (Lefebvre et al., 2025).
What a manager's duplicate button misses
McKeown and Mir (2021) searched MEDLINE, Embase, PsycINFO and CENTRAL, retrieved 3,130 references, found 1,238 duplicates by hand, and ran the default de-duplication of Ovid, EndNote X9, Mendeley, Zotero, Covidence and Rayyan against that benchmark. Unique references wrongly removed as duplicates ranged "from 0 in Ovid, to 208 in EndNote". Run the button, then read every pair it proposes to merge.
How do systematic reviews keep count, and should a thesis?
A number at every stage, and a thesis should keep a smaller version. Item 16a of the PRISMA 2020 statement asks for the search and selection "from the number of records identified in the search to the number of studies included in the review, ideally using a flow diagram" (Page et al., 2021a). The explanation paper lists what the diagram holds: records identified; records excluded before screening, for example as duplicates or as ineligible by machine classifiers; records screened; records excluded on title or abstract; reports retrieved for detailed evaluation; potentially eligible reports that could not be retrieved; retrieved reports excluded, with the primary reasons; and the studies and reports included (Page et al., 2021b). Cochrane says why the counting happens as you go: "it is crucial to keep track of the number of references and subsequently the number of studies so that a flow diagram can be constructed", in reference management software, a document or a spreadsheet (Lefebvre et al., 2025).
The thesis version
No examiner expects a flow diagram from a narrative chapter, but keep four numbers per chapter: found, read, kept, cited. A chapter with 260 sources read and 40 cited can be defended; a chapter with no numbers cannot.
Where does a reference manager fit beside a checker?
The manager holds the sources while the thesis is written; the checker tests the list it produces. EdCitation vs Zotero and ZoteroBib sets out what Zotero does, a library with PDFs and plug-ins for Word, LibreOffice and Google Docs, and what it does not: it stores what a page or registry supplied and never looks an entry up to see whether it exists. EdCitation vs Mendeley and EndNote finds the same of both: they keep the library and insert citations as you write, EndNote flags retractions among its own items, and neither tests a finished list against the publisher's record.
When should I stop reading?
When new searches and citation chains return sources you already hold, and every main idea has two or more sources under it. That is a judgement, and the handbooks say so. The Cochrane chapter on searching says it "is often difficult to decide in a scientific or objective way when a search is complete", that the ability develops with experience, and that the stopping rules proposed have had "little formal evaluation" (Lefebvre et al., 2025). It offers signs, not rules. For a thesis, saturation is when the matrix stops changing, not when the reading feels finished; write down the date and the reason.
What are the weekly and end-of-chapter checks?
Weekly: every source read this week has a row and its tags, every entry added was built from an identifier, and the four counts are current. At the end of each chapter, in this order:
- Filter the reference list to the chapter's tag and read it once, from the last entry to the first.
- Look for duplicates by identifier, then by first author and year, and merge each pair, keeping the entry built from the record.
- Check every preprint for a published version and every conference paper for a journal version, and decide which is cited.
- Match every citation in the text to an entry and every entry back to a citation.
- Look every entry up in the publisher's record and correct what differs.
- Screen the list for retractions; retracted papers in your reference list has the decision.
- Record the four counts and the date, and note which sources were dropped and why.
How EdCitation helps
EdCitation is the best tool for keeping a long list right, because it looks every reference up in the publisher's record and never writes one, and because it runs the chapter's checks, citations matched both ways and a retraction screen, on a whole list at once, free. Where McKeown and Mir (2021) found EndNote's default removing 208 real references as duplicates, a check against the record has nothing to misjudge.
A free account keeps the reading list, the sources you bookmarked to read, one press from being cited, and reference lists set in any of the six styles, APA 7, MLA 9, Chicago 18 author-date, Harvard, IEEE and Vancouver, with your own tags to filter by and a style for a single reference when one has to differ. A list copies into Word with italics and hanging indents and downloads for Word, LaTeX, Zotero, EndNote and Mendeley; filtered to one tag, it copies what is showing, the chapter's list. Cite a source builds each entry from a DOI, a title, a PubMed ID or an ISBN in the six styles, screened for retraction, free with no account. Verify references, also free, takes the chapter's list pasted or the whole file uploaded: each entry comes back verified, doubtful (shown as "check this") or not found, retracted papers are flagged, an index that did not answer is shown as "could not check" and never as not found, and with the file uploaded every in-text citation is matched to the list and every entry back to the text.
References from a file runs the same check and keeps the chapter in My papers, so the next run shows what was fixed; three checks a month come with a free account, and Pro, $8 a month, removes the limit. The Library is a catalogue of textbooks and scholarly books by discipline, kind, language and year, kept in My Books, with more than 5 million books to read or borrow on Max at $24 a month. Pricing has the plans; a year at once saves 25%.
What it does not do: it never writes any part of the review, keeps no PDFs, and does not insert citations into Word as you type. A reference manager stays beside it for those three jobs.
Quick questions
What is a synthesis matrix for a literature review?
A table with one row per source and a column for each thing you need from it: the claim, the method and sample, the finding with its page, and where it sits in the argument. Some library guides draw it the other way round.
Should I use tags or folders for thesis sources?
Tags. A source on two themes sits in one folder and is lost to the other; with a tag for each it is found from both, and a list filtered to one tag is a chapter's reference list.
What counts as a duplicate reference?
Two records of the same report, such as one article saved twice. The PRISMA 2020 explanation paper treats records of merely similar reports, such as one abstract sent to two conferences, as unique (Page et al., 2021b).
When have I read enough for a literature review?
When new searches and citation chains return papers you already hold and every main idea has at least two sources. The Cochrane Handbook says the decision is hard to make objectively and comes with experience (Lefebvre et al., 2025).
References
- American Psychological Association. (n.d.). Guide to what's new in the Publication Manual of the American Psychological Association, seventh edition. Retrieved 23 September 2026, from https://apastyle.apa.org/instructional-aids/whats-new-7e-guide.pdf
- American Psychological Association. (2020). ISO standard references. APA Style. https://apastyle.apa.org/style-grammar-guidelines/references/examples/iso-standard-references
- Lefebvre, C., Glanville, J., Briscoe, S., Featherstone, R., Littlewood, A., Metzendorf, M.-I., Noel-Storr, A., Paynter, R., Rader, T., Thomas, J., & Wieland, L. S. (2025). Chapter 4: Searching for and selecting studies. In J. P. T. Higgins, J. Thomas, J. Chandler, M. Cumpston, T. Li, M. J. Page, & V. A. Welch (Eds.), Cochrane handbook for systematic reviews of interventions (version 6.5.1). Cochrane. https://www.cochrane.org/authors/handbooks-and-manuals/handbook/current/chapter-04
- McKeown, S., & Mir, Z. M. (2021). Considerations for conducting systematic reviews: Evaluating the performance of different methods for de-duplicating references. Systematic Reviews, 10, Article 38. https://doi.org/10.1186/s13643-021-01583-y
- Page, M. J., McKenzie, J. E., Bossuyt, P. M., Boutron, I., Hoffmann, T. C., Mulrow, C. D., Shamseer, L., Tetzlaff, J. M., Akl, E. A., Brennan, S. E., Chou, R., Glanville, J., Grimshaw, J. M., Hróbjartsson, A., Lalu, M. M., Li, T., Loder, E. W., Mayo-Wilson, E., McDonald, S., . . . Moher, D. (2021a). The PRISMA 2020 statement: An updated guideline for reporting systematic reviews. BMJ, 372, Article n71. https://doi.org/10.1136/bmj.n71
- Page, M. J., Moher, D., Bossuyt, P. M., Boutron, I., Hoffmann, T. C., Mulrow, C. D., Shamseer, L., Tetzlaff, J. M., Akl, E. A., Brennan, S. E., Chou, R., Glanville, J., Grimshaw, J. M., Hróbjartsson, A., Lalu, M. M., Li, T., Loder, E. W., Mayo-Wilson, E., McDonald, S., . . . McKenzie, J. E. (2021b). PRISMA 2020 explanation and elaboration: Updated guidance and exemplars for reporting systematic reviews. BMJ, 372, Article n160. https://doi.org/10.1136/bmj.n160
- Utah State University Libraries. (2026). Synthesis: The synthesis matrix. Retrieved 23 September 2026, from https://libguides.usu.edu/c.php?g=648048&p=10935720
- Williams College Libraries. (2026). Literature review: A self-guided tutorial: Using a synthesis matrix. Retrieved 23 September 2026, from https://libguides.williams.edu/literature-review/synthesis-matrix