How to evaluate source credibility: CRAAP, lateral reading, SIFT and the checks scholars use

This guide is part of our series on essays, theses and sources. It covers the evaluation methods, what peer review does and does not guarantee, and the three problems that catch researchers most often: predatory journals, retracted papers and preprints. For where to find sources in the first place, see how to find sources for a research paper.

What makes a source credible?

A source is credible when its authors have relevant expertise, a checking process stands behind it, its evidence can be traced, it is current for your question, and its purpose is clear. No single sign is enough. A peer-reviewed paper can be retracted; a government report can be out of date; a blog post by a leading researcher can be the best explanation available.

Being primary or secondary is a separate question from being credible. See primary vs secondary sources for that distinction, and levels of evidence for how study design affects how much a finding can bear.

What is the CRAAP test?

The CRAAP test is a five-part checklist (Currency, Relevance, Authority, Accuracy, Purpose) created by Sarah Blakeslee at the Meriam Library, California State University, Chico [1]. It is the most widely taught source checklist in college libraries.

Verdict: credible for that country's prevalence in 2019; find newer data and comparable surveys for the other two countries.

What are the limits of the CRAAP test?

CRAAP checks are done on the page, and a well-designed site can pass them while hiding who is behind it. An "About" page can claim expertise, list credentials and show a professional design. Research by Wineburg and McGrew found that checking a site by reading its own pages closely, the way many checklists encourage, often led readers to trust sources they should have questioned [2]. Use CRAAP for scholarly sources and structured documents. For websites, add lateral reading.

What is lateral reading?

Lateral reading means leaving a source to see what other sources say about it, instead of evaluating it from the inside. The term comes from a study by Sam Wineburg and Sarah McGrew of the Stanford History Education Group, who watched professional fact-checkers, historians and Stanford undergraduates evaluate websites [2].

The fact-checkers did best. Within seconds of landing on an unfamiliar site, they opened new browser tabs and searched for the organization and the author. Historians and students tended to read vertically, staying on the site and judging it by its look, its "About" page and its references. In one task, many participants could not tell a long-established professional medical body from a small advocacy group with a similar name, because both sites looked authoritative [2].

Later classroom assessments by the same group found that most students struggled to judge online sources, for example by failing to notice who had sponsored content [3].

How to read laterally (worked example, illustrative)

You find a site, "Institute for Healthy Soil Policy", with a report saying a fertilizer ban will halve crop yields.

1. Leave the site. Open a new tab and search the institute's name plus "funding", then read what news reports and registry records say about it.
2. Check the author. Search the report's author: do they publish in peer-reviewed soil science journals?
3. Decide. If you find that the institute is funded by a fertilizer trade group, the report may still contain data, but you should not rely on it alone. Find peer-reviewed field studies on the same question.

What is the SIFT method?

SIFT is a four-move routine for checking online claims quickly: Stop, Investigate the source, Find better coverage, Trace claims to the original context. It was developed by Mike Caulfield, building on his open textbook Web Literacy for Student Fact-Checkers and on lateral reading [4][5].

The last move matters most in academic work. A statistic in a news article or a review usually comes from a study. Trace it, open the study, and check that the number, population and conditions match. Our guide on how to check a claim against the research walks through that step in detail.

CRAAP or SIFT: which should you use?

Does peer review make a source credible?

Peer review makes a source more likely to be sound, but it does not guarantee that the findings are correct. In peer review, editors send a manuscript to independent experts who judge its methods, analysis and conclusions before publication.

Peer review catches many problems, but reviewers rarely see raw data, cannot rerun experiments, and can miss errors, weak statistics and fraud. Richard Smith, a former editor of The BMJ, summarized the evidence as showing that peer review is slow, expensive, inconsistent and poor at detecting errors, while still being the system science relies on [6].

How to check:

• Find the journal's peer review policy on its website ("Peer review process", "For authors" or "Editorial policies").
• Check the article type. Editorials, letters, commentaries and news items in a journal are often not peer reviewed.
• Read the paper anyway. Peer review is a minimum, not a verdict. Our guide on how to read a research paper covers critical reading.

How do you spot a predatory journal?

A predatory journal takes publication fees but skips or fakes peer review and misrepresents its practices. An international panel defined predatory journals and publishers as entities that "prioritize self-interest at the expense of scholarship and are characterized by false or misleading information, deviation from best editorial and publication practices, a lack of transparency, and/or the use of aggressive and indiscriminate solicitation practices" [7].

How to check a journal:

1. Look it up in the Directory of Open Access Journals (DOAJ) if it is open access [8]. DOAJ vets journals before listing them.
2. Check indexing claims directly. For health journals, search the NLM Catalog to see whether the journal is indexed in MEDLINE.
3. Check the publisher's memberships, such as the Committee on Publication Ethics (COPE), on the member organization's own site.
4. Use the Think. Check. Submit. checklist [9], designed for authors choosing a journal but useful for readers too.

Open access is not a warning sign in itself; many leading journals are open access.

How do you check if a paper has been retracted?

Search the paper's title in the Retraction Watch Database, and check the article page on the publisher's site for a retraction notice, expression of concern or correction. A journal retracts a paper when its findings are unreliable, because of error or misconduct, or when it breaches publication ethics. Guidelines from the Committee on Publication Ethics describe when journals should retract, and that retracted articles should stay online, clearly marked [10].

The Retraction Watch Database is the largest public list of retractions. Crossref acquired it in 2023 and made it openly available [11].

A real example. Wakefield and colleagues' 1998 paper in The Lancet, which suggested a link between the MMR vaccine and autism, was fully retracted by the journal in 2010 [12]. It is still discussed and sometimes cited. If you mention it, cite it as a retracted paper and cite the retraction notice.

Retracted papers keep collecting citations, so check every reference in your final list, not just the ones you doubt.

Are preprints credible sources?

Preprints are research papers posted publicly before peer review, so they are early findings, not settled ones. Servers such as arXiv (physics, mathematics, computer science), bioRxiv (biology) and medRxiv (health sciences) screen submissions for basic suitability but do not peer review them. medRxiv states on its site that its preprints have not been peer reviewed and should not guide clinical practice [13].

Use a preprint when:

• It is the newest or only evidence on a fast-moving question.
• You label it as a preprint in the text and the reference.
• You have checked for a peer-reviewed version. Preprint pages often link to the published article; you can also search the title.

If a published version exists, cite it instead, and check whether the conclusions changed during review.

A five-minute credibility check for any source

1. Stop. What is the source, and what claim do I want it to support?
2. Investigate. Read laterally about the author and publisher (SIFT), or check authority and purpose (CRAAP).
3. Check the process. Peer reviewed? Journal legitimate? Preprint?
4. Check its status. Retracted, corrected or under an expression of concern?
5. Trace and match. Does the original study say what you are about to claim, for the same population and conditions?

How to check sources with AI for Research

AI for Research works from real scholarly records, and it shows you the status of each paper so you can decide. The judgment stays with you.

1. Search with status shown. Find Papers searches over 250 million scholarly works. It flags retracted papers and preprints, and lets you filter by study type, year and peer-reviewed sources.
2. Check your reference list. Paste your references into the citation checker. It checks that each reference exists and flags retractions, corrections, expressions of concern, preprints and wrong years or authors. This check costs nothing.
3. Check your claims. Claim check shows whether the studies you cite support each claim in a paragraph, and lists the papers it used. It reports what the papers say; it does not decide whether a claim is true.
4. Read the original. Upload a paper to Chat with PDF and ask what population, methods and limits it reports. Answers cite the page.
5. Re-check formatting. The free citation formatter re-formats a pasted list and flags lines it cannot match to a real record, a quick way to catch references that may not exist.

Search, the citation check, citations and the free tools cost nothing. The Free plan includes 45 credits a month for AI features. See pricing. Limits: retraction and preprint flags come from the scholarly index and can lag behind a publisher's notice, so check the publisher's page for any paper your argument depends on. The tools do not evaluate websites; use lateral reading for those.

FAQ

What makes a source credible?

A credible source is written by people with relevant expertise, published through a process that checks it, based on evidence you can trace, current enough for your question, and honest about its purpose. In academic work that usually means peer-reviewed articles, scholarly books and official reports. Credibility is not a label on the source: you judge it for each claim you use it to support.

What is the CRAAP test?

The CRAAP test is a checklist for evaluating sources, created by librarian Sarah Blakeslee at California State University, Chico. It asks about Currency (when it was published), Relevance (whether it fits your need), Authority (who wrote it), Accuracy (whether the evidence holds up) and Purpose (why it exists). It works well for scholarly sources, but checks on the page alone can be fooled by a professional-looking website.

What is lateral reading?

Lateral reading means leaving a website to find out what other sources say about it, instead of judging it from its own pages. Researchers at the Stanford History Education Group found that professional fact-checkers did this within seconds, opening new tabs to search the organization and author, and judged sources more accurately than historians and Stanford students who read the site from top to bottom.

What is the SIFT method?

SIFT is a four-step routine for checking online information, developed by Mike Caulfield. Stop before you read or share; Investigate the source, briefly, by reading laterally; Find better coverage, looking for a more trusted source on the same claim; and Trace claims, quotes and media back to their original context. Each step takes a minute or less.

Can I cite a retracted paper?

Only to discuss the retraction itself, and then you should say clearly that the paper was retracted and cite the retraction notice. Do not cite a retracted paper as evidence for a finding. Check every source before you submit, because retracted papers keep being cited for years. The Retraction Watch Database and the publisher's article page show whether a paper has been retracted.

Are preprints credible sources?

Preprints can be useful, but they have not been peer reviewed, so treat them as provisional. Cite one only when it is the best evidence available, label it as a preprint in your reference, and check before you submit whether a peer-reviewed version has been published. If it has, cite the published version, and note whether the findings changed.

Next steps

• Know what kind of source you have: primary vs secondary sources.
• Weigh the evidence: levels of evidence and study types.
• Check a specific claim: how to check a claim against the research.
• Build the argument: what is a thesis statement.

Sources

1. Blakeslee, S. (2004). The CRAAP test. LOEX Quarterly, 31(3), 6–7. https://commons.emich.edu/loexquarterly/vol31/iss3/4
2. Wineburg, S., & McGrew, S. (2019). Lateral reading and the nature of expertise: Reading less and learning more when evaluating digital information. Teachers College Record, 121(11), 1–40. https://doi.org/10.1177/016146811912101102
3. McGrew, S., Breakstone, J., Ortega, T., Smith, M., & Wineburg, S. (2018). Can students evaluate online sources? Learning from assessments of civic online reasoning. Theory & Research in Social Education, 46(2), 165–193. https://doi.org/10.1080/00933104.2017.1416320
4. Caulfield, M. (2019, June 19). SIFT (the four moves). Hapgood. https://hapgood.us/2019/06/19/sift-the-four-moves/
5. Caulfield, M. (2017). Web literacy for student fact-checkers. Pressbooks. https://webliteracy.pressbooks.com/
6. Smith, R. (2006). Peer review: A flawed process at the heart of science and journals. Journal of the Royal Society of Medicine, 99(4), 178–182. https://doi.org/10.1177/014107680609900414
7. Grudniewicz, A., Moher, D., Cobey, K. D., et al. (2019). Predatory journals: No definition, no defence. Nature, 576(7786), 210–212. https://doi.org/10.1038/d41586-019-03759-y
8. Directory of Open Access Journals. (n.d.). DOAJ. https://doaj.org/
9. Think. Check. Submit. (n.d.). Think. Check. Submit. https://thinkchecksubmit.org/
10. COPE Council. (2019). Retraction guidelines. Committee on Publication Ethics. https://doi.org/10.24318/cope.2019.1.4
11. Crossref. (2023). Crossref acquires Retraction Watch data and opens it for the scientific community. https://www.crossref.org/news/2023-09-12-crossref-acquires-retraction-watch-data-and-opens-it-for-the-scientific-community/ ; database: http://retractiondatabase.org/
12. The Editors of The Lancet. (2010). Retraction: Ileal-lymphoid-nodular hyperplasia, non-specific colitis, and pervasive developmental disorder in children. The Lancet, 375(9713), 445. https://doi.org/10.1016/S0140-6736(10)60175-4 (Original article: Wakefield, A. J., et al. (1998). The Lancet, 351(9103), 637–641. https://doi.org/10.1016/S0140-6736(97)11096-0)
13. medRxiv. (n.d.). About medRxiv. Cold Spring Harbor Laboratory. https://www.medrxiv.org/content/about-medrxiv