How to Tell If a Source Is Credible
A source is credible when you can trust it for a specific claim: someone accountable has relevant expertise, the evidence can be opened, and the purpose is clear. This page is the accept / caution / reject decision, not a tour of every checklist. The wording below comes from the ACRL Framework, the CRAAP handout at CSU Chico’s Meriam Library, Mike Caulfield’s SIFT moves, the Stanford History Education Group’s 2016 and 2019 civic-online-reasoning reports, the free Civic Online Reasoning curriculum, and the MLA Style Center’s 2 October 2024 note that a search overview is not a work.
By CitationEasy Editorial Team27 min read
On this page
How to Tell If a Source Is Credible: Quick Answer
Leave the page before you decide. Search the author or organization in a new tab, match the name you find to the name on the page, then come back and check date, purpose, and whether a key fact opens in another source. A polished layout is not evidence. A .org suffix is not a seal. A generated search box is not a source. Peer review is a filter, not a lifetime guarantee.
Decision templates (fill from the source in front of you)
Accept and cite:
I can name the accountable person or office, their relevant expertise, a publication or revision date that fits this field, and a checkable locator (DOI, dataset, statute, or transcript). I will use this source for [this claim], not for a claim it does not support.
Use with caution:
The publisher is real, but the item is outdated, one-sided, not peer-reviewed, or only a summary of someone else’s work. I will name that limit in the sentence, or I will open the original and cite that instead.
Reject for this claim:
I cannot tell who is accountable, the purpose is to sell or recruit, the item has been retracted, a search overview invented the citation, or a sixty-second search shows the publisher is not what the About page pretends. I will not launder the claim through a blog that still treats it as current.
The Civic Online Reasoning curriculum, built from how professional fact-checkers actually work, reduces the first minute to three questions. Ask them off the page:
Civic Online Reasoning (Inquiry Group / Stanford History Education Group)
Who’s behind the information?
Who’s behind the information?
What’s the evidence?
What’s the evidence?
What do other sources say?
What do other sources say?
Run the full CRAAP questions when you already know the publisher is real and need letter-by-letter checks. Use fact-checking when the publisher is fine and the claim still might be wrong. This URL answers only: is this source credible enough to cite for the job you are asking it to do?
What this page is for
CitationEasy already has a CRAAP walkthrough, a claim-checking page, a scholarly versus popular type map, and a bias guide. Those pages share some official names. They do different jobs. This one is the three-way verdict, with cases where a checklist run inside the page fails: lookalike organizations, lobbyist fronts, Wikipedia, a retracted journal article, a climate site whose About page hid its funders, and a search overview that is not a work.
What Is a Credible Source?
A credible source is one you can reasonably trust for the job you are asking it to do. That last clause matters. The Association of College and Research Libraries treats authority as something communities construct, not a sticker a website earns forever:
ACRL Framework for Information Literacy for Higher Education (2016)
Authority Is Constructed and Contextual
Information resources reflect their creators’ expertise and credibility, and are evaluated based on the information need and the context in which the information will be used. Authority is constructed in that various communities may recognize different types of authority. It is contextual in that the information need may help to determine the level of authority required.
In practice, a Centers for Disease Control and Prevention data table can be an excellent source for U.S. mortality statistics and a weak source for a literary interpretation. A survivor’s oral history can be a primary source of the highest authority for what that person experienced and a poor source for national prevalence numbers. Credibility is a judgment about fit, not a ranking of brands.
For most college papers, a credible source usually has three things you can verify: a named author or accountable organization with relevant expertise, a publication or revision date, and evidence other people can check — a DOI, a methods section, a reference list, a dataset, or a public record. A polished homepage is not one of those three things.
The three-way credibility decision
- Accept and cite. Expertise, date, and checkable evidence all hold. Use it for the claim it actually supports.
- Use with caution. The source is real, but it is outdated, one-sided, not peer-reviewed, or only a summary of someone else’s work. Name the limitation in your prose, or go find the original.
- Reject for this claim. You cannot tell who is accountable, the purpose is to sell or recruit, the item has been retracted, a generated overview invented the citation, or a lateral search shows the publisher is not what it pretends to be.
Why Students Misjudge Credibility
The most-cited evidence on this problem is not a blog tip. Between January 2015 and June 2016, the Stanford History Education Group administered 56 tasks to students in 12 U.S. states and analyzed 7,804 responses, from middle school through selective and open-admission universities. Their report, Evaluating Information: The Cornerstone of Civic Online Reasoning (22 November 2016; PDF opened 17 September 2026), summed up the result in one word:
Stanford History Education Group (2016), executive summary
Overall, young people’s ability to reason about the information on the Internet can be summed up in one word: bleak.
The failures were specific. In the middle-school homepage task, more than 80 percent of 203 students treated a Slate item labeled “sponsored content” as a real news story. One open-web task sent students to MinimumWage.com, which linked to the New York Times and called itself a project of the Employment Policies Institute, “sponsoring nonpartisan research.” The PDF is blunt: “Ninety-three percent of students were snared.” Among Advanced Placement history students sent to the same site, “only nine percent of high school students in an Advanced Placement history course were able to see through MinimumWage.com’s language to determine that it was a front group for a D.C. lobbyist.” Lead author Sam Wineburg later put the lesson this way in a Stanford Graduate School of Education release (22 November 2016):
Sam Wineburg, Stanford Graduate School of Education, 22 November 2016
Many people assume that because young people are fluent in social media they are equally perceptive about what they find there. Our work shows the opposite to be true.
Three years later the same group tested a national sample of 3,446 U.S. high-school students (June 2018–May 2019). Students’ Civic Online Reasoning: A National Portrait did not find that the 2016 “bleak” result had been fixed. The official abstract states both of the next sentences:
Stanford History Education Group and Gibson Consulting (November 2019)
Ninety-six percent of students did not consider why ties between a climate change website and the fossil fuel industry might lessen that website’s credibility. Instead of investigating who was behind the site, students focused on superficial markers of credibility: the site’s aesthetics, its top-level domain, or how it portrayed itself on the About page.
Nearly all students floundered. Ninety percent received no credit on four of six tasks.
In that same 2019 sample, 52 percent treated a grainy video that claimed to show ballot stuffing in the 2016 Democratic primaries as “strong evidence” of U.S. voter fraud. The footage was shot in Russia. Among more than 3,000 responses, only three students tracked the video to its source. Professional fact-checkers in the follow-up work did something students almost never did: they left the page. Wineburg and Sarah McGrew call that lateral reading — opening new tabs to ask who is behind a site before you finish reading it. Reading the About page and admiring the layout is vertical reading, and it is how both the Employment Policies Institute task and the CO2 Science task were designed to trap people. If you take only one habit from this guide, take that one: do not decide credibility from inside the source.
This Skill Can Be Taught
The 2016 and 2019 reports diagnose the problem. Later Civic Online Reasoning work tests whether a short, curriculum-embedded sequence changes what students do. Sarah McGrew and Joel Breakstone taught ninth-grade biology and world geography teachers a set of lessons built around one core question, real web pages, and formative assessment. Their AERA Open article (2023) reports the result in one sentence:
McGrew and Breakstone, AERA Open (2023), abstract
Pretest/posttest data (N =574) showed statistically significant growth in students’ ability to evaluate the credibility of online content.
That is not a claim that six lessons make anyone a professional fact-checker. A related Stanford Graduate School of Education write-up of the Civic Online Reasoning classroom trials is careful: students who received six 50-minute lessons were twice as likely to spot a questionable website as they had been on the pretest, and they still earned only about half the available points. The free lesson bank lives at Civic Online Reasoning. This page is the student-facing verdict those lessons train: who is behind the information, what the evidence is, and what other sources say.
Credible vs Reliable vs Authoritative vs Scholarly
Instructors use these words as if they were synonyms. They are related, and they are not the same. Mixing them up is how a student ends up citing a beautiful popular magazine as if it were a scholarly source, or discarding a government dataset because it is not “scholarly.”
| Word | The question it answers | A source can fail this and still be useful |
|---|---|---|
| Credible | Can I trust this source for this claim? | Yes — a blog can be a primary source of someone’s public statement. |
| Reliable | If I checked again, would I get the same facts? | A breaking news story can be credible as early reporting and still get numbers revised. |
| Authoritative | Does this community treat the creator as an expert here? | A Nobel economist is not automatically authoritative on nutrition. |
| Scholarly | Was this written by researchers for researchers, usually with peer review? | A CDC table is authoritative without being a journal article. |
Most assignment sheets that say “use credible scholarly sources” want the last column: peer-reviewed journal articles, university-press books, and the occasional grey-literature report from a known agency. They do not want you to treat every .org as scholarly. For the type-by-type contrast, use Scholarly vs popular sources. For the mechanics of refereeing, use How the peer-review process works.
Two Official Frameworks: CRAAP and SIFT
Librarians still teach two complementary methods. One is a checklist you run on the source. The other is a set of moves that take you off the source. You need both. A checklist without a lateral search is how students blessed MinimumWage.com and CO2 Science. A lateral search without a checklist is how students accept a real journal article that is ten years out of date for a fast-moving clinical question.
The CRAAP test (Meriam Library, CSU Chico)
Librarian Sarah Blakeslee and colleagues at the Meriam Library, California State University, Chico, published the CRAAP test as a list of questions, not a score. The official handout (dated 17 September 2010; PDF opened 17 September 2026) opens with the problem this page is about:
Meriam Library, California State University, Chico
When you search for information, you're going to find lots of it . . . but is it good information? You will have to determine that for yourself, and the CRAAP Test can help. The CRAAP Test is a list of questions to help you evaluate the information you find. Different criteria will be more or less important depending on your situation or need.
| Letter | Official gloss | The credibility question |
|---|---|---|
| Currency | The timeliness of the information | Is this recent enough for this field and claim? |
| Relevance | The importance of the information for your needs | Would you be comfortable citing this in the paper you are actually writing? |
| Authority | The source of the information | Who is accountable, and are they qualified on this topic? |
| Accuracy | The reliability, truthfulness and correctness of the content | Can you verify the claim in another source? |
| Purpose | The reason the information exists | Is this here to inform, teach, sell, entertain, or persuade? |
Those five glosses are the library’s wording, not a paraphrase. The full question lists, field-by-field currency notes, and a longer walkthrough live on How to evaluate sources. This page uses CRAAP as the vocabulary for the decision, then spends its space on cases where the checklist alone is not enough.
SIFT and lateral reading (Mike Caulfield)
Mike Caulfield’s SIFT method is the short list of things to do when a page is unfamiliar. He published the four moves under a Creative Commons license on 19 June 2019:
Mike Caulfield, “SIFT (The Four Moves),” 19 June 2019
The four moves: Stop, Investigate the source, find better coverage, trace the original context.
On the first move
Don't read it or share media until you know what it is.
CRAAP (stay and inspect)
- Best once you already know the publisher is real
- Forces you to check date, audience, and purpose
- Catches an outdated but genuine article
- Weak against a site designed to look scholarly
SIFT (leave and recontextualize)
- Best in the first 60 seconds on an unfamiliar URL
- Asks what other people say about the source
- Catches front groups and lookalike names
- Weak if you never return to read the methods
Caulfield’s investigation example is worth keeping: take sixty seconds to learn whether you are reading a Nobel economist or a video “put out by the dairy industry” before you decide what the claims mean. The claim-checking half of SIFT — find better coverage, then trace the quote or figure to its original context — is unpacked with tools in How to fact-check sources.
Worked Examples: Decide, Don’t Just Describe
Abstract criteria are easy to agree with and hard to apply. The seven cases below are real sources. Each one ends with the three-way decision, not a vibe.
Example 1: A CDC FastStats table — accept
Open NCHS FastStats: Leading Causes of Death (page opened 17 September 2026). The publisher is the National Center for Health Statistics, a unit of the U.S. Centers for Disease Control and Prevention. The live table lists heart disease at 683,491 U.S. deaths and cancer at 619,876, then names its source and review date in the library’s own lines:
CDC / NCHS FastStats, live page lines
Source line
Source: Mortality in the United States, 2024, data table for figure 4
Review line
Last Reviewed: February 5, 2026
A lateral search for “CDC NCHS” returns the same federal statistical agency, not a lookalike. Cite the table for the numbers it publishes. If you need methods, open the named NVSS release rather than stretching FastStats into a causal story.
Decision
- Authority: A federal statistical agency whose job is to publish these numbers.
- Accuracy: Figures can be checked against the cited NVSS release, Mortality in the United States, 2024.
- Purpose: To inform. No product, no membership pitch.
- Currency: Use the date on the table (last reviewed 5 February 2026), not the date you found the URL.
- Verdict: Accept for U.S. vital statistics. Do not stretch it into a claim it does not make — FastStats will not tell you why a cause rose in one county.
Example 2: A lookalike medical organization — reject as the AAP
Stanford’s college tasks used a pair of organizations whose names are easy to confuse. The American Academy of Pediatrics is the large professional association of pediatricians in the United States. Its live About page (opened 17 September 2026) states the founding in one sentence:
American Academy of Pediatrics, About the AAP
The AAP was founded in 1930 by 35 pediatricians to serve as an independent forum to address children's health needs.
The American College of Pediatricians is a much smaller advocacy organization founded in 2002. Both can buy a professional-looking site. Only one is the body most physicians and journalists mean when they say “the AAP.”
American Academy of Pediatrics (aap.org)
- Founded in 1930 (AAP About page, opened 17 September 2026)
- The organization newsrooms and textbooks usually mean
- Clinical guidance widely cited in pediatrics
- Lateral search: Wikipedia, news, and medical schools point here
American College of Pediatricians (acpeds.org)
- Founded in 2002
- Advocacy group, not the AAP
- Name and seal designed to be easy to mix up
- Lateral search: fact-checkers and medical coverage flag the distinction
Verdict: Do not treat acpeds.org as “the pediatricians’ association.” If an assignment wants professional-society guidance, use the Academy and say so. If you quote the College, name it in full so a reader does not hear “AAP.” The point is not that advocacy groups are unusable — see Bias in sources — it is that a name that sounds official is not authority.
Example 3: Wikipedia — use as a map, not a citation
Wikipedia’s own help page is blunter than most student blogs. From Wikipedia:Citing Wikipedia (page opened 17 September 2026):
Wikipedia:Citing Wikipedia
Normal academic usage of Wikipedia is for getting the general facts of a problem and to gather keywords, references and bibliographical pointers, but not as a source in itself. Remember that Wikipedia is a wiki. Anyone in the world can edit an article, deleting accurate information or adding false information, which the reader may not recognize. Thus, you probably shouldn't be citing Wikipedia.
Verdict: Use with caution as a finding aid. Follow the footnotes to the book, article, or dataset, then cite that source. Cite Wikipedia itself only when Wikipedia is the object of study — a primary source of how a page changed — and only if the instructor agrees. The longer workflow is in Using Wikipedia for research.
Example 4: A retracted journal article — reject
Peer review is a filter, not a lifetime guarantee. Andrew Wakefield and colleagues published “Ileal-lymphoid-nodular hyperplasia, non-specific colitis, and pervasive developmental disorder in children” in The Lancet in 1998 (DOI 10.1016/S0140-6736(97)11096-0). After the UK General Medical Council’s Fitness to Practise Panel judgment of 28 January 2010, the journal’s editors issued a full retraction60175-4/abstract) (DOI 10.1016/S0140-6736(10)60175-4; print issue 6 February 2010). The live abstract states both of the next sentences:
The Editors of The Lancet, retraction notice
In particular, the claims in the original paper that children were "consecutively referred" and that investigations were "approved" by the local ethics committee have been proven to be false.
Therefore we fully retract this paper from the published record.
Decision
- Looks scholarly: Named authors, a famous journal, volume and issue, a DOI.
- Lateral check: Search the title plus “retraction,” or look the DOI up in the Retraction Watch Database and on the publisher page.
- Verdict: Reject as support for any medical claim. You may discuss it as a case of scientific misconduct or media amplification, with the retraction in the citation. Never launder a retracted finding through a secondary blog that still treats it as current.
Example 5: An industry research front — reject as independent research
MinimumWage.com, the site in the 2016 Stanford college and AP-history tasks, linked to the New York Times and described itself as a project of the Employment Policies Institute, “sponsoring nonpartisan research.” The 2016 PDF says a sixty-second Google search for the institute plus the word “funding” is what almost no student did. The site can still contain a true statistic. The credibility failure is the frame: it is not an independent research institute just because the About page says so.
Verdict: Reject as a neutral authority on labor economics. If a figure on the page matters, trace it to the government series or journal article it came from and cite that. Purpose is doing the work here, not spelling quality.
Example 6: A climate site with hidden funders — reject as independent science
The 2019 national portrait sent students to the homepage of CO2 Science (co2science.org). The About page said the mission was to “disseminate factual reports and sound commentary” on carbon dioxide and the environment. Students who earned a mastery score left the site and found that the Center for the Study of Carbon Dioxide and Global Change, which runs the page, has been funded by fossil-fuel companies, including ExxonMobil — the finding the report says fewer than two percent of students reached. Over 96 percent stayed on the page and graded its .org suffix, graphs, or About copy.
Decision
- Looks independent: A .org domain, graphs, and a mission about “factual reports.”
- Lateral check: Search the organization name plus “funding” or “Exxon.” Compare what independent coverage says with the About page.
- Verdict: Reject as a neutral authority on climate science. If a figure on the page matters, trace it to the underlying dataset or journal article and cite that. Superficial markers — aesthetics, a .org suffix, a solemn About page — are how this task was designed to fail vertical readers.
Example 7: A Google AI Overview — reject as the source
A generated search box can paste a tidy paragraph and a handful of links above the ordinary results. That box is not a publisher, not a peer-reviewed article, and not a stable document. The MLA Style Center answered the student question directly on 2 October 2024:
MLA Style Center, 2 October 2024
No. Google’s AI Overviews feature is a form of search results, and as we note in this post, “[s]earch results are not a work.”
If you want to cite something from the AI Overviews feature, click through to the source and cite it instead.
The same post walks a real trap: a search for “scare quotes MLA style” surfaced “Bob experienced the ‘catastrophe’ of having his tooth pulled.” That sentence is an official MLA Style Center example — and the Overview also offered “I ‘liked’ her post on Facebook,” which the underlying post uses as an example of when not to use scare quotes. The generated box mixed a usable example with a counter-example. How to cite a saved chat, if your instructor allows one, is a separate problem, covered in How to cite ChatGPT and AI tools.
Decision
- Looks convenient: A paragraph, a few links, and no author byline.
- Lateral check: Click through to the named page. Read the claim in context. If the identifier does not resolve, there is no source.
- Verdict: Reject the overview as the thing you cite. Accept only the document you actually opened — and only after it clears the same accept / caution / reject test as any other URL.
Domain Suffixes Are Not a Credibility Seal
CRAAP still asks whether a URL “reveal[s] anything” (.com, .edu, .gov, .org, .net). That is a clue, not a verdict. The Stanford college tasks showed students treating .org as “a Good Housekeeping seal.” It is not. The .org top-level domain is unrestricted: advocacy groups, companies, and private individuals can register it. A .edu host can serve a student paper, a retired faculty page, or a lab that has not been updated since 2009. A .gov host is a stronger signal for official U.S. public information, and even then you still check the date and the office.
What a domain can and cannot tell you
- .gov: Usually an official government publisher. Still check which office wrote the page and when it was last revised.
- .edu: An institution of higher education owns the domain. The page may still be a class project or an opinion.
- .org: Once marketed as “non-profit.” Anyone who pays the fee can use it. CO2 Science and the Employment Policies Institute both used .org.
- .com / .net: No quality information at all. Reuters and a supplement store both use .com.
- Country codes (.ac.uk, .gov.uk, .edu.au): Read the second level. ac and gov are meaningful; a random .co.uk blog is not.
Peer Review Is Necessary. It Is Not Sufficient.
For empirical claims in the sciences and social sciences, a peer-reviewed article is still the default credible genre. Reviewers can still miss fraud, underpowered samples, and conflicts of interest. Predatory journals sell the appearance of review. High-impact journals retract papers. The Wakefield Lancet paper in Example 4 had volume, issue, and a DOI. Impact factor is a journal-level citation average, not a grade for the article in your PDF — the citation metrics guide is the place for that distinction.
Before you treat a journal as scholarly, confirm three things: the journal is the one you think it is (watch for hijacked titles), it is indexed in a library database or in the Directory of Open Access Journals rather than only on its own website, and the specific article has not been retracted or marked with an expression of concern. Ulrichsweb, available through many campus libraries, is the usual place to check the “refereed” flag. How to access academic journals and library databases cover the access path; this page is only the credibility gate.
Predatory Journals: When “Peer-Reviewed” Is a Sales Pitch
A predatory journal charges authors for publication while providing little or no genuine review. Citing one imports unreviewed claims into your paper. The cross-sector campaign Think. Check. Submit. is the official checklist publishers and libraries point authors to. Its homepage question is the right one for readers as well as authors:
Think. Check. Submit.
Are you submitting your research to a trusted journal or publisher? Is it the right journal or book for your work?
Turn the same questions around when you cite: do you recognize the publisher? Do colleagues in the field publish there? Is the editorial board made of real people who know they are listed? Does the journal appear in DOAJ, Scopus, Web of Science, or your library’s catalog — or only in a flattering email? Think. Check. Submit. deliberately does not publish a forever-list of “good” and “bad” titles, because new journals launch constantly. The checklist is the method.
AI-Generated Text and Invented Citations
A chatbot can write a paragraph that sounds like a literature review and invent articles, DOIs, and page spans that do not exist. That output is not a credible source for the claims it makes about the world. The same rule covers a generated search overview (Example 7). If an instructor lets you use a model as a tool, you still have to open every citation it proposes. Paste each DOI into a DOI validator or https://doi.org; if the identifier does not resolve, the citation is not a source. How to cite a retrievable AI transcript is a separate problem, covered in How to cite ChatGPT and AI tools.
Credibility rule for generated text
Treat model output as an unreviewed draft, never as evidence. The credible source is the article, dataset, or document you actually opened.
When a Non-Scholarly Source Is Still Credible
Students sometimes throw out everything that is not a journal. That is the wrong half of the ACRL sentence. Authority is contextual. These non-scholarly sources are often the right credible source for a specific job:
- Official statistics and legal texts. Census tables, statutes, court opinions, and agency guidance are credible as records of what the state published. Cite the document, not a news rewrite. See grey literature.
- News reporting of a public event. A named correspondent at a newsroom with editors is credible for what was said at the press conference. It is not a substitute for the study the reporter summarized. Separate news from opinion.
- Trade and professional publications. A nursing magazine can be the right source for practice guidance your scholarly journal has not yet reviewed.
- Primary testimony. Interviews, letters, social posts, and oral histories are credible as evidence of the speaker’s words. They are not automatically credible as evidence that the words are true. See primary vs secondary sources.
- Open versions of scholarly work. A green open-access accepted manuscript in a university repository can be as citable as the paywalled version of record if you identify the version. See open access research.
How Many Credible Sources Do You Need?
There is no universal “one or two sources per page” rule, and repeating one invents a standard your instructor did not write. The assignment sheet wins. A two-page response paper may need three carefully read articles. A literature review may need several dozen. What does travel across assignments:
- Prefer sources you have opened, not titles a generator or a chatbot listed.
- If the prompt says “scholarly,” make scholarly sources the majority, then add news, law, or data only where the claim requires them.
- One excellent methods paper beats five undated websites that repeat each other.
- Counter-evidence from a credible source is a feature. Ignoring it is bias, not focus.
- When a topic is too new for journals, say so, use preprints and agency briefs as preprints and briefs, and ask the instructor before you lean on them.
Tools That Actually Check a Source
Use tools to verify, not to outsource the judgment. A high citation count is not a credibility certificate; a clean website is not one either.
| Check | Tool | What a pass looks like |
|---|---|---|
| Does this DOI exist? | doi.org or the DOI validator | The identifier resolves to the publisher page. |
| Is the journal refereed? | Ulrichsweb (via your library) or the publisher site | The record says refereed / peer-reviewed, and the title matches. |
| Is this open-access journal real? | DOAJ | The journal is listed; the ISSN matches. |
| Has the paper been retracted? | Retraction Watch Database and the publisher PDF | No retraction or expression of concern for this DOI. |
| Who is citing it? | Google Scholar | Citations come from real journals, not only the same authors. |
| Who funds or owns this site? | A plain web search for the organization plus “funding” or “about” | Independent coverage agrees with the About page — or explains why it does not. |
| How do I practice the three questions? | Civic Online Reasoning | You can name who is behind the page, what the evidence is, and what other sources say. |
Google Scholar is a discovery layer, not a credibility layer. It indexes preprints, student papers, and predatory titles alongside Nature. Use it to find work and to see who cited a paper, then evaluate the paper. The search tactics live in the Google Scholar guide.
Quick Credibility Checklist
Sixty seconds, then a closer read
Before you read the argument
- Stop. Do you already know this publisher’s reputation?
- Open a new tab and search the organization or author — not the article title alone.
- Match the name you found to the name on the page (AAP is not ACPeds).
- Note the date. Is it current enough for this field?
- If the item is a generated overview, click through. Search results are not a work.
Before you add it to the bibliography
- Can you name the author or accountable organization and their relevant expertise?
- Is the purpose to inform, or to sell, recruit, or campaign?
- Are key facts backed by evidence you can open (DOI, dataset, statute, transcript)?
- If it is a journal article, is it peer-reviewed and free of a retraction notice?
- Would you be comfortable if your instructor opened the same URL?
- Have you cited the version you actually read, in the style the assignment named — APA, MLA, Chicago, or another?
Frequently Asked Questions
- How do you know if a source is credible?
- Leave the page first. Search the author or organization, confirm they exist and have relevant expertise, then come back and check date, purpose, and whether the key claims can be opened in another source. Stanford’s 2016 and 2019 civic-online-reasoning studies found that students who stay on the page and judge the layout get the decision wrong.
- What makes a source credible for a research paper?
- For most college papers, credibility means an accountable expert or agency, a date that fits the field, and evidence a reader can verify. Scholarly journal articles and university-press books are the usual default. Government data, laws, and well-edited news can be credible for the claims they are actually good at.
- Are .edu and .org websites credible?
- Not automatically. A .edu domain means a college or university owns the host; the page may still be a student project. A .org domain can be registered by anyone. In Stanford’s 2019 national portrait, students treated CO2 Science’s .org suffix as a quality seal and missed the fossil-fuel funding a sixty-second search would have shown.
- Is Wikipedia a credible source?
- Wikipedia’s own citing guide says you should use it to gather keywords and references, not as a source in itself, because anyone can edit a page. Follow the footnotes to the underlying book or article and cite that. Cite Wikipedia only when the encyclopedia is the object of study and your instructor agrees.
- What are examples of credible sources?
- Typical accept-list examples: a peer-reviewed article you opened in a library database, a university-press book, the CDC FastStats leading-causes table last reviewed 5 February 2026, a statute or court opinion, and a named news report of a public event. Typical reject-list examples: an undated commercial blog, a retracted paper used as current evidence, a lookalike advocacy group treated as a professional academy, a Google AI Overview, and a chatbot citation that does not resolve.
- Can I use a source that is biased?
- Yes, if you name the perspective and do not pretend it is neutral. Bias is not the same as fabrication. An advocacy report can be a credible source of that organization’s position and a poor source of population statistics. Pair it with independent evidence. See the bias guide for types and tells.
- Are preprints credible sources?
- A preprint is a manuscript that has not yet passed peer review. It can be the only public version of a fast-moving finding. Cite it as a preprint, do not upgrade it to a journal article in your bibliography, and prefer the version of record once it exists. Ask your instructor before a preprint carries a central claim.
- How do I check if a journal is peer-reviewed?
- Search the title in Ulrichsweb through your library, or read the journal’s own editorial policy and confirm the title is indexed in DOAJ, Scopus, Web of Science, or a subject database — not only on a stand-alone website. Then check the specific article for a retraction notice in the Retraction Watch Database.
- Is a Google AI Overview a credible source?
- No. The MLA Style Center’s 2 October 2024 note says search results are not a work. Click through to the linked page, read the claim in context, and cite that document if it clears the same credibility test. Generated overviews can mix a real example with a counter-example from the same post.
- Is ChatGPT a credible source?
- No. Generated text is unreviewed and can invent citations. Use a model, if your course allows it, to brainstorm or outline, then open real sources. If you quote a saved conversation, follow your style’s AI rules and still support factual claims with documents you read.
Cite the source you actually opened
Once a source clears this check, cite the version you read — the article with the DOI, the agency table, the news report — in the style your assignment names. The generator formats the reference. It does not decide whether the source was credible.
Related Guides
- How to evaluate sources (CRAAP walkthrough)
- How to fact-check sources (SIFT in practice)
- Scholarly vs popular sources
- Bias in sources
- Using Wikipedia for research
- How the peer-review process works
- How to find scholarly sources
- Grey literature
- Citation metrics (impact factor and more)
- How to cite ChatGPT and AI tools
- How to find a DOI
- Peer review (glossary)
- Predatory journal (glossary)
- Retraction (glossary)