How to Evaluate a Source: The CRAAP Test and Beyond

Why evaluating sources is a skill, not a formality

Every source you cite lends your argument credibility or costs it some. A marker who spots one weak source starts doubting the rest; a reader who checks a claim and finds it unsupported stops trusting the writer. Evaluating sources is how you make sure each piece of evidence can carry the weight you put on it. It starts with finding credible sources in the first place, but searching well does not remove the need to check.

It is also harder than it used to be. Professional-looking websites cost little to build, statistics travel without their original context, and AI tools can produce fluent summaries that cite papers which do not exist. The habits below take a few minutes per source and catch most problems before they reach your bibliography. If an AI tool suggested the source, read the guide to catching AI hallucinations to confirm the source exists at all.

The CRAAP test, letter by letter

The CRAAP test is a checklist from academic librarianship, usually credited to librarians at California State University, Chico, and it remains a good way to make sure you have asked the obvious questions. Here is what each letter asks, with a concrete warning sign for each.

Beyond CRAAP: read laterally

The CRAAP test's weakness is that it reads a source from the inside. A site can display an impressive author biography, a recent date and a long list of references and still mislead you, because each of those signals is something the site itself controls. Lateral reading closes that gap.

Lateral reading means leaving the source to find out about it. Research by the Stanford History Education Group compared how professional fact-checkers, historians and university students judged unfamiliar websites, and the fact-checkers reached sounder conclusions faster because they quickly opened new tabs to see what others said about a site instead of studying the site itself. You can copy their habit.

In practice, search the organization's name together with words like “funding” or “criticism,” check a reference source on who runs it, and see whether independent experts cite it. A few minutes of this often tells you more than a long close reading. The guide to finding credible sources applies the same habit to unfamiliar websites, alongside the signals to look for on the page itself.

The SIFT moves for fast online checks

The digital literacy educator Mike Caulfield distilled lateral reading into four moves, known as SIFT. SIFT and CRAAP complement each other. SIFT is fast and catches untrustworthy publishers; CRAAP's questions about relevance and currency still matter once you know a source is legitimate, because a reputable source can still be the wrong one for your claim.

The four moves are quick enough to use on almost every source you meet online:

1. 1Stop. Before you read closely or share, ask whether you know this source and its reputation. A strong emotional reaction is a reason to slow down.
2. 2Investigate the source. Spend a minute finding out who is behind it, using other sites rather than its own about page.
3. 3Find better coverage. If the claim matters more than this particular source, look for the same claim in sources you already trust, and use those instead.
4. 4Trace claims, quotes and media to their original context. Follow a statistic back to the report it came from and a quotation to the full speech or interview, and confirm the original says what the copy claims.

How to evaluate a research study

Peer review is a valuable filter, but it does not guarantee a paper is right; the guide to peer-reviewed sources explains what it does and does not check. Check the journal itself, too. Predatory journals imitate real ones while charging authors to publish with little or no review. Warning signs include a vague, very broad scope, flattering email invitations, promises of unusually quick acceptance and an editorial board you cannot verify. When in doubt, see whether the journal is indexed in the main databases for its field, or ask a librarian.

When a source is an empirical study, add these questions to your checklist:

• Design: does the method fit the claim? A survey can show an association; on its own it cannot show cause.
• Sample: who or what was studied, how many, and how were they chosen? Results from one group may not transfer to another.
• Measures: how were the key ideas measured, and how much depends on self-report?
• Effect size: a statistically significant result can still be too small to matter in practice.
• Limitations: what do the authors themselves say the study cannot show?
• Funding and conflicts of interest: usually declared near the end; they do not invalidate a study but deserve attention.
• Later reception: has the finding been repeated, challenged, corrected or retracted since?

A worked example: evaluating a policy report

Suppose you are writing about the minimum wage and find a polished report from a hypothetical policy institute arguing that raising it destroys jobs. On the CRAAP checklist it looks fine: it is recent, on-topic, written by economists, full of charts and open about its purpose of informing policy.

Lateral reading changes the picture. A quick search shows the institute is funded largely by industry groups with a stake in the outcome, and economists publishing in peer-reviewed journals disagree about the effects of moderate increases. None of that proves the report wrong. It means you should present it as one side of a contested debate (“an industry-funded institute argues…”), set it against peer-reviewed evidence, and avoid quoting its conclusion as settled fact.

That is the general lesson. Evaluation rarely ends in a simple “use” or “discard.” More often it tells you how to use a source — with which caveats, for which claims, and alongside what other evidence.