Credible Sources for Research: the CRAAP Test, Step by Step

Why a source found in a hurry isn’t enough

A search engine ranks results by its own algorithm, not by how reliable the content is: the top link isn’t automatically the strongest one for a school assignment, a paper or a thesis. The CRAAP test gives a student a practical checklist instead of a gut call, and it covers five aspects of a source: when it was published, who it’s written for, who wrote it, how well its claims are backed by evidence, and why it was written.

The CRAAP test, criterion by criterion

Librarian Sarah Blakeslee designed it in 2004 at California State University, Chico’s Meriam Library, for the UNIV 001 course; from there it spread to other schools and universities across the US, as far as Australia. The acronym packs in five questions:

• Currency: when was the source published or updated, and does the date matter for this topic? A ten-year-old figure about a fast-moving technology carries different weight than one about a closed historical event.
• Relevance: does the source match the level, general-audience or specialist, that the assignment calls for? A general encyclopedia entry doesn’t stand in for a scientific paper when a technical figure is needed.
• Authority: who wrote or published the text, and with what credentials or affiliation? A site with no listed author and no way to trace who runs it is already a warning sign.
• Accuracy: are the claims backed by verifiable data or citations, or do they stay opinions with nothing behind them?
• Purpose: does the text aim to inform, or to sell, persuade or entertain? A commercial purpose doesn’t automatically disqualify a source, but it should factor into how you read what it says, and what it leaves out.

Practical example: a student researching ocean plastic pollution finds two pages: a blog post from a company selling “plastic free” products, and a report from an environmental agency with a stated methodology and references at the bottom. Running Purpose and Accuracy on both, the second page holds up: it explains how the data was gathered and has no product to sell at the end.

These checks build on the ones already covered for media literacy: how to recognize authoritative, up-to-date sources, applied here to the specific case of a school or college assignment that needs a bibliography.

Primary, secondary or tertiary source

The same label changes meaning depending on the subject. In the humanities, a primary source is a document from the period under study, a sonnet, a letter, a political manifesto: in a paper on nineteenth-century Europe, a treatise written in that era is primary, while a critical essay analyzing it today is secondary. In the sciences, a primary source is typically an original empirical study published in a peer-reviewed journal, the experiment itself, not a piece commenting on it elsewhere.

A tertiary source, such as a textbook or an encyclopedia entry, summarizes and organizes other people’s work without adding original research. It’s useful for getting a general picture at the start of a project, less so for backing a specific claim in the paper, where it’s better to trace the claim to a primary source or an authoritative secondary one.

Google Scholar doesn’t certify peer review

One of the most common misunderstandings among students starting a research project is assuming that “found on Google Scholar” means “reviewed by other experts.” Google Scholar indexes peer-reviewed articles alongside preprints that haven’t been reviewed yet, theses, conference papers, books and other so-called grey literature, without labeling which is which: it’s a tool for discovering content, not a body that certifies its quality.

The same cross-checking applies to a text written with a generative AI system: fluent writing is no guarantee of accuracy or of verified sources behind it, and the CRAAP test’s Authority criterion applies just as well there, asking who, or what, actually wrote the text. The Accuracy criterion, in particular, is what helps take apart content built to mislead rather than inform: if the claims aren’t backed by verifiable data, the rest of the source doesn’t count for much.

The domain doesn’t make the source

The .edu and .gov extensions are reserved for educational and government institutions respectively, but they only show what kind of organization owns the site: they don’t certify that a specific page, perhaps written by a student on a university server, is authoritative on that particular topic. The reverse also holds: solid content can appear on a generic .com or .org domain, and it’s the source itself that needs evaluating, not the address hosting it.

Before you cite it, a three-step check

Before a source goes into a bibliography, it’s worth closing the loop with three practical questions: does the date hold up for the topic, does the author have traceable credentials, and does another independent source confirm the same claim. Two students typing the same search phrase can see different results partly because recommendation algorithms work inside a personalized filter bubble: another reason to run the CRAAP test on a source before citing it.