How to Evaluate Sources: SIFT and CRAAP

To evaluate a source, find out who made it and whether other reliable sources back up what it says, and do most of that checking somewhere other than the page itself. Many university libraries now teach this as SIFT: Stop, Investigate the source, Find better coverage, and Trace claims to their original context (Princeton University Library). SIFT is built on lateral reading, a habit Stanford researchers saw in professional fact checkers, who leave an unfamiliar site and open new tabs to see what others say about it (Stanford GSE). The older CRAAP test asks about a source’s currency, relevance, authority, accuracy and purpose (Meriam Library, CSU Chico). It still helps with research papers, but use it after you’ve checked the source from the outside.

SIFT: four quick moves

Mike Caulfield developed SIFT and published it in 2019 as “SIFT (The Four Moves)” (Princeton University Library).

Move	What you do
Stop	Pause before you read, share or use it, and ask whether you know this source
Investigate the source	Spend a minute learning who is behind it and what their expertise and agenda are
Find better coverage	Look for the best coverage of the claim and what experts agree on
Trace claims, quotes and media	Follow it back to the original and check that it was represented accurately

Stop

Pause when something seems false or gives you a strong emotional jolt, and don’t share it until you know what it is (UC Merced Library). Ask whether you know the source and its reputation, and notice when you’re drifting into rabbit holes that won’t settle the question (Princeton University Library).

Investigate the source

A source’s expertise and agenda shape how you should read it. About sixty seconds spent finding out where something comes from tells you whether it’s worth your time (Princeton University Library). A .org address alone doesn’t make a site reliable (UC Merced Library).

Find better coverage

When you care more about the claim than about this particular source, isolate the claim, find the best source you can on it, and see what reputable sources agree on. You don’t have to accept that consensus, but start from it (Princeton University Library). Fact-checking sites such as Snopes and PolitiFact can help (UC Merced Library).

Trace claims, quotes and media

Find the original context and check whether the claim, quote or image was presented accurately. With media, ask whether it was edited, manipulated or clipped in a misleading way (Princeton University Library).

An earlier version, Caulfield’s “four moves and a habit,” was taught at colleges through the Digital Polarization Initiative. Its habit was to check your emotions (Brodsky et al., 2021).

Lateral reading: how fact checkers do it

Sam Wineburg and Sarah McGrew of the Stanford History Education Group watched 10 professional fact checkers, 10 historians with PhDs and 25 Stanford undergraduates search the live web. One task compared bullying information from the American Academy of Pediatrics with information from the American College of Pediatricians. Most historians rated both sites reliable, and most students rated the American College of Pediatricians site as more trustworthy. The fact checkers quickly worked out where that group stood, and overall they reached sounder conclusions in a fraction of the time (Stanford GSE). The study was later published in Teachers College Record (Digital Inquiry Group).

Historians and students read vertically, staying on the site and judging it from the inside. Fact checkers glanced at the site, then opened more tabs to see what others said about it (Stanford GSE). To try it yourself (Princeton University Library):

1. Open a new tab and search for the organization. Adding -site: with its address leaves out the site’s own pages.
2. See what Wikipedia and major news outlets say about it.
3. Remember that a site’s About page is usually written to make it look good.
4. Check specific claims on fact-checking sites. If a page says “research shows,” look for that research.

Lateral reading can be taught. College civics students taught Caulfield’s moves were more likely than a comparison group to read laterally and to judge trustworthiness correctly, though the authors noted it’s unknown whether the gains last (Brodsky et al., 2021). A cluster-randomized study in high school government classes across an urban district found that students improved significantly too (Digital Inquiry Group).

The CRAAP test

Sarah Blakeslee introduced the CRAAP test in LOEX Quarterly in fall 2004 (LOEX Quarterly). Meriam Library at California State University, Chico publishes it as a list of questions and notes that different criteria matter more or less depending on your needs (Meriam Library, CSU Chico):

Criterion	Sample questions
Currency	Has it been revised or updated? Does your topic need current information? Do the links work?
Relevance	Does it answer your question? Who is the audience? Is it at the right level?
Authority	What are the author’s credentials? Is the author qualified on this topic? Is there contact information?
Accuracy	Where does the information come from? Is it backed by evidence? Was it reviewed or refereed? Can you verify it elsewhere?
Purpose	Is it meant to inform, teach, sell, entertain or persuade? Is it fact, opinion or propaganda? Are there biases?

Princeton’s guide credits the test to “University of California Chico” (Princeton University Library), but the original handout is from California State University, Chico (Meriam Library, CSU Chico).

How the methods compare

Compare	SIFT	Lateral reading	CRAAP test
Where you look	Mostly off the page	Off the page	Mostly on the page
Time	Quick moves	Quick	Longer checklist
Best for	Everyday claims, posts, news	Unfamiliar sites and organizations	Deciding if a source fits a research paper
Weak spot	Less help with fit and relevance	Depends on searching well	Can keep you stuck on one page

The methods mainly disagree about where to look. Lateral reading has you leave the page fast, while checklists like CRAAP encourage close reading of the page itself (Brodsky et al., 2021). CRAAP examines one resource at a time, which can cut it off from its context (Princeton University Library). In a 2019 report, Stanford researchers said checklist methods can mislead students by focusing them on a single site instead of the wider web (Stanford Report).

The URL question shows the gap. CRAAP asks whether an ending such as .org or .edu reveals anything (Meriam Library, CSU Chico), yet in the Stanford study a nonprofit URL was one of the cues that fooled vertical readers (Stanford GSE). The lateral reading research is newer and comes from watching people judge real websites, so give it more weight on credibility. Use SIFT or lateral reading first. If you’re writing a paper, then use CRAAP’s currency and relevance questions to decide whether a source you already trust fits your project.

Quick checks by type of source

News

• Look for labels such as “op-ed” or “opinion contributor.” Opinion pieces can sound authoritative while misrepresenting their sources (Princeton University Library).
• Compare several outlets to see which facts they all report, and with breaking news, watch how the facts get updated (same source).
• Find out who owns the outlet, and be wary of an angry or alarming tone that discourages questions (same source).
• Check who funds a site. In a 2019 Stanford assessment of 3,446 high school students, more than 96% didn’t consider that a climate website’s ties to the fossil fuel industry might affect its credibility, though a basic search revealed them (Stanford Report).

Studies, and news stories about studies

Questions from the National Institutes of Health (NCCIH):

• Was it tested on people, or only in animals or a lab?
• Were enough people studied? Small studies give weaker results.
• Was the effect big enough to matter to you?
• Is this the first study to report it? One study rarely proves anything.
• Was it funded by a group that could profit from the result?

Also look for a correction, an “expression of concern” or a retraction notice on the article page (Princeton University Library).

Images and video

A reverse image search can show where an image came from and whether credible outlets have reported it as manipulated. In Google Lens, upload the image or paste its URL, then check the matches and the “About this image” panel (Princeton University Library). Question captions too. In the 2019 Stanford assessment, 52% of students called a Facebook video “strong evidence” of U.S. voter fraud, but it showed ballot stuffing in Russia (Stanford Report).

Common traps

• Trusting the look of a site. A polished name and logo, lots of scholarly references or a nonprofit URL fooled readers in the Stanford study (Stanford GSE).
• Clicking the first search result. Fact checkers scanned the results before choosing, while others clicked top-ranked links without thinking about how sites game search rankings (same source).
• Sharing in the heat of the moment. A strong emotional reaction is a reason to stop (UC Merced Library).
• Going easy on claims you like. Check those hardest; our guide to confirmation bias explains why that’s difficult.

For the rest of the toolkit, see how to improve your critical thinking and our list of logical fallacies.