How to Evaluate an Online Source

The mistake almost everyone makes

Faced with an unfamiliar website, most people evaluate it by studying it. They look at the design, read the About page, check whether it has a professional logo, and see whether it cites sources.

Every one of those signals is controlled by the site itself. A site that wants to appear credible can produce all of them cheaply. Reading down the page tells you what the publisher chose to say about themselves.

Wineburg and McGrew compared how professional fact checkers, academic historians, and undergraduates evaluated unfamiliar websites. The fact checkers reached better conclusions faster, and the reason was procedural rather than intellectual: they left the site almost immediately and looked for what independent sources said about it. The researchers called this lateral reading, in contrast to the vertical reading the other groups performed [1].

The method

Step 1: Leave the page within thirty seconds

Note the site or account name. Then open a new tab. You are not looking for the site’s own account of itself.

Step 2: Ask who is behind it

Search the organization’s name together with a neutral term such as “funding”, “founded”, or the name of a person listed on the masthead. You are looking for coverage, registry entries, academic references, or encyclopedia entries produced by someone other than the site.

Three outcomes are common, and each is informative:

• A clear record exists. You learn the organization’s history, ownership, and orientation. This is the best case, and it takes under a minute.
• Almost nothing exists. A site presenting itself as an established institution with no independent footprint is a meaningful signal. It does not prove bad faith, but it removes the basis for treating it as established.
• A contested record exists. Different sources characterize it differently. Note the disagreement rather than picking the first result.

Step 3: Ask what it is for

Organizations publish for reasons. A trade association exists to advance an industry’s interests. An advocacy group exists to advance a position. A university research centre exists to produce and publish research. A commercial content site exists to sell attention.

None of these purposes disqualifies a source. They tell you what kind of selection pressure to expect. An industry association’s data may be accurate and its framing still systematically favourable.

Step 4: Trace the specific claim

Evaluating a source and evaluating a claim are different tasks, and the second is often faster. If the page cites a study, find the study. If it cites a report, find the report. Check three things: that it exists, that it says what is claimed, and that its scope matches the use being made of it.

Most misleading content fails at this step rather than at step two. The underlying source is real; the characterization of it is not.

Step 5: Write down what you concluded and what you still do not know

Two sentences is enough. “This is an advocacy organization founded in 2019, funded by a source I could not identify, citing a real 2023 study that it describes accurately. I do not know who funds it.”

That is a usable result. It is not a verdict of trustworthy or untrustworthy, and it should not be.

What lateral reading does not do

It does not tell you whether a specific factual claim is true. It tells you what kind of source you are dealing with and what interests are in play, which changes how much independent corroboration you should want before accepting the claim.

It also does not scale to every item you encounter. Use it when the stakes justify two minutes: before sharing, before acting, before citing.

Evidence that this is teachable

Guess and colleagues tested a short digital media literacy intervention, derived from a publicly circulated list of tips for spotting false news, in surveys in the United States and in a field experiment in India. Exposure to the intervention was associated with an increase in respondents’ ability to discern mainstream from false news headlines in the United States and in the online sample in India, though the effect was not observed in the in-person rural India sample [2].

Two things follow. Short, practical instruction does measurably change discernment for many people. And the effect is not universal across contexts, which is a reason to teach a method rather than a list of warning signs.

Worked example

You encounter an article on a site called the Centre for Regional Economic Progress, arguing that a proposed local tax will cost 4,000 jobs.

Reading laterally: a search returns no independent coverage of the organization, but does return a state business registry entry showing incorporation eight months ago. The article credits a study to a named consultancy. Searching the consultancy plus the study title finds the study, which models a scenario in which the tax is applied at three times the proposed rate.

You now know the number is real, the study is real, and the application of it to this proposal is not supported. That conclusion took about three minutes and did not require you to have an opinion on the tax.

Practical exercise

Pick three sources you have relied on in the past month, at least one of which you consider reliable. Spend two minutes on each, using steps one to three only. Write two sentences per source.

The reliable source is the important one. Lateral reading is not a tool for confirming suspicion. It is a tool for knowing what you are dealing with, including when the answer is reassuring.

Summary checklist

• I left the page before forming a judgment about it.
• I looked for what independent sources say about the publisher.
• I identified the organization’s purpose and likely interests.
• I traced the specific claim back to the source it cites.
• I wrote down both what I concluded and what I still do not know.

References

Numbered citations in the text above correspond to the entries below.

1. 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://journals.sagepub.com/doi/10.1177/016146811912101102
2. Guess, A. M., Lerner, M., Lyons, B., Montgomery, J. M., Nyhan, B., Reifler, J., & Sircar, N. (2020). A digital media literacy intervention increases discernment between mainstream and false news in the United States and India. Proceedings of the National Academy of Sciences, 117(27), 15536-15545. https://doi.org/10.1073/pnas.1920498117

Next step

Try it on three sources in the Think Lab

Topics covered: Media LiteracyDigital Literacy