Evaluating sources: how to judge what belongs in your essay
Evaluating a source means judging whether it can carry the specific claim you want to make, and that judgement rests on five questions: authenticity, currency, independence, bias and relevance. At university level the useful question is not whether a source is true but what kind of statement it can support – an opinion piece is an excellent source on what someone thinks and a terrible source on what is the case. Always start by working out whether the source is primary, secondary or tertiary.
Key takeaways
- Source evaluation judges fitness for a specific claim, not truth in the abstract.
- The five criteria are authenticity, currency, independence, bias and relevance.
- Peer reviewed means reviewed, not correct – it lowers the error rate and guarantees nothing.
- Never cite a source you have not read in the original as though you had.
What does it mean to evaluate a source?
Evaluating a source is deciding how much weight a claim can bear given where it came from. It is not an exercise in suspicion and not a sorting into true and false, but a test of fit: which question can this particular source answer?
That distinction is the whole point. An advert from a pharmaceutical company is a poor source on whether a drug works and a first-rate source on how the company markets it. The source does not get better or worse – the question does.
What is the difference between primary, secondary and tertiary sources?
A primary source is first-hand material: the original study, the legislation, the interview, the dataset. A secondary source interprets or summarises primary material, such as a review article or a textbook. A tertiary source compiles secondary sources, such as an encyclopaedia or Wikipedia.
The rule in an essay is to argue from primary sources and use secondary sources to orient yourself. Tertiary sources belong in the reading phase and never in the reference list.
The five criteria
The criteria come from historical method but apply across every discipline. Each one is a question you put to the source, not a score you give it.
| Criterion | The question you ask | Warning sign |
|---|---|---|
| Authenticity | Is the source what it claims to be? | No named author, no publisher, no contact details. |
| Currency | How close is it to the event or the research front? | No date at all, or a figure recycled for twenty years. |
| Independence | Is it independent or repeating someone else? | Five sources that all trace back to one original report. |
| Bias | Does the author gain from a particular conclusion? | Commissioned research with undisclosed funding. |
| Relevance | Does it answer your question specifically? | Right topic, wrong population, wrong country or wrong decade. |
What does peer reviewed actually mean?
Peer reviewed means two or three researchers in the field read the article before publication, and nothing more. It is a filter that lowers the error rate, not a certificate of correctness. Ioannidis (2005) argued in one of the most cited methods papers ever written that a substantial share of published findings fail to replicate, and the psychology replication project (Open Science Collaboration, 2015) reproduced the original effect in roughly a third to a half of the studies examined, depending on the measure used.
The practical conclusion: treat a single peer-reviewed article as a reasonable but provisional claim, and lean on systematic reviews and meta-analyses when you need something that holds.
How do I spot a predatory journal?
Predatory journals charge a publication fee without real review and look like genuine journals at a glance. The signals are consistent: an email invitation to publish, review promised within days, an editorial board with unverifiable affiliations, an implausibly broad subject scope, and the fee mentioned before the guidelines. Check the journal in the Directory of Open Access Journals or your library's list before citing it – your subject librarian will do that check for free in minutes.
How do I evaluate a news article about research?
Find the original. A news article about a study is a secondary source, usually based on a press release, which is itself based on a paper. Every step can drop a caveat. Sumner et al. (2014) found that much of the exaggeration in health news was already present in the universities' own press releases – the distortion happens before the journalist. Find the study, read the abstract and the methods, and cite the study.
Which number traps should I watch for?
Three recur in student essays. The first is correlation reported as cause, usually through the phrase "leads to". The second is a relative figure with no absolute one: a doubled risk from 1 in 100,000 to 2 in 100,000 is statistically true and practically irrelevant. The third is the sample – a survey answered by those who wanted to answer measures something other than the population. Always state the sample size in your own text when you cite a study.
How do I write the evaluation into the essay?
Source evaluation belongs in the methods section, and it has to be concrete. State which databases you searched, which search strings you used, which inclusion criteria you applied, and which source types you excluded and why. A line saying "sources were critically evaluated" earns nothing; a line saying "articles published before 2015 were excluded because the measurement instrument changed in 2014" is a defensible methodological choice. Our guide to choosing a method and the dissertation guide cover how the section is built.
What does this look like step by step?
Start by writing the claim you want to support in one sentence. Search an academic database rather than a general search engine. Read the abstract and methods before the results. Check whether the source is primary. Apply the five criteria. Record the full reference immediately, because it is always harder to find again than you expect – our Harvard guide shows what the entry should contain.
Once you have a stack of articles to get through, reading them actively beats reading them passively. Upload them to Memmo and let the summary tool give you a way in to each one before you read the original – the technique is described in our article on reading academic texts.
Frequently asked questions
How old can a source be?
It depends on the field, not on a fixed number of years. In medicine and engineering five years is often the limit for empirical results, while in the humanities a hundred-year-old text may be the obvious primary source. Ask instead whether anything has happened in the field since publication that changes the conclusion.
Is peer review enough on its own?
No. Peer review filters out obvious flaws but guarantees neither correctness nor replicability. Treat it as a minimum requirement, not as your only criterion.
Can I cite an AI tool as a source?
No. An AI answer is not a source, because a reader cannot retrieve or verify it. Use it to find concepts and search terms, then cite the underlying sources you have read yourself.
What do I do when two credible sources disagree?
Report the disagreement rather than quietly dropping one side. Compare samples, methods and dates, and say what the difference might come from. An essay that shows where the field is unsettled marks higher than one that pretends it is settled.
Are government reports reliable sources?
They are usually reliable on facts and statistics, but they have a remit, and the bias criterion applies to them too. Use them as primary sources on what was measured and decided, and be more cautious with their conclusions about the effectiveness of their own programmes.