Students working at table

Principles of Citing Sources: Appraising and Choosing Sources

At a Glance

  • It can be difficult to distinguish between a scholarly source and a popular one. In general, scholarly sources have been approved by a group with recognized expertise in the field under discussion. All other sources can be generally described as popular sources.
  • While writing, there are times you might rely on popular sources and times where scholarly sources may be better to cite.
  • What qualifies as evidence or as a reliable source varies across different disciplines, and sometimes among different courses in the same discipline. It’s very important to check with your teachers about appropriate sources for the research you’ve been assigned.
  • Using generative AI complicates nearly every aspect of research writing. No instructor at Yale will consider AI output to be authoritative, so you cannot cite it as a scholarly source. At the same time, while many instructors will direct you not to use AI at all, Yale College requires that if you do use it—you attribute and cite any such usage. If you are reading this page for help doing careful research and scholarship, you should probably avoid using AI as an important source.

On This Page

Scholarly vs. Popular Sources

There are two types of sources that you can cite: scholarly sources and popular sources.

  1. Scholarly sources: Scholarly sources are those that have been approved by a group with recognized expertise in the field under discussion. Books published by university presses fall into this category, as do articles published in peer-reviewed journals—journals where the editors send pieces out to be read by experts in the field before deciding to publish them. The Yale library subscribes to several databases that specialize in scholarly sources (such as Academic Search Premier). If you use sources for facts or ideas in your writing, some research projects will demand that you rely heavily, if not exclusively, on scholarly sources. Scholarly sources are not infallible, but their publication process includes many steps for verifying facts, for reducing political bias, and for identifying conflicts of interest (for instance, informing readers when a drug company has funded research on its own product).
  2. Popular sources: In a narrow sense, every other source that cannot be classified as scholarly could be called a popular source. But this does not mean that all popular sources are of equal reliability. Nor does it mean that you should use only scholarly sources for all of your writing at Yale. Depending on the research context, some projects will permit a mix of scholarly and popular sources. For example, in a history seminar about World War II, you would usually be expected to consult the most definitive, academic studies of the period. But in a literary essay analyzing Joseph Heller’s Catch-22, you may be able to rely on popular history sources for facts about World War II because the focus of your paper is on interpreting the novel.
Man works on laptop

To understand this difference, it may help to recognize that when you write for a departmental class, you are writing for an audience more expert than the common reader. When writing about Republican and Democratic voting patterns on stem cell research, science reports from a general interest magazine like Newsweek might be considered sufficiently authoritative. But when you write about stem cells for a Biomedical Engineering class, your teachers and classmates are considered more expert than the average non-scientist reader. For this audience, your sources must have stronger science credentials and must generally be working scientists, not journalists.

Classifying Journalistic and Online Sources

Journalistic sources are not scholarly sources, but they can still be useful sources for research. Responsible journalism can provide timely reporting, interviews, documents, data, and investigations that may not be available in scholarly publications. Its claims are ordinarily produced through different procedures from scholarship, however, and should be evaluated accordingly.

A publication’s reputation can offer useful context, but it should not substitute for examining the particular article. The same publication may contain reported news, analysis, opinion, reviews, sponsored material, and material supplied by outside organizations. Before using an article, identify what kind of writing it is and how its claims are supported. Useful signs of accountable reporting include a named author, clearly identified sources, links or references to relevant documents and data, a distinction between reporting and opinion, editorial oversight, and a visible process for correcting errors. None of these features guarantees that an account is accurate, but they make its claims easier to assess and verify.

Every source also has a stake in what it presents. A news organization may have commercial or political incentives; an institution may be defending its policies; an advocacy group may be trying to persuade; and an individual may be reporting from limited experience. Having a point of view does not by itself make a source unusable. Ask whether the source’s interests are apparent, whether its evidence supports its claims, and whether important factual assertions can be confirmed through independent sources.

As recently as twenty years ago, instructors would routinely distinguish between print and online sources in terms of their reliability. But those categories are no longer very meaningful. Instead of treating online sources as inherently less reliable than print sources, determine who produced the material, what kind of publication it is, what evidence it provides, and what editorial or review process stands behind it.

Some sources that may be unreliable for factual claims may still be valuable for other purposes. A partisan article, organizational website, advertisement, or social-media post may offer important evidence of what a person or group believes, how it represents itself, or how it attempts to influence an audience. Use such material as evidence of that viewpoint or rhetorical activity—not automatically as independent confirmation that its factual claims are true.

Using Sources as Objects of Analysis

Music book open

Any source can be cited if it’s being used as what we might call an “object” source—for instance, if the text is an object of analysis, such as the poem in a literary criticism paper. In these instances, you are not relying on the source for authoritative information about a topic, but are instead investigating the source itself and using paraphrase and quotation to argue about the source’s meaning. Sources that serve as a project’s object of analysis can also be called primary sources. Depending on the course, academic papers might use movies, commercials, buildings, magazine advertisements, popular songs, or dialogue recorded on a city bus as their primary source. It’s still necessary to identify the origin of your source, but if your paper makes clear that you do not accept the text as a source of fact but are instead developing your own ideas about the text and how it works, you can profitably use almost anything in an essay. Essays or other works that discuss and analyze a primary source are called secondary sources. Often, papers written for a class will cite only one or two primary sources and multiple secondary sources. 

Assessing the Reliability of Sources of Opinion

Depending on the research context, you may have sources that are not as reliable as scholarly sources but that still add texture or authority to your argument. In the humanities, for instance, critics’ opinions often demonstrate bias (even celebrate it). But because of their authors’ professional experience, such opinions can be used as ideas to extend your own analysis—or to serve as counter-arguments that clarify your own claims. When critics highlight the same element of the text that you focus on, for instance (the ending of a play, the entrance to a building, the color scheme in a movie), their attention can serve as evidence that the component is significant, even as you use your paper to argue for a different interpretation. See Why Cite? for more discussion of counter-argument. Some magazines—like Discover—are more authoritative than a general news magazine but less authoritative than a scholarly journal. In the sciences, studies done by interested parties would normally not be accepted uncritically (such as pollution studies commissioned by chemical companies). But they may still identify key elements of the topic that require attention.

Finally, almost any source can be used as evidence that someone believes the idea you quote. Although that may seem self-evident, such evidence can be surprisingly helpful when developing an argument, especially for articulating the research problem of your essay. If the Harvard Crimson publishes a negative article about Yale’s football team, you probably can’t trust that it’s an unbiased assessment. But such an article can still be cited as evidence that “some people have negative opinions about Yale football.” It can also be cited as evidence that “Football matters enough at some Ivy schools to merit coverage in the campus newspaper.” If you compare it to other articles in the issue that describe cross-campus cooperation, then the football article might be used to suggest “Feelings are divided at Harvard about its rivalry with Yale.” In the sciences and social sciences, you might discuss popular, non-expert representations of a key issue, explaining where they go astray—and therefore why your paper is necessary.

Generative AI Tools

Generative AI tools can produce a variety of outputs, including texts, code, images, and other content based on vast datasets that include unverified information. When using generative AI tools, it’s important to evaluate them the same way you would any other source. Although generative AI tools can seem helpful when doing research, you should always approach them with a critical eye since they may mislead you or cause you to plagiarize.  

AI-generated material tends to lack the depth, context, and peer-reviewed validation that we see in scholarly sources. AI tools can generate what are called hallucinations —outputs that are inaccurate, false, or nonsensical. Hallucinations can range from minor inaccuracies to entirely fabricated sources, including realistic-looking citations to articles, journals, or books that do not exist. Therefore, it’s crucial to assess the accuracy, reliability, and relevance of AI-generated output before incorporating it into your work, just like any other source.