References?
Do I need a reference for that?
When was the last time a coauthor, colleague, or friend asked you, “Do you have a reference for that?” For researchers, this is probably the equivalent of everyday questions like “How do you know?” ‘Who told you that?” or “What’s your point?” How did you respond? My typical replies are “yes,” “no,” or “do I need one?” and “Who wants to know?” The context is that “You know what you want to say, but sometimes you need some backup.”
When you are writing a manuscript for publication, the key question is, “Do I need one?”—and you need to answer it before anyone has to ask. That means you must be clear about the purpose of references. In my view, they serve two main functions: to provide backup and to offer clarification or illumination.
I once considered submitting to a journal whose author guidelines limited how references could be used. The rules did not put a cap the total number of citations, but they did tell authors to leave out any sources that did not clearly support the rationale for the study in the Introduction or underscore the significance of the results in the Discussion. The journal stressed that citations should be used to substantiate claims based on prior data or to connect your findings to earlier work. It discouraged the use of references to provide a broad overview of the field or how much you know about it.
As for how recent the references should be, the suggestion was to prioritize original research from the last 5 years and to include review papers, which often point to older but still “key” publications. Some reviewers and editors are suspicious of citations that are more than 10 years old (perhaps because they themselves have published newer work that you did not cite). During peer review, you may also be asked to add more references when you revise and resubmit. Reviewers may simply disagree with you about what counts as common knowledge and what should be supported by a citation. That’s not a big problem—just identify an appropriate reference and add it. Sometimes reviewers request extra citations so readers can more easily explore certain issues you mention. And occasionally—especially if their research background is close to yours—they will want you to cite a particular statement because it relates to something they have published, thereby generating a citation to their own work. This really does happen!
Even if you are good at searching for references that you need for a systematic review or meta-analysis, or to use in an article about your original research, your favorite AI assistant will find things much faster than you can. It is trained to look past the keywords (or metadata) that you use to guide the your database search (e.g., PubMed or Web of Science) search. AI systems are trained to follow “plain text” instructions to find potentially usable references. It searches for words used in the title and abstract, author names and affiliations, keywords, and if not subject to copyright restrictions, the full text of the articles in its “memory.” Remember AI is an assistant or “droid” that helps you do things. Think of it as something like Siri, Alexa, or Hey Google but closer to C3PO or R2D2. “It” may ask for clarification if it finds inconsistencies, but the quality of its work is only as good as the questions you ask. Why not ask your AI assistant how he/she/it follows plain text instructions to search for the references you are looking for? Then try asking a few questions for which you already know the answers. You know, just like a pop quiz.
One last thought. Why do references actually matter? Because they give your work context and credibility. They sharpen your hypothesis, make your reasoning transparent, and shape how your findings are understood. In short, they’re the bridge between what’s already known and what you’re adding. For me, their real power isn’t in propping up your research, but in illuminating it—showing clearly what’s new, what’s different, and why your contribution counts. You know, your article will be someone else’s reference.