How to Find Credible Sources for a Research Paper

What counts as a credible source?

Credible means a qualified author wrote it, somebody reviewed it before it went out, and a real organisation has its name on the masthead and can be held to account if it's wrong. Four buckets do most of the work: peer-reviewed journal articles, academic books and book chapters, reports from recognised institutions (governments, intergovernmental bodies, major research institutes), and primary data like statistics, court rulings, or archival documents.

Credibility isn't a yes/no. It's a dial. A working paper from a serious economist beats a tabloid piece. A peer-reviewed version of the same paper beats the working paper. Good researchers weight their citations by where each one sits on that dial. Treating every source as equal is how bad arguments get built.

Primary, secondary, and tertiary sources

Before you open a database, work out which kind of source you actually need. Most papers want a mix of all three. The balance shifts depending on your field and the argument you're making.

Where to search for credible sources

Google is almost never where you should start academic work. The databases below index peer-reviewed scholarship across disciplines, and most are free either to the public or through a library subscription you already have.

General-purpose academic databases

• Google Scholar — broadest coverage; good first pass
• Semantic Scholar — AI-powered relevance and citation graphs
• OpenAlex — open metadata for 200M+ scholarly works
• Crossref — the DOI registry; reliable for canonical metadata
• CORE — open-access full-text aggregator

Discipline-specific databases

• PubMed — biomedicine, nursing, public health
• arXiv — physics, mathematics, computer science preprints
• PsycINFO — psychology and behavioural sciences
• JSTOR — humanities and older journal archives
• IEEE Xplore — electrical engineering and computing
• ERIC — education research

Short on time? Paste the claim into our Find Source tool. It hits Crossref, Semantic Scholar, OpenAlex, PubMed, and arXiv in parallel and hands back the most relevant peer-reviewed papers in seconds.

Using Google Scholar effectively

Google Scholar is the most-used academic search engine on the planet. Most students still treat it like regular Google. A handful of small habits change the quality of what you pull out.

Use quotation marks for phrases

Quotes force an exact-phrase match instead of a loose word-bag. Try self-determination theory against "self-determination theory" — the first dumps thousands of vaguely related hits, the second returns work that actually uses the phrase.

Filter by year

Machine learning, medicine, climate science: cap results at the last five years using the sidebar filter. For foundational theory in slower-moving fields, skip the filter. Sometimes you actually want the 1970s paper.

Follow the citation trail

Scholar prints a citation count under every result. Big numbers are a strong tell for foundational work. The real trick is the Cited by link. Click it and you walk forward through time, finding newer papers that built on the one you already trust. That's forward citation searching. Related articles does the same job sideways.

How to evaluate a source (the CRAAP test)

Got a candidate paper? Run it past a quick quality check before you cite. The CRAAP test, originally from California State University Chico, is the heuristic most academic writing courses still teach.

A source doesn't need a perfect score on every line. A ten-year-old paper can still be the right citation if it's the foundational study. A corporate white paper is fine to cite if you're upfront about who wrote it and why. CRAAP is a prompt for judgment. It isn't a pass-fail filter.

Red flags: sources to avoid

Predatory journals

Some "open-access" outlets will print almost anything for a fee. No real peer review. Tells: aggressive solicitation emails, suspiciously fast turnaround, an editorial board you can't verify, journal names that sound like an established title with one word swapped. Cross-check anything you're unsure about against the Directory of Open Access Journals (DOAJ) or the Committee on Publication Ethics (COPE) member list.

Retracted papers

Papers get retracted. Fraud, error, ethics violations. They keep showing up in search results anyway. Before you cite, run the title through Retraction Watch and check the publisher's page for a retraction notice.

Wikipedia as a citation

Wikipedia is a great starting point. It's not a citable one in most academic settings. Use it to get oriented and pick up keywords, then follow the footnotes to whatever the article is drawing on. Cite those.

Content farms and SEO-optimised blogs

Generic "Top 10 facts about X" posts almost never cite anything primary. They mostly recycle each other's mistakes. If a page won't name its author or where it got its claims, assume you can't cite it.

Anonymous AI-generated content

Treat unsourced AI text the way you'd treat an anonymous blog. Brainstorm with it, summarise with it, never cite it. When you do cite, cite the underlying source the model pulled from. And only after you've checked it exists and actually says what you think it says.

How many sources do you need?

No universal rule here. Instructors and journals want enough sources to prove you've read the literature. They don't want a padded bibliography.

Quality beats count, every time. Ten tightly-chosen, on-target sources will outperform thirty citations dropped in for weight. Instructors can spot padding from the first page.

A practical search workflow

Don't think of source hunting as one search. Think of it as a staged process. The workflow below holds up for essays, theses, and review articles.

1. Orient. Read a Wikipedia entry, a textbook chapter, or a handbook article to map the subfield and collect keywords.
2. Anchor. Find one or two recent review articles or meta-analyses on your topic. These are compressed maps of the literature and are full of useful citations.
3. Backward search. Read the bibliographies of your anchor papers and pull the papers that keep being cited.
4. Forward search. Use Google Scholar's "Cited by" link on your best anchor paper to find newer work that builds on it.
5. Fill gaps. Run targeted database searches for any sub-claims that still lack direct support.
6. Evaluate and cull. Apply the CRAAP test. Drop anything redundant or weak.
7. Organise. Store everything in a reference manager (Zotero, Mendeley, or CiteGenie's library) so you can cite as you write rather than at the end.

Quick summary

Finding credible sources is a handful of small habits stacked together. Search in the right databases. Use quotes. Walk the citation trail. Run a CRAAP check before you commit. Once that loop is automatic the rest of the paper gets easier, because a well-sourced draft mostly writes itself.