How to Find and Screen Sources for a Research Paper

A folder full of downloaded articles can feel like progress until you start writing and realise that half of them do not answer your question. Some repeat the same study. Others mention your topic but offer little evidence you can use.

Finding sources is only the first step. You also need to decide which ones belong in your paper, assess what they contribute and keep enough notes to explain your choices. This guide walks through that process using a student research example.

1. Start with a Question You Can Answer

Before searching, turn your broad topic into a focused question. “Sleep and students” could lead to thousands of loosely related results. A more useful question is:

This identifies a population, a factor of interest and an outcome. It also asks about an association rather than assuming that one causes the other.

Check your assignment brief before setting limits. Does it require primary research, recent peer-reviewed articles, a particular country or a minimum number of sources? These requirements should shape your search. If you are still shaping your question, see our guide to writing a strong research question.

2. Turn the Question into Search Terms

Databases may describe the same idea using different words. Break your question into concepts and list alternatives for each.

In databases that support Boolean searching, use OR between alternatives and AND between different concepts. Cornell University Library’s research tutorial explains this approach to combining keywords.

An example starting search is:

("sleep duration" OR "sleep length" OR "hours of sleep")
AND ("university students" OR "college students" OR undergraduates)
AND ("academic performance" OR "academic achievement" OR grades OR GPA)

Treat this as a starting point. Check your database’s instructions because phrase searching, field options and other search features vary. Review relevant results for additional terms, then refine your search.

If you get very few results, remove an unnecessary restriction or try a broader term. If you get too many unrelated results, search specific fields such as title and abstract, or narrow your question.

3. Choose Where to Search

Start with your university library’s subject guide. It can help you identify databases suited to your discipline and access material through your institution. A librarian can also help when your searches return little useful evidence.

Google Scholar can supplement your database searches and help you follow references or find papers that cite a useful article. Check the actual publication before treating a search result as suitable evidence. Our guide to finding credible academic sources explains how to judge whether a source can be trusted.

Once you find a relevant paper, examine its reference list. Following these references can reveal earlier research and foundational work. Record sources discovered this way so you remember how you found them.

4. Keep a Search Log

Record each useful search while you are working. Include the database, search date, exact search terms, filters and result count.

The following log is fictional and illustrates the format; it does not report a real search.

Record the exact result count when available. If a platform displays an estimate, label it as approximate. Keep the count returned by the search separate from the number of records you actually saved.

A search log helps you avoid repeating unsuccessful searches and makes it easier to explain or update your approach later.

5. Agree Your Selection Criteria

Decide what qualifies as useful evidence before you start choosing papers. For the sleep example, you might include research that examines university students, measures sleep duration and reports an academic outcome.

You might exclude studies focused only on schoolchildren or papers that discuss sleep without examining academic performance.

Use date and language limits only when they fit your assignment and question. An older source may still be important, and a recent publication is not automatically better evidence. If you change your criteria after exploring the literature, record what changed and why.

Avoid selecting only papers that support the argument you hoped to make. Relevant findings that challenge your expectations can strengthen your analysis.

6. Screen Titles and Abstracts

Use an initial screening pass to decide which papers deserve closer reading. Mark each record as:

• Retrieve for full-text review: it appears relevant.
• Exclude: it clearly falls outside your criteria.
• Unsure: you need more information.

For example, a study of sleep duration and university examination scores would be worth retrieving. A study of sleep in preschool children would fall outside the example population. A vague abstract should remain “unsure” until you can investigate further.

An abstract is a summary. Do not use it alone to make detailed claims about a study’s methods, limitations or findings.

7. Read the Full Text and Assess Quality

Check whether each retrieved paper actually meets your criteria, then assess how much confidence you can place in its evidence.

Ask whether the methods suit the research question, how participants were selected, how key variables were measured and whether the conclusions follow from the results. Note limitations that affect how you will use the source. For help breaking a paper down, see how to analyze an article, and for a refresher on research designs, see how to choose a research methodology.

In the sleep example, self-reported grades and official university records provide different kinds of measurement. A study conducted at one institution may have limited relevance elsewhere. A cross-sectional association also does not, by itself, establish causation.

Keep relevance and quality separate in your notes: a paper may directly address your topic while still requiring cautious interpretation.

8. Record Decisions and Keep Counts Clear

Give each saved source an identifier, such as S01 or S02. Record its citation, screening decision, reason and brief notes about its contribution.

These entries are fictional examples. “Not retrieved” means you could not assess the full text; keep it separate from a paper you read and excluded.

Check for duplicate records when combining searches. Also remember that two different publications can report the same underlying study. They are distinct reports, but should not automatically count as independent studies.

For formal systematic reviews, PRISMA guidance distinguishes records, reports and studies, and separates reports not retrieved from exclusions after assessment. Ordinary coursework may need a much simpler record. Follow your course’s requirements for the type of paper you are writing.

9. Turn Your Reading into an Argument

Once you have selected your sources, group them around themes, findings, methods or disagreements. Compare what they show rather than writing one disconnected summary per article.

For example, you could compare studies that use self-reported sleep with those using objective measures, or discuss whether findings differ across student populations. Keep citation details and page numbers alongside your notes so you can check every claim when drafting.

For help organising the resulting paper, see How to Write a Research Paper: Format & Structure. If your course asks for a source-by-source write-up, see our guide to writing an annotated bibliography, and to synthesise your sources by theme, see how to write a literature review.

Keep Your Search and Screening Records in One Place

If you want a ready-made structure for these steps, The Apex Scholars’ Research Search & Screening Tracker includes a printable PDF, an editable Word workbook and five CSV logs covering searches, records, title-and-abstract screening, full-text screening and count reconciliation.

Use it to document your searches, record decisions and review your totals while you work. You carry out the searching and screening yourself; the tracker provides the forms for organising the process.

Explore the Research Search & Screening Tracker →

The tracker is a manual recordkeeping tool. It does not automatically search databases, remove duplicates or assess sources, and it does not guarantee PRISMA compliance.

Frequently Asked Questions

How many sources should a research paper have?

Follow your assignment brief. Choose enough relevant evidence to support your discussion and address significant competing findings. A longer reference list does not compensate for weak analysis.

Can I cite a paper if I have only read its abstract?

Obtain the full text before using it to support detailed claims. If it remains unavailable, find accessible evidence or ask your instructor how to handle the situation. Never imply that you assessed a paper you could not read.

Do I need a PRISMA diagram for a normal research paper?

Only if your assignment or chosen review method requires one. PRISMA provides reporting guidance for systematic reviews; keeping a search log alone does not make a paper a systematic review.

What should I do when sources disagree?

Compare their populations, methods, measurements and limitations. Explain the disagreement and judge what the evidence supports rather than choosing whichever conclusion you prefer.

Further Reading

• Cornell University Library: Getting Started with Library Research
• Cornell University: Research tutorial on searching with keywords and Boolean operators
• PRISMA: Official reporting guidance and resources