03 · What You Need to Know
Deciding What to Read Is Part of Research Reading
Do not begin by asking whether the paper is "good"
A common instinct is to ask, "Is this a good paper?" before deciding whether to read it. That question is usually too broad.
A methodologically rigorous paper may have little relevance to your research question. A paper with important limitations may still deserve close attention because it introduced an influential method, shaped a debate, produced a surprising result, or is repeatedly cited in the literature you are studying.
Likewise, a paper does not become worth reading merely because it was published in a prestigious journal, has accumulated many citations, or appears near the top of a database search.
A more useful first question is:
What could this paper contribute to the task I am performing?
That shifts the decision from an abstract judgment of quality toward a purpose-driven judgment of relevance and potential value.
First decide whether the paper is actually relevant
The title and abstract are usually the most efficient places to begin. Guidance on reading scientific manuscripts recommends using them for initial screening because a paper returned by a keyword search may still fall outside the true scope of what the researcher needs.
Research on how scientists and health researchers read papers similarly suggests that the abstract is commonly used to determine whether further reading is worthwhile. In one survey of researchers reading IMRaD papers, almost all respondents reported reading the abstract first, and lack of relevance was the reason given by those who stopped after it.
At this stage, ask whether the paper genuinely intersects with your question. Depending on your research, relevance may involve the phenomenon being studied, population, context, variables, intervention, theoretical perspective, method, outcome, time period, or another characteristic.
A paper can share several keywords with your topic and still answer a substantially different question.
Watch Out
Do not confuse keyword similarity with research relevance. Search engines and databases retrieve records partly through words and metadata; your task is to determine whether the study itself addresses something that matters to your research purpose.
Relevance is not all-or-nothing
A paper does not have to match your study perfectly to be useful. In fact, insisting on exact similarity may exclude literature that provides valuable theory, methods, measures, contrasting evidence, or contextual insight.
It helps to distinguish several ways a paper might be relevant.
| Type of relevance |
What the paper may contribute |
Possible reason for deeper reading |
| Direct empirical relevance |
Evidence addressing a closely related research question |
You may use or compare its findings |
| Methodological relevance |
A design, instrument, procedure, analytical method, or protocol |
You may evaluate or adapt its approach |
| Theoretical relevance |
A concept, model, mechanism, or explanatory framework |
It may shape how you frame or interpret your study |
| Contextual relevance |
Evidence from another population, setting, discipline, or period |
It may help establish boundaries, differences, or transferability |
| Scholarly relevance |
An influential, foundational, contradictory, or frequently discussed contribution |
You may need to understand an important conversation in the literature |
This distinction matters because different forms of relevance call for different reading priorities. A methodological paper may deserve close reading even if its substantive topic differs from yours. A directly relevant empirical paper may deserve scrutiny because you expect to cite its findings.
Use the abstract to screen, not to finish the evaluation
Abstracts are designed to summarize papers concisely and can often tell you enough to decide that a paper is irrelevant. They are less reliable for deciding that a paper is sufficiently understood.
A favorable abstract should therefore trigger another question: What do I still need to verify before deciding how much attention this paper deserves?
You might need to establish exactly who was studied, what was measured, which research design was used, whether the result you care about was actually analyzed, or whether the authors' conclusion corresponds to the evidence relevant to your question.
This is where selective inspection becomes useful. You do not necessarily need to read every paragraph yet. Move to the sections that resolve the uncertainty preventing you from making the next decision.
This approach builds naturally on using the opening moments with a paper for orientation rather than immediately committing to a complete reading.
Check whether the study actually answers the question you think it answers
Titles and abstracts necessarily compress complicated studies. Once a paper appears relevant, identify its actual research question or objective and compare it with your reason for reading it.
Suppose you are interested in whether an educational technology improves student learning. A paper titled around "student outcomes" might measure only satisfaction or intention to use the technology. Those outcomes may be legitimate, but they do not answer the same question as academic performance.
Likewise, a study investigating an association between two variables does not necessarily provide evidence that changing one causes a change in the other.
The closer the paper comes to influencing an important claim in your work, the more carefully you should verify this alignment.
Inspect enough of the methods to identify consequential mismatches
A highly relevant research question does not automatically make the study useful for every purpose. The methods tell you what kind of evidence the study can actually provide.
You may need to check the research design, participants or data source, sampling strategy, measures, intervention or exposure, comparison condition, analytical approach, and other features relevant to your question.
At the screening stage, you are not necessarily performing a complete methodological appraisal. You are looking for information that could materially change your decision to continue.
For example, you might discover that the study concerns secondary-school students when your review concerns university students, that an outcome was measured in a way incompatible with your question, or that the paper uses a research design you specifically need to examine.
If methodology itself is why the paper interests you, then the methods may deserve close attention much earlier.
Look at the actual results when a particular finding is what interests you
An abstract may report that an intervention "significantly improved" an outcome, that two variables were "associated," or that participants expressed a particular view. Those summaries can establish initial interest, but they may not tell you enough about the evidence.
If the finding itself is why you might keep the paper, inspect the relevant results, table, figure, or qualitative findings.
Ask what was actually observed. How large was the difference or association? What comparison was made? How much uncertainty surrounds the estimate? Were there multiple outcomes? Does the table tell a more complicated story than the abstract suggests?
You do not have to resolve every analytical issue at this point. You are determining whether the result is sufficiently relevant or consequential to justify deeper reading.
A surprising or contradictory paper may deserve more attention, not less
Researchers naturally gravitate toward studies that fit what they already expect. A paper reaching a different conclusion can therefore be easy to dismiss.
That is precisely when closer inspection may be valuable.
A contradictory finding might result from a different population, measurement strategy, research design, analytical decision, implementation context, or theoretical assumption. Alternatively, it may expose a weakness in an interpretation you had begun to take for granted.
You should not read every contradictory paper simply because it disagrees with you. But disagreement itself is not a reason for exclusion. When the paper is credible and directly relevant, understanding why its findings differ can sharpen your interpretation of the literature.
Foundational papers may deserve reading even when they are old
Publication date can be relevant, especially in rapidly changing areas, but recency is not equivalent to importance.
An older paper may have introduced a theory, measure, methodological approach, dataset, or distinction that later research continues to use. If contemporary papers repeatedly invoke an earlier source for a concept central to your work, reading the original can be more informative than relying indefinitely on second-hand summaries.
The reverse also applies. A recent publication is not automatically more useful merely because it is recent.
Ask what role the paper plays in the literature rather than treating publication year as a simple quality filter.
Pay attention when you keep encountering the same paper
Repeated appearance across relevant literature can be a useful signal that a paper deserves investigation. Perhaps multiple authors cite it for the same theoretical proposition, instrument, methodological procedure, or empirical finding.
That does not prove the cited paper is correct. Citation counts and repeated citation should not be treated as measures of research quality. They can, however, tell you that the paper has become part of the scholarly conversation you are trying to understand.
At that point, reading the original source may help you determine whether later authors are representing it accurately and whether the original evidence supports the proposition for which it is repeatedly cited.
The intended use of the paper should determine the required reading depth
This is the most important decision criterion.
If you only need a broad understanding of an unfamiliar topic, a limited reading may be enough. If you need one methodological detail, you may read that part closely without reading every other section with equal intensity.
But if you intend to use the paper as central evidence, critique it, reproduce its method, compare its findings with your own, or build an argument around it, superficial reading becomes harder to justify.
Low-consequence use
The paper provides orientation, terminology, a possible lead, or background context.
High-consequence use
Your argument, method, interpretation, or research decision depends materially on what the paper establishes.
The higher the consequence of misunderstanding the paper, the stronger the case for reading it in depth.
"Worth reading in full" does not necessarily mean reading every word
A full reading should not be confused with mechanically processing every sentence at identical intensity.
Even when a paper deserves serious attention, you may move quickly through familiar background material and spend substantially more time on a complex analysis, an important table, or a methodological limitation. You may revisit some sections several times and barely return to others.
Multi-pass approaches to scientific reading formalize this idea by progressively increasing reading depth. S. Keshav's three-pass approach, for example, begins with a broad assessment and reserves the most intensive engagement for papers that require deeper understanding.
So the decision is not simply "skim or read every word." There are degrees of engagement between those extremes. Understanding how much of a particular paper you actually need to understand is part of deciding how far to go.
Formal evidence reviews require a different kind of screening decision
If you are conducting a systematic review, scoping review, or another evidence synthesis with predefined eligibility criteria, deciding whether a paper receives full-text assessment is not merely a personal judgment about whether it seems interesting.
Screening should follow the review's protocol and eligibility criteria. Title and abstract screening is commonly used to identify records that may be eligible, followed by full-text assessment when eligibility cannot be determined confidently from the initial record.
That is a different task from deciding what to read deeply for ordinary scholarly learning. The underlying logic is similar, however: use an initial screen to avoid unnecessary full-text work while preserving papers that require further examination.