Should You Replicate an Existing Study or Ask a New Research Question?

Does Good Research Always Need a New Question?

You have found an existing study that is closely related to what you want to investigate. Now you face a deceptively difficult choice: should you test that study again, or should you develop a different research question?

The temptation is often to choose whatever appears more novel. Researchers are routinely asked what is "new" about their work, which can make replication seem less ambitious than asking a question nobody has examined before. But novelty and scientific value are not the same thing. A published finding may still be uncertain, dependent on a particular population or setting, or supported by too little independent evidence to justify treating it as established.

At the same time, repeating an existing study merely because it is easy to reproduce is not automatically useful. The more important question is what uncertainty your study would resolve.

Choose the Question the Evidence Actually Needs

The Difference Is About What You Are Trying to Learn

Replication Is Not Simply Repeating Someone Else's Work

A common description of replication is that a researcher repeats an earlier study to see whether the same result appears again. That description is useful as a starting point, but it can be too mechanical.

Nosek and Errington offer a more informative way to think about replication: a replication produces new evidence that is diagnostic of a claim made by prior research. In other words, the central issue is not whether every procedural detail has been copied. It is whether the new study provides evidence that changes how confidently you should believe, qualify, or reject an existing claim.

The National Academies of Sciences, Engineering, and Medicine similarly distinguishes replication from computational reproducibility. In its terminology, reproducibility involves obtaining consistent computational results using the same data and computational procedures, whereas replication involves a new study with new data aimed at the same or a similar scientific question.

This distinction matters because changing something about a previous study does not necessarily create a genuinely new research question. A different sample, setting, instrument, or procedure might instead test how well an existing claim survives under different conditions. Depending on the purpose and design, such changes may relate to different forms of replication rather than an entirely new line of inquiry.

Replication Makes Sense When the Existing Claim Is Still Uncertain

Suppose one study reports an association between a particular teaching strategy and student achievement. The study may be carefully conducted, but one study alone usually cannot establish that the finding will recur across new samples and circumstances. A replication can provide additional evidence about the credibility and possible boundaries of that claim.

This is one reason replication is scientifically important. A result appearing in a peer-reviewed article does not transform it into a settled fact. Findings can depend on sampling variation, measurement choices, analytical decisions, contextual conditions, or other features of the original investigation. Independent evidence helps researchers determine how much confidence a claim deserves.

A replication may therefore be particularly informative when:

• an influential claim depends heavily on one or a small number of studies;
• previous studies have produced inconsistent findings;
• the original finding has important theoretical, practical, clinical, educational, or policy implications;
• there is uncertainty about whether the result holds under conditions relevant to your field;
• the evidence supporting the claim is methodologically limited or unusually uncertain; or
• subsequent research has treated the original finding as established without much independent testing.

In these situations, asking something entirely new may actually leave an important weakness in the literature untouched. The scientific contribution comes from reducing uncertainty, not from maximizing the distance between your question and everything previously published. Whether replication deserves priority also depends on when additional evidence would be more useful than novelty.

A New Question Makes Sense When Repeating the Existing Test Would Add Little

Replication is not automatically the better choice simply because replication is valuable in science. The relevant question is the expected informational value of another test.

Imagine that a phenomenon has already been examined repeatedly using strong studies conducted across several populations and settings, with results that collectively provide reasonably consistent evidence. Another closely similar replication could still contribute information, but its incremental value may be modest. A more consequential unresolved question might concern why the phenomenon occurs, under what conditions it changes, what mechanism explains it, or what consequence follows from it.

A new research question may therefore be preferable when the existing claim is already supported sufficiently for your purpose and the literature contains a more important unanswered problem that your study is capable of addressing.

This does not mean that a certain number of previous replications automatically closes a question. Evidence accumulates unevenly. Ten weak studies do not necessarily settle a claim more convincingly than several rigorous and appropriately varied investigations. The decision requires judgment about the quality, relevance, independence, and consistency of the existing evidence.

Do Not Confuse a Different Setting With a New Question

Researchers sometimes create an apparently new study by changing the location, institution, population, or demographic group while leaving the underlying question largely unchanged.

For example, suppose a published study asks whether perceived instructor support predicts student engagement among university students. You propose examining the same relationship among students at another university. Your location is new, and your participants are new, but the underlying scientific question may not be.

That is not necessarily a problem. The new population might provide valuable evidence about whether the original relationship generalizes. The important issue is to describe the contribution accurately rather than manufacture novelty from geography.

Whether a population change constitutes replication, an extension, or a meaningfully different investigation depends on what theoretical claim the population change is intended to test. A change in participants should therefore have an evidential rationale, not merely a logistical one. This becomes especially important when deciding what changing the study population actually changes about the research question.

Replication and Extension Are Not the Same Decision

Sometimes neither "repeat the study" nor "ask something completely new" accurately describes what you want to do.

You may want to test the original relationship while deliberately adding something that addresses a further question. For example, an original study might examine whether academic self-efficacy predicts persistence. Your study might examine that relationship again while also testing whether the association differs according to a theoretically relevant moderator.

That design may move toward an extension because it attempts to generate evidence beyond the original claim. The boundary is not always perfectly sharp, and terminology varies across disciplines. A useful way to clarify the design is to ask whether your principal inferential target remains the original claim or whether your study is primarily intended to establish an additional claim. The distinction between replication and extension studies deserves explicit consideration before you describe the contribution.

Novelty Is Not the Same as Contribution

The pressure to identify a "research gap" can encourage researchers to search for something that has never been studied, regardless of whether studying it would materially improve knowledge. This can produce increasingly narrow combinations of populations, variables, technologies, and settings whose principal virtue is that nobody appears to have published the exact combination before.

A literature gap is not automatically a knowledge gap worth filling.

Replication illustrates why. A researcher may ask a question that has already been asked and still make a substantial contribution because the reliability, generalizability, or boundary conditions of the previous answer remain uncertain. Conversely, a question may be technically unprecedented yet scientifically minor.

Ask What Each Possible Outcome Would Teach You

One of the strongest ways to choose between replication and a new question is to imagine the possible results before conducting either study.

If you replicate the previous study and obtain a result consistent with the original finding, what would you learn? If the result differs, what would you learn? Would either outcome meaningfully change your understanding of the claim?

This reasoning follows the view that a replication should produce evidence that is diagnostic of an existing claim. A result consistent with prior evidence can strengthen confidence under the conditions tested. A result that is inconsistent can reveal uncertainty, possible boundary conditions, methodological differences, or reasons to reconsider the original interpretation. An inconsistent replication should not automatically be reduced to a verdict that the original study was "wrong."

Now apply the same test to the proposed new question. What would each plausible result add to current knowledge? If the answer is considerably more consequential than what another replication could establish, the new question may deserve priority.

This counterfactual exercise is surprisingly effective because it forces you to identify the contribution before seeing the results. It also prevents the rationale from becoming a post hoc story written after the statistics have already spoken, a habit reviewers tend to notice with almost supernatural efficiency.

Choosing Between Replicating a Finding and Asking Something New

Notice that neither option is inherently superior. The replication asks, in effect, "How much confidence should we place in this existing claim?" The new study asks, "What else do we need to understand about this phenomenon?"

There is also a possible middle ground. The researcher could design an extension that evaluates the original relationship while introducing a theoretically justified additional question. But adding variables simply to make a replication appear novel can obscure the study's purpose. The distinction becomes particularly important when considering whether a new variable changes the study from replication to extension.

Common Mistakes When Choosing Replication or Novelty

Make the Decision From the Literature, Not From the Desire to Look Novel

Before deciding on your research question, map the evidence surrounding the phenomenon rather than searching only for an untouched topic. Identify the central claim, how many independent studies have tested it, how strong those studies are, whether their findings agree, which populations and conditions have been examined, and what uncertainty remains consequential.

Then compare the informational value of your alternatives.

Whatever you choose, write the rationale around the unresolved uncertainty. A replication does not need to masquerade as unprecedented research to be worthwhile. In fact, being explicit about what is being retested can make the logic of the contribution considerably clearer. The challenge is to justify replication without pretending that the question itself is completely novel.

Likewise, if you choose a new question, explain why answering it advances understanding beyond the evidence already available. "No study was found" is an observation about the literature. The research rationale begins with why that absence matters.

Before You Choose Replication or a New Question

Questions About Choosing Replication or a New Study

Do Not Choose Novelty Until You Know What Needs to Be Known

How to Cite This Guide

This guide is intended to be read, shared, and used in research, teaching, and academic work. If you draw on its ideas, explanations, or other content, please acknowledge the source by citing the guide. Doing so gives appropriate credit and helps your readers locate the original resource.