Is Another Primary Study Needed or Would Better Synthesis Be Enough?

Do you need more evidence, or do you need to understand the evidence you already have?

You review a topic and find a messy literature. Some studies report an association, others do not. Samples vary. Measures differ. Findings are scattered across disciplines or settings. No single paper seems to provide a satisfying answer.

The instinctive response is often to conduct another study.

But another primary study and a better synthesis solve different problems. A new primary study generates new observations. Evidence synthesis organizes, evaluates, and, when appropriate, combines observations that have already been generated. If the information needed to answer the question substantially exists but remains fragmented or poorly integrated, collecting another dataset may add another piece without assembling the puzzle.

The decision therefore begins by diagnosing what is actually missing.

First determine whether the problem is missing data or unsynthesized evidence

How to tell whether new primary research would add more than synthesis

Primary studies and evidence syntheses answer different informational needs

A primary study collects or generates original data to answer a research question. Depending on the field, this might involve an experiment, survey, cohort, interview study, observation, trial, measurement exercise, or another form of empirical investigation.

Evidence synthesis works at a different level. It asks what can be concluded from a body of existing research. Systematic reviews use explicit methods to identify and evaluate studies relevant to a defined question. Depending on the evidence and purpose, synthesis may be quantitative, qualitative, narrative, configurative, or take another systematic form. Meta-analysis is one statistical approach to combining compatible quantitative results; it is not synonymous with systematic review.

The distinction matters because producing more evidence does not automatically solve a synthesis problem. Ten unintegrated studies do not necessarily become more informative when an eleventh is added.

A collection of studies is not yet a body of evidence

Reading several individual papers can tell you what those papers found. It does not necessarily tell you what the literature as a whole supports.

Studies may differ in sample characteristics, measures, comparators, analytical choices, designs, follow-up periods, and risk of bias. Their estimates may also differ simply because of sampling variability. Looking at conclusions paper by paper can therefore produce a misleading impression of agreement or disagreement.

A systematic synthesis changes the unit of reasoning. Instead of asking, “How many papers found a statistically significant result?” it asks questions such as: What estimates do the studies provide? How comparable are they? How precise are those estimates? How credible are the designs? Is there meaningful heterogeneity? Is relevant evidence missing from the published record? How certain should we be about the resulting conclusion?

Cochrane describes systematic reviews as a means of bringing together relevant primary research using systematic methods so that decisions can be informed by an up-to-date understanding of the evidence. It also recommends that systematic review typically precede new primary research because synthesis can reveal whether knowledge gaps remain and identify weaknesses that future studies should address.

More primary studies may be unnecessary when the answer is already distributed across existing studies

Imagine that 15 reasonably comparable studies have examined the same relationship, but no one has systematically evaluated them together. Individual findings vary. Some report positive associations, several have wide confidence intervals, and others report little evidence of an association.

It would be premature to infer from this pattern alone that another study is needed. The apparent inconsistency may change once effect estimates, sample sizes, precision, design differences, and risk of bias are examined together.

A synthesis might reveal that the studies actually converge around a similar estimate despite different significance tests. It might instead reveal substantial heterogeneity associated with identifiable methodological or contextual differences. Either finding would be more informative for planning future research than simply adding another isolated estimate.

Ask whether the existing studies are capable of answering your question

The first practical test is not the number of studies but their relevance.

Suppose you want to know whether an educational intervention improves long-term retention. You find 30 studies, but all measure performance immediately after the intervention. The literature may be large while containing almost no direct evidence about long-term retention.

Likewise, dozens of cross-sectional studies cannot by themselves supply longitudinal observations that were never collected. A meta-analysis of self-reported behavior cannot transform those observations into objectively measured behavior. A synthesis of studies lacking an important comparator cannot recreate that comparator.

In such cases, the existing literature may justify collecting new data because the information required to address the question is genuinely absent.

Ask whether synthesis could change what you think the literature says

Synthesis is particularly valuable when conclusions based on individual studies are unstable, fragmented, or difficult to reconcile.

For quantitative evidence, an appropriate meta-analysis may improve precision by combining compatible estimates, although statistical pooling should never be automatic. Cochrane explicitly emphasizes that reviewers must first consider whether studies are sufficiently appropriate to combine. Differences in participants, interventions or exposures, outcomes, methods, and risk of bias can make a single pooled estimate unhelpful or misleading.

Even when statistical pooling is inappropriate, systematic synthesis can still be valuable. Researchers can characterize patterns, investigate sources of heterogeneity, compare methodological choices, identify recurring limitations, assess the certainty of evidence, and establish which questions remain unanswered.

This means “better synthesis” does not necessarily mean “calculate a meta-analysis.” The appropriate synthesis depends on the question and evidence.

An old synthesis may no longer be enough

Finding an existing systematic review does not automatically settle the matter. Reviews can become outdated as new studies appear, methods improve, or relevant questions change.

Cochrane notes that adding new studies can alter conclusions, improve precision, demonstrate broader applicability, or enable new comparisons and subgroup analyses. An update can therefore be more useful than either starting another review from scratch or immediately launching a new primary study.

Before choosing primary research, ask whether the apparent uncertainty exists because the best synthesis is outdated. If several important studies have appeared since its search date, updating the synthesis may reveal that the uncertainty has already narrowed.

Sometimes synthesis reveals the need for a very specific new study

The relationship between synthesis and primary research is not adversarial. Often the best synthesis is precisely what tells you what primary research should come next.

For example, a review might reveal that estimates are reasonably consistent among adults but that evidence for adolescents is indirect. It might show that short-term outcomes are well established while long-term outcomes remain poorly measured. It might reveal that most studies compare an intervention with no treatment when the practically relevant question concerns comparison with an active alternative.

Now the justification for new research becomes more precise. Instead of “more studies are needed,” the evidence points toward a genuinely missing comparison, a measurement problem worth correcting, or another identifiable deficit.

Synthesis cannot rescue fundamentally inadequate evidence

A synthesis inherits important properties of the studies it contains. Combining weak studies does not automatically produce strong evidence.

If all available studies share serious risks of bias, use inappropriate measures, or provide only indirect evidence for the question of interest, a review may document those weaknesses more convincingly but cannot remove them. Similarly, greater statistical precision is not necessarily greater validity. A very precise pooled estimate can still be systematically wrong if its constituent evidence is biased.

Frameworks such as GRADE therefore evaluate certainty using considerations that include risk of bias, inconsistency, indirectness, imprecision, and publication bias. The details of certainty assessment vary by question and evidence type, but the broader lesson is important: deciding whether more research is needed requires examining the quality and relevance of accumulated evidence, not simply its volume.

The most useful sequence is often synthesis first, targeted research second

Evidence-based research advocates systematic and transparent use of prior studies when justifying and designing new research. Lund and colleagues argue that researchers should use the available scientific knowledge, together with relevant end-user perspectives, to establish whether a new study is necessary and how it should be designed.

This does not imply that every project in every discipline must conduct a full systematic review from scratch. Existing high-quality and sufficiently current syntheses may already provide the necessary foundation. The required level of synthesis should be proportionate to the question, field, stakes, and state of the evidence.

What should be avoided is using an unsystematic reading of a few papers to declare that primary research is necessary when the answer may already be sitting, unassembled, in the literature.

When another survey may add less than synthesizing the surveys already conducted

The synthesis has not eliminated primary research. It has changed the question from “Can I conduct another study?” to “What study would resolve something the existing evidence cannot?” That is a much stronger basis for research design.

Common mistakes when choosing between synthesis and another study

Diagnose the evidence problem before choosing the research design

The practical decision is not whether primary research is inherently more valuable than synthesis. Neither occupies a higher rung of some universal research hierarchy. The useful choice depends on what information is missing.

The ideal outcome of reviewing existing research is therefore not necessarily a declaration that “more research is needed.” It is a defensible account of what kind of research, if any, would be most informative next.

Before launching another primary study, check whether synthesis could answer the question

Questions about choosing primary research or evidence synthesis

Do not collect new data to solve a problem of unsynthesized evidence

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.