Will the Study Change What Someone Should Do?

If the study produces new evidence, will anyone have reason to act differently?

Many research proposals promise to “inform practice,” “guide policy,” or “support decision-making.” Those phrases sound useful, but they can conceal a basic unanswered question: what decision would actually change because of the evidence?

A study can produce a credible and practically meaningful result without changing what anyone should do. Perhaps one option remains preferable under every plausible result. Perhaps the uncertain variable being studied is not what drives the decision. Or perhaps the difference between alternatives is too small to outweigh costs, harms, feasibility, or other considerations.

If decision relevance is part of the study's justification, it is worth identifying the decision before collecting the evidence.

Decision-relevant evidence can change which action is preferable

Start with the decision, not merely the variable you can measure

Name the decision and the decision-maker

“This research may inform practice” is difficult to evaluate because neither the practice nor the decision is specified.

A more useful formulation identifies who is choosing among what.

A university might be deciding whether to adopt a learning platform. A clinician may be choosing between treatments. A funding agency may be deciding whether another trial deserves investment. A researcher may be deciding whether a preliminary finding justifies a larger study.

These decisions require different evidence.

Once the decision is explicit, you can ask what information could change it.

Evidence matters when the preferred action depends on something uncertain

Suppose two interventions are available. Intervention A is currently preferred because it appears more effective after accounting for its costs and disadvantages. However, its true effectiveness remains uncertain.

If sufficiently different plausible values of effectiveness would make Intervention B preferable, reducing uncertainty about effectiveness has potential decision value.

Now suppose Intervention A remains preferable across the entire plausible range. Learning its effectiveness more precisely may improve knowledge, but it is less likely to change the current decision.

This is the central intuition behind value of information analysis. In decision theory, additional information has value when reducing uncertainty can improve the expected consequences of a decision. Formal value-of-information methods are widely used in health economics and related decision problems to assess whether further evidence is worth collecting and which uncertainties deserve priority.

A result can matter scientifically without changing the decision

Suppose a new study demonstrates convincingly that an intervention works through a different mechanism than previously believed. That finding could change scientific understanding substantially.

But if both mechanisms imply exactly the same choice between available interventions, the immediate practical decision may remain unchanged.

The reverse can also occur. A study might contribute little to theory yet estimate a cost, risk, or effect with enough precision to change which policy is preferable.

Many studies do both. They do not have to.

The decision usually depends on more than the focal outcome

An educational technology may improve test scores but require substantial licensing fees, faculty training, technical support, student data collection, and curriculum changes.

A study showing a positive learning effect therefore does not automatically establish that adoption is the better decision.

Decision-makers may need to consider several outcomes and constraints, including benefits, costs, harms, feasibility, equity, acceptability, opportunity costs, and uncertainty. Which considerations matter depends on the context.

This is why a technically correct answer can still be practically useless. The study may answer one component accurately while omitting the information needed to choose among actions.

Ask whether the evidence could cross a decision boundary

Imagine that an institution will adopt a new program only if its expected benefits justify its costs and implementation burden.

Current evidence places the program near that boundary. A new study capable of moving the estimated benefit meaningfully upward or downward could change the preferred action.

Now imagine that the program is extraordinarily costly and existing evidence suggests only negligible benefits. Even a moderately more favorable estimate may leave the adoption decision unchanged.

The same study design can therefore have very different decision value depending on where the current uncertainty lies relative to the choice being made.

Expected value of information makes this reasoning explicit

Formal value-of-information analysis asks how much better decisions could become if uncertainty were reduced. Several related quantities can be used depending on the problem.

The expected value of perfect information considers the value of eliminating relevant uncertainty entirely. The expected value of partial perfect information considers eliminating uncertainty about particular parameters. The expected value of sample information considers the expected value of the imperfect information obtainable from a specific proposed study.

These methods can help determine whether further research is worthwhile and which study designs have the greatest expected decision value. They are especially developed in healthcare decision analysis, although the underlying reasoning is more general.

Most research projects will not require a formal calculation. The conceptual lesson is still powerful: additional evidence is valuable for a decision only to the extent that reducing uncertainty can improve the choice.

Changing what someone should do is not the same as predicting what they will do

Evidence can support one action while decision-makers choose another.

People and institutions may face legal restrictions, limited budgets, political pressures, organizational constraints, conflicting values, or implementation barriers. They may also interpret evidence differently.

The research question should therefore avoid promising behavioral change merely because evidence becomes available.

“Would this evidence change which action is best supported?” is generally more defensible than “Will decision-makers change their behavior?” unless behavior itself is the object of study.

Sometimes the correct decision is to collect more information

Research decisions are themselves decisions.

A pilot study may not determine whether an intervention should be adopted, but it may determine whether a full-scale trial should proceed. A feasibility study may reveal that recruitment is too difficult. An early experiment may identify which mechanism deserves a stronger test.

In these cases, the action changed by the evidence is the next research action. That is a legitimate form of decision relevance and connects directly to whether the study will help determine what research should happen next.

When additional evidence could change an adoption decision

The study is decision-relevant because credible outcomes can lead to different preferred actions.

Now change the scenario. Suppose the new platform is so expensive that the existing system remains preferable even under the most optimistic plausible effectiveness estimate. A new effectiveness study may still have scientific value, but it has little immediate value for this particular adoption decision because resolving that uncertainty does not alter the choice.

Why useful evidence does not automatically produce a different action

Reverse-engineer decision-oriented research from the choice it is meant to inform

If your proposal claims that the findings will guide practice, policy, management, or another decision, write down the decision before finalizing the study.

For complex or high-stakes decisions, formal decision analysis and value-of-information methods may be appropriate. These approaches can compare the expected benefit of additional evidence with the cost of obtaining it and help prioritize which uncertainty deserves further research.

For less formal research planning, you can still use the underlying logic. Identify the decision, identify what makes the decision uncertain, and ask whether the evidence your study can realistically produce could change the preferred action.

If the answer is no, the study may still be scientifically worthwhile. Just do not justify it by claiming a decision consequence that its findings cannot plausibly produce.

Before claiming that your research will inform a decision, check the decision

Questions about research evidence and decision-making

Know the decision before claiming that the evidence will inform it

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.