How Do You Compare the Cost of Research With the Cost of Remaining Uncertain?

When is reducing uncertainty worth what the research will cost?

A research proposal can address an important uncertainty and still be a poor investment. A study may cost millions, require years of work, expose participants to burdens or risks, and consume resources that could have supported other valuable research.

Yet choosing not to conduct research also has consequences. Decisions continue to be made with existing evidence. If that evidence leads people toward an inferior option, the resulting losses may accumulate across individuals, organizations, or entire populations.

The relevant comparison is therefore not simply “research costs money” versus “knowledge is valuable.” It is whether the expected improvement produced by the information from a feasible study is large enough to justify the full cost of obtaining it.

Compare what the study could save or improve with what it consumes

Both conducting research and remaining uncertain have opportunity costs

The cost of uncertainty comes from decisions made without better information

Uncertainty itself is not a bill that can simply be added to a budget. Its cost arises from the possibility that decisions made using current information will be inferior to decisions that could have been made with better information.

Suppose current evidence favors Option A. Because uncertainty remains, there are plausible circumstances in which Option B is actually preferable. If A is chosen and B would have produced a better outcome, the difference represents an opportunity loss associated with the decision under uncertainty.

Across the range of uncertainty, the expected value of these losses represents the potential benefit available from better information.

This is the basic logic behind asking what knowing the answer to a research question would actually be worth.

The expected value of perfect information provides an upper bound

The expected value of perfect information (EVPI) estimates the expected improvement in the decision if all relevant uncertainty represented in the decision problem could be eliminated before choosing an action.

EVPI is useful as an upper bound. If perfect information is worth less than the cost of a proposed study, that study cannot be justified solely by the decision value of resolving the uncertainty represented in the analysis.

A real study should be evaluated using the information it can actually produce

Perfect information is hypothetical. Real studies have finite samples, imperfect measurements, sampling variation, and design limitations. They reduce uncertainty rather than eliminate it.

The expected value of sample information (EVSI) estimates the expected improvement in decisions associated with the information that could be generated by a particular study design and sample size.

This makes EVSI more directly relevant to the research investment decision. Two studies investigating the same uncertainty may have different EVSI because one is larger, uses better measurements, targets a more informative population, or collects evidence more directly relevant to the decision.

Expected net benefit of sampling compares information value with research cost

Once the expected value of a proposed study's information has been estimated, it can be compared with the expected cost of generating that information.

The expected net benefit of sampling (ENBS) expresses this comparison. In a simplified form:

In formal applications, the precise calculation and relevant cost categories depend on the decision context and modelling framework. The simple expression nevertheless captures the central logic: additional information is not worth obtaining at unlimited cost.

Research cost is more than the study budget

Direct financial expenditure is the most visible research cost, but it is not the only one.

A study can require researcher time, participant effort, clinical or organizational capacity, data infrastructure, equipment, administrative resources, and scarce expertise. Some research also exposes participants to burdens or risks that need ethical consideration rather than simple monetary valuation.

Resources used for one project are unavailable for competing research, services, or interventions. This is an opportunity cost. A study can therefore be affordable in accounting terms yet still represent poor resource allocation if those resources could generate substantially greater benefits elsewhere.

The cost of delay may belong in the comparison

Research takes time. If a decision is postponed until results become available, people may continue receiving an inferior intervention, an effective policy may remain unavailable, or inefficient resource use may persist.

The consequences of delay can sometimes outweigh the expected benefit of waiting for better information. In other situations, acting immediately may make future research difficult or impossible, which can increase the value of obtaining evidence before widespread adoption.

The relevant choice may therefore involve several strategies: act now, conduct research before acting, or implement while continuing to collect evidence.

This is why deciding whether to act with imperfect evidence or wait for more research requires more than comparing a study budget with the value of its results.

Population size can make research worth far more than its project cost

The value of research often depends on how many future decisions can benefit from the information. A study may cost a substantial amount but produce only a small expected improvement for each individual decision. If the evidence will inform decisions for a large population over several years, those small benefits can accumulate.

Population-level value-of-information analysis therefore considers the number of people expected to face the decision and the period during which the information remains useful.

The reverse is also possible. A study may address a high-stakes uncertainty but apply to very few future decisions, limiting its total expected information value.

Research costs and uncertainty costs need a common decision framework

Directly comparing “$1 million of research” with “30% uncertainty” is meaningless because the quantities are measured on different scales.

The uncertainty must first be translated into its expected consequences for outcomes or net benefit. The research must then be evaluated according to how much of those expected consequences it could reduce.

In health economic applications, costs and health outcomes can be incorporated into a common net-benefit framework. Other fields may use different outcome measures, utility functions, or multi-criteria decision approaches depending on what the decision values.

Where monetization is inappropriate or impossible, a structured qualitative comparison can still be useful. Researchers can make explicit what is gained, what is consumed, who bears each consequence, and what important effects cannot reasonably be expressed in a single metric.

The cheapest study is not necessarily the best value

Suppose Study A costs 200,000 units and Study B costs 500,000. Choosing A because it is cheaper ignores how much useful information each design produces.

If Study A is too small to materially reduce decision uncertainty while Study B resolves much more of the relevant uncertainty, the more expensive study could have substantially greater net value.

Research design can therefore be treated as an optimization problem: among feasible designs, which produces the greatest expected net benefit after considering both information value and cost?

A positive research result is not required for research to have been worthwhile

The value of research is assessed before the study outcome is known. A well-designed study may ultimately confirm the current decision rather than change it.

That does not mean the research had no value. Before the study was conducted, several results were possible. Its expected value came from reducing uncertainty across those possible results and allowing decisions to respond appropriately to whatever evidence emerged.

Conversely, a study that happens to produce a surprising decision-changing result was not necessarily a good investment if, prospectively, the study was extremely costly and had little expected chance of providing useful information.

Some research benefits and costs resist simple monetization

Formal value-of-information methods are particularly well developed in health economics, where decision models often use explicit measures of costs and health outcomes. Other research contexts may not permit such complete quantification.

Research can generate theoretical knowledge, methodological advances, infrastructure, trained researchers, reusable datasets, or benefits for future questions. It can also impose ethical, distributive, or social consequences that should not be reduced casually to a monetary figure.

A decision framework should therefore be as explicit as possible about what is included and excluded. Quantification can improve transparency, but it does not eliminate the need for judgment.

When an expensive study can still be cheaper than remaining uncertain

Now suppose the same study would inform only a small one-time decision and its expected information value were 250,000 units. An 800,000-unit study would then be difficult to justify solely for that decision. The study cost has not changed; the value of reducing the uncertainty has.

Common mistakes when comparing research costs with uncertainty

Evaluate research as an investment in better information

When deciding whether to conduct a study, separate three questions. How costly is current uncertainty? How much of that uncertainty can the proposed study realistically reduce? What will obtaining that information cost?

The strongest case for research occurs when the uncertainty has meaningful consequences, the proposed study is capable of reducing it, and the expected improvement exceeds the resources and other consequences required to obtain the evidence.

This comparison can also reveal when research itself may have negative net value because it consumes more resources than the information is expected to justify.

The point is not to minimize research expenditure. It is to allocate research effort where the expected value of reducing uncertainty is greatest relative to what must be given up to obtain that information.

Before deciding whether research is worth its cost

Questions about research cost and the cost of uncertainty

Research is worth its cost when the information it produces is expected to be worth more

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.