01 · The Question
Should We Study the Questions We Are Most Uncertain About?
Research exists partly because we do not know things. It therefore seems reasonable to prioritize questions for which the current answer is especially uncertain.
But consider two unresolved questions. Researchers may be almost completely uncertain about the first, yet every plausible answer would have little effect on what anyone does or understands. For the second, existing evidence may already favor one answer, but the remaining uncertainty concerns a decision with substantial consequences.
Which deserves additional research?
The comparison reveals why uncertainty is relevant to research priority but cannot determine it by itself . Research becomes valuable not merely because uncertainty exists, but because reducing a consequential uncertainty can improve what we know or what we choose to do.
03 · What You Need to Know
The Important Question Is Not Simply “How Uncertain Are We?”
Uncertainty Creates an Opportunity for Research, Not an Automatic Priority
If an answer were already known with sufficient confidence, additional research addressing exactly the same uncertainty would usually have less informational value. In that sense, uncertainty creates an opportunity for new evidence to contribute.
Yet there are countless things researchers do not know. Some are scientifically consequential. Others are minor unresolved details. Still others are unknown because nobody has had much reason to investigate them.
The existence of uncertainty therefore tells you that knowledge could potentially be improved. It does not tell you how valuable that improvement would be.
A useful prioritization question is not simply, “How uncertain is the answer?” but, “What is at stake because we remain uncertain?”
Distinguish Evidence Uncertainty From Decision Uncertainty
This distinction is particularly important when research is intended to inform a choice.
Evidence uncertainty
We do not know the precise value, magnitude, relationship, mechanism, or effect with high confidence.
Decision uncertainty
Because of what we do not know, we are uncertain which available decision or course of action is preferable.
The two can occur together, but they need not.
Imagine that researchers are uncertain whether an intervention improves an outcome by 8%, 10%, or 12%. That may represent meaningful statistical or evidential uncertainty. But if the intervention would be preferred across that entire range, obtaining a more precise estimate may not change the decision.
Now imagine that the plausible range is from a small harmful effect to a substantial beneficial effect. The uncertainty is consequential because different plausible values support different choices.
Value-of-information analysis formalizes this distinction in decision contexts by examining the expected benefit of reducing uncertainty so that better decisions can be made. It does not treat uncertainty as valuable to eliminate merely because uncertainty exists.
Research Priority Depends on Where the Plausible Answers Lead
One practical way to judge uncertainty is to examine the range of answers that remain credible given current evidence.
Current situation
Would more research potentially matter?
Why?
High uncertainty, but all plausible answers imply essentially the same conclusion
Possibly, but priority may be limited
Greater precision may not change an important interpretation or decision
High uncertainty and plausible answers imply very different conclusions
Potentially high value
Better evidence could substantially change what is believed or done
Moderate uncertainty near an important decision boundary
Potentially high value
A relatively modest reduction in uncertainty could change the preferred choice
Low uncertainty and the remaining plausible alternatives have minor consequences
Often lower priority
Additional evidence may provide little consequential information
Low probability that the current conclusion is wrong, but being wrong would be extremely consequential
May still deserve research
The consequences of error can make residual uncertainty important
The last situation is particularly important. Research priority cannot be inferred from uncertainty without also considering the consequences of being wrong .
More Uncertainty Does Not Necessarily Mean More Value of Information
In formal decision analysis, value of information concerns the expected improvement that could result from resolving uncertainty. This depends not only on the probability that the current decision is suboptimal, but also on the losses associated with making the wrong decision.
Two questions can therefore involve similar levels of uncertainty while having very different values for additional evidence.
Suppose researchers are equally uncertain between two instructional formats and between two treatments for a serious disease. If choosing the inferior instructional format produces only a negligible difference while choosing the inferior treatment could substantially affect health outcomes, the same apparent degree of uncertainty does not imply the same research priority.
That does not mean health research automatically outranks educational research. It means that uncertainty must be interpreted in relation to the consequences attached to the particular alternatives.
Uncertainty About an Inconsequential Detail Can Remain Inconsequential
Researchers sometimes become uncomfortable when confidence intervals are wide, estimates vary among studies, or a mechanism remains incompletely specified. Scientifically, those uncertainties may be worth acknowledging. They do not all require additional research.
Imagine that two measurement approaches differ slightly, but either produces sufficiently accurate information for the purpose at hand. Determining which is marginally more precise could reduce uncertainty without changing any substantive inference.
In such cases, reducing uncertainty may still have scientific value even when the decision remains unchanged , but that value needs its own justification. “We are still uncertain” is not enough.
What Matters Is the Uncertainty Relevant to the Question
A broad research problem may contain many uncertain parameters, mechanisms, assumptions, and contextual factors. Trying to reduce all of them equally can waste research effort.
Some uncertainties drive the conclusion; others barely affect it.
Formal value-of-information methods can examine which uncertain parameters contribute most to decision uncertainty and therefore where additional information could have the greatest value. Outside formal modelling, researchers can apply the same intuition more modestly: identify which unknowns actually determine the conclusion you care about.
This can prevent a common form of research accumulation in which investigators repeatedly measure variables that are uncertain but not particularly consequential.
Ask Whether the Proposed Study Can Actually Reduce the Uncertainty
High uncertainty may provide a strong reason to want better evidence. It does not establish that your proposed study will provide it.
A small, biased, poorly measured, or otherwise weak study may leave the relevant uncertainty largely unchanged. In some cases, it can add another imprecise estimate without clarifying why existing studies disagree.
The priority of a research project therefore depends partly on the expected information gain from the actual study, not merely the amount of uncertainty present before the study begins.
Watch Out
Do not argue that a study is necessary merely because “the literature is inconclusive.” Identify why it is inconclusive and explain how the proposed design will reduce the uncertainty that previous evidence could not resolve.
Sometimes the Rational Decision Is to Act Despite Uncertainty
Decisions often cannot wait until research eliminates uncertainty. Policymakers, clinicians, educators, organizations, and researchers routinely choose among alternatives using incomplete evidence.
Formal value-of-information analysis recognizes this explicitly. The question is not whether uncertainty exists, but whether additional research has sufficient expected value to justify obtaining it rather than acting on current evidence.
If further research is expensive, slow, unlikely to resolve the uncertainty, or unlikely to change the preferred action, proceeding with the best-supported current decision may be reasonable.
This is why the cost of answering a question must eventually be compared with the value of knowing the answer .
04 · A Practical Example
More Uncertain Does Not Always Mean More Worth Studying
Hypothetical Example
Choosing Which Uncertainty About an Online Course to Investigate
A university has implemented an online course and is considering two possible research questions.
Question A Researchers are highly uncertain whether students prefer one of two visually similar dashboard layouts. Existing evidence provides almost no indication which layout students prefer.
Question B Evidence moderately favors providing structured instructor feedback within 48 hours rather than within seven days, but uncertainty remains about whether the improvement in learning is large enough to justify the additional instructor workload.
Question A contains more uncertainty in the ordinary sense. Researchers genuinely do not know which dashboard students prefer. Yet if either layout works adequately and the choice has little bearing on learning or use, resolving that uncertainty may make little difference.
Question B contains less uncertainty because current evidence already leans toward one option. But the remaining uncertainty crosses an important decision: whether the expected educational benefit warrants a substantial recurring staffing commitment.
Additional evidence about Question B may therefore have greater research value despite the lower initial uncertainty.
The lesson is not that preferences are unworthy of study or that staffing questions always take priority. It is that the amount of uncertainty cannot determine priority without considering what different answers would change.
06 · What This Means for You
Prioritize Consequential Uncertainty, Not Uncertainty in the Abstract
When uncertainty is part of your justification for a study, make the argument more specific than “the evidence is limited” or “results are mixed.”
A simple decision framework
If uncertainty is high and different plausible answers would lead to meaningfully different conclusions
Additional research may have substantial value, especially if your study can discriminate among those possibilities.
If uncertainty is high but every plausible answer would change little
Do not assume that reducing uncertainty deserves high priority. Identify another scientific reason why greater precision matters.
If uncertainty is relatively low but the consequences of being wrong are substantial
Further research may still be justified because residual uncertainty carries meaningful expected consequences.
If your proposed study is unlikely to reduce the uncertainty materially
Improve the design, investigate the source of uncertainty, or reconsider whether this study is the appropriate next research step.
You can also apply the “what would change if we knew?” test . Imagine that tomorrow you received substantially better information. Would an important decision change? Would one explanation become meaningfully more plausible? Would subsequent research take a different direction?
If the answer is no, the uncertainty may be scientifically real without being a high research priority.
07 · A Quick Checklist
Before Prioritizing a Question Because the Answer Is Uncertain, Check:
Before using uncertainty as the justification, check:
Describe what is actually uncertain rather than saying only that the literature is inconclusive.
Identify the range of answers that remains reasonably plausible given current evidence.
Ask whether different plausible answers would lead to meaningfully different conclusions or decisions.
Consider the consequences of making the wrong inference or decision under current uncertainty.
Identify which particular unknowns are driving the consequential uncertainty.
Explain how your proposed study would reduce those uncertainties rather than merely add another estimate.
Consider whether the expected improvement in knowledge or decisions warrants the cost of obtaining additional evidence.
Do not assume that uncertainty must be eliminated before a reasonable decision can be made.
11 · Cite this Guide
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.
Recommended (Field Guide)
APA
MLA
Chicago
Copy Citation
