How Decision-Making Breaks | Bias, Evidence and Risk

Did you know? A meeting can contain charts, confident experts and pages of statistics—and still answer the wrong question. Everyone argues about which supplier is cheapest, but nobody checks whether the material is suitable for the job.

This is a familiar way decision-making breaks. Data do not rescue a poorly defined decision. Nor does confidence prove that a chosen explanation is correct.

One-minute answer: decisions can fail when the problem is misidentified, relevant evidence is missing or distorted, alternative explanations are ignored, consequences and values are unclear, or changing circumstances do not trigger a review. Biases can contribute, but a disappointing outcome does not by itself prove that the decision process was irrational.

The first broken step: solving the wrong problem

Imagine a school team asking, ‘Which new educational app should we buy to raise comprehension scores?’ There may be good options. But the question assumes an app is the missing intervention.

A more basic question is: ‘What prevents students from answering comprehension questions independently?’ Perhaps vocabulary depth is weak, students cannot use evidence or there is insufficient feedback. A purchase made before investigating those possibilities may address the wrong need.

The same principle applies to business and public policy. A manager may blame low sales on advertising when delivery reliability is the underlying issue. A city may respond to missed bus connections by adding signs when the timetable itself needs adjustment.

The US National Academies’ 2026 guide to decision-making names misidentification of the problem, inadequate data, misleading presentation, biases and failure to adapt as common ways factual competence can break down. These are reasons to examine the process before celebrating its apparent thoroughness.

Six points where reasoning can go off course

Unclear objective: the people involved have not agreed on what a satisfactory result would look like.
Anchoring on the first explanation: early information is treated as the baseline, while later contradictory evidence receives too little attention.
Selective evidence: the team searches mostly for reasons to support its preferred option instead of looking for conditions under which it would fail.
Missing stakeholders: people who would experience the consequences are left out, so practical constraints remain invisible.
Unexamined trade-offs: the proposed benefit is discussed without comparing costs, fairness, safety, time or what is sacrificed.
No revision rule: the decision is implemented but nobody has established what result would justify modifying it.

These weaknesses can reinforce each other. A team commits to a favoured answer, stops seeking competing evidence and then treats later problems as disloyal objections rather than useful information.

Biases are shortcuts, not proof that people are foolish

The human brain often relies on quick patterns and experience. That is useful in ordinary life: recognising a familiar route or detecting an obvious mistake should not require a long calculation every time.

Difficulties arise when a helpful shortcut is applied to a complex or unusual situation without checking whether its assumptions still hold. The first plausible explanation can become the only one considered.

The National Academies’ 2026 discussion of heuristics and cognitive bias explains why quick judgement and deliberate analysis both have roles in decision-making. A practical response is to notice when a decision deserves slower checking rather than treating all intuition as bad.

A classroom example: the ‘careless student’ label

Hypothetical example. A child loses marks on several Mathematics papers. A parent says, ‘The problem is carelessness.’ A tutor begins assigning longer timed worksheets. Weeks pass without much improvement.

On closer inspection, the child is making a specific error when subtracting a negative number. The original explanation may have been appealing because the mistakes appeared in simple questions, but it did not identify the mathematical cause.

A stronger decision process starts with the actual working, checks whether the student can explain the operation and compares an untimed example with a fresh timed question. The next intervention is then based on the cause supported by evidence.

The same principle appears in How Examinations Fail | When Scores Stop Measuring Understanding. A test score is information; it is not a complete diagnosis of the learner.

Why ‘we have the data’ is not enough

An analysis can be technically accurate while irrelevant to the choice. Imagine comparing the average travel time on two bus routes when the real decision concerns accessibility for passengers using wheelchairs. Averages alone may overlook whether a route can be used.

Useful evidence must match the decision’s purpose and include relevant conditions, uncertainty and affected people. A large sample does not solve a wrong measurement. A beautiful chart cannot reveal an unasked question.

Before using a result, ask which population was measured, whether the data are current, what is excluded and whether the measure represents the outcome that matters.

What to do when experts disagree

Disagreement between qualified people may reveal genuine uncertainty, different assumptions or different values rather than one side acting irresponsibly. Ask what evidence would change each person’s recommendation and which uncertainties are material to the decision.

Technical expertise deserves attention in its proper domain. A specialist can estimate a bridge’s structural condition; residents can explain how a closure affects their travel. The best decision may need both kinds of information.

The National Academies’ decision-making guide describes seeking internal and external advisers from different perspectives, as well as recognising interactions across a large organisation.

The opposite trap: endless analysis without a choice

Checking evidence matters, but a decision can also fail when people postpone action indefinitely while waiting for impossible certainty. Some opportunities expire. Some risks increase while a choice is delayed.

A good process distinguishes decisions that are easy to reverse from those involving serious, irreversible consequences. It also considers whether more information could change the outcome and whether obtaining that information is feasible.

A classroom teacher choosing a ten-minute practice exercise may not need a formal committee. A government considering a high-consequence infrastructure intervention needs much stronger evidence, legal review and public accountability.

Seven questions that repair a shaky decision

What decision are we actually making? Write it as one clear question without embedding a favoured answer.
What matters? Identify the intended outcomes, non-negotiable constraints and relevant values.
What alternatives exist? Include a realistic status-quo or limited-action option.
What do we know? Separate observed facts, estimates, assumptions and missing information.
What would prove us wrong? Actively seek disconfirming evidence and the strongest plausible alternative explanation.
Who is affected? Include appropriate expertise and the experience of those who will bear consequences.
When will we revisit it? Choose a trigger, owner or review point for learning from results.

This is a simple thinking framework, not a guarantee that every decision will produce a good outcome. Sometimes an appropriate, evidence-informed choice leads to an unfavourable result because the future contains uncertainty.

What is a decision log good for?

A lightweight decision log records the problem, date, options considered, assumptions, evidence, reason for choosing, responsible person and review condition. It helps a team distinguish ‘what we knew then’ from hindsight after outcomes become clear.

If a decision turns out poorly, the log can reveal whether crucial information was ignored, the environment changed or a reasonable risk materialised despite sensible preparation. The purpose is disciplined learning, not retroactive blame.

For the wider operational link, see How Strategy Fails | When Plans Ignore Feedback and Reality. A choice becomes useful only when responsibilities, execution and feedback remain aligned.

Warning signs that a team is reasoning in a circle

Questions are framed to make one solution seem inevitable.
People use ‘obviously’ or ‘everyone knows’ instead of examining the supporting evidence.
Strong objections are dismissed without documenting their substance.
An important decision depends on a statistic whose definition nobody can explain.
The team cannot name any result that would trigger reconsideration.
Success is declared from completed activity rather than the outcome the decision was meant to improve.

A warning sign is an invitation to investigate, not proof that a decision is doomed. Good judgement requires evidence, context, values and an appropriate degree of caution.

Frequently asked questions

Does every bad outcome prove a bad decision?

No. A well-reasoned decision under uncertainty can still have an unfavourable outcome. Judge both the evidence available at the time and the quality of the process.

Is relying on intuition always wrong?

No. Experience and pattern recognition can be valuable. Higher-stakes, unfamiliar or disputed situations often deserve explicit checking of assumptions and alternatives.

How can students practise better decisions?

Have them state a claim, identify supporting evidence, consider at least one alternative explanation and say what new fact would change their view.

What if the team cannot agree?

Clarify who has legitimate authority, what criteria matter and which uncertainties need evidence. Respectful disagreement about values or trade-offs may remain even when facts are settled.

The final idea

Decision-making breaks when the confidence of an answer outruns the quality of the question and evidence behind it. Repair begins by naming the real decision, testing alternatives and leaving room to learn when circumstances change.

Explore the other How Things Fail guides

How Examinations Fail | When Scores Stop Measuring Understanding
How Decision-Making Breaks | When Bias Overrides Evidence (this guide)
How Innovation Stalls | When Good Ideas Cannot Scale
How Early Warning Systems Fail | When Alerts Do Not Reach People

To connect these explanations with the wider series, see How Education Fails, How Strategy Fails and the unchanged Secondary 1 Mathematics Tutor Clementi | Small Groups Tutorials.

Evidence and further reading

The situations above are illustrative teaching examples, not claims about identifiable students or institutions. Good responses depend on the actual evidence, duties and circumstances.

How Decision-Making Breaks | When Bias Overrides Evidence