How do I decide what evidence would change my mind about a preferred option?

Write the change rule before looking for the result

Start with the reason your preferred option is currently ahead. Reduce it to one claim that matters to the choice, then write what observation would weaken that claim and what action the observation could justify. This makes the rule inspectable before a favorable or unfavorable result invites a new explanation.

The 2026 Test and Learn guidance recommends agreeing decision criteria before testing, assessing evidence against agreed success criteria, and considering in advance what evidence would justify continuing. The CDC evaluation framework likewise recommends discussing how potential positive and negative findings could affect decisions before the evidence is used. These are formal evaluation sources; the useful low-stakes transfer is the discipline of deciding the evidence-to-action link early.

Use the COUNTER card

COUNTER is a Forever Free Compass framework, not an external research finding. Complete it for one important reason behind the preferred option rather than trying to pre-write every possible future fact.

1. Current claim: write the decision-driving reason you presently believe, including the source or observation behind it.
2. Observable counter-signal: name what you could actually see, verify, or learn that would weaken that reason.
3. Use in the decision: state whether the signal could change the ranking, cross a boundary, or alter only the next step.
4. Needed evidence quality: match the source, method, amount, and independence of evidence to the cost and consequence of being wrong.
5. Transfer boundary: define the people, setting, time, version, and conditions for which the evidence would be relevant.
6. Explanation alternatives: name at least one other reason the same observation might occur and what would distinguish it.
7. Response: pre-write whether you would pause, adapt, seek stronger evidence, reduce the commitment, or switch options.

Make the counter-signal relevant to the decision

A counter-signal matters only if it bears on the actual reason for preference. If you prefer an option because it fits a fixed time boundary, evidence about popularity does not test that reason. If you prefer it because you expect a certain opportunity to remain available, credible evidence about availability may matter directly.

The Government Analysis Functional Standard says the question, scope, assumptions, intended use, data fitness, uncertainty, and limitations should be explicit and proportionate to the decision. Apply that discipline by tracing each proposed counter-signal to a named claim and decision consequence. Interesting evidence that cannot change a criterion, boundary, assumption, or next action is context, not a change rule.

Match evidence strength to the stakes

Do not invent one universal number of observations. A low-cost, reversible choice may need only enough credible evidence to justify a smaller next step. An expensive, irreversible, public, or high-consequence choice needs stronger sources, methods, and independent checks. If the claim is causal, a convenient before-and-after observation may not support it.

The Green Book recommends simple what-if questions for lower-cost, lower-risk uncertainty and more advanced analysis when costs and risks are high. It also warns that estimating probabilities can introduce spurious accuracy. The practical lesson is to describe the evidence rule plainly—source, context, observation, and consequence—instead of manufacturing a precise threshold that the available evidence cannot justify.

Protect the interpretation from preference

Record mixed and unfavorable evidence, not only the result that supports the current favorite. Keep the source, date, context, limitation, and alternative explanations beside the observation. If the evidence comes from a different person, setting, version, or time, explain why it transfers before using it.

The CDC framework emphasizes objectivity, transparency, clear limitations, and preventing positive or negative findings from being selectively overemphasized or used outside their agreed purpose. The Government Analysis Functional Standard similarly calls for balanced uncertainty and avoiding unwarranted confidence in a particular option. A written COUNTER card cannot remove bias, but it makes later reinterpretation easier to notice.

Change the next action at the right level

Evidence can change confidence without deciding the whole choice. A weak but relevant counter-signal may justify checking again. A credible result may justify adapting the option or lowering the commitment. Evidence that crosses a non-negotiable boundary may justify stopping even when other signals look favorable. State the response in advance, but preserve room to explain why genuinely new context changes the rule.

Use COUNTER only for ordinary, low-stakes reflection. It is not a validated research, evaluation, statistical, safety, legal, medical, financial, employment, housing, or safeguarding method. Do not use a personal evidence rule to bypass consent, confidentiality, professional duties, formal review, or qualified guidance for consequential decisions.