Research replication
Decisions under uncertainty
Open the exact fictional inputs behind each worked study. Project cases use their companion model; auxiliary cash bridges retain a separate, verifiable calculation.
Return to the paper↗These are illustrative scenarios, not actual project or supplier data. Inputs and method boundaries govern each result.
Illustrative study
A useful signal can change the chosen action
Demonstrate how conditional actions create gross and net information value.
What changes
- Parameter: prior and likelihood interpretation
- Range: The canonical prior policy sweep plus the original eight separately frozen contrasting studies.
- Parameter: test cost and admissible action set
- Range: Useful, no-change, expensive, imperfect, inverse, perfect-price, restricted-action and reduced-upside fixtures.
Limits of the study
- Binary states and a single binary signal; state completeness is an assumption.
- Payoffs and likelihoods are illustrative, not empirical estimates.
- Risk-neutral expected monetary value; no utility, causal identification, delay or joint-signal engine.
Open this case in the model↗Illustrative study
An informative signal can leave the policy unchanged
Distinguish uncertainty reduction from an actionable branch change.
What changes
- Parameter: prior and likelihood interpretation
- Range: The canonical prior policy sweep plus the original eight separately frozen contrasting studies.
- Parameter: test cost and admissible action set
- Range: Useful, no-change, expensive, imperfect, inverse, perfect-price, restricted-action and reduced-upside fixtures.
Limits of the study
- Binary states and a single binary signal; state completeness is an assumption.
- Payoffs and likelihoods are illustrative, not empirical estimates.
- Risk-neutral expected monetary value; no utility, causal identification, delay or joint-signal engine.
Open this case in the model↗Illustrative study
A useful signal can cost too much
Keep gross information value separate from the purchase decision.
What changes
- Parameter: prior and likelihood interpretation
- Range: The canonical prior policy sweep plus the original eight separately frozen contrasting studies.
- Parameter: test cost and admissible action set
- Range: Useful, no-change, expensive, imperfect, inverse, perfect-price, restricted-action and reduced-upside fixtures.
Limits of the study
- Binary states and a single binary signal; state completeness is an assumption.
- Payoffs and likelihoods are illustrative, not empirical estimates.
- Risk-neutral expected monetary value; no utility, causal identification, delay or joint-signal engine.
Open this case in the model↗Illustrative study
A rare success state meets a weak negative-state filter
Expose prior incidence and asymmetric conditional likelihoods.
What changes
- Parameter: prior and likelihood interpretation
- Range: The canonical prior policy sweep plus the original eight separately frozen contrasting studies.
- Parameter: test cost and admissible action set
- Range: Useful, no-change, expensive, imperfect, inverse, perfect-price, restricted-action and reduced-upside fixtures.
Limits of the study
- Binary states and a single binary signal; state completeness is an assumption.
- Payoffs and likelihoods are illustrative, not empirical estimates.
- Risk-neutral expected monetary value; no utility, causal identification, delay or joint-signal engine.
Open this case in the model↗Illustrative study
A negative association still carries usable information
Challenge the informal rule that a positive signal must be favourable.
What changes
- Parameter: prior and likelihood interpretation
- Range: The canonical prior policy sweep plus the original eight separately frozen contrasting studies.
- Parameter: test cost and admissible action set
- Range: Useful, no-change, expensive, imperfect, inverse, perfect-price, restricted-action and reduced-upside fixtures.
Limits of the study
- Binary states and a single binary signal; state completeness is an assumption.
- Payoffs and likelihoods are illustrative, not empirical estimates.
- Risk-neutral expected monetary value; no utility, causal identification, delay or joint-signal engine.
Open this case in the model↗Illustrative study
Perfect information exactly meets its price
Provide a transparent zero-net-value boundary.
What changes
- Parameter: prior and likelihood interpretation
- Range: The canonical prior policy sweep plus the original eight separately frozen contrasting studies.
- Parameter: test cost and admissible action set
- Range: Useful, no-change, expensive, imperfect, inverse, perfect-price, restricted-action and reduced-upside fixtures.
Limits of the study
- Binary states and a single binary signal; state completeness is an assumption.
- Payoffs and likelihoods are illustrative, not empirical estimates.
- Risk-neutral expected monetary value; no utility, causal identification, delay or joint-signal engine.
Open this case in the model↗Illustrative study
Evidence cannot buy an inadmissible response
Remove the high-upside action rather than pretend information restores authority.
What changes
- Parameter: prior and likelihood interpretation
- Range: The canonical prior policy sweep plus the original eight separately frozen contrasting studies.
- Parameter: test cost and admissible action set
- Range: Useful, no-change, expensive, imperfect, inverse, perfect-price, restricted-action and reduced-upside fixtures.
Limits of the study
- Binary states and a single binary signal; state completeness is an assumption.
- Payoffs and likelihoods are illustrative, not empirical estimates.
- Risk-neutral expected monetary value; no utility, causal identification, delay or joint-signal engine.
Open this case in the model↗Illustrative study
The opportunity can change while research is commissioned
Use explicitly supplied reduced-upside payoffs to show that a changed decision perimeter requires a new fixture.
What changes
- Parameter: prior and likelihood interpretation
- Range: The canonical prior policy sweep plus the original eight separately frozen contrasting studies.
- Parameter: test cost and admissible action set
- Range: Useful, no-change, expensive, imperfect, inverse, perfect-price, restricted-action and reduced-upside fixtures.
Limits of the study
- Binary states and a single binary signal; state completeness is an assumption.
- Payoffs and likelihoods are illustrative, not empirical estimates.
- Risk-neutral expected monetary value; no utility, causal identification, delay or joint-signal engine.
Open this case in the model↗Illustrative study
Exact decision-value purchase price
Separate exact economic equality from positive and negative net information values near the same test price.
What changes
- Change: Only the aggregate test price changes across the three precision studies.
- Held constant: Prior, test likelihoods, complete response menu and all payoffs.
- Interpretation: The comparison tests arithmetic sign; a small computed surplus is not a reason to ignore assumption uncertainty.
Limits of the study
- Every probability and payoff is an authored illustration, not calibrated evidence.
- Risk-neutral binary-state utility and a single test remain the method boundary.
Open this case in the model↗Illustrative study
Price strictly below the same decision value
Separate exact economic equality from positive and negative net information values near the same test price.
What changes
- Change: Only the aggregate test price changes across the three precision studies.
- Held constant: Prior, test likelihoods, complete response menu and all payoffs.
- Interpretation: The comparison tests arithmetic sign; a small computed surplus is not a reason to ignore assumption uncertainty.
Limits of the study
- Every probability and payoff is an authored illustration, not calibrated evidence.
- Risk-neutral binary-state utility and a single test remain the method boundary.
Open this case in the model↗Illustrative study
Price strictly above the same decision value
Separate exact economic equality from positive and negative net information values near the same test price.
What changes
- Change: Only the aggregate test price changes across the three precision studies.
- Held constant: Prior, test likelihoods, complete response menu and all payoffs.
- Interpretation: The comparison tests arithmetic sign; a small computed surplus is not a reason to ignore assumption uncertainty.
Limits of the study
- Every probability and payoff is an authored illustration, not calibrated evidence.
- Risk-neutral binary-state utility and a single test remain the method boundary.
Open this case in the model↗