{"schema_version":1,"assessment_id":"eoa_inverse_innovation_exp03_opportunity320_20260801","source_experiment_id":"eoa_inverse_innovation_exp03_full320_20260801","cell_id":"deadweight_loss_reduction__statistics_experimental_design","archetype_slug":"deadweight_loss_reduction","domain_slug":"statistics_experimental_design","title":"Prespecified priority-weighted alpha allocation for heterogeneous hypothesis families","opportunity_summary":"Evaluate whether outcome-independent, prespecified alpha weights can improve decision-weighted power over equal Bonferroni-style allocation at the same sample size and verified familywise error rate, without unacceptable losses for protected hypotheses or represented groups. The proposal is structurally testable, but the prevalence of binding misallocation, magnitude of benefit, stakeholder demand, and distinctiveness from existing methods are unestablished.","adopter_authorizer":"The study sponsor and accountable scientific-statistical governance body, with ethics or regulatory approval when protocol registration, participant risk, or protected comparisons are affected.","scores":{"meaningful_impact":{"score":3,"rationale":"If multiplicity allocation is genuinely binding, the proposal could reduce consequential false negatives, delayed decisions, or additional sampling while preserving the error cap. The packet provides no evidence about how often this condition occurs or the realized magnitude of benefit."},"stakeholder_pull":{"score":2,"rationale":"Investigators, sponsors, statistical governance bodies, and downstream decision-makers are identifiable potential beneficiaries, but the packet contains no adoption inquiry, expressed demand, prevalence evidence, or willingness to accept priority weighting."},"incremental_advantage":{"score":4,"rationale":"Relative to equal allocation or increasing sample size, the intervention directly targets the proposed allocation wedge and can be compared at unchanged sample size and familywise error. Advantage remains conditional on defensible priorities and robust gains after protected-comparison losses are counted."},"distinctiveness_plausibility":{"score":2,"rationale":"The candidate specifies a coherent safeguard and evaluation architecture, but prior art is explicitly unsearched and the packet itself identifies established weighted multiple-testing and gatekeeping methods as the key unresolved boundary. A distinctive incremental contribution is therefore not yet supported."},"technical_implementability":{"score":4,"rationale":"Synthetic simulation, locked archived-design replay, prespecified weighting rules, and direct checks of familywise error and power incidence are technically bounded. Implementability falls short of very favorable because dependence misspecification, utility construction, and gaming resistance require substantial validation."},"adoption_authority_feasibility":{"score":3,"rationale":"The proposal names accountable sponsors, scientific-statistical governance, and applicable ethics or regulatory reviewers, and supplies halt and rollback rules. Feasibility is mixed because approval requirements and authority conflicts around contested priorities or protected floors are not resolved."},"evidence_readiness":{"score":4,"rationale":"The packet provides explicit baselines, same-error and same-sample comparisons, problem and intervention falsifiers, negative tests, and a safe non-live first step. Archived-design availability, performance tolerances, and utility definitions remain unspecified."},"safety_net_benefit":{"score":4,"rationale":"Outcome-informed weighting, error-rate increases, incomplete reporting, and removal of minimum protections are prohibited; specified triggers revert analysis to approved equal allocation and require reporting the deviation. Safeguards do not by themselves resolve institutional bias or cumulative underpowering of lower-weighted groups."},"scalability":{"score":3,"rationale":"A validated design and simulation workflow could be reused across multi-hypothesis experiments, but weights, protected floors, dependence structures, governance, and contested decision values must be adapted study by study. No cross-domain reproducibility evidence is supplied."}},"score_confidence":"MODERATE","costs":{"first_evidence":{"band_2026_usd":"50K_TO_250K","scope":"A preregistered simulation study plus replay of a small set of locked archived designs, including statistical labor, data preparation or access, sensitivity analysis, fairness incidence checks, gaming tests, and an auditable report.","confidence":"LOW","assumptions":["A small number of archived designs can be accessed without purchasing unusually expensive data rights.","No participants are enrolled and no live decisions are changed.","Several weighting rules, plausible effect distributions, and dependence structures are evaluated.","Senior statistical review and partner coordination are included."]},"initial_deployment_startup":{"band_2026_usd":"50K_TO_250K","scope":"Preparation for one prospective governed pilot, including weight-elicitation procedures, protected-comparison floors, simulation tooling, protocol and registration changes, independent statistical review, training, and approval coordination.","confidence":"LOW","assumptions":["The weighting method does not require new experimental equipment or major software infrastructure.","The host organization already has protocol, statistical, and compliance functions.","Costs represent incremental implementation resources rather than the underlying experiment's full cost.","Regulatory review, if required, is an amendment-scale process rather than a new product approval."]},"operational_launch":{"band_2026_usd":"50K_TO_250K","scope":"Incremental resources to execute and evaluate one prospective pilot through locked analysis and complete reporting, including monitoring, independent error-control verification, group-level incidence analysis, and rollback readiness.","confidence":"LOW","assumptions":["The underlying study is already funded and is excluded except for costs caused by the allocation intervention.","One hypothesis family and a limited number of protected comparisons are involved.","No safety trigger requires major protocol redesign or additional enrollment.","Both weighted and approved equal-allocation analyses can be retained for evaluation."]},"annual_recurring":{"band_2026_usd":"50K_TO_250K","scope":"Ongoing maintenance for a small portfolio of studies, including rule governance, simulation reruns, statistical audits, documentation, training, monitoring of group-level losses, and periodic recalibration.","confidence":"LOW","assumptions":["Adoption covers a limited organizational portfolio rather than a large multinational trial network.","Weights remain prospectively governed and require review for each new hypothesis family.","Existing statistical software and compliance systems can be extended.","The estimate excludes ordinary study conduct and participant recruitment."]}},"research_burden":"HIGH","earliest_credible_horizon":"3_TO_12_MONTHS","pipeline_gates":{"recognizable_externally_supportable_problem":{"status":"YES","reason":"The candidate identifies an observable discrepancy: high-priority hypotheses can have lower power under equal allocation than under valid alternatives at the same familywise error rate and sample size, with explicit conditions that would falsify the diagnosis."},"identifiable_adopter_or_authorizer":{"status":"YES","reason":"The study sponsor and accountable scientific-statistical governance body are explicitly assigned decision authority, with ethics or regulatory approval where protocol or participant risk is affected."},"distinct_testable_incremental_claim":{"status":"YES","reason":"The incremental claim is that prespecified outcome-independent weighting improves decision-weighted power over equal allocation at the same verified familywise error rate and sample size without unacceptable hypothesis- or group-level losses."},"bounded_next_evidence_step":{"status":"YES","reason":"Synthetic simulation and locked archived-design replay are non-live, bounded comparisons against equal allocation and include falsification by absent robust gain, excess error, or unacceptable protected losses."},"no_unresolved_safety_or_authority_stop":{"status":"YES","reason":"The proposed first step changes no live decision and is bounded by explicit exclusions, halt triggers, protected floors, complete reporting, and reversion to approved equal allocation. Bias and gaming remain research questions but do not stop the non-live evidence step."},"implementation_cost_scope_and_range":{"status":"UNCERTAIN","reason":"The candidate defines the analytical and governance activities, but the number and accessibility of archived designs, complexity of approval, portfolio size, and whether prospective evaluation requires additional enrollment are unspecified, so only low-confidence broad resource bands are possible."}},"blocking_evidence":["No evidence establishes that equal alpha allocation is binding in a material set of candidate experiments rather than sample size, variance, or small effects being the limiting factor.","Prior art is unsearched, so no distinctive contribution beyond existing weighted multiple-testing or gatekeeping approaches is established.","No locked rule for eliciting priorities, utilities, or protected-comparison floors has been validated as outcome-independent and resistant to institutional bias or gaming.","No simulation or archived replay yet demonstrates robust improvement across plausible effects and dependence structures at the same verified familywise error rate.","Archived-design access, prospective adopter willingness, approval requirements, and acceptable hypothesis- and group-level loss tolerances are unknown."],"next_evidence_step":"Preregister a non-live benchmark using synthetic data and a small set of locked archived designs. Compare equal Bonferroni allocation, several prospectively specified priority-weighting rules, and the nearest rival of equal allocation with increased sample size across plausible effect and dependence scenarios. Verify familywise error, decision-weighted power, ordinary hypothesis-level power, group-level incidence, robustness to misspecification, and gaming sensitivity. Falsify advancement if equal allocation is not binding, no weighting rule yields a robust same-sample improvement, error control fails, or any protected comparison crosses its approved loss floor.","research_questions":["In which archived or planned hypothesis families is multiplicity allocation demonstrably binding rather than sample scarcity, measurement variance, or weak effects?","Can substantive priorities and minimum protections be fixed independently of outcomes and audited for investigator gaming?","How should decision-weighted power be defined without converting contested scientific or group values into unsupported utility estimates?","Do gains persist across plausible effect sizes, dependence structures, family definitions, and misspecified assumptions?","Which hypotheses or represented groups lose power, and what prospective floors or veto rights are acceptable to accountable governance?","What methods already exist for weighted multiple testing, gatekeeping, independent hypothesis weighting, and related safeguards, and what incremental claim remains?","Will sponsors, statisticians, ethics reviewers, and affected communities authorize a prospective comparison with complete reporting and rollback?","What monitoring tolerances should trigger reversion when error, protected-comparison power, auditability, or group incidence deteriorates?"],"recommendation":"PRIOR_ART_RESEARCH","uncertainty_constraints":["Closed-book assessment provides no external evidence of prevalence, realized impact, adopter demand, market size, prior art, or exact cost.","Impact and advantage apply only when unequal decision priorities are defensible and multiplicity allocation is the binding constraint.","Distinctiveness cannot be inferred from the proposal's specification quality or from its structural transfer framing.","Cost bands are low-confidence resource equivalents and exclude the ordinary cost of conducting the underlying experiment unless caused by this intervention.","Aggregate decision-weighted gains cannot override familywise error validity, complete reporting, ethics, or minimum protections for hypotheses and represented groups."],"closed_book_prior_art_boundary":"The packet explicitly labels prior art as UNSEARCHED and identifies novelty evidence as weak. This assessment therefore makes no claim that prespecified weighted alpha allocation, gatekeeping, incidence analysis, or the proposed safeguard architecture is new, rare, or superior to known methods. External prior-art review is required before attributing distinctiveness or selecting a contribution claim."}