F1 Interval Selection Guide Method

An F1 Interval Selection Guide Method is a decision support method that provides evidence-based recommendations for choosing appropriate confidence interval methods based on sample size, coverage requirements, and computational constraints.

• AKA: F1 CI Method Selection Framework, Interval Method Decision Tree, F1 Confidence Interval Recommendation System, Evidence-Based F1 CI Guide.
• Context:
  • It can typically recommend Wilson+CC for n < 20 based on Brown et al. (2001) findings.
  • It can typically suggest Wilson or BCa for n = 20-40 for optimal coverage.
  • It can typically allow standard Wilson or delta method for n > 40.
  • It can often warn against Wald F1 Confidence Interval Method showing 85% actual coverage.
  • It can often document trade-offs between computational cost and coverage accuracy.
  • It can often incorporate recent research findings like Lam (2024) comparisons.
  • It can range from being a Simple F1 Interval Selection Guide Method to being a Comprehensive F1 Interval Selection Guide Method, depending on its decision complexity.
  • It can range from being a Conservative F1 Interval Selection Guide Method to being a Liberal F1 Interval Selection Guide Method, depending on its coverage priority.
  • It can range from being a Automated F1 Interval Selection Guide Method to being a Manual F1 Interval Selection Guide Method, depending on its implementation.
  • It can range from being a Static F1 Interval Selection Guide Method to being a Adaptive F1 Interval Selection Guide Method, depending on its update mechanism.
  • ...
• Example(s):
  • Sample Size Based Decisions, such as:
    • n=15: Recommend BCa bootstrap or Wilson+CC (avoid Wald).
    • n=30: Wilson or Agresti-Coull acceptable, BCa if computational feasible.
    • n=100: Any method except Wald acceptable, Wilson still preferred.
  • Coverage Requirement Decisions, such as:
    • Must achieve ≥95%: Use Wilson+CC or BCa, avoid Wald and basic bootstrap.
    • Can tolerate 93%: Wilson without CC acceptable.
    • Computational speed critical: Agresti-Coull over BCa.
  • Boundary Proximity Decisions, such as:
    • F1 near 0 or 1: Must use asymmetric method (Wilson, BCa).
    • F1 in [0.3, 0.7]: Even Agresti-Coull acceptable.
    • Unknown F1 range: Default to Wilson for safety.
  • ...
• Counter-Example(s):
  • Ad-Hoc Selection Method, which chooses without evidence.
  • Single Method Approach, which uses same method regardless of context.
  • Trial-and-Error Selection, which tests multiple methods post-hoc.
• See: Decision Support Method, F1 Confidence Interval Construction Method, Wilson Score F1 Confidence Interval Method, BCa Bootstrap F1 Confidence Interval Method, Agresti-Coull F1 Confidence Interval Method, Coverage Probability Validation Method, Sample Size Planning, Brown-Cai-DasGupta Recommendation, Lam 2024 Study, Evidence-Based Practice.

---