1[2 {"id":"C1","text":"CR0 under-covers in every design cell, failing (<0.90) in all G=5 and G=10 cells and reaching only 0.9165-0.9275 at G=30.","results_keys":["coverage","failure_cells"],"status":"supported","caveat":"CR0 at G=10, ICC=0 (0.8965) is within one MC SE of the 0.90 threshold."},3 {"id":"C2","text":"The CR0 variance estimator is biased downward, with E[V_CR0]/Var(ybar) of about 0.48-0.66 at G=5 and 0.93-0.94 at G=30; the Monte Carlo estimates agree with the analytic expectation within 2.5 MC SE.","results_keys":["rel_bias_var","rel_bias_var_exact","rel_bias_var_mcse"],"status":"supported","caveat":"Specific to the 8/16/32/64 unbalanced design and Gaussian DGP."},4 {"id":"C3","text":"CR1 (G/(G-1) scaling with t_{G-1}) has the highest average coverage at G=5 (0.8905) but still fails for ICC 0.2 and 0.5 at G=5.","results_keys":["best_procedure_small_G","mean_coverage_G5","coverage","failure_cells"],"status":"supported","caveat":"CR1 vs WCB difference at G=5 ICC=0.5 (0.8755 vs 0.8510) is a few MC SE; paired precision not separately quantified."},5 {"id":"C4","text":"The Rademacher wild cluster bootstrap-t fails at G=5 (coverage 0.851-0.858) despite having the widest mean intervals, but is consistent with nominal 95% at G=30 for all ICC values and closest to nominal at G=10 for ICC>0.","results_keys":["coverage","nominal_cells","failure_cells","data_issues"],"status":"supported","caveat":"At G=5 Rademacher weights yield only 16 distinct |t*| values; other weights (Webb) or enumeration not studied. Mean half-width comes from table1.csv."},6 {"id":"C5","text":"11 of 27 (procedure, cell) pairs are failure regimes and only 5 are consistent with nominal coverage under the 2-MC-SE rule.","results_keys":["n_failure_cells","failure_cells","nominal_cells"],"status":"supported","caveat":"Classification uses per-pair 2-SE rule without multiplicity adjustment; borderline pairs could change class under another seed."},7 {"id":"C6","text":"Ignoring clustering (naive iid SE) gives coverage of only about 0.29-0.49 whenever ICC>0.","results_keys":["baseline_iid_coverage"],"status":"supported","caveat":"Baseline only; not one of the three studied procedures."}8]9