Paired Comparison: Pre-fix MERGED vs Post-fix Wave 1 (Full 10k)

Same 10,000 Sobol parameter sets. Two bugs fixed: TSQ tunnel-exit (b_years_since_quit=1 at exit) and initiation zeroed for ages 31+. Updated STS targets and fresh ABM stochasticity estimates.

Pre-fix: MERGED_10000_complete_R11cov (scored with old V_s)
Post-fix: 20260703_0711_postfix_wave1 (scored with fresh V_s from postfix stoch pilot)

Summary

	Design A			Design B
Metric	Pre-fix	Post-fix	Change	Pre-fix	Post-fix	Change
Best Q_STPM	90.0	35.6	-54.4	70.0	20.4	-49.6
Q_STPM p5	191.6	109.7	-81.9	177.4	88.9	-88.5
Q_STPM median	741.8	526.4	-215.3	606.0	408.2	-197.8
NROY (Q<34.3)	0	0	+0	0	6	+6
Jackson pass	1212 (12.1%)	1167 (11.7%)	-45	1147 (11.5%)	1151 (11.5%)	+4
Prevalence falls	1098 (11.0%)	2059 (20.6%)	+961	843 (8.4%)	1861 (18.6%)	+1018
Prev delta mean	+4.16 pp	+0.97 pp	-3.20 pp	+4.32 pp	+1.09 pp	-3.22 pp
% Q improved	—	99%		—	100%
% prev improved	—	99%		—	98%

Q_STPM: Pre-fix vs Post-fix (paired scatter)

Each dot is one parameter set. Points below the diagonal improved after the bug fixes.

ΔQ_STPM Distribution

Histogram of (post-fix Q) − (pre-fix Q). Negative = improved.

Prevalence Delta: Pre-fix vs Post-fix (paired scatter)

Each dot is one design. Points below y=0 have falling prevalence post-fix.

Prevalence Shift Distribution

Histogram of (post-fix prev delta) − (pre-fix prev delta). Negative = prevalence improved (shifted more downward).

Generated 2026-07-04 | Same 10k Sobol parameter sets | Pre-fix: MERGED_10000_complete_R11cov | Post-fix: 20260703_0711_postfix_wave1