Add make_boxa_variations.m producing 6 variation folders (data.csv + analyze.m
+ result.txt) for the paper LME over windows 0-5, 6-10, 0-10:
boxa_a2 B2 vs A2 + Box-A (b2 vs a2+a)
boxa_b2 B2 + Box-A vs A2 (b2+a vs a2)
plus variations/boxa_summary.csv (residual / Satterthwaite / random-slope
interaction p). Early window (0-5) is obs-level significant (res p~0.03-0.04)
but n.s. under the honest random-slope test (rs p~0.13); later windows n.s.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Experiments DB Dev