Security, Agency, and Voluntary Contribution — Pilot Operational Note v0.5: Randomization and Inference Freeze

Design note for a 100-household feasibility and variance-estimation pilot on freely chosen unpaid help under four factorial arms (00, 01, 10, 11). Revision v0.5 freezes unstratified individual-household randomization with equal allocation and names the primary inference rule: the randomized contrast of freely chosen hours under arm 11 versus arm 01, estimated by a baseline-adjusted factorial model with HC3 covariance fitted within each joint completion and pooled by Rubin's rules. Randomization inference is demoted to optional secondary methodology until a missing-data-compatible procedure is frozen and validated. It separates the exact marginal arm-mean decomposition (T + C + P) from the HC3 coefficient, specifies how partial diaries enter through joint completion of total and freely chosen hours, and keeps the detectable-effect surface labelled pre-pilot. This is a design proposal, not a registered protocol. All simulation figures are synthetic, and no participant outcomes are reported. Revision v0.5 (6 October 2026). Only the latest revision is deposited; earlier v0.4 notes and the v0.4→v0.5 boundary memo are superseded. The document carries the author's CONFIDENTIAL — PROPRIETARY / RESEARCH HYPOTHESIS banner; the author approved public release of confidential-marked research. Source: author's iCloud folder "Scientific research published final update" (deposited 2026-10-20).