Testing Zodiac Marriage Taboos in a Chinese Lineage Genealogy, 1507-1855: Frozen Main Analysis Set and Extraction Pipeline

This is the frozen main analysis set and the complete extraction/audit pipeline for a preregistered study testing whether traditional Chinese zodiac marriage taboos (六冲 "six clash", 六害 "six harms") and preferences (六合 "six harmonies", 三合 "trine harmonies") left any trace in real historical marriages. For centuries, Chinese matchmaking practice has classified zodiac (&23646;相) pairs as inauspicious or auspicious and steered marriage decisions accordingly — a practice that continues commercially today. No population-scale test of whether these taboos left any behavioral trace in real marriages has ever been published. Lineage genealogies are the only known data source recording lunar birth dates for both husbands and wives, couple by couple, across centuries. Both outcomes are pre-declared as informative: a behavioral trace would be a large-scale finding on culture shaping mate selection; a null trace is a falsification result for the folk system. Data: 12,262 person rows transcribed from the biographical volumes of the Longxi Chen lineage genealogy (Shanghai Library open-access platform, STJP000475), covering branch lineages across volumes v05, v07, v10, v12, v13, v14 and v15 (partial), birth years 1507–1855. Couple construction yields 4,152 husband–wife links; 3,817 couples have parseable birth years for both spouses (3,132 primary-marriage couples under the pre-registered first-wife rule). Matching quality validated by age-gap distribution (median husband older by 5 years; |gap| ≤ 15 y in 89.4%). A hold-out verification set of ≈2,500 pages remains unextracted until after the main analysis (pre-registered out-of-sample replication). Frozen before any outcome computation (2026-10-27): the 12×12 couple-branch cross-tabulation has never been computed as of this deposit. The confirmatory analysis runs once, after OSF preregistration. trial_rows.json SHA-256 = 849FBD4B2ACF3DEF6FEAC71CE06A0D2AF9B3F3272F1700F2648E53DC3F290C13. Contents: trial_rows.json (frozen data), pipeline_and_audit_code.zip (extraction pipeline, parsing/QA code, pre-registration audit scripts, batch provenance log), README.md (source, data dictionary, methods), marriage_freeze_stats.md (couple-audit statistics), SHASUMS.txt.