This record contains an extracted, reconciled, and analyzed dataset of family-court custody-related case filing statistics from Puerto Rico's judicial branch, spanning sixteen consecutive fiscal years (FY2010–FY2025), the complete Python extraction and analysis pipeline used to produce it, and a supplementary analysis relating this caseload series to household-composition trends drawn from IPUMS USA's Puerto Rico Community Survey (PRCS) microdata. The caseload data was compiled by parsing the "Movimiento de casos de relaciones de familia" (family relations caseload) tables from the Anuario Estadístico del Poder Judicial de Puerto Rico, the annual statistical report published by the Oficina de Administración de los Tribunales (Office of Court Administration). It tracks new filings, dispositions, and pending caseloads for custody, patria potestad (parental authority), and related compound causes of action across the study period. This dataset was compiled to support empirical analysis in a forthcoming law review article examining Puerto Rico's non-adoption of the Uniform Child Custody Jurisdiction and Enforcement Act (UCCJEA) and its implications for interstate and territorial custody disputes; a citation will be added to this record's "Related identifiers" once the article is placed. The dataset establishes that Puerto Rico's underlying custody caseload is large and stable — averaging approximately 2,090 new Custodia filings per year (SD=170) across the full study period — situating the doctrinal jurisdictional gap examined in the article within an active, ongoing docket rather than a hypothetical or rare edge case. What is included pr_custody_filings_extracted.csv — the final tidy custody caseload dataset, one row per fiscal year per cause-of-action category, including filing counts, table-format provenance, a data-quality flag (comparable_across_years), and a plain-text note explaining any reconciliation issues or methodological caveats specific to that row pr_household_composition_annual.csv — annual, survey-weighted household-composition indicators for Puerto Rico (2010–2019, 2021–2024; 2020 is absent from the source survey — see Known Limitations), derived from IPUMS USA PRCS microdata (not redistributed here; see IPUMS citation and extract specification below) pr_household_caseload_merged.csv — the two series merged by year, with the approximate fiscal-year/calendar-year alignment used and documented UCCJEA.ipynb — the complete, documented Python/Colab notebook performing extraction, reconciliation, and statistical analysis: descriptive statistics, Mann-Kendall trend tests, stationarity tests, segmented regression/Chow-test structural break analysis, and the household-composition merge and correlation analysis dataset_metadata.json — machine-readable metadata including package versions used, SHA-256 checksums of all 16 source PDFs, and a summary of known manual overrides pr_custody_trends.png, pr_custody_event_study.png, pr_custody_breakpoint_comparison.png, household_vs_caseload_trend.png — generated visualizations The 16 source PDF reports (Anuarios Estadísticos, FY2010–FY2025), reproduced here for archival completeness and reproducibility, sourced from Puerto Rico's Poder Judicial public website IPUMS data citation and reproducibility Household-composition indicators are derived from IPUMS USA, Puerto Rico Community Survey samples (SAMPLE codes ending "02"), 2010–2019 and 2021–2024. In accordance with IPUMS's data license, the raw extract is not redistributed in this record; only the aggregate annual indicators computed from it (pr_household_composition_annual.csv) are included. Researchers wishing to reproduce this component should obtain their own extract directly from IPUMS USA (https://usa.ipums.org) using the variable list documented in Cell 15 of the accompanying notebook, and cite: Steven Ruggles, Sarah Flood, Matthew Sobek, Daniel Backman, Annie Chen, Grace Cooper, Stephanie Richards, Renae Rodgers, and Megan Schouweiler. IPUMS USA: Version 15.0 [dataset]. Minneapolis, MN: IPUMS, 2024. https://doi.org/10.18128/D010.V15.0 Methodology summary Puerto Rico's Anuario reports use two distinct table layouts across the study period: a 13-column "materia y causa" format nesting specific causes of action under aggregate subject headers (FY2010–FY2019), and an 11-column flat "causa de acción" format with no header aggregation (FY2020–FY2025). The extraction pipeline infers table format on a per-page basis (not per fiscal year), since FY2019 — the year Puerto Rico's judiciary migrated case-management systems from SIMEC to SUMAC-Civil mid-year — contains pages in both formats representing two genuinely non-overlapping case populations that must be summed to obtain an accurate annual total. Every extracted row is checked against internal arithmetic identities expected to hold by construction. A distinction is drawn between hard reconciliation failures (which exclude a row from the analytic sample) and informational discrepancies — small, consistent closing-balance mismatches that plausibly reflect genuine case-reclassification adjustments in the underlying court records rather than extraction error, and which are retained with a disclosed note rather than excluded. One data point (FY2019 "Custodia," the pre-migration SIMEC-system page) could not be reliably parsed by the automated pipeline despite multiple extraction strategies; its value was obtained by direct manual transcription from the source PDF, cross-verified against the table's internal arithmetic, and is applied via a documented, disclosed override rather than silently patched. The household-composition merge joins PRCS calendar-year indicators to Puerto Rico fiscal-year (July–June, labeled by ending year) caseload figures on census_year == fiscal_year, an approximation disclosed and stress-tested (not asserted as exact) in the notebook, since no precise calendar-to-fiscal-year mapping exists for survey-based annual estimates. Full methodological detail, including the rationale for each design decision, is provided in inline documentation throughout the notebook. Key empirical findings Custody ("Custodia") filings were stable across the study period: mean 2,090/year, median 2,040/year, SD 170, no significant linear or Mann-Kendall trend over the full 16-year span (p>0.5 for both). A segmented regression identified a statistically significant structural break in the Custodia filing trend at fiscal year 2018 (Chow test F=5.96, p=0.016): a significant upward pre-2018 trend (+68 cases/year, p=0.007) reversed to approximately flat following FY2018 (slope change of −91 cases/year, p=0.009). This break coincides with three contemporaneous, confounded events: Hurricane Maria (landfall September 2017), the enactment of Puerto Rico's Law 102-2018 (governing relocation of minors), and the SIMEC-to-SUMAC-Civil case-management system migration (FY2019). This dataset's design cannot individually attribute the break to any one of these three developments. A robustness check testing an alternative break point at fiscal year 2020 (isolating Puerto Rico's 2020 Civil Code revision) found no comparable structural break (Chow test F=1.97, p=0.18), suggesting the 2020 Civil Code revision is not an independent driver of the FY2018 pattern. Household composition does not explain the Custodia filing pattern. Puerto Rico's share of single-parent households fell steadily from 53.3% (2010) to 40.3% (2024), and households with a child under 5 fell from 12.1% to 5.6% over the same period — a substantial demographic shift consistent with sustained net out-migration. Custodia filings show no comparable trend and do not track this decline: correlations between household-composition measures (both as population shares and as survey-weighted counts) and Custodia filings are weak and not statistically significant (|r| 0.18 across all tested measures, n=14 overlapping years). This null result is offered as evidence against a purely demographic explanation for custody caseload levels, not as evidence of a specific alternative cause. An exploratory secondary measure — Custodia filings per 1,000 single-parent households — shows an approximate 45% increase from FY2010 to FY2024 (2.79 to 4.05 per 1,000), suggesting the intensity of custody litigation relative to the shrinking household base may have risen over the study period. This measure is reported as descriptive and exploratory only (n=14, low statistical power); it is consistent with, but does not independently confirm, the FY2018 structural break finding above. Known limitations This dataset measures aggregate custody caseload volume; Puerto Rico's public statistical reporting does not isolate the subset of filings specifically involving interstate or mainland-U.S.-connected jurisdictional disputes, which is the precise phenomenon of doctrinal interest in the accompanying article. The aggregate volume is offered as evidence of scale and administrative significance, not as a direct count of interstate disputes. The "Patria potestad" category is not directly comparable across the FY2010–2019 and FY2020–2025 periods (see notebook Cell 6/7); it is therefore excluded from the household-composition merge, which uses "Custodia" only. The PRCS series is missing 2020 entirely (ACS/PRCS data collection was suspended that year due to COVID-19 field-operation disruptions), and the fiscal-year/calendar-year join is an approximation whose sensitivity is tested (±1-year shift) in the notebook; at least one correlation examined changes sign under this shift, underscoring that the household-composition findings should be read as exploratory. With n=16 annual observations for the caseload series (and n=14 for the household-merged series), stationarity tests (ADF, KPSS), structural-break tests, and correlation analyses are inherently low-powered; results are reported with explicit caveats in the notebook and should be treated as descriptive rather than confirmatory. License Derived dataset (CSV, JSON): CC-BY 4.0 Source PDFs: Puerto Rico government public records Household-composition indicators are derived from IPUMS USA data under IPUMS's data license; see IPUMS citation above