NYC Subway Accessibility, Mapped: Gallery and Engine Cross-Walk for the subway-access Study
Appendix artifact

cross_walk.json

Download
artifacts/cross_walk.json·4.9 KB·json
{
  "n_rows": 18,
  "rows": [
    {
      "upstream_section": "§3.1 Data sources",
      "topic": "MTA stations + ACS tracts + elevator uptime + tract geometries",
      "ff_engine": "-",
      "portfolio_topic": "Data provenance sidecars"
    },
    {
      "upstream_section": "§3.2 Accessibility model",
      "topic": "800 m Euclidean catchment on tract centroids",
      "ff_engine": "nyc_geo_toolkit.catchment (spatial extra)",
      "portfolio_topic": "Walking-distance proxies and the network-distance upgrade path"
    },
    {
      "upstream_section": "§3.3 Need and gap scores",
      "topic": "Equal-weight composite of disability, senior, poverty rates",
      "ff_engine": "factor_factory.engines.inequality (composite indices)",
      "portfolio_topic": "Composite need indices and sensitivity to weighting"
    },
    {
      "upstream_section": "§3.4 Reliability-weighted coverage",
      "topic": "Discount nominal coverage by observed uptime fraction",
      "ff_engine": "factor_factory.tidy.Panel (reliability-weighted outcomes)",
      "portfolio_topic": "De jure vs. de facto service delivery"
    },
    {
      "upstream_section": "§3.5 DiD specification",
      "topic": "Twoway fixed-effects panel with tract + period effects",
      "ff_engine": "factor_factory.engines.did.twfe",
      "portfolio_topic": "Staggered rollout treatment effects"
    },
    {
      "upstream_section": "§3.5 DiD — staggered adoption robustness",
      "topic": "CS / SA / BJS heterogeneity-robust DiD estimators",
      "ff_engine": "factor_factory.engines.did.{callaway_santanna,sun_abraham,borusyak_jaravel_spiess}",
      "portfolio_topic": "Cross-estimator agreement as identification check"
    },
    {
      "upstream_section": "§3.5 SAR panel",
      "topic": "Spatial autoregressive extension of the DiD panel",
      "ff_engine": "factor_factory.engines.spatial.spatial_lag",
      "portfolio_topic": "Spatial spillovers and SUTVA violations"
    },
    {
      "upstream_section": "§3.6 Spatial analysis (weights matrix)",
      "topic": "Row-standardized 2 km distance-based weights over 2,317 tracts",
      "ff_engine": "factor_factory.engines.spatial._base (weights)",
      "portfolio_topic": "Spatial weights construction and sensitivity"
    },
    {
      "upstream_section": "§4.1 System-wide coverage",
      "topic": "Descriptive coverage counts, gap populations",
      "ff_engine": "-",
      "portfolio_topic": "Headline-number communication"
    },
    {
      "upstream_section": "§4.2 Borough disparities",
      "topic": "Stratified coverage + mean-distance-to-nearest by borough",
      "ff_engine": "factor_factory.engines.inequality.theil (between-borough decomposition)",
      "portfolio_topic": "Between-group vs. within-group inequality"
    },
    {
      "upstream_section": "§4.3 Reliability analysis",
      "topic": "Station uptime ranking; fragile-station surfacing",
      "ff_engine": "factor_factory.engines.changepoint.ruptures_adapter (uptime regime shifts)",
      "portfolio_topic": "Operational reliability as outcome"
    },
    {
      "upstream_section": "§4.4 Temporal progression",
      "topic": "Coverage share across 7 panel periods (2017-2023)",
      "ff_engine": "factor_factory.engines.stl.sktime_stl",
      "portfolio_topic": "Rollout curves and pre-trend inspection"
    },
    {
      "upstream_section": "§4.5 Treatment-control balance",
      "topic": "Welch's t-tests on pre-treatment covariates",
      "ff_engine": "factor_factory.engines.did._base (balance diagnostics)",
      "portfolio_topic": "Identifying assumptions — parallel trends, no anticipation"
    },
    {
      "upstream_section": "§4.6 Model diagnostics",
      "topic": "Jarque-Bera, skewness, kurtosis, distance-decay plots",
      "ff_engine": "-",
      "portfolio_topic": "Aggressive diagnostics"
    },
    {
      "upstream_section": "§4.7 OLS equity regression",
      "topic": "Gap score ~ disability + senior + poverty (HC1 SE, VIF)",
      "ff_engine": "factor_factory.engines.panel_reg.pyfixest_adapter",
      "portfolio_topic": "Vertical equity quantified"
    },
    {
      "upstream_section": "§4.8 Moran's I",
      "topic": "Global spatial autocorrelation on gap, need, disability",
      "ff_engine": "factor_factory.engines.spatial.morans_i",
      "portfolio_topic": "Spatial autocorrelation as clustering evidence"
    },
    {
      "upstream_section": "§5 Discussion",
      "topic": "Limitations, policy implications, future research",
      "ff_engine": "-",
      "portfolio_topic": "Honest limitations"
    },
    {
      "upstream_section": "Appendix D — engine audit",
      "topic": "factor-factory engine cross-check of primary results",
      "ff_engine": "factor_factory.engines.{panel_reg,spatial,did}",
      "portfolio_topic": "Engine-audit-as-appendix pattern (directional agreement as check)"
    }
  ]
}