Skip to content

Limitations

Despite our best efforts to create a multifaceted view of neighborhood accessibility, this analysis has several metholodological and conceptual limitations that should be considered when interpreting the results.

Data Limitations

Data Quality and Completeness Issues

  • OpenStreetMap (OSM) Inconsistencies

    • OSM coverage varies substantially across neighborhoods.
    • Sidewalks, bike lanes, and amenities are incompletely mapped, especially in North and West Philadelphia.
    • Some features (e.g., childcare centers, social services, private clinics) may be absent or misclassified, affecting proximity calculations.

  • NDVI Resolution and Representativeness

    • NDVI was extracted from Landsat 8, which has a 30m resolution after atmospheric correction. At this resolution, block-level vegetation differences—especially in dense rowhome neighborhoods like Passyunk Square—are smoothed out.
    • Seasonal NDVI variation wasn’t modeled, making environmental quality time-bound to a particular imagery window.

  • Tree Canopy Data (PPR) Age

    • The 2015 tree canopy dataset used is nearly a decade old.
    • Significant canopy loss and regrowth since then are not captured, likely misrepresenting real conditions.

Spatial Scale and Aggregation Issues

  • Tract to Neighborhood Aggregation

    • Aggregating tract-level data to neighborhoods may obscure intra-neighborhood variability.
    • Some neighborhoods contain diverse tracts with varying accessibility profiles that are averaged out.
    • Walkshed or service areas might be more appropriate for certain analyses.
    • Mixed geographic untis (tracts vs. neighborhoods) complicate interpretation.

Methodological Simplifications

  • Nearest-Distance Approach

    • Using straight-line (Euclidean) distances to nearest amenities ignores real-world travel paths, barriers, and street network connectivity.
    • Should consider using network-based distances in future analyses.

  • Arbitrary Weighting of Components

    • The equal weighting of components in composite scores may not reflect their true importance to accessibility.
    • Future work could explore data-driven weighting schemes or stakeholder input.

  • Equity Considerations

    • The analysis does not explicitly account for socioeconomic or demographic disparities in accessibility.
    • Future iterations should integrate equity-focused metrics to better capture differential access across populations.

Other Accessibility Dimensions Missing

  • Affordability and Social Access

    • The analysis focuses on physical proximity but does not consider affordability, quality of services, or cost barries, which are critical to true accessibility.

Technial Limitations

  • No Sensitivity Analysis

    • No robustness checks or sensitivity analyses were conducted to assess weight choices, distance thresholds, and normalization choices (min-max vs. z-score).

Summary

Overall, results should be interpreted as approximate indicators, not definitive statements about Philadelphia’s accessibility landscape. The analysis provides a useful exploratory framework, but future work would benefit from:

 - Higher-quality and more complete datasets

 - Network-based travel time measures

 - Updated environmental and canopy data

 - Integration of theory from mobility justice and urban accessibility literature

 - Equity-centered modeling

 - Temporal accessibility measures

 - Sensitivity testing of index construction