If transit-based food access varies systematically by neighborhood demographics, that matters for distributional analysis. The federal executive-policy context is historical, not current: Executive Order 12898 directed agencies to identify and address certain disproportionate environmental effects beginning in 1994, and Executive Order 13985 established a broader federal equity policy in 2021. Executive Order 13985 was revoked on January 20, 2025, and Executive Order 12898 was revoked on January 21, 2025.1 Transportation policies have historically contributed to racial and social inequities, from highway construction that destroyed minority neighborhoods to transit service that failed to meet community needs.2
The article compares majority-minority and majority-white neighborhoods across a reported 9,039 residential census tracts.3 That population conflicts with a related statewide page.
The article reports shorter grocery distances but higher mobility-desert rates in majority-minority tracts. The related page reports lower mobility-desert rates, so the substantive direction is unresolved.
The tables below preserve the article-reported values for reconciliation. They should not be treated as verified comparisons.
This article reports a distance-transit paradox, but a related statewide page reports the opposite mobility-desert direction for similar stated inputs. The population, rates, and adjusted estimate are under reconciliation.
The Distance-Transit Paradox
The article's core comparison is preserved below as an article-reported claim under reconciliation. The removed chart did not depict the variables its caption claimed.
Research in other urban areas has documented similar patterns. The loss of public transportation service in food desert areas led to decreases in grocery store trips among below-median income households, with corresponding increases in trips to less healthy food sources like dollar stores.4
The Descriptive Pattern
The article reports the comparisons below, but the mobility-desert direction conflicts with a related page. The table is retained for reconciliation, not as an established pattern.
| Metric | Majority-Minority | Majority-White |
|---|---|---|
| Tracts | 4,156 | 4,883 |
| Mean distance to grocery | 0.72 mi | 0.85 mi |
| Median distance to grocery | 0.48 mi | 0.62 mi |
| Traditional food desert rate | 11.8% | 16.2% |
| Mobility desert rate | 13.2% | 11.1% |
| Mean vulnerability index | 0.352 | 0.291 |
What This Shows
Article-reported distance comparison: The article reports closer stores and lower traditional food-desert rates in majority-minority tracts. The underlying tract population is under reconciliation.
Conflicting transit comparison: This page reports more mobility deserts in majority-minority tracts, while a related page reports fewer. No interpretation is supported until the populations and outputs are reconciled.
Article-reported vulnerability comparison: The article reports a 0.06-point index difference. Its meaning depends on the unresolved population and index construction.
Why the Pattern Exists
The following are hypotheses that could be tested if the article-reported disparity survives reconciliation:
Factor 1: Development Patterns
Majority-minority neighborhoods in California are often located in older urban cores or inner suburbs developed before car-centric planning dominated. Historical housing discrimination, including redlining practices from the 1930s-1960s, concentrated minority populations in specific urban areas with higher-density development patterns. Research across 102 U.S. urban areas documented how discriminatory housing policies shaped current food environments, with effects persisting decades after the policies ended.5 These areas have:
- Higher retail density (more stores per square mile)
- Smaller lot sizes and more mixed-use zoning
- Transit infrastructure built for an earlier era
Density and changing transit networks are candidate explanations, not findings established by this page.
Factor 2: Income Correlation
The article reports lower median incomes in its majority-minority tract group. Those figures inherit the unresolved population:
| Characteristic | Majority-Minority | Majority-White |
|---|---|---|
| Median household income | $72,400 | $98,300 |
| Poverty rate | 16.8% | 10.2% |
| Car ownership rate | 87.2% | 92.8% |
These patterns reflect broader national trends. Research has documented that people who are lower-income, non-White, older, or disabled have increased likelihood of not owning a vehicle and of being dependent on public transit, with food-insecure households facing particularly acute challenges accessing healthy, affordable groceries via transit.6
Income, vehicle access, and the vulnerability index are potential confounders. This page does not currently reproduce how much they explain because the outcome direction is unresolved.
Factor 3: Transit Investment Patterns
Historical transit investment has not been uniform across neighborhoods. Research consistently documents that transit service quality correlates with neighborhood demographics, with lower-income and minority neighborhoods receiving less frequent service.2 Analysis of transit systems across U.S. cities found that neighborhoods most dependent on public transportation—those in minority areas—received the lowest level of service based on transit coverage and frequency.7 A recent national survey found that approximately half of U.S. municipalities have public transit planning, with opportunities to improve food access through transit equity interventions, particularly in underserved communities.8
This page cannot determine whether transit investment patterns differ by neighborhood demographics in California. The conflicting mobility-desert direction also prevents using the article's rate difference as supporting evidence.
Methodological Considerations: Correlation vs. Causation
Before examining regression results, we must address a critical limitation: racial composition could be endogenous to transit access.
The Residential Sorting Problem
Neighborhoods are not randomly assigned racial compositions. People choose where to live based on housing affordability, proximity to jobs, school quality, and transit access. If transit-dependent households disproportionately select into neighborhoods with better transit, and those households are disproportionately minority, this sorting could produce better measured transit access in majority-minority neighborhoods even without differential agency treatment. Other sorting mechanisms could move the association in the opposite direction.
The Reverse Causality Problem
Causation could run backward: poor transit makes a neighborhood less attractive, leading to demographic change; good transit attracts higher-income residents, leading to gentrification and demographic change. Both directions could operate simultaneously.
The Omitted Variables Problem
Historical factors affect both current racial composition and current transit infrastructure. Redlining (1930s-1960s) shaped residential segregation and infrastructure investment. Highway construction destroyed minority neighborhoods and created transit barriers. Zoning policies enforced segregation and influenced transit-oriented development.
These historical processes make it impossible to separate the effect of racial composition from the effect of historical discrimination using current cross-sectional data alone.
Implications for Interpretation
The regression table that follows preserves an article-reported association for reconciliation. The conflicting population and direction mean it should not be used to describe the statewide pattern or a causal effect.
Transit subsidies, fare assistance, and demand-responsive services remain policy options to test. This unreconciled comparison cannot rank locations or predict their effects.
Accounting for Income and Density Differences
The article reports estimating the following model to examine whether the mobility-desert difference persists after accounting for income and density:
Mobility_Desert = β₀ + β₁(Majority_Minority) + β₂(Income) + β₃(Density) + ε
Important Caveat: This regression estimates associations, not causal effects. We cannot rule out that other factors—historical discrimination, political representation, geographic constraints, residential sorting—drive both racial composition and transit access.
Results
| Variable | Coefficient | Clustered SE | Article-reported p-value |
|---|---|---|---|
| Majority-Minority (dummy) | +0.018 | 0.007 | Withdrawn |
| Median Income (per $10K) | -0.008 | 0.002 | <0.01 |
| Population Density (per 1000) | -0.003 | 0.001 | <0.01 |
Article-reported estimate under reconciliation: This page preserves a coefficient of +0.018 and a clustered standard error of 0.007. Their ratio is about 2.57, which is incompatible with the p-value previously displayed here under conventional normal or t inference. Because no matching model output documents the inferential calculation, the p-value is withdrawn. The coefficient and standard error remain visible only for reconciliation, alongside the opposite descriptive direction on a related page.
Interpretation limit: We cannot interpret the 1.8 percentage point coefficient as the causal effect of racial composition on transit access. A causal interpretation would require a design with a defensible source of exogenous variation plus supported identifying assumptions. Panel data, an instrument, or a policy change is not sufficient by itself.
Article-reported relative comparison: The article converts 1.8 percentage points to approximately 16% of its stated baseline. That calculation inherits the unresolved population and direction.
The article reports that income and density controls reduce its raw difference. No matching public output currently reproduces that result.
Note on clustered standard errors: The article reports clustering standard errors at the county level (58 clusters) rather than treating each tract as independent.9 Tracts within the same county may share transit systems, policy environments, and geographic constraints. With a limited number of clusters, asymptotic cluster-robust inference may be unreliable. No public output currently verifies the reported standard errors or their relationship to naive tract-level estimates.
Regional Variation
The pattern is not uniform across California:
| Region | Majority-Minority Mobility Desert Rate | Majority-White Mobility Desert Rate | Difference |
|---|---|---|---|
| Bay Area | 8.4% | 6.2% | +2.2 pp |
| Los Angeles | 14.8% | 12.3% | +2.5 pp |
| Central Valley | 18.6% | 15.8% | +2.8 pp |
| San Diego | 11.2% | 9.8% | +1.4 pp |
| Other | 15.4% | 13.1% | +2.3 pp |
The article reports regional differences from +1.4 to +2.8 percentage points and the same direction in each region. Those comparisons inherit the unresolved statewide population and outcome direction and are not verified robustness evidence.
What This Analysis Cannot Determine
Several important questions remain beyond the scope of this descriptive analysis:
Causation
The correlation between racial composition and mobility deserts does not establish that race causes worse transit access. Multiple pathways could produce this pattern:
- Historical discrimination in transit planning
- Income differences that correlate with both race and transit
- Residential sorting based on preferences and constraints
- Land use patterns established before current residents arrived
Distinguishing these mechanisms requires evidence matched to each mechanism. Descriptive panels or qualitative planning records can document timing and process; a causal estimate additionally requires a supported identification strategy.
Individual Experience
Census tract analysis captures neighborhood-level patterns. It does not measure:
- How individual residents actually travel to grocery stores
- Whether transit-dependent residents have adapted (carpooling, delivery, shopping elsewhere)
- Quality of available transit service beyond stop proximity
- Variation within tracts
Alternative Explanations
If the article-reported 1.8 percentage point estimate survives reconciliation, possible explanations to test include:
- Omitted variables (geographic constraints, political factors, transit agency boundaries)
- Measurement error in income or density controls
- Non-linear relationships not captured by the linear model
- Actual disparities in transit service provision
Statistical Limitations
Limited clusters: With 58 county clusters, our clustered standard errors rely on asymptotic approximations that may not hold. P-values should be interpreted cautiously.
Cross-sectional design: We observe each tract at a single point in time, preventing analysis of how changes in demographics relate to changes in transit access. Panel data could document changes over time, but would not establish a causal effect without a supported identification strategy.
Limitations
Majority-Minority as Binary
The analysis treats majority-minority as a binary variable, but demographic composition is continuous. Tracts that are 51% minority and 49% minority are treated as different categories despite near-identical composition.
The article reports continuous and alternative-threshold estimates as robustness checks. No matching public output reproduces them, and they inherit the unresolved population and outcome direction. They therefore cannot establish a consistent or monotonic relationship.
Finer-grained analysis using specific racial/ethnic group breakdowns (Hispanic, Black, Asian populations separately) might reveal whether the pattern varies by group, though sample sizes for some groups limit statistical power at the tract level.
ACS Data Limitations
Race/ethnicity data comes from ACS 5-year estimates, which have sampling error particularly for smaller tracts. The majority-minority classification may misclassify some tracts near the 50% threshold.
Single Time Point
Demographic composition changes over time (gentrification, demographic shifts, migration). The classification reflects 2019-2023 composition, which may not match current or historical patterns.
Ecological Fallacy
Tract-level analysis does not capture individual-level relationships. A majority-minority tract where white residents have poor transit access and minority residents have good access would show as a majority-minority tract with average access.
Data and Methods
Material status: Article only. No matching public package currently derives the tract population, tables, model estimates, or regional comparisons on this page.
Data sources:
- Race/ethnicity: ACS 2019-2023, Table B03002
- Transit stops: Cal-ITP statewide GTFS
- Grocery distances: Calculated from population-weighted centroids
- Vulnerability index: Composite of food access, poverty, renter status, minority percentage, sprawl
Classification:
- Majority-minority: > 50% non-Hispanic non-white
- Mobility desert: Grocery within 1 mile, transit stop > 0.5 miles or < 2 stops within 0.5 miles
Statistical analysis:
- Descriptive comparison of means
- Linear probability model with income and density controls
- Regional stratification
Software: The article reports using Python 3.11 and statsmodels for regression estimation and clustered standard errors.
Notes
[1] Executive Order 12898, Federal Actions To Address Environmental Justice in Minority Populations and Low-Income Populations (February 11, 1994); Executive Order 13985, Advancing Racial Equity and Support for Underserved Communities Through the Federal Government (January 20, 2021); Executive Order 14148, Initial Rescissions of Harmful Executive Orders and Actions (January 20, 2025), which revoked Executive Order 13985; and Executive Order 14173, Ending Illegal Discrimination and Restoring Merit-Based Opportunity (January 21, 2025), which revoked Executive Order 12898. ↩
[2] Bullard, R. D., & Johnson, G. S. (1997). Just Transportation: Dismantling Race and Class Barriers to Mobility. New Society Publishers. Taylor, B. D., & Morris, E. A. (2015). Public transportation objectives and rider demographics: Are transit's priorities poor public policy? Transportation, 42(2), 347-367. https://doi.org/10.1007/s11116-014-9547-0 ↩
[3] Majority-minority defined as census tracts where non-Hispanic white population is less than 50% of total population. Based on ACS 2019-2023 Table B03002. ↩
[4] Arnold, S. (2025). Public transportation access and food insecurity. Journal of Regional Science, 65, 1449-1467. https://doi.org/10.1111/jors.70010 ↩
[5] Li, M., & Yuan, F. (2022). Historical redlining and food environments: A study of 102 urban areas in the United States. Health & Place, 75, 102775. https://doi.org/10.1016/j.healthplace.2022.102775 ↩
[6] Bayly, R., Pustz, J., Stopka, T. J., Metzger, J., & Waters, M. C. (2025). Transit bus access to healthy, affordable food: A novel geographic information system (GIS) and community-informed analysis. SSM - Population Health, 29, 101753. https://doi.org/10.1016/j.ssmph.2025.101753 ↩
[7] Powder, J. (2020). For Blacks and other minorities, transportation inequities often keep opportunities out of reach. Hopkins Bloomberg Public Health Magazine. https://magazine.publichealth.jhu.edu/2020/blacks-and-other-minorities-transportation-inequities-often-keep-opportunities-out-reach ↩
[8] Smarsh, B. L., Park, Y. S., Lee, S. H., Harris, D. M., & Blanck, H. M. (2025). Public transit supports for food access: 2021 National Survey of Community-Based Policy and Environmental Supports for Healthy Eating and Active Living (CBS HEAL). Preventing Chronic Disease, 22, 240458. https://doi.org/10.5888/pcd22.240458 ↩
[9] Cameron, A. C., & Miller, D. L. (2015). A practitioner's guide to cluster-robust inference. Journal of Human Resources, 50(2), 317-372. https://doi.org/10.3368/jhr.50.2.317 ↩
Tags: #FoodSecurity #Equity #Race #TransitAccess #Demographics #California #Disparities #PublicPolicy
Next in this series: Transit Data Quality Under Reconciliation.
Frequently Asked Questions
Do majority-minority neighborhoods have better or worse food access in California?
This article reports closer stores but higher mobility-desert rates in majority-minority tracts. A related page reports the opposite mobility-desert direction for similar stated vintages, so the result is under reconciliation.
How large is the racial disparity in mobility desert rates?
The article reports a 1.8 percentage point higher rate after controls. That estimate is under reconciliation with a related page that reports an 8.2 percentage point lower rate.
Which California region shows the largest racial disparity in mobility deserts?
The article reports regional disparities from 1.4 to 2.8 percentage points. Those comparisons inherit the unresolved statewide population and direction.
How does income explain the racial gap in transit access?
The article reports income differences and a 1.8 percentage point adjusted gap. The model population and direction are under reconciliation.
How to Cite This Research
Cholette, V. (2025, December 14). Who gets left behind: Transit access and race in California. Too Early To Say. https://tooearlytosay.com/research/transit-equity/transit-equity-race/Copy citation