How Census Tract Child Poverty is Measured
Census tract child poverty rates are the ONLY methodology available for measuring neighborhood-level poverty. These estimates come from the American Community Survey (ACS) 5-Year data, published annually by the U.S. Census Bureau.
What is the ACS 5-Year Estimate?
- A rolling 5-year sample of approximately 3.5 million U.S. households per year
- Uses the Official Poverty Measure (OPM) -- cash income compared to poverty thresholds
- The OPM is the federal standard (not the Supplemental Poverty Measure / SPM)
- At the census tract level, sample sizes are small, producing wide confidence intervals
- Estimates represent a 5-year average, not a point-in-time snapshot
- Published with a margin of error representing a 90% confidence interval
OPM vs. SPM
- Official Poverty Measure (OPM): Used for ENOUGH. Based on pre-tax cash income vs. 1960s-era thresholds adjusted for inflation.
- Supplemental Poverty Measure (SPM): Not available at tract level. Accounts for government benefits, geographic cost of living.
- OPM is the only measure published by Census at the tract level
- It tends to overcount poverty in areas with high benefit receipt and undercount in high-cost areas
- For ENOUGH purposes, OPM is the sole available methodology
Key Limitation
ACS 5-Year estimates are the only nationally standardized, publicly available source for census tract-level poverty measurement. There is no alternative data source that provides this level of geographic granularity. Any tract-level poverty threshold must rely on these estimates and contend with their inherent statistical uncertainty.
The Problem: Statistical Noise in Small-Area Estimates
Distribution of Margins of Error Across Maryland Tracts
Year-Over-Year Poverty Rate Shifts (2023 to 2024 ACS)
The Scale of Uncertainty
Why This Matters: Threshold Volatility
Between the 2023 and 2024 ACS 5-Year releases, 189 census tracts experienced a child poverty rate shift of more than 10 percentage points. These are not genuine changes in community economic conditions over one year -- they are artifacts of the 5-year rolling sample methodology.
A community that was measured at 32% child poverty one year could appear at 22% the next -- not because families improved their economic situation, but because a different subset of households was sampled. Without the MOE provision, such a community would lose ENOUGH eligibility despite no real change in conditions.
The Proposed Methodology: Upper-Bound Approach
How it Works
Step 1
Check if the 2024 ACS point estimate exceeds 30%. If yes, the tract qualifies (Path 1).
Step 2
If the point estimate is below 30%, calculate the upper bound (estimate + margin of error). If above 30%, proceed to Step 3.
Step 3
Confirm the previous year (2023 ACS) point estimate also exceeded 30%. If yes, the tract qualifies via the MOE provision (Path 2).
The Dual-Year Safeguard
Why the Previous-Year Requirement?
Requiring the previous year's estimate to also exceed 30% ensures that the MOE provision captures tracts with genuine sustained poverty that appears to have improved due to sampling variability, rather than tracts that were never truly above the threshold. A tract that was at 35% last year and is now at 28% (+/- 8%) likely still has poverty above 30%. A tract that was at 15% last year and is now at 22% (+/- 12%) does not qualify, even though its upper bound exceeds 30%.
Impact of the Proposed Methodology
Tracts Qualifying by Criteria
Practical Effect
The MOE provision primarily serves as a stability mechanism. It prevents communities from losing ENOUGH eligibility due to statistical noise rather than genuine economic improvement. The dual-year requirement ensures this provision only applies to tracts with demonstrated, sustained high poverty.
Tracts Over 30% by Cycle (Demonstrating Volatility)
Note: The number of tracts above 30% fluctuates across ACS releases (175 to 167 to 180) even though statewide poverty has not changed dramatically. This illustrates the sampling noise that the MOE provision addresses.
The Case for Approval
Summary Argument for the Secretary
The proposed upper-bound methodology is a reasonable, evidence-based refinement that:
- Addresses a known data limitation. ACS tract-level estimates have high margins of error (~49% of tracts have MOE exceeding 10 percentage points). The Census Bureau itself publishes MOE specifically so users can assess estimate reliability.
- Prevents harm from statistical noise. 189 tracts shifted more than 10 percentage points between the 2023 and 2024 ACS releases. Communities should not lose access to anti-poverty programming because of sampling artifacts.
- Includes a safeguard against abuse. The dual-year requirement (previous year must also exceed 30%) ensures only tracts with demonstrated sustained poverty qualify. This is not an open-ended expansion.
- Aligns with statistical best practices. Using confidence interval bounds for threshold decisions is standard practice in federal statistics. The Census Bureau, HHS, and other agencies routinely recommend considering MOE when making determinations based on ACS estimates.
- Is conservative in application. Only 36 additional tracts qualify through the MOE provision beyond those already qualifying via the point estimate (216 total vs. 180 via Criteria 1 alone). This is a modest 20% expansion, not a dramatic change.
- Promotes program continuity. Communities already receiving ENOUGH funding should not have their eligibility stripped due to year-over-year sampling variation when poverty conditions have not genuinely improved.