Methodology
Data Source: American Community Survey
All district statistics come from the American Community Survey (ACS). The ACS is a large, high-quality nationwide survey conducted by the U.S. Census Bureau. In 2024, the response rate exceeded 80%, and the total sample included nearly 2 million households.
ACS 5-Year Estimates
We use the ACS’s 2024 5-Year Estimates, which draw from surveys conducted from 2020 through 2024. The 2024 5-Year Estimates are the latest 5-year data available; the 2025 5-Year Estimates should be available in early 2027.
Ideally, we would use more recent data rather than aggregating five years of survey responses, some of which are up to six years old. However, only 5-year estimates have a large enough sample to compute statistics at the local district level.1 On the city homepages, the four summary cards use ACS 1-year estimates instead, since citywide sample sizes are large enough to support the more current release.
Geographic Aggregation
Plain-English Summary
Local district statistics are not precomputed by the Census Bureau. It is also not possible to simply aggregate survey responses from people within a given district: the Census Bureau does not reveal respondents’ precise geographic locations. To calculate local district statistics, we aggregate data from smaller geographic areas that fall within or overlap a given district, using the most granular geographic information the Census Bureau releases. Some of these Census-defined areas do not perfectly align with district boundaries: part of the area may be inside the district, while another part is outside of it. When this occurs, we weight the data by the fraction of the area’s population that falls within the district boundaries. While this geographic mismatch is unavoidable, it has little impact on computed district-level statistics. For a subset of metrics where we can test the difference between perfect boundary matches and imperfect overlap, average errors are less than half of one percentage point.
Aggregating Across Census Geographies
Local district statistics are not precomputed by the Census Bureau. To calculate them, we take a weighted average of smaller geographic units, called census tracts, that fall within or overlap a district.
As a simple example, consider calculating the percentage of renter-occupied households in a district with four census tracts, each of which is fully within the district boundary:
To find the district-level renter percentage, we take the weighted average of the tracts’ renter share, weighting each tract by its total number of households:
District renter % = (popA × renter%A) + (popB × renter%B) + (popC × renter%C) + (popD × renter%D) popA + popB + popC + popD
= (1,500 × 40%) + (1,000 × 60%) + (2,000 × 30%) + (1,500 × 80%) 1,500 + 1,000 + 2,000 + 1,500 = 50.0%
Dealing with Imperfect Overlap Between Tracts and Districts
In practice, census tract boundaries do not always align with district boundaries: some tracts include residents from multiple districts. For split tracts, we use the data for the full tract, including residents living both inside and outside the district boundary. We account for the fact that some residents live outside the district by weighting the tract by the fraction of its population falling within the district boundaries, rather than using the tract’s total population.
For example, the district boundary below fully encompasses Tract A but only partially overlaps Tract B. When calculating the district’s renter share, we know that 35% of households in Tract A are renter-occupied and that 55% of households in Tract B are renter-occupied, but we do not know the renter share for the portion of Tract B inside the district. However, we do know how much of Tract B’s population is inside versus outside the district. We account for this by weighting Tract B’s renter share by population of Tract B inside the district (1,400 households), rather than its full population (2,400 households).
District renter % = (popA × renter%A) + (popB, in district × renter%B) popA + popB, in district
= (1,800 × 35%) + (1,400 × 55%) 1,800 + 1,400 = 43.8%
We calculate the share of each split census tract’s population within the district using block-level data from the 2020 decennial Census. Census blocks are very small and almost always fall entirely within a single district.2 The figure below shows how block-level population data can be used to weight split tracts, using the example from Tract B above.
Importantly, only a handful of basic variables, such as population, are available at the block level. While we have total population data at the block level from the 2020 decennial Census, renter share and most other ACS variables are available only at higher levels of aggregation. As a result, we must use the tract-wide renter percentage rather than the renter percentage for the portion of the tract within the district boundary.
Sources of Error and Imprecision
Sampling Error
The ACS surveys a sample of households rather than every household in the country. Because the ACS captures only a slice of the population, any statistic it produces is an estimate — and like all estimates, it comes with some uncertainty. If the ACS were reconducted with a new sample of households, it would yield slightly different results. This sample-to-sample variation is called sampling error, and it exists for all surveys.
A survey’s margin of error quantifies the uncertainty from sampling error. It defines a range around the published estimate within which the true population value is likely to fall. For example, if a district has an estimated renter-occupied household share of 42%, with a margin of error of ±4 percentage points, the true figure is likely somewhere between 38% and 46%.3 ACS margins of error are reported at the 90% confidence level, meaning that if the survey were repeated many times, the true value would fall within that range about 90% of the time. Each district page has a button that allows users to toggle margins of error on or off for each estimate. Technical details on how we compute district-level margins of error from ACS tract and block group data are described on this page.
Geographic Aggregation Error
When a tract overlaps a district boundary, our calculations necessarily incorporate some data from people living outside the district. This introduces geographic aggregation error — a source of imprecision distinct from sampling error. Unlike sampling error, geographic aggregation error is not directly quantifiable from the published ACS data.4 The degree to which a district is mismeasured by incorporating non-district residents depends on how different those non-district residents are from district residents, which is unknown.5
While we cannot directly test geographic aggregation error in the ACS estimates, we can test it using block-level decennial Census data. Since Census blocks are very small, we can use them to produce district-level estimates without geographic aggregation error. Using the same data, we can compute district-level estimates with geographic aggregation error by applying the same methodology we use on the ACS data.6 Comparing district estimates from block-level data aligned perfectly with district boundaries to the same data purposefully computed with geographic aggregation error allows us to quantify how much imperfect geographic overlap between Census and district boundaries can skew estimates.
For each district, we show the geographic mismatch between Census boundaries and district boundaries, as well as validation tests using decennial Census data with and without geographic aggregation error. Importantly, decennial Census data is only available for a small number of measures, including race, age, owner- versus renter-occupied households, and household types.7 However, geographic aggregation error in measures included in the decennial Census may differ from the error in other ACS measures. For instance, the portion of a split tract inside the district may have a similar renter share to the portion outside the district, a measure available in both ACS and decennial Census data, while rent prices may differ substantially across the same boundary, a measure available only in the ACS.
Self-Report Errors
ACS data come from self-reported responses, which are subject to error. The degree to which this matters varies by question. For instance, most people can accurately report their age, but when asked the year their building was constructed, respondents may give their best estimate rather than a precise answer. More broadly, the accuracy of survey data depends on the reliability of self-reported responses. For questions like building age, answers should be understood as rough approximations rather than precise measurements. We try to account for this in how we present the data. For instance, our housing age variable uses wide bins that convey the rough age distribution of the housing stock, such as relatively new, middle-aged, and older buildings, rather than precise construction dates.
Median and Percentile Calculations Are Approximations
When calculating medians and percentiles, such as median household income and 25th percentile home values, we average tract-level values within a district rather than computing them directly across the full district, which is not possible from published ACS data. Averaging tract-level medians and percentiles is an approximation of the true district-wide values. To calculate the true median or percentiles, we would need either a Census geography that directly matched district boundaries or the underlying microdata with address-level geographic identifiers. This simplification will increase the influence of extreme values. For instance, if one tract within a district has a very high median income, averaging across tracts gives that high-income tract more influence than it would have if the median were computed directly. As a result, the reported median income would be higher than the district’s true median income.
Footnotes
The Census does not release precise geographic information from the American Community Survey to protect the privacy of respondents. To calculate district-level statistics, we have to aggregate data across the geographies that the Census Bureau does release. The smallest published geographic units typically include only 1,200 to 8,000 people, and there are not enough survey respondents within those units in a single year. The 5-year estimates combine five years of survey responses to provide more stable and reliable estimates for smaller geographies.↩︎
In the rare cases where a block straddles a district boundary, we weight the block’s population by the share of the block’s area falling within the district.↩︎
Not all values within this range are equally plausible. The published estimate is the single most likely value; plausibility decreases gradually toward the edges of the range, following a bell-curve distribution.↩︎
Arguably, sampling error also has some unquantifiable bias. Traditional margins of error are calculated based on an assumption of a random sample. In practice, not everyone who receives the ACS answers the survey, and people who respond may differ from people who do not respond. This breaks the logic of random sampling that margins of error are based on and makes the degree of total sampling error also fundamentally unknowable. This helps explain how political polls are wrong by margins well exceeding their published margin of error. However, the ACS is a very high-quality survey, with high response rates and methods designed to reduce sampling bias.↩︎
Population shifts within split tracts since the 2020 Census (which is the data source we use to weight split tracts by the fraction of the population inside the district) may also be a source of error.↩︎
Specifically, we aggregate data from Census blocks up to the tract and block-group levels, then use those data the same way we use ACS data: weighting units that split district boundaries by their block-level population.↩︎
The ACS is far more comprehensive and surveys people more frequently than once every 10 years, which is why we use it despite the presence of geographic aggregation error.↩︎