The City Health Dashboard’s racial and ethnic diversity index is an entropy-based score from 0 to 100 that measures how evenly a city’s residents are spread across racial and ethnic groups. Because the score depends on how groups are distributed, not on any single group’s share, it gives a different picture than a simple non-white percentage. Higher scores mean residents are spread more evenly across groups, while lower scores mean one group dominates the population.
How is racial and ethnic diversity measured?
The index is calculated from American Community Survey 5-year estimates, specifically table DP05, using 2024 data. Each city receives one score, so cities can be compared on the same scale.
Reading the 0-to-100 scale
| Score | What it means |
|---|---|
| 0 | One group makes up the whole population |
| 100 | All groups are equally represented |
Scores between the two ends show how close a city is to an even split across groups. No score is inherently good or bad. A higher value indicates a more even mix, not a better place to live, so the score should be read alongside other local data.
What the probability means
Boston Indicators defines its diversity index as the probability that two randomly chosen residents belong to different groups. The City Health Dashboard’s entropy score is a related but different measure. A higher score means a higher chance that two people picked at random come from different groups. A score near 0 means almost any two residents are likely to share the same dominant group.
Non-white share versus diversity: San Antonio
A city’s non-white share does not determine its diversity score. San Antonio is 75% non-white, yet 64% of its residents are Hispanic or Latino. Because one group accounts for most of the non-white population, the non-white figure overstates how varied the population is. This is why the entropy score and a single-category percentage answer different questions.
What data do cities use to compare diversity?
Most published comparisons of U.S. city diversity rely on one public source: the American Community Survey (ACS), run by the U.S. Census Bureau. Boston Indicators (The Boston Foundation) also reports city-level diversity figures, including a long-run series for the 50 largest U.S. cities.
5-year estimates are the main input. Instead of relying on a single survey year, they combine responses collected over five years, which produces more stable figures for smaller areas.
Table DP05 is the demographic profile table that supplies the race and ethnicity variables. It covers the Census race categories and Hispanic or Latino origin, which are recorded as separate questions.
| Element | What it provides | What to check |
|---|---|---|
| ACS 5-year estimates | Multi-year survey base for the City Health Dashboard index | The period label of the release |
| Table DP05 | Race and Hispanic or Latino origin variables | Whether a source uses DP05 or a different table |
| “Other” category | Catch-all for groups the index merges | The group breakdown behind the score |
What do the available city scores show?
The fact sheet gives scores for only two cities, so this section can rank just those two against the 50-city benchmark. Oakland, CA has the highest score cited, at 77 on the Boston Indicators diversity index (a different formula from the City Health Dashboard entropy score). Boston scores 70.2. Neither figure comes with a full ranking of all US cities, so these two values are examples of published scores, not a ranking of the most diverse cities.
| Measure | Value |
|---|---|
| Oakland, CA diversity index score | 77 |
| Boston diversity index score | 70.2 |
| Boston score in 1980 | 48.7 |
| Average score, 50 largest US cities, 1980 | 45.7 |
| Average score, 50 largest US cities, 2017 | 61.4 |
Boston’s score rose by 21.5 points between 1980 and the latest figure cited (70.2 − 48.7). Over the same period the average for the 50 largest cities rose by 15.7 points between 1980 and 2017 (61.4 − 45.7). Because the year of Boston’s 70.2 figure is not stated, its 21.5-point rise since 1980 cannot be compared with the 15.7-point rise in the 50-city average between 1980 and 2017. These figures cover different periods, so they do not show whether Boston’s diversity rose faster than the typical large-city trend. The two figures use different end years, so the comparison is approximate.
When comparing cities on this index, check three things before drawing conclusions:
- Scale direction: a higher score means residents are spread more evenly across groups.
- Benchmark: a city’s score is only meaningful next to a reference point, such as the 50-city average.
- Year of the figure: trend claims require two dated values from the same measure.
Why do city rankings disagree?
Two publishers can rank the same cities differently and both be using accurate data. The disagreement usually comes from three choices: how residual groups are grouped, which year of data is used, and which formula turns population counts into a single score. Each choice changes the ordering.
| Factor | What it changes in the ranking |
|---|---|
| Grouping of residents | Groups placed in an “other” category are not scored separately, which changes the score for the city |
| Data year | Scores reflect the population at the time of the survey, so they describe different periods |
| Index formula | Scores from different formulas are not on the same scale and do not measure the same thing |
The “other” category. The City Health Dashboard index lumps some groups into an “other” category. Two cities with the same real mix of residents can therefore score differently depending on how many people fall into residual groups.
Data vintage. A ranking built on a different survey year, or on a different averaging window, will compare cities at different points in time. Long-run comparisons add another layer: the average diversity index of the 50 largest U.S. cities rose from 45.7 in 1980 to 61.4 in 2017, so a ranking from an earlier year can place cities differently from one published later.
Differing formulas. The City Health Dashboard uses an entropy score running from 0 to 100. Boston Indicators describes its index as the probability that two randomly chosen residents belong to different groups. These are related but not identical measures, so a city’s position can shift between publishers even when both use the same census source.
Does diversity mean the same as segregation?
No. A diversity index describes how mixed a city’s population is as a whole, while segregation describes how unevenly each group is spread across its neighborhoods. A city can score high on one and still be divided on the other.
| Question | Diversity index | Dissimilarity index |
|---|---|---|
| What it measures | Probability that two randomly chosen residents belong to different groups | Unevenness of each group’s spread across neighborhoods |
| Uses where residents live | No | Yes |
| Answers | How mixed the city is | How separated the groups are |
The City Health Dashboard diversity index ranges from 0 to 100, but it does not capture where groups live. It cannot show whether residents of different backgrounds share streets, schools, or blocks. Two cities with the same score can have very different neighborhood patterns.
A dissimilarity index is the standard way to measure segregation between groups. It compares each group’s distribution across neighborhoods with its distribution across the whole city. It is a separate calculation from the diversity score, so it should be used when the question concerns residential separation rather than population mix.
Mixed cities can still be highly segregated. Boston scored 70.2 on the diversity index, up from 48.7 in 1980, and Oakland, CA scored 77. Neither score says anything about how groups are distributed block by block.
City, metro or neighborhood: which scale to compare?
The diversity index describes whatever geographic unit it is calculated for, so the boundary you choose determines the result. The City Health Dashboard score runs from 0 to 100. A city score and a metro score for the same place can differ because they draw on different populations, so any comparison should name the unit explicitly.
Each scale answers a different question:
| Scale | Best used for | Main limit |
|---|---|---|
| City | Comparing municipalities, such as Oakland, CA (77) and Boston (70.2) | Follows municipal borders, which may not match where people live and work |
| Metro | Regional housing and labor-market context | Can average out large differences between the central city and its suburbs |
| Neighborhood | Local questions about a specific district | Small populations produce less stable estimates |
Small samples are the main constraint at neighborhood level. Five-year estimates pool survey responses over several years to stabilize figures for smaller areas, but a neighborhood with few residents still yields estimates with wider uncertainty. Treat small differences between neighborhoods as unreliable unless the gap is large.
To match scale to question:
- Use the city scale for comparisons between municipalities with similar administrative status.
- Use the metro scale when the question concerns the wider region people commute through.
- Use the neighborhood scale only for local questions, and only when the population is large enough to support the estimate.
How to compare two cities yourself
Use one source for both cities wherever possible. Mixing figures from different publications, or from different years, produces a comparison that looks precise but is not. The workflow below relies on two sources cited in this article: the City Health Dashboard (NYU Grossman School of Medicine) and Boston Indicators (The Boston Foundation).
1. Build a side-by-side input table
Record every figure in one table before drawing conclusions. Use one row per city and keep the source and period in the same row as the number.
| City | Figure as published | Source | Period given in the source |
|---|---|---|---|
| Oakland, CA | Diversity index: 77 | Boston Indicators (The Boston Foundation) | Not stated in the source sheet |
| Boston | Diversity index: 70.2 | Boston Indicators (The Boston Foundation) | Up from 48.7 in 1980 |
| San Antonio | 75% non-white; 64% Hispanic or Latino | Boston Indicators (The Boston Foundation) | Not stated in the source sheet |
Add a separate line recording the method: the formula used (entropy score or probability index), the grouping scheme and the table. Do not compare values from different formulas directly.
2. Verify each figure’s publication year
- Separate the data year (when the survey values were collected) from the publication year (when the report came out). They are not always the same.
- Compare only figures that share a data year. A 1980 value and a 2017 value describe different moments, even when both appear in one report.
- The Boston Indicators material gives the average index of the 50 largest U.S. cities as 45.7 in 1980 and 61.4 in 2017. Use those as paired values, not as a single current figure.
- If you cannot confirm the year of a figure, mark it as unverified in your table instead of guessing.
3. Cross-check against group share percentages
A diversity score is not a population share, so check each score against the underlying percentages. Do not add, subtract, or substitute one for the other.
Frequently asked questions
Does a higher diversity score mean a city is a better place to live?
No. The score shows only how even spread residents are across racial and ethnic groups, and no score is inherently good or bad. Read it alongside local data on housing, services and transport.
Why can’t I compare Oakland’s 77 with City Health Dashboard scores?
Oakland’s 77 and Boston’s 70.2 come from Boston Indicators, which uses a different formula, so the figures are not directly comparable. The source also does not state the year of either figure.
Which data sits behind the City Health Dashboard score?
The score is calculated from American Community Survey 5-year estimates, specifically table DP05, using 2024 data. The index also places some groups into an “other” category, so smaller groups may not appear separately.
