How Much of a City Income Ranking Is Noise?
Rank large US cities by median household income and almost every neighbor in the list turns out to be a statistical tie. What the survey can tell apart is far coarser than the ranking implies.
Across the 883 US cities of 50,000 or more with a published median household income, 875 of the 882 pairs that sit next to each other in the ranking (99%) cannot be told apart once the survey’s margin of error is counted. The typical city is statistically tied with 107 others.
Key findings
- Sammamish, WA ranks first at $239,690 (±$15,260), but cannot be told apart from cities ranked as low as #3.
- 780 of the 883 cities (88%) are tied with 50 or more others. Only 5 can be told apart from every other city in the group.
- For the median city, the cities it cannot be told apart from are spread across 160 places in the ranking.
- New Brunswick, NJ has the least certain position: ranked #651, it is tied with cities from #473 to #810.
The top of the ranking
| City | Rank | Median household income | Margin of error | Tied with | Cannot be told apart from ranks |
|---|---|---|---|---|---|
| Sammamish, WA | 1 | $239,690 | ±$15,260 | 2 | #1 to #3 |
| Cupertino, CA | 2 | $234,707 | ±$12,596 | 2 | #1 to #3 |
| Palo Alto, CA | 3 | $231,101 | ±$9,859 | 2 | #1 to #3 |
| Dublin, CA | 4 | $214,385 | ±$9,481 | 0 | #4 to #4 |
| San Ramon, CA | 5 | $196,161 | ±$11,535 | 5 | #5 to #10 |
| Bethesda, MD | 6 | $192,237 | ±$6,260 | 6 | #5 to #12 |
| Newton, MA | 7 | $190,304 | ±$8,593 | 9 | #5 to #14 |
| Pleasanton, CA | 8 | $190,124 | ±$9,119 | 9 | #5 to #14 |
| Mountain View, CA | 9 | $189,917 | ±$10,319 | 9 | #5 to #14 |
| Sunnyvale, CA | 10 | $186,170 | ±$5,402 | 9 | #5 to #14 |
| Fremont, CA | 11 | $181,506 | ±$3,676 | 7 | #7 to #14 |
| Hoboken, NJ | 12 | $180,579 | ±$10,044 | 9 | #6 to #16 |
| Santa Clara, CA | 13 | $178,958 | ±$8,034 | 8 | #7 to #16 |
| Milpitas, CA | 14 | $178,798 | ±$10,931 | 9 | #7 to #16 |
| Bellevue, WA | 15 | $165,576 | ±$7,893 | 14 | #14 to #28 |
Where a rank means least
| City | Rank | Median household income | Margin of error | Tied with | Cannot be told apart from ranks |
|---|---|---|---|---|---|
| New Brunswick, NJ | 651 | $65,810 | ±$10,751 | 318 | #473 to #810 |
| Berwyn, IL | 473 | $78,408 | ±$6,616 | 190 | #338 to #651 |
| Palm Desert, CA | 485 | $77,513 | ±$7,228 | 200 | #338 to #651 |
| Columbus, IN | 486 | $77,456 | ±$4,545 | 155 | #338 to #651 |
| Arden-Arcade, CA | 490 | $77,321 | ±$4,493 | 153 | #338 to #651 |
| Indio, CA | 491 | $77,167 | ±$6,265 | 183 | #338 to #651 |
| Highland, CA | 492 | $77,120 | ±$6,818 | 193 | #338 to #651 |
| Westchester, FL | 556 | $72,262 | ±$6,986 | 214 | #421 to #699 |
| St. George, UT | 494 | $76,508 | ±$4,339 | 149 | #376 to #651 |
| Hanford, CA | 495 | $76,461 | ±$5,174 | 165 | #376 to #651 |
The test is pairwise at the ACS’s 90% confidence level, with no correction for making this many comparisons, and it treats each city’s estimate as independent. Read the counts as a measure of how coarse the survey is, not as exact figures. A tie says the survey cannot separate two cities; it does not say their incomes are the same. Read a rank as a neighborhood rather than a position.
Methodology
Cities of 50,000 or more residents with a published median household income and margin of error in the American Community Survey 5-Year estimates, ranked from highest to lowest. Two cities are called statistically tied when the gap between their medians is smaller than the margin of error of that gap, computed the way the Census Bureau recommends: the square root of the sum of the two published margins, each squared. ACS margins are at the 90% confidence level, so a tie means the survey cannot distinguish the two at that level. It does not mean their true incomes are equal. Ties are counted pairwise and are not transitive, so A can be tied with B and B with C while A and C are not tied, and no adjustment is made for the number of comparisons. Cities whose median is top-coded at $250,000 or more are excluded, since a top-coded figure has no usable margin.
Sources
- U.S. Census Bureau — American Community Survey 5-Year Estimates (Table-Based Summary File) (2024)
Related
Every figure here is recalculated from the same measurements the profiles use. See the methodology for how the underlying data is collected and normalized, and the editorial policy for how these reports are written.