Start here
The short version
Oakmont isn't its own city or town in the eyes of the U.S. Census Bureau, so there's no ready-made "Oakmont" in the data. To draw the community accurately, we trace it from 75 individual census blocks that together match Oakmont's boundary.
From there, two different government datasets fill in the picture, and they don't work the same way. One is an exact head-count. The other is a survey that produces estimates. Things like how many people live here, how old they are, and who owns versus rents come from the exact count. Other things, like income, education, and home values, only exist as survey estimates for a larger area.
That's the whole story in one line: exact counts where we can get them, careful estimates where we can't. The rest of this page explains why, in as much or as little detail as you want.
The two sources
Two kinds of Census data
Almost every number here comes from one of two U.S. Census Bureau programs. Knowing which is which explains everything else.
The Decennial Census
A head-count · counts only · reaches the block
A 100% head-count taken every ten years (2020 is the latest). Because it counts everyone rather than sampling, it carries no sampling error, and it's published all the way down to a single city block. But it asks only a handful of questions, so it gives us counts: population, age, sex, race, Hispanic origin, and whether homes are owned, rented, or vacant.
It isn't flawless. The Bureau adds a small amount of deliberate statistical noise to block-level figures to protect individuals, so a single block can be off by a few people — more on that below.
The American Community Survey
Estimated · rich detail · never reaches the block
A rolling survey sent to a sample of households, published as five-year averages. Because it's a sample, its numbers are estimates with a margin of error, and they are never published for an area as small as a block. But it asks far more: income, education, home value, marital status, and where people were born. It's where all the richer detail comes from.
The trade-off, in a phrase: exactness versus richness, and geography versus detail. The next section is the geography half.
Where data lives
The geography ladder
Census data is published in nested pieces, from the whole country down to a single block. The smaller the piece, the fewer datasets reach it.
The head-count reaches the block, so Oakmont's 75 blocks can be traced exactly. The survey never does — no ACS figure is published for a block, at any margin of error.
The survey does reach one step below the tract, to the block group, for some of its tables. That sounds like it should help, and for Oakmont it doesn't. Oakmont's blocks are spread across all six block groups in the two tracts, and only three of the six lie wholly inside Oakmont. Those three hold about half its residents; the other three bring in roughly 1,250 people who live outside the community. Using block groups would mean covering half of Oakmont, or importing nearly as many outsiders as the tracts do — and the tables this report leans on most, like education by age and marital status by age, aren't among those published that small.
So the tracts it is. Oakmont sits inside Census Tracts 1516.01 and 1516.02, which also take in non-Oakmont residents along the edges.
Both maps are at the same scale. The tracts (left) are the area the survey figures cover, and they reach past Oakmont; the 75 Decennial blocks (right) trace the community itself. That gap is exactly why counts use the blocks and estimates use the tracts.
There's one more hurdle. Oakmont sits inside the City of Santa Rosa and has no Census "place" code of its own — those go to cities and towns, not to neighbourhoods within them — and there's no machine-readable map of the Oakmont Village Association boundary to hand the Census API. So the exact view isn't a built-in geography; it's a hand-picked list of the 75 blocks that best trace the community's edge.
A common question
Two places inside the line
Two parts of Oakmont come up often enough to deserve a straight answer: Wild Oak, which isn't age-restricted, and Oakmont Gardens, the senior-living community on White Oak Drive. Both sit inside the boundary these figures count. Neither can be separated out — and it turns out that matters far less than you'd expect.
Why they can't be separated
The Decennial count reaches the block, and that's the end of the line — the Census publishes nothing smaller. Both of these areas are smaller than the block they sit in, and they share that block with ordinary Oakmont homes. Removing the block would take the neighbours with it.
Wild Oak is 41 addresses in a block that holds 263 — dropping it would delete 222 Oakmont homes to remove 41. Oakmont Gardens is 169 addresses in a block of 281, alongside 112 other homes. Splitting a block by hand would mean inventing the split, which is exactly the shortcut this project avoids.
Wild Oak
At roughly one home in eighty, Wild Oak cannot move a community-wide figure. Even in the extreme case where every Wild Oak resident were under 55, setting them aside would shift Oakmont's 55-and-over share from 92% to 93.1% — about a point, and only under an assumption we know to be false. The real effect is smaller than that.
Oakmont Gardens
Oakmont Gardens is the only multi-unit property of any size in the community — the next largest address holds six homes. Because its residents are overwhelmingly older, setting them aside would nudge the age figures down, not up, and only by a fraction of a point either way.
Tenure is the one place it genuinely matters. Every home at Oakmont Gardens is rented, which makes it roughly 24% of all 707 rented homes in Oakmont. Set it aside and the owner-occupied share rises from 77.9% to about 82.2%. That is the single largest effect either area has on any published number, and it is worth knowing when you read the ownership figures.
How these were measured
Census data can't size either area, so this uses Sonoma County's public address-point file, counting the addresses on Wild Oak Drive and Court, and the unit addresses at 301 White Oak Drive. As a check on the method, those address points total 3,524 across the footprint against 3,427 Census housing units — close enough agreement to trust the shares. The population figures are ranges rather than single numbers, because the split of a block's residents between an area and its neighbours isn't published; the range spans two reasonable ways of estimating it.
Worth keeping in proportion: every income, education and home-value figure on this site is a tract estimate, and the tracts already reach past Oakmont to take in an estimated 1,000 non-Oakmont residents along Los Alamos Road, Bennett Valley and the Highway 12 corridor. That is many times either area discussed here, and no change to the block list would affect it.
Putting it together
Which number comes from where
Here is exactly which dataset and geography each part of the Community Report draws on, and why.
| What you see | Dataset | Geography | Why |
|---|---|---|---|
| Population, age & sex | 2020 Decennial (DHC) | Exact 75 blocks | 100% head-count, published at the block |
| Race & Hispanic origin | 2020 Decennial (DHC) | Exact 75 blocks | 100% head-count, published at the block |
| Owner / renter, occupancy | 2020 Decennial (DHC) | Exact 75 blocks | 100% head-count, published at the block |
| Household & per-capita income | 2020 ACS 5-year | Tracts 1516.01 + 1516.02 | Survey estimate, never published as small as a block |
| Education | 2020 ACS 5-year | Tracts 1516.01 + 1516.02 | Survey estimate, never published as small as a block |
| Home value | 2020 ACS 5-year | Tracts 1516.01 + 1516.02 | Survey estimate, never published as small as a block |
| Marital status, place of birth | 2020 ACS 5-year | Tracts 1516.01 + 1516.02 | Survey estimate, never published as small as a block |
This is why a single report mixes two geographies: the exact 75-block count for anything the Decennial measures, and the two-tract ACS estimate for anything only the survey measures. Every figure on the report is tagged with its source for the same reason.
Reading the report
Who each figure counts
Every section of the Community Report is tagged with the population it covers. They aren't all the same, because the Census doesn't publish them all the same way. Here is what each tag means and why it is what it is.
| Report section | Counts | Why that group |
|---|---|---|
| Who are we, Age & gender, Race | Everyone living in Oakmont | The Decennial counts every resident of the blocks — including younger spouses, family and caregivers. The age bars start at 55; the under-55 residents are in the table, not omitted. |
| Marital status | Residents 55 and over | Table B12002 crosses marital status with age at a real 55 line, so this matches who actually lives here. The all-ages figures are kept underneath. |
| Where residents were born | Residents 55+, then all residents | Table B06001 has an age cut but only four categories; B05002 has the regional detail but no age. No table has both, so both are shown rather than losing either. |
| Education | Residents 45 and over | Attainment is published with age breaks at 45 and 65 — there is no 55 line. 45+ is the closest available cut, and it drops most of the younger non-Oakmont fringe. |
| Income, sources of income, owners vs. renters, households | Households in the two tracts | No income table is crossed with age at any useful cut, and the survey is never published as small as a block. These figures cover more than Oakmont; see below. |
| Home value, and rent | Owner-occupied homes for value; rented homes for rent | Only owners are asked what their home is worth, so the value figure covers the 74% of occupied homes that are owner-occupied. The rent figure beside it covers rented homes — a different set of homes in the same section. |
| Work and hardship | People in the two tracts | Unemployment counts people in the labour force, not households; the poverty rate counts people the Census can assess. Both are person-level, and both cover the wider tract area. |
Why the income figures cover more than Oakmont
Income, education, home value and household size come from the survey, which is never published as small as a block. Oakmont's two tracts reach past the community to take in an estimated 1,000 non-Oakmont residents — the rural Highway 12 corridor, Los Alamos Road, Bennett Valley and the Bennett Ridge hillsides. Those households are younger and differently housed, so tract-based shares read slightly lower than an Oakmont-only count would. There is no way to correct this: the survey is not published for individual blocks, and the block groups it does reach cut across Oakmont’s edge rather than following it.
Household or family?
A household is everyone living in one home, including a person living alone. A family is a householder plus relatives. The family median runs highest; the household median sits below it because it also counts Oakmont's many one-person homes. Neither is wrong — they answer different questions.
What counts as income, in short
The full definition, with the Census Bureau’s own wording, is further down this page.
The Census measures money income: cash a household receives, counted before taxes. It does include Social Security, SSI, pensions and regular retirement-account withdrawals (IRA, Roth IRA, 401(k)), VA payments, and interest and dividends — whether or not any of it is taxable. It does not include proceeds from selling a home, withdrawals from savings, money borrowed against a home (reverse mortgage, HELOC), inheritances, life insurance payouts, or one-time gifts.
That gap matters in a retirement community: a household living comfortably on savings and home equity can still report a modest income here.
Why amounts sometimes read "not published" or "not disclosed"
The sources-of-income table shows an average per receiving household, because the Census publishes no median by income source at any geography — only totals, from which an average can be derived. Averages are pulled upward by a few large values, so read them as a rough centre rather than a typical household.
Where no amount appears, the reason differs and the wording says which: not published means the Census collects no dollar figure for that source anywhere (SNAP benefits, for example); not disclosed means the figure exists but too few households reported it to publish without identifying them; and a dash means the survey found no household reporting that source — which is not quite the same as none existing, since the survey reaches only a sample.
A note on marital status
In 2020 the federal ACS did not consistently record same-sex married couples as married, which understates marriage among Oakmont's same-sex couples.
For the data-minded
The fine print
Why not just get ACS data for the 75 blocks?
This is the most common question, and the answer is that block-level ACS demographics don't exist. Because the survey reaches only a sample of households, publishing results for a very small area would be both unreliable and a privacy risk. The Bureau therefore stops at the block group, and publishes only a subset of its tables even at that level. Income, education, and home value for a single block are never published, at any margin of error. Approaches that appear to show block-level ACS demographics are really filling the gap with non-Census sources such as member surveys or commercial address data; those aren't comparable to Census figures, so this site doesn't use them. Excluding blocks to trace a boundary is legitimate only for the Decennial count, which is exactly where we apply it.
Differential privacy
The 2020 Decennial Census adds a small amount of statistical noise to block-level counts to protect individuals (the Bureau's "differential privacy" system). For a single block this can nudge a number by a few people; summed across all 75 Oakmont blocks the effect is much smaller than on any single block, which is why the boundary totals are dependable at community scale even though no individual block is guaranteed to the person. Treat a single block's count as approximate.
Margins of error
Every survey figure is an estimate with a published margin of error: the Bureau's statement that the true number is very likely within that much either way. The report and the comparison page show the central estimate alone, for readability, so read those as "about," not "exactly." Explore the Data shows the margin beside every figure and includes it in the spreadsheet exports.
Two caveats on those margins. This site sums counts across two tracts, and summed margins combine as the square root of the sum of their squares rather than by simple addition — that is the Bureau's own formula, and it is what the explorer applies. Medians cannot be summed at all; they are combined here as a population-weighted average, and a weighted average of two medians has no published combined margin. The explorer shows the wider of the two tract margins in its place, which is a floor rather than an exact figure.
Five-year surveys that share years are not independent samples, which is why What’s Changed compares 2015–2019 against 2020–2024 — two surveys with no year in common.
What the Census counts as income
The income figures on this site measure what the Census Bureau calls money income: cash a household actually receives over twelve months, counted before taxes. It is not a measure of taxable income, and the definition draws no line between taxable and non-taxable money.
Counted: wages and salary; self-employment income; interest, dividends, rent, royalties and income from estates or trusts (interest counts whether or not it is tax-exempt, so municipal bond interest is in); Social Security and Railroad Retirement; Supplemental Security Income (SSI); public assistance; retirement, survivor and disability pensions, including regular withdrawals from an IRA, Roth IRA, 401(k) or 403(b); and "all other income," which covers Department of Veterans Affairs payments, unemployment and workers' compensation, alimony and child support, and regular contributions from people outside the household.
Not counted: capital gains; money from selling property, including a home; withdrawals from savings; money borrowed, which is where reverse mortgages and home equity lines fall; tax refunds; one-time gifts and lump-sum inheritances; life insurance and other lump-sum payouts; and the value of non-cash benefits such as SNAP, housing subsidies or employer-paid health insurance.
That distinction matters more in Oakmont than in most places. A retired household drawing down savings, or borrowing against a home it owns outright, can be living comfortably and still report a modest income in these tables. Read the income figures as cash received, not as a measure of wealth or of what a household can afford. The definitions above are the Census Bureau's own, from the ACS Subject Definitions, "Income in the Past 12 Months."
The variables, by table
Every figure traces to a public Census table. The main ones:
- P12 age & sex (Decennial). The site reports the 55+ share (total minus the under-55 bands) because Oakmont is a 55+ community.
- P3 race, P4 Hispanic origin (Decennial).
- H1 housing units, H3 occupancy, H4 tenure / owner vs. renter (Decennial).
- B19013 household income, B19301 per-capita income (ACS).
- B15001 educational attainment, ages 45+ (ACS).
- B25077 median home value (ACS), owner-occupied homes only.
All of these are public. You can pull them yourself from the Census API, or use Explore the Data, which mirrors every published table for these geographies and downloads as a spreadsheet.