What percentage of kids attend private school in San Francisco?
A simple question that's surprisingly hard to answer
Private school enrollment is higher in San Francisco than in the rest of California. This is often attributed to San Francisco’s fraught racial history. This post looks at the available statistics to see how the proportion of students attending private school has changed over time.
Historical Data
A previous post looked at the detailed enrollment reports of each private school in San Francisco since 2000. If we want to look back further, we need to rely on the census.
The decennial census provides direct estimates for total k-12 enrollment and for private and public school enrollment for the years 1960 - 2000. Unlike the census’s population figures which are based on actual counts and attempt to reach 100% of the population, the education figures are samples based on questionnaires given to a fraction (usually 20%) of the population. These questions are (since 2010) now asked as part of the American Community Survey which is updated annually but relies on a much smaller sample size (2 million households across the country).
Unfortunately, if we compare the census estimates for public school enrollment with the actual figures as reported by the California Department of Education (CDE), we find that the census estimates have sometimes been well outside the normal margin of error. In 1980 and 2000, the census overestimated public school enrollment by 8.2% and 9.7% respectively. If the census estimate for public school enrollment can be so wrong, we have to be skeptical about its estimates for private school enrollment (and total k-12 enrollment, which is just the sum of the two).
An approach that doesn’t involve so much sampling error is to estimate the size of the school-age population using the census age counts. For example, we could take the under 18 population and subtract the under 5 population to approximate the school-age population. Of course, this is not precise either because people don’t start school on their 5th birthdays, not everyone who is of school age attends school, and the proportion of high school dropouts has varied over time. But it does provide a rough estimate of the number of people available to attend school. If we subtract the known public school population, we get what should be an upper bound on the size of the private school population.
We can see this in the chart above. The top two lines are the estimates of the total number of school-age kids in the city and the bottom two lines are the estimates of the number attending public school. Some things to note:
Although the total population of San Francisco has risen by 29% since its nadir in 1980, the school-age population has fallen by 30%. But it did (probably) rise in the 2010s for the first time since the 1950s.
Counts are more accurate than survey-based samples. When they differ significantly, as they do in 2000, I believe that the census count of the school age population (i.e. the population aged 5-17 inclusive) is more accurate than the census estimate of the k-12 population and that the Department of Education count of public school enrollment is more accurate than the census estimate of the same. Why they should differ by so much is unclear.
Loosely speaking, the number attending private school is the gap between the top pair of lines and the bottom pair of lines. Here are estimates of private school population derived from these figures.
I don’t place much confidence in any one of these estimates by itself. We have seen that the census bureau estimates of public school enrollment can be off by a lot so the red line, the bureau’s estimate of private school enrollment, should be at least as error prone. The blue line should be considered an upper bound, not an actual estimate, because it makes the false assumption that all children attending private schools in San Francisco actually reside in San Francisco. The yellow line assumes that all children aged 5-17 who are not in SFUSD are in private school. That should also be an upper bound because some children are not enrolled in any school.
One observation that reduces my confidence in these estimates is that the gap between the blue and yellow lines, at over 6000, could only be explained if 25% of all students at private schools in the city were commuting from outside the city. There are some schools for which that is true (e.g. Lick-Wilmerding which says that 31% of its students hail from outside the city) but I’d be shocked if it were true as an average across all k-8 and high schools.
We can also convert these numbers, shaky as they are, into estimates of the proportion of San Francisco students attending private school. It turns out this has been remarkably stable, staying solidly in the middle of the 20%-30% range and is not much different in 2020 from where it was back in 1950.
Interpretation
To put these numbers in context, San Francisco in 1950 was still 89% White with public schools that were de facto segregated. Nevertheless, private schools served close to 30% of the city’s students. The largest Catholic schools (e.g. Saint Ignatius, Sacred Heart) were long established by this point.
The baby boom was under way and “White flight” (the phenomenon of white people moving out of urban areas) had started. While the overall population of the city fell during the 1950s, the number aged 5-17 rose by 26,000 during the decade. For scale, all of SFUSD’s elementary schools today serve about 25,000 students. Imagine the strain all these extra students must have placed on school infrastructure. New schools needed to be built and new teachers hired in a city with a declining population. Some of the Catholic k-8 schools in the western part of the city (e.g. St. Stephen, St. Thomas More) were opened during this time. Others reached their peak enrollment levels.
The city’s population fell by about 25,000 during the 1960s, pushing the school-aged population down by about 4% but public school enrollment rose slightly during the decade, indicating that the proportion going to private school fell.
The introduction of busing caused public school enrollment to crater during the 1970s. Most of those who fled SFUSD in those years moved out of the city but a small fraction moved to private schools, which experienced enrollment increases of between 3% and 12% (depending on the calculation method). This pushed the percentage in private schools above 30% for the first and only time.
The proportion reverted to its long-term average during the 1980s when private school enrollment fell 27% - 33%. The numbers were then fairly stable between 1990 and 2010.
The last decade confuses me. The 2020 census shows an increase of about 5,700 in the Under 18 population. That means the school age population should have increased by something between 2800 and 3800 (the difference depends on whether you assume the U5/U18 percentage is as of the 2019 ACS 5-year survey or as of the 2010 census). But we know that enrollment in the city’s private schools is up by fewer than 500 while public school enrollment has gone up by about 1800 over the decade. Where are the other 500-1,500 kids? Meanwhile, the American Community Survey estimates that the number of residents attending private schools rose by 4,500 during the 2010s. That seems completely unsupportable by enrollment records.
Methodology Notes
I used the population aged 5-17 to represent the the School Age population because it covers a thirteen-year span (the normal length of time for a k-12 education) and is available for all census years. It might be more accurate to use the population aged 5.5 to 18.5 or 6.000 - 18.999 since most 12th graders are 18 by April 1, the date of the census, but the Census stopped publishing counts of the population by single years of age in 2000. Now they only publish 5-year buckets (e.g. 5-9 and 10-14) and the total Under 18. In the years where I could calculate it, the effect of pushing the age window forward six months was that the estimate of the school age population was over 1200 higher in 1980 and nearly 700 lower in 1960. The difference would not be as dramatic today.
The census bureau has not yet published detailed age reports for the 2020 census. They have published total population and total adult population figures, from which simple subtraction gives the U-18 figures. To exclude the Under 5s, I used the 2019 American Community Survey which said that Under 5s were 33.1% of the Under 18s.
The American Community Survey will not be publishing estimates of public and private school enrollment for 2020 because of the pandemic. I used their 2019 figures.
I used the school enrollment figures for the school year ending in the census year. For example, I used the 2019-20 school year figures to compare with the 2020 census. The public school numbers include all charter schools except Five Keys.