SF Kindergarten Madness

Michael Siliski
5 min readDec 2, 2017

--

San Francisco has a notoriously confusing public school student assignment process. I get asked all the time about how we navigated this. It’s application time for the 2018–2019 school year, so here are my notes about how the process works.

N.B. This is mostly based off research I did 2 years ago when we were applying to kindergarten for the 2016–2017 year. After a quick look around, it doesn’t seem like anything has changed, but if anything here is outdated (or just wrong!) please let me know.

Assignment in a nutshell

Background

  • Most schools have attendance area (AA) preference, others are city-wide (CW). AA preference means you get priority if you live in the school’s assigned area. CW means you don’t. Generally CW indicates K-8 (as opposed to the more common K-8) or a special language program. Here’s the list of AA vs CW schools.
  • AA schools give preference to: 1. Siblings, 2. Pre-K/TK, 3. CTIP (low test score area), 4. Attendance area. CW schools have the same preferences except #4 is omitted. Details here.
  • The assignment system runs an independent lottery for every school. You enter your child in the lottery for a school by listing it on your preference form. You can play in as many lotteries as you want.

Assignment process

  1. Each lottery first runs in preference order: first all the siblings get in, then all the Pre-K kids, etc. At a certain point, the school is full, and it comes down to chance at whatever stage of preference you’re at.
  2. If you win multiple lotteries, you get your top choice. Schools will have open slots from people who won multiple lotteries and got their top assignment, so repeat as long as there are open slots. If you win no lottery, you get put in a pool of people assigned to an open spot at the end. I’m not 100% sure, but I’ve read that that process is by proximity.
  3. After all the assignments, there’s a second pass where the system looks for pairs of children who can both be made happier by swapping choices. So if you listed Grattan #1 and Rooftop #2 and I listed Rooftop #1 and Grattan #2, and we each happened to win the lottery for our #2 choice but not our #1 choice, the system would swap us so we both got our #1 choice. [Update: the swap was turned off for 2019–2020 assignments.]

Appeals & waiting pool

After round 1 (March Placement Period), there’s an enrollment period. I heard ~25% of the elementary-aged population goes to private school but almost all play in the public school lottery, so not all the assigned spots actually get filled.

Then there is a 2nd round (May Placement Period) where you list all the schools again, and if you get a higher choice than your current assignment, you get to take it. Description of various assignment periods and schedule.

After that, there’s a waiting pool where you choose one school you would prefer over your current assignment. Waiting Pool Process.

School preference strategy

This is a pretty good summary of the strategy we settled on. In short, your school preference order is your own, so the questions are (a) how many schools to list, and (b) whether there is any better strategy than listing them in true preference order.

How many schools to list

List enough schools until you truly have no preference between all the schools you didn’t list, or are 100% sure you’ll get into one of the ones you listed. If you fall off the end of your list, who knows where you’ll end up. Your list might be short if your attendance area school is on there and has enough capacity to take everyone in the area. Mine wasn’t, since our school, Grattan, doesn’t have the capacity. I listed 40+ schools since the marginal cost of listing one more school was extremely low (I just picked schools off a map after the first 10 or so) compared to the value of getting that school over the worst case.

Ordering strategy

There’s some debate over whether you are incentivized to list your school choices in your actual order of preference. SFUSD would say you are, but the researchers who developed the algorithm have claimed that without actually looking at the algorithm, it’s impossible to know (and the SFUSD is pretty closed about the actual algorithm).

I think there is probably some actual value, but it’s so slight it’s not worth worrying about. [Update: the swap was turned off for 2019–2020 assignments, and both of these are predicated on the swap. I haven’t thought through exactly what that means for strategy though.] Consider two cases:

  1. This case comes down to the swapping algorithm, which we don’t know. Basically the question goes like this: Let’s say my true preference is Grattan #1, New Traditions #2, Clarendon #3. I have a 50% chance of winning the Grattan lottery (with AA preference), a 0% chance of winning the New Traditions lottery (no AA preference), and a 2% chance of getting into Clarendon (city-wide). Is there a benefit to listing Clarendon #2, to improve my chances of swapping into Grattan vs swapping into New Traditions? Would it outweigh the potential loss of getting Clarendon over New Traditions? Impossible to say.
  2. Say I had the same preference, but a 20% chance of getting into New Traditions. Then if my preference for New Traditions over Clarendon was slight, I might prefer to list Clarendon first so that the 0.4% of the time I won both the New Traditions and Clarendon lotteries, I would get assigned to Clarendon, which would maximize my swap value.

Regardless, the value seems pretty slight and debatable, and we kind of just listed schools in preference order. Where I didn’t really care between schools, I gave a slight edge to more popular city-wide ones that might have more swap value. The only sure thing is that everyone should have these super popular city-wide schools on their list, because the non-zero chance of winning the lottery adds up across several schools and gives you a better chance of swapping up.

Wrapping up

I hope that’s helpful. Please do let me know if you have any input/feedback, I’d be happy to incorporate it.

References

Some of these are for past years, but I assume the equivalents are available with a bit of searching.

--

--

Michael Siliski
Michael Siliski

Written by Michael Siliski

products, software, data, cities, housing, mobility, San Francisco, Boston sports, minds, music, coffee, spirits, funny stuff, beautiful stuff, amazing stuff

No responses yet