SFUSD weighs six student‑assignment models, asks board to set measures for evaluation

San Francisco Unified School District staff and outside researchers presented six candidate student‑assignment systems at an ad hoc Board committee meeting and asked the board to agree on measurable objectives to compare them. The presentation, led by SFUSD staffer Miss O'Keefe and Stanford researchers Muriel Niederle and Clayton Featherstone, outlined options ranging from local attendance‑area assignment with limited citywide choice to choice‑based systems tuned by two preference “dials” — local preference and academic preference.

Board members and researchers framed the redesign as one element of a broader strategy to make every school a “quality school,” noting that assignment policy alone cannot close achievement gaps. Staff proposed three priorities for judging options: equitable access to district opportunities, reducing racial isolation and concentrations of underserved students, and transparency in the assignment process. To operationalize those priorities, staff proposed measures including the number of ethnically isolated schools (using a 60% threshold as an example), counts of schools with high shares of students achieving below basic, rates of on‑time participation in the enrollment process, and counts of underenrolled schools.

Researchers described three baseline assignment mechanisms and five simulations built from kindergarten application data. They emphasized two design principles: simplicity (parents can safely rank true preferences) and non‑wastefulness (choose systems that place more parents in ranked schools when outcomes are otherwise equal). In the simulations, a local assignment with only citywide transfers produced roughly 39% of students receiving their first choice and 52% receiving their first or second choice; open local assignment increased access to higher‑API schools for students from low‑API neighborhoods (about 45% in one scenario versus 32% in the restricted citywide case). Parent‑guided systems using academic preference showed higher opportunity figures (around the mid‑50s percent in the researchers' summary) and, in some tuned settings, fewer ethnically isolated schools than the current diversity‑index model.

The researchers cautioned that the data have limits: changing assignment rules can alter family behavior (for example, who applies, where families move or which programs they select), so simulation results are informative about comparative outcomes under current behavior but not definitive predictions after policy changes. They also noted that about 20% of families submit rankings late under the current calendar; moving the round‑1 deadline could raise on‑time participation by roughly 10 percentage points for some groups but could not capture all late entrants.

Board members pressed staff and researchers on several technical points: which schools were treated as citywide in the simulations (staff said a preselected list of 11 met the parameters used), how geocoded student data were used to infer theoretical local schools, and whether programmatic choice (selecting a program rather than a specific campus) might improve equity and predictability. Several commissioners asked staff to disaggregate simulation outputs by race and ethnicity and to include academic outcome measures (for example, counts of students from low‑API neighborhoods assigned to higher‑API schools) in subsequent reports.

Staff outlined next steps: run narrowed simulations responsive to board feedback, disaggregate results by race/ethnicity, add an academic outcome metric, and begin community engagement (parent meetings, large public meetings, and a randomly selected qualitative study) so the board can receive consolidated findings in January and consider policy readings in February and March. Staff also said they had secured federal and foundation grants to support technical assistance and qualitative research.

The committee did not vote on any policy. Staff set follow‑up meetings for November and December and said a January report will include the refined simulations and community feedback. The board asked staff to return cost and transportation analyses and to recommend targeted outreach strategies for groups that underparticipate in round‑one application windows.