Berkeley law professor tells SF school committee data link school composition and achievement gap

Professor Goodwin Liu, a visiting scholar at Berkeley Law, told a San Francisco Unified School District ad hoc student assignment committee in December that district and national research show a strong correlation between a school’s racial composition and its academic performance.

Liu told board members the district is demographically diverse overall but that many individual schools are racially identifiable: “half the schools are around that number are racially identifiable in the district,” he said. He showed district graphs and national studies — including the Moving to Opportunity experiment and a Texas longitudinal study — that he said suggest higher concentrations of Black or Hispanic students correlate with lower Academic Performance Index (API) scores for those student groups.

Liu cautioned that correlation is not causation and said a fuller analysis would require student-level subgroup data. He recommended examining Black and Hispanic students’ outcomes within schools of differing racial compositions rather than relying only on overall school API scores. “If you really are serious about closing the achievement gap, it means that the groups who are traditionally found on the underside of the achievement gap need to be making progress at an even greater rate,” he said.

Liu outlined several possible mechanisms linking composition and outcomes. One is teacher experience and turnover: his slides showed a negative relationship between the share of Black and Latino students and average years of teacher experience in a school. He cited research indicating that less experienced teachers are concentrated in higher-poverty, higher-minority schools and that turnover may undermine school climate and program continuity. He also summarized research on incentive pay, citing analyses (noted in his presentation) suggesting very large pay increases would be needed to equalize retention rates across schools.

Liu pointed to exceptions and caveats. He named KIPP as an outlier — a high-Black- and Hispanic-enrollment school with strong API results — and emphasized that some studies show only modest achievement gains from housing desegregation because families who move do not always change where they send children to school. “Economic desegregation and housing did not necessarily result in significant student achievement gains,” he said of the Moving to Opportunity findings.

Board members used the presentation to ask for local, disaggregated analyses. Commissioner Feuer asked about the magnitude of effects in the Texas study; Liu said the study’s authors reported large swings — on the order of up to a grade level in learning in some comparisons — and offered to re-check exact figures. Commissioners and public commenters repeatedly requested San Francisco–specific breakdowns controlling for socioeconomic status, within-school subgroup trends, and teacher and principal turnover data.

Staff said they will supply more detailed local data and run simulations of different assignment models. Ms. O’Keefe, staff lead on the assignment work, told the committee staff are collecting additional datasets and running simulations with outside consultants and that Jan. 29 and Feb. 12 have been scheduled for further community meetings. She also confirmed staff would make materials available on the district web page and that the public may submit input to studentassignmentideas@sfusd.edu.

The presentation framed the technical and policy choices ahead: district leaders must weigh priorities such as equity, choice, predictability and neighborhood proximity and then ask staff to test how different trade-offs would affect school composition, capacity and transportation costs. The committee requested targeted local analyses — including subgroup API trends by school composition and measures of teacher/principal experience and turnover — to inform policy options the board will consider in coming months.