Ad Hoc Committee on Student Assignment September 14, 2020 Orla - - PowerPoint PPT Presentation
1 Ad Hoc Committee on Student Assignment September 14, 2020 Orla OKeeffe Chief Policy & Operations Henry OConnell Project Manager 2 Tonights Agenda 1. Staff Presentation (45 mins) a. Introduction and Framing b. Draft
September 14, 2020
1
Orla O’Keeffe Chief Policy & Operations Henry O’Connell Project Manager
2
3
What does the Board think?
SAN FRANCISCO UNIFIED SCHOOL DISTRICT
4
5
6
Synthesized Goals
Equity Lens
Stated in Resolution
7
8
9
A Race Equity Culture is one that is focused on proactive counteraction of race inequities inside and
Culture is the foundational work when organizations seek to advance race equity; it creates the conditions that help us to adopt anti-racist mindsets and actions as individuals, and to center race equity in our life and in our work. A Race Equity Culture is the antithesis of dominant culture, which promotes assimilation over integration and dismisses opportunities to create a more inclusive, equitable environment. The work of creating a Race Equity Culture requires an adaptive and transformational approach that impacts behaviors and mindsets as well as practices, programs, and processes.
Equity in the Center. https://www.equityinthecenter.org/wp-content/uploads/2019/06/Awake-to-Woke-to-Work-Glossary-of-Terms-.pdf
Anti-racism is the active, conscious, and non-neutral process of identifying and eliminating racism by changing systems,
and attitudes, so that power is redistributed and shared equitably. The heart of an anti-racist system is personal, professional, and system-wide accountability.
10
SAN FRANCISCO UNIFIED SCHOOL DISTRICT
11
12
Does this accurately capture what changes the Board wants to make and why? If… (implementation of change idea) then... (short-term impact) so that…. (long term impact)
If the Student Assignment Policy: 1. Diversity: Creates assignment zones that are racially, ethnically, socioeconomically, linguistically, and academically diverse; and 2. Diversity: Assigns students so that every school mirrors the diversity of its zone; and 3. Choice: Gives all students access to the range of programs in SFUSD; 4. Equity: Prioritizes younger siblings and historically underserved students; and 5. Predictability: Provides all students with certainty that they will be placed at an elementary school in their assignment zone; and If the Educational Placement Center
for families to enroll; and IF transportation routes and schedules are aligned with and support assignment zones; and IF each and every elementary school facilitates positive interaction across difference, and provides equitable access to resources and opportunities that exist within the school;
experience with SFUSD will be a simple and predictable enrollment process; and
will help create more diverse enrollment and will increase enrollment in currently under-enrolled schools; and
stability in enrollment patterns will reduce transportation costs and support multi-year planning and budgeting; and
equitably allocated across elementary schools (students, programs, staff, funding); and
confidence in SFUSD will increase and there will be strong community connections to local schools; and
school will be diverse and integrated
receive what they need to develop to their full potential.
13
SAN FRANCISCO UNIFIED SCHOOL DISTRICT
14
15
Concept #1: Initial Assignment + Choice Concept #2: Choice in Small Zones Concept #3: Choice in Medium Zones
Goals Predictability, Diversity, Proximity Predictability, Diversity, Proximity Diversity, Predictability, Proximity Student Assignment Automatic assignment, then optional choice process Choice Choice Geographical Constraints Attendance Areas (1 school) Zones (3 - 5 schools) Zones (8-12 schools) Portfolio of Schools 1. Attendance Area Schools 2. Citywide Schools 1. Zone Schools 2. Citywide Schools 1. Zone Schools
16
Community Input Simulation Results Diversity
(Racial/ethnic, socioeconomic)
Predictability
(Range of possible
different assignments in 10 simulations)
Proximity
(Distance/ community cohesion)
Choice
(Not a goal - measure of disruption to status quo)
Concept 1 Initial assignment
3.3/5 Mixed
Depends on metric
Slightly better Slightly better Mixed
Depends on metric
Concept 2 Small zones
3.5/5 Non-contiguous Better (generally) Slightly better Better Slightly worse Contiguous Worse (always*) Much Better
Concept 3 Medium zones
2.5/5 Better (generally) Slightly better Worse (generally) Slightly Worse
2019-20 Assignments (baseline)
*Simulations have not yet found a set of small contiguous zones that do better on measures of racial/ethnic and socioeconomic diversity than our current system.
17
Policy
How well do options work in support of SFUSD’s goal of Access and Equity? Data (demographics,
choice, capacities, etc.)
Feedback from the Board of Education Community Input Research and Case Studies Simulations of Policy Outcomes How well do options achieve the Board’s policy goals of diversity, predictability, and proximity?
18
want to go to school close to home, but they want to be confident in the quality of their local schools first
people don’t want too much (overwhelming) or too little (restrictive) choice
and low-income families
difficult to give feedback without knowing what the boundaries would look like
19
and families who live on the west side
demographic group, including African American, Latinx, and low-income families
○ Despite this, African American, Latinx, and low-income families were skeptical that any of the concepts would work for them, including Concept 2
○ Some felt this was too restrictive (no choice outside of zones),
choice (felt like the current system, on a slightly smaller scale).
20
21
Concept 1: Initial Assignment + Choice Families in the Southeast had serious concerns about Concept 1.
Concept 1 was only popular among higher income families.
Participants noted that a proximity-based student assignment system “makes winners and
have the money to live in the nicer neighborhoods.” [Starr King Elementary School]
African American and Latinx families rated Concept 1 as the least likely to meet their needs.
22
Concept 2: Small Zones (3-5 schools) There were questions about concept 2 in the Southeast, but there wasn’t intense opposition. This could be workable with some changes or creatively drawn zones.
Most demographic groups rated Concept 2 as the most likely to meet their families needs.
Despite this, African American, Latinx, and low-income families were still skeptical that any of the concepts would work for them. How might we adjust Concept 2 to better meet the needs of African American and Latinx families?
Concept 3: Large Zones (8-12 schools) Families throughout the city disliked Concept 3.
23
Most demographic groups did not believe Concept 3 would meet their needs.
24
Policy
How well do options work in support of SFUSD’s goal of Access and Equity? Data (demographics,
choice, capacities, etc.)
Feedback from the Board of Education Community Input Research and Case Studies Simulations of Policy Outcomes How well do options achieve the Board’s policy goals of diversity, predictability, and proximity?
25
Background/expertise in matching markets & algorithm design ○ Itai Ashlagi, Associate Professor of Management Science and Engineering ○ Irene Lo, Assistant Professor of Management Science and Engineering
Chemistry
Management Science and Engineering
Science
Computational and Mathematical Engineering
Mathematics
Science and Engineering
26
○ Concept 1 can have a small, positive effect on district goals but doesn’t shift outcomes much from the status quo ○ Restricting choice within zones (Concepts 2 and 3) can improve diversity, but outcomes depend heavily on how the zone boundaries are drawn ○ When drawing zone boundaries there is a tradeoff between diversity and predictability/proximity
27
28
Available Schools and Programs Tiebreakers Requests Students School Assignments
Simulations assume some things are fixed:
applied in past years
new policy
And some things can be changed:
to (e.g. zones, citywide schools)
different goals (e.g. % of students getting their first choice vs. % of students getting one of their choices).
Assignment Algorithm
1. DATA: The Stanford research team ran simulations for each of the concepts based on students who participated in kindergarten student assignment in 2018-2019, using the preference lists they submitted in their first round of participation in 2018-2019, the capacities from 2018-2019.
in 2019-2010 . The algorithm lets students rank schools in order of preference, and tries to assign them to their preferred schools, giving priority to students based on tiebreakers for sibling, CTIP1, and zone priority.
29
INITIAL ASSIGNMENT TO ATTENDANCE AREA. Students are initially assigned to their attendance area school, and have the option to participate in choice. ZONE BOUNDARIES WITH RESTRICTED CHOICE. Draw new zone boundaries and restrict students’ access to programs depending on the zone they live in. DIFFERENTIAL ACCESS TO CHOICE. Give targeted student groups more access to choice, e.g. CTIP1 students, redefined CTIP/target student groups, students in homeless/foster/public housing
elementary school’s student population, within an identified zone, reflects the diversity of the zone OTHER ALGORITHMS. Examples include the transfer mechanism used until 2018-2019, as well as other parameters in the algorithm such as tiebreakers.
30
attendance area school.
than the number of students in the attendance area who rank it highly) Process: We first assigned all students who ranked a program their attendance area school in their top 1 (or top 3, top 5, top 10) to their most preferred program at their attendance area school. We then ran the choice algorithm for the remaining students at the remaining seats at all programs. Finding: Providing an initial assignment can have small, positive effects on our goals
31
○ To best achieve the district’s goals, zone boundaries need to be redrawn. Existing boundaries, such as the attendance areas and middle school feeder zones, lack socio-economic and ethnic diversity. ○ The size and shape of zones affect which of the district’s goal they are best able to achieve. There is a tradeoff between small or contiguous zones, which improve predictability/proximity, and large or non-contiguous zones, which have the most effect on diversity. ○ There is significant scope for improving on the zone boundaries that we show today. However board and community feedback is important for focusing the zones we explore and appropriately balancing tradeoffs between predictability, proximity ,and diversity.
32
Concept #2: Choice in Small Zones Concept #3: Choice in Medium Zones
Predictability, Diversity, Proximity Diversity, Predictability, Proximity Choice Choice Zones (3 - 5 schools) Zones (8-12 schools) 1. Zone Schools 2. Citywide Schools 1. Zone Schools
Concepts 2 and 3 differ in (1) how zones are drawn (small vs medium), and (2) how much access students get to schools
Today’s simulations focus on the effects of how zones are drawn: 1. Drew ~100 potential zone boundaries 2. Selected ~20 zone boundaries to simulate 3. Simulated choice algorithm where students have access only to GE programs in their zone and all citywide and language programs.
33
Middle School Feeder Zones
Existing boundaries, such as the attendance areas and middle school feeder zones, lack diversity in socio-economic status and ethnicity, and don’t balance capacity with the student population.
DESCRIPTIONS OF METRICS %FRL Students: Percentage of students living in zone receiving free
Neighborhood SES: Measure of SES combining median household income, poverty and adult educational attainment; averaged across students living in zone Ethnic Diversity: Distance between distribution of ethnicities of students living in zone and district distribution Capacity Deficit: #students in zone - #seats in zone, including citywide seats geographically located in zone
For student measures, darker blue areas are higher need and/or less diverse.
% FRL Students Neighborhood SES Ethnic Diversity Capacity deficit Metrics by MSF Zone
34
had roughly the same number of students
○ Contiguous zone = all blocks border each other ○ Non-contiguous zone = some blocks do not border others
and ethnic diversity of the underlying student population: ○ Average FRL compared to district average ○ Average AALPI Score compared to district average ○ Average Diversity Index compared to district average
for medium zones (6-8 zones) and small zones (10-13 zones)
Maps are for illustrative purposes only
35
Maps are for illustrative purposes only
Non- contiguous Contiguous Small Zones
(12-13 zones: more colors)
Medium Zones
(6-8 zones: fewer colors)
36
hundreds of zone boundaries that satisfy the criteria
socio-economic diversity and ethnic diversity and selected 14 small zones, 12 medium zones on the frontier according to these measures
Max FRL percentage
Max distance vs Max FRL percentage (Each blue dot represents a zone map) Distance
Note: Today’s results are based on simulations where we chose zones based on diversity measures for the student populations living in the zones. We are now also able to choose zones based on diversity measures that capture the effects of choice.
SMALL, CONTIGUOUS ZONES are likely to improve
proximity and predictability and provide simplicity. However, they tend to be less diverse.
Non- contiguous Contiguous Small Zones
(12-13 zones, 4-5 schools each)
EXAMPLE ZONES Medium Zones
(6-8 zones, 10-12 schools each)
Ethnic Diversity
Hellinger: Distance between distribution of ethnicities of students living in zone and district
indicates larger distance. %FRL: Percentage of students living in zone receiving free or reduced lunch. Darker blue indicates higher concentration of such students.
Socio-economic Diversity
MEDIUM, NON- CONTIGUOUS ZONES are best able to
disrupt existing residential patterns of socioeconomic disparity between zones, and can improve diversity
(i.e. make the shades of blue lighter and more similar).
Maps are for illustrative purposes only
38
cohesion and provide less choice
zones need to be drawn and evaluated very carefully.
39
Predictability of assignment is better with small zones than medium zones. Proximity is improved by small zones, especially contiguous, and can worsen with medium zones.
Predictability
%Assigned
Top 3 # Distinct Assignments Better predictability Worse predictability
METRICS # Distinct Assignments from 10 runs of the algorithm % Assigned
the programs they ranked and have access to
Small (4-5 schools)
Medium: (10-12 schools)
METRICS Community Cohesion: % Students attending program with fewer than 3 students from their block group. Distance (miles): Average travel distance for students from assigned program
Distance Better proximity
Proximity
Community Cohesion 2019-20 Algorithm Small contiguous Small non-contiguous Medium non-contiguous Medium contiguous
40 Ethnic Diversity Measure 1 Ethnic Diversity Measure 2
Ethnic Diversity
More Diverse
2019-20 Algorithm
Less Diverse
Socio-Economic Diversity
2019-20 Algorithm SES Diversity Measure 1 SES Diversity Measure 2
More Diverse Less Diverse
Non-contiguous or medium zones can give more diverse outcomes.
worsens with small zones
worsens with contiguous zones The specific shape of the zones significantly affects diversity, and so zones need to be drawn and evaluated very carefully.
socio-economic diversity and ethnic diversity Choice can lead to re-segregation within the zone even if populations within the zones are diverse. In addition to drawing diverse zones we must consider other levers such as guardrails and differential access to choice.
Small contiguous Small non-contiguous Medium non-contiguous Medium contiguous
41
All concepts shown today reduce student choice. Assuming that students prefer some of the programs that would no longer be available to them, moving to one of the concepts would likely reduce:
1st Choice Top 3 Choice
Rank (all listed programs) In-Zone Rank (all listed accessible programs) Less Choice Less Choice
1st Choice Top 3 Choice
2019-20 Algorithm 2019-20 Algorithm
42
generate hundreds of zone boundaries that satisfy the criteria
measures of socio-economic diversity and ethnic diversity capturing effects of choice and use this to narrow down on possible zone boundaries
Ethnic Diversity vs Socio-Economic Diversity (Each blue dot represents a zone map) SES Diversity
Ethnic Diversity More Diverse
Less Diverse
43
Today’s focus:
TO ATTENDANCE AREA. Students are initially assigned to their attendance area school, and have the
choice.
WITH RESTRICTED
boundaries and restrict students’ access to programs depending on the zone they live in. There are other levers that can be explored, including:
zones, perform more targeted search, and incorporate citywide schools.
access to choice, e.g. CTIP1 students, redefined CTIP/target student groups, students in homeless/foster/public housing
each elementary school’s student population, within an identified zone, reflects the diversity of the zone
language programs. Language program zones could be used with GE program zones to increase predictability and proximity, and may also help with diversity.
mechanism used up until 2018-2019. Our simulations indicate that the tradeoff is typically between algorithms that increase choice versus algorithms that increase diversity.
44
SAN FRANCISCO UNIFIED SCHOOL DISTRICT
45
46
Community Input Simulation Results Diversity
(Racial/ethnic, socioeconomic)
Predictability
(Range of possible
different assignments in 10 simulations)
Proximity
(Distance/ community cohesion)
Choice
(Not a goal - measure of disruption to status quo)
Concept 1 Initial assignment
3.3/5 Mixed
Depends on metric
Slightly better Slightly better Mixed
Depends on metric
Concept 2 Small zones
3.5/5 Non-contiguous Better (generally) Slightly better Better Slightly worse Contiguous Worse (always*) Much Better
Concept 3 Medium zones
2.5/5 Better (generally) Slightly better Worse (generally) Slightly Worse
2019-20 Assignments (baseline)
*Simulations have not yet found a set of small contiguous zones that do better on measures of racial/ethnic and socioeconomic diversity than our current system.
Tonight:
○ Are we ready to take Concept 1 off the table? ○ Zones are a powerful lever to increase all of our policy goals. However, the way zones are drawn matters a lot, and there are tradeoffs between proximity and diversity. Are we willing to draw non-contiguous, or oddly shaped zones?
Future Meetings:
○ Many families value proximity for its logistical benefits (easy pickup/dropoff) and community cohesion (kids go to school with their neighbors). How might we accomplish both via busing? ○ How might we improve on our goals by including K8 schools as part of a zone, and drawing service areas for language pathways and special education? ○ Even if the zone is diverse, choice can lead to re-segregation within the
algorithm) to limit self-segregation? ○ Choice emerged as especially important to African American, Latinx, and low-income families. How might we prioritize choices for focal communities (e.g. by revising CTIP)?
47
Policy
How well do options work in support of SFUSD’s goal of Access and Equity? Data (demographics,
choice, capacities, etc.)
Feedback from the Board of Education Community Input Research and Case Studies Simulations of Policy Outcomes How well do options achieve the Board’s policy goals of diversity, predictability, and proximity?
48
○ Board almost took Concept 1 off the table in December ○ Southeast community has expressed opposition. ○ Simulations showed small, positive effects, but the benefits were concentrated among families who already liked their attendance area schools. ○ Capacity constraints make this logistically difficult to implement
○ They seem to do well on metrics of diversity, predictability, and proximity. ○ However, diversity metrics are very sensitive to how the maps are drawn -- if zones are too proximate, then diversity decreases
49
What does the Board think?
SAN FRANCISCO UNIFIED SCHOOL DISTRICT
50
○ Zones, pathways, and how we might incorporate
○ Staff Recommendation ○ Community engagement plan
51
SFUSD, Stanford and UC Berkeley hosting a series of online conversations about the history of SFUSD’s school desegregation efforts, the benefits of school integration, and the implications of school choice.
○ History of Student Assignment in SFUSD
○ Research about School Integration
○ Research about School Choice
○ Panel discussion on policy recommendation Sign up at bit.ly/SFUSDresearch
52
Staff just launched a weekly Student Assignment Newsletter and Blog which will be active in September and October and the Board discusses the student assignment policy.
To read the blog and sign up for the newsletter, visit: www.sfusd.edu/studentassignment
53
2020-21 School Year 2021-22 SY 2022-23 SY
Fall Winter Spring Summer Fall Winter Spring Summer Fall
Start of School
Engagement
Committee of the Whole
Reading and Action Launch enrollment
2023-24 SY
Aug Sep Oct Nov Dec Spring
* Might require more time depending on the scale of change
Series - History
Series - Integration
Series, Choice
Recommendation
○ Board almost took Concept 1 off the table in December ○ Southeast community has expressed opposition. ○ Simulations showed small, positive effects, but the benefits were concentrated among families who already liked their attendance area schools. ○ Capacity constraints make this logistically difficult to implement
○ They seem to do well on metrics of diversity, predictability, and proximity. ○ However, diversity metrics are very sensitive to how the maps are drawn -- if zones are too proximate, then diversity decreases
55
What does the Board think?
SAN FRANCISCO UNIFIED SCHOOL DISTRICT
56
SAN FRANCISCO UNIFIED SCHOOL DISTRICT
57
58